LOX-1: A potential driver of cardiovascular risk in SLE patients

Traditional cardiovascular disease (CVD) risk factors, such as hypertension, dyslipidemia and diabetes do not explain the increased CVD burden in systemic lupus erythematosus (SLE). The oxidized-LDL receptor, LOX-1, is an inflammation-induced receptor implicated in atherosclerotic plaque formation in acute coronary syndrome, and here we evaluated its role in SLE-associated CVD. SLE patients have increased sLOX-1 levels which were associated with elevated proinflammatory HDL, oxLDL and hsCRP. Interestingly, increased sLOX-1 levels were associated with patients with early disease onset, low disease activity, increased IL-8, and normal complement and hematological measures. LOX-1 was increased on patient-derived monocytes and low-density granulocytes, and activation with oxLDL and immune-complexes increased membrane LOX-1, TACE activity, sLOX-1 release, proinflammatory cytokine production by monocytes, and triggered the formation of neutrophil extracellular traps which can promote vascular injury. In conclusion, perturbations in the lipid content in SLE patients’ blood activate LOX-1 and promote inflammatory responses. Increased sLOX-1 levels may be an indicator of high CVD risk, and blockade of LOX-1 may provide a therapeutic opportunity for ameliorating atherosclerosis in SLE patients.


Introduction
Systemic lupus erythematosus (SLE) is a chronic, autoimmune disease that leads to multiorgan damage and degradation of connective tissue primarily through inflammation. Although cardiovascular damage related fatalities remain the leading cause of all mortality worldwide, SLE associated inflammatory risk factors independently contribute to a rapid acceleration of premature atherosclerosis [1,2]. Independent clinical studies show strong evidence that patients with SLE have at least two-to three-fold higher risk of heart disease and report an elevation of sLOX-1 in 35% patients with SLE. SLE patients with high sLOX-1 levels had elevated hsCRP, triglycerides, oxLDL, piHDL and impaired HDL CEC compared to the low sLOX-1 group. We also show that modified lipids from SLE patients function as LOX-1 ligands contributing to inflammatory immune activation through proinflammatory cytokine secretion from monocytes and enhanced NET formation by LDGs. These data establish a mechanism of action whereby sLOX-1 levels can be predictive of vascular inflammation, highlighting the potential for LOX-1 receptor blockade as a target for preventing binding of various atherogenic ligands and ameliorating cardiovascular damage in SLE.

sLOX-1 is elevated in SLE patients and is higher in patients up to 40 years of age
sLOX-1 was measured in age-and sex-matched healthy donors (n = 72) and SLE patients (n = 273) (S1 Table) using an MSD-based ELISA assay. We found serum sLOX-1 to be significantly increased in a subset of SLE patients compared to healthy donors ( Fig 1A), with the mean for SLE at 580.9 ± 36.1 pg/mL compared 274.9 ± 30.34 pg/mL in healthy donors (p<0.0001). A cut-off at 532 pg/mL sLOX-1 (mean plus one standard deviation of healthy donors) was used to define high and low sLOX-1 levels. Patients in the high sLOX-1 group were younger (39.82 ± 1.442 years) compared to the low sLOX-1 group (44.53 ± 1.035 years; p<0.0014) ( Fig 1B and Table 1). There was no difference in disease duration between the two groups ( Table 1), indicating that high sLOX-1 is associated with an earlier onset of SLE. Age of SLE diagnosis was indeed significantly lower (25.55 ± 0.87 years) in the high sLOX-1 group versus low sLOX-1 group (31.83 ± 0.9143; p<0.0001) ( Fig 1C and Table 1).

sLOX-1 is associated with inflammatory CVD risk in SLE patients
Stratification using sLOX-1 levels showed no difference in the 10-year risk for cardiovascular disease based on traditional Framingham Risk Scores (FRS) calculated using lipid risk factors (high sLOX-1 3.623 ± 0.5456 versus low sLOX-1 3.456 ± 0.3062) ( Fig 1D and Table 2). Neither HDL nor LDL levels varied between the two groups ( Table 2). Elevated hsCRP levels in the pathogenesis of atherosclerotic vascular disease is a hallmark of chronic inflammation and large-scale clinical trials have used hsCRP greater than or equal to 2 mg/L for defining an increased risk of CVD [33]. High sensitivity C-reactive protein (hsCRP) levels in the high sLOX-1 group (5.936 ± 1.362 mg/L; p<0.00001) were significantly higher than the low sLOX-1 group (2.187 ± 0.456 mg/L) ( Fig 1E). The median hsCRP values for high sLOX-1 SLE patients was 2 mg/L compared to 1.3 mg/L in the low sLOX-1 patients ( Table 2). Multivariate regression analysis adjusting for FRS did not change the positive correlation between sLOX-1 and hsCRP (β = +0.25; p = 3.07e -05 ) ( Table 2), indicating that age, sex, systolic blood pressure, hypertension, smoking, diabetes, total cholesterol, HDL cholesterol and BMI did not account for the association of higher sLOX-1 levels with hsCRP. Besides hsCRP levels, triglycerides were increased in the high sLOX-1 group (136.3 ± 7.298, n = 99 versus 113.5 ± 4.699 mg/dL for low sLOX-1; p = 0.0095) and correlated significantly (r s = +0.21; p = 0.0004) with sLOX-1 levels, even after adjustment for FRS (S2 Table).

High sLOX-1 is associated with normal complement levels and lower SLE disease severity
Laboratory tests to detect specific autoantibodies directed against nuclear or cytoplasmic antigens (serum anti-nuclear antigens (ANA), dsDNA, and extractable nuclear antigens (ENA)) were analyzed in the two sLOX-1 groups. Fewer patients were seropositive for ENA, which measures antibodies to saline-extracted antigens, anti-RNP, anti-SmRNP, anti-Ro, anti-La, anti-Sm, Scl-70, anti-Jo-1, in the high sLOX-1 group versus low sLOX-1 group, and there was a similar trend for ANA and anti-dsDNA. High sLOX-1 patients also had lower prevalence of hypocomplementemia (Table 1) and higher C3 and C4 levels in comparison to patients with low sLOX-1 (Fig 2A and 2B and Table 1). Complement levels also show moderate positive association with sLOX-1 as measured by correlation r s = +0.36 (p = 1.30e -09 ) ( Table 1). Platelets and white blood cell (WBC) counts were increased in the high sLOX-1 group (Fig 2C and  2D and Table 1). Erythrocyte sedimentation rate (ESR) which is an indicator of inflammation in SLE is comparable in the two comparison groups ( Fig 2E). Interestingly, patients with low disease activity ( Fig 2F) exhibited significantly higher sLOX-1 levels compared with mild and moderate disease patients. In support of this observation, a negative correlation was observed between sLOX-1 levels and SLEDAI scores (Table 1; r s = -0.25, p = 3.11e -05 ). This is consistent with lower frequencies of autoantibodies, normal complement and cell counts in the high sLOX-1 patients. There was no evidence to suggest that the sLOX-1 levels were influenced by medications such as oral corticosteroids, hydroxycholoroquine, azathioprine, mycophenolic acid, methotrexate or biologics-usage (S3 Table). Although high sLOX-1 correlated with high hsCRP, sLOX-1 levels did not positively associate with individual clinical manifestations, overall disease activity, medications or hematological abnormalities.

Patients with high sLOX-1 have impaired HDL functionality, high proinflammatory HDL and high oxidized LDL levels
In a study on SLE individuals, 86.7% of patients with atherosclerotic plaques had increased piHDL plasma levels compared with 40.7% of those without plaques, suggesting that piHDL reflects increased CVD risk [9]. We determined piHDL index, which is a measure of the ability of HDL to inhibit LDL oxidation. HDL from high sLOX-1 patients had an index of 1.552 ± 0.08214 (p<0.015), compared to 1.3 ± 0.05396 in the low sLOX-1 patients ( Fig 3A). Further, we confirmed previous published studies indicating that proinflammatory SLE HDL, functions as a ligand for LOX-1 [24], inhibiting nuclear translocation of the anti-inflammatory transcription factor ATF3 in macrophages, an effect reversed with an LOX-1 receptor blocking antibody (S1 Fig).
In further studies of HDL functionality, we evaluated the ability of SLE or healthy HDL to accept cholesterol particles from macrophages in CEC assays. We found that CEC in the high sLOX-1 SLE HDL (0.104 ± 0.005856%) was impaired compared to low sLOX-1 SLE HDL (0.1193 ± 0.004774%; p = 0.04) when normalized to HDL cholesterol content ( Fig 3B). In addition, we observed that the plasma level of oxidized LDL (oxLDL) was significantly higher in high sLOX-1 patients (72902 ± 5345 mU/L) compared to low sLOX-1 patients (59489 ± 3396 mU/L, p = 0.03) (Fig 3C). These findings are consistent with previous reports showing that both piHDL and oxLDL, which function as ligands for the LOX-1 receptor are elevated in the plasma of SLE patients compared to controls [9]. These data show for the first time that a significant positive correlation exists between oxLDL levels and sLOX-1 levels in SLE patients (r s = +0.23, p<0.0003), even after adjusting for FRS in multivariate regression analysis (Fig 3D  and S2 Table). Interestingly, neither age, CEC efflux or oxLDL showed differences when patients were grouped based on high or low hsCRP levels (S2 Fig). This suggests that sLOX-1 is a more sensitive measure of distinguishing an SLE population with dysregulated lipoprotein that is responsible for a higher risk for CVD related events.

Cleaved sLOX-1 levels strongly correlate with TACE activity in SLE patients
To determine if membrane and sLOX-1 are concordantly induced, we assessed their levels following stimulation with oxLDL and/or DNA-IC. Monocytes increased membrane and soluble LOX-1 upon oxLDL, DNA-IC and ox-LDL + DNA-IC stimulation at 24h (Fig 5A and 5B). A strong correlation was found between membrane and soluble LOX-1 levels (r s = +0.57, p = 0.008), indicating that the source of sLOX-1 may be membrane LOX-1 ( Fig 5C). Previously, LOX-1 has been shown to be cleaved by ADAM activity [30,31]. To determine if this is  the case, we assessed whether activity of TNF-α converting enzyme (TACE) is increased upon stimulation with oxLDL, DNA-IC and oxLDL+DNA-IC (Fig 5D). Following stimulation of monocytes, TACE activity positively correlated with sLOX-1 levels in culture supernatants ( Fig 5E). In addition, sLOX-1 cleavage was inhibited in the presence of either TACE inhibitor (TAPI-1, 100 μM) or 50 μg/mL of the LOX-1 receptor blocking antibody (Fig 5F). The physiologic relevance of this relationship was also observed in SLE patients, where TACE activity levels were higher in the high sLOX-1 groups (low sLOX-1 mean = 60.4 ± 4.409, n = 167, high sLOX-1 mean = 79.75 ± 3.851, n = 101) and correlated significantly with sLOX-1 levels (r s = 0.422, p = 3.01e -13 ) (Fig 5G and S2 Table). This data indicates that upregulation of surface LOX-1 and TACE activity lead to increased sLOX-1 levels in SLE patients.

oxLDL promotes low-density granulocytes to form extracellular traps
Low density granulocytes (LDGs) are prevalent in SLE patients but not healthy controls [36], and these cells are sensitive to the generation of NETs which can induce endothelial injury and promote inflammatory responses [37]. Interestingly, we found that LDGs (which have similar phenotype to PMN-myeloid derived suppressor cells (MDSCs) [38, 39] expressed high levels of LOX-1 in SLE patients, and slightly lower levels in healthy donors ( Fig 6A). LOX-1 expression on CD14+ monocytic MDSCs were much lower and there was no difference between SLE and healthy donors ( Fig 6B).  6C). The impact of oxLDL on NET formation was inhibited by the anti-LOX-1 receptor blocking antibody or a ROS inhibitor (DPI; 10 μM).

Discussion
We report for the first time that sLOX-1 may be useful in identifying SLE patients with cardiovascular risk. sLOX-1 levels were two-fold higher in SLE patients. We postulate that sLOX-1 is driven by activation of inflammatory pathways associated with either SLE disease activity or atherogenesis secondary to SLE diagnosis. In our studies, sLOX-1 levels were positively associated with hsCRP levels, proinflammatory HDL, oxLDL and impaired HDL efflux, rather than traditional risk factors and SLE disease activity.
Although the mechanistic explanation for the dissociation between serum sLOX-1 levels and SLE disease activity require further studies, our data suggests that different aspects of inflammation contribute to SLE and CVD. Consistent with our findings, in an independent cohort of SLE patients with low disease activity and without pre-existing CVD, preclinical vascular damage was observed and associated with type I IFN activity [41]. One of the largest studies to examine cardiovascular risk in SLE was conducted on 1,784 patients with a total of 9,485 person-years follow-up showed cardiovascular risk was associated with current rather than past disease activity, and CV events may be precipitated by acute changes in disease activity [42]. This is not necessarily discordant with our data since the sLOX-1 levels may reflect underlying vascular inflammation.
It was also notable that the patients with high sLOX-1 levels were younger than those with low sLOX-1 levels. Both groups had a similar time since diagnosis, which indicates that the high sLOX-1 group also had an earlier onset of disease. This may appear unexpected since oxLDL and LOX-1 increase with age and cardiovascular disease in the general population. However, SLE patients have a greater risk of cardiovascular disease and the disparity is particularly evident in young females. Women under the age of 44 with SLE have >52-fold risk of myocardial infarction. Coronary events are rare in women under 55 years, whereas 54% of cardiac events in a female SLE patient population occurred under the age of 44 [43]. Therefore, the increased sLOX-1 levels in this younger SLE population may reflect subclinical atherosclerosis and the enhanced risk of CVD which is consistent with high sLOX-1 levels associated with higher hsCRP, oxLDL and impaired HDL functionality.
The MDC cohort enabled the first large, prospective study showing relevance for high levels of sLOX-1 with plaque size and increased risk for future ischemic stroke. An analysis of stroke risk factors in the MDC cohort on 4703 subjects showed that increasing sLOX-1 levels in plasma correlated significantly with increases in hsCRP and size of carotid plaques. A separate CPIP (Carotid Plaque Imaging Project) cohort further validated significant correlation between plaque LOX-1 and sLOX-1. Furthermore, proinflammatory cytokines and ox-LDL levels correlated with plaque as well as soluble LOX-1 content [32]. In a recent study with 173 psoriasis patients by Dey et al., sLOX-1 was shown to be correlated with non-calcified plaque burden irrespective of hyperlipidemia. Additionally, an improvement in psoriasis was associated with a decrease in sLOX-1 levels [29].
Our data showed that the SLE patients with high sLOX-1 levels had significant increase in hsCRP, a well-known clinical marker of inflammation in cardiovascular disease [44]. hsCRP has been successfully used as a patient stratification marker for ACS and reduction in hsCRP has been shown to result in CV event reduction [33, 45,46]. sLOX-1 is elevated in the systemic and coronary circulation of patients with acute coronary syndrome (ACS) [25,26] and is proposed as a marker for presence of active and vulnerable inflammatory atherosclerotic lesions [47]. A recent "liquid biopsy" analysis of proteins released locally by the vasculature indicated that LOX-1 is the most abundant protein released by vessels with atherosclerotic plaques compared to controls [48].
Based on evidence from our current study, we postulate that even in low disease activity patients, sLOX-1 may be indicative of low-grade inflammation contributing to atherogenesis or subclinical atherosclerosis. sLOX-1, in fact, performed better than hsCRP in predicting higher oxLDL levels and impaired CEC in SLE patients (Fig 3 and S1 Fig). It has been apparent that CRP is not a reliable predictive marker of cardiovascular disease or underlying inflammation in SLE patients as it is in other conditions. This likely reflects the increased type I IFN activity in SLE patients which has been shown to suppress CRP levels [49]. Given this limitation of CRP as a CV marker for SLE patients, sLOX-1 may serve as a better predictive marker of dyslipidemia and cardiovascular risk. Longitudinal studies in SLE cohorts will be required to confirm this finding.
Dysregulated lipoproteins also represent a significant family of atherogenic LOX-1 ligands and are usually a hallmark of CVD inflammatory disease. piHDL is the HDL incapable of removing reactive oxygen species (ROS) from LDL [10]. In our high sLOX-1 group, we observed higher piHDL levels, a measure that McMahon et al. established as a 28-fold increased risk predictor of plaque deposition [50]. 50% of women with SLE had piHDL, as compared with fewer than 10% of age-matched healthy women and 87% of SLE patients with plaque on carotid ultrasound had piHDL, as compared with 41% of those without carotid plaque. Normal HDL also functions in maintaining efficient cholesterol efflux, a mechanism that regulates reverse cholesterol transport. In light of multiple failures in past trials involving HDL cholesterol-raising therapies [51], a large-scale longitudinal study following 2924 individuals over 9.4 years was undertaken to demonstrate an inverse relationship of HDL cholesterol efflux capacity (CEC) with the incidence of cardiovascular events [52,53]. In SLE, ATP-binding cassette A1 and G1 (ABCA1/ABCG1)-dependent CEC has been shown to be impaired [15,54]. In our study, high sLOX-1 patients were significantly impaired in ABCA1-dependent CEC compared to the low sLOX-1 patients. Based on our data, sLOX-1 levels can distinguish a subset of SLE patients with the likelihood of having impaired HDL cholesterol, and potentially increased oxidized LDL (oxLDL) levels.
Atherosclerotic lesions develop when low-density lipoproteins (LDLs) are oxidized into oxLDL phospholipids by ROS generation from endothelium dysfunction [55] combined with the inability of HDL to inhibit their oxidation. oxLDL is then phagocytized by macrophages leading to the formation of foam cells that necrotize and give rise to atherosclerotic plaque [56]. Our studies for the first time establish that higher oxLDL levels are prevalent in SLE patients with high sLOX-1 and that oxLDL levels positively and significantly correlate with sLOX-1 levels. Mechanistically, higher levels of oxLDL can potentiate increased activation of LOX-1 pathway. oxLDL engages in a positive feedback loop enhancing LOX-1 levels on macrophages and endothelial cells [57]. We detected significantly upregulated LOX-1 on the surface of monocytes and LDGs in SLE patients with their numbers strongly correlating with sLOX-1 levels. Our in vitro studies showed that oxLDL can induce membrane LOX-1 expression on monocytes and further drive cleavage of sLOX-1 in a TACE-dependent manner. Both LOX-1 blocking antibody and the TACE inhibitor, TAPI-1, could reverse sLOX-1 cleavage. This is consistent with previous studies on human monocyte-derived macrophages where sLOX-1 released by CRP stimulation has been shown to be attenuated by a TACE inhibitor or TACE siRNA [31]. TACE activity in human plasma is measurable [58,59]. Detection of TACE activity in serum of SLE patients primarily, revealed a strong positive correlation of TACE activity with sLOX-1 levels.
Functionally, oxLDL was pathogenic to monocytic cells through its ability to prime them to be more proinflammatory with DNA-IC stimulation. We found that co-stimulation with DNA-IC leads to TNF-α, IL-1β and IL-6 secretion that can potentially feed forward into the LOX-1/sLOX-1 induction pathway since proinflammatory cytokines such as TNF-α and IL-1β have been shown to induce LOX-1 expression [60,61]. Breaking this feed forward loop can be beneficial, for example, IL-1β β blockade (CANTOS trial) has been shown to successfully reduce inflammation-driven CVD [62,63].
LOX-1 was also increased on CD15 + /HLADR -LDGs also known as granulocytic MDSCs, a specialized subset of polymorphonuclear cells found to be elevated in SLE, cancer and inflammatory diseases [64][65][66]. Our data reveals that oxLDL can prime LDGs in a LOX-1-dependent manner to promote the generation of NETs. This process typically yields the release of DNA, myeloperoxidase (MPO), neutrophil elastase (NE), proteinase 3 (PR3); secondary neutrophil granules: lactoferrin, pentraxin 3 and gelatinases [67]. The presence of NETs has been documented in human plaques with a superficial erosion-like morphology [68]. Plasma from patients with eroded plaques exhibit high MPO, produced primarily by neutrophils, compared with those with ruptured lesions [69]. In SLE patients with plaques, LDG-specific NET associated gene signatures, including MPO, correlated with noncalcified plaque burden [70]. In fact, LDGs through NET formation were also shown to enhance HDL oxidation, implying loss of atheroprotective capacity [15]. Thus, NET-induced HDL oxidation [70] can convert HDL into a LOX-1 activating ligand [24]. Since our studies also show that LOX-1 activation with oxLDL promotes the generation of NETs, it is conceivable that that NET-induced lipid oxidation may amplify LOX-1 mediated NET formation. These results justify additional studies to evaluate LOX-1-mediated LDG activation and NET generation in atherosclerotic conditions.
In this cross-sectional study, we have provided indirect evidence that sLOX-1 levels may represent a useful biomarker for cardiovascular risk in SLE patients. The size of the study was underpowered to provide direct linkage between LOX-1 levels and cardiovascular events. It was notable that 10 patients that were identified with a prior or recent history of cardiovascular disease trended towards higher LOX-1 levels 750.3± 742.9 pg/mL compared to the SLE population as a whole 580.9 ± 36.1 pg/mL. In lieu of a longitudinal study to evaluate soluble LOX-1 levels, flares in disease activity and the development of cardiovascular disease, a study has been initiated to evaluate soluble LOX-1 levels and sub-clinical atherosclerosis in SLE patients. Prior studies of this sort have indicated that sLOX-1 levels correlate with carotid plaque inflammation and risk of ischemic stroke [32], and track with the burden of non-calcified coronary plaques in psoriasis patients [29].
We have provided insight into the regulation and LOX-1 expression, and experimental evidence which indicates that oxidized lipids may drive LOX-1 dependent atherogenic pathways in SLE patients. Perturbed activation of the LOX-1 pathway may explain the increased risk of cardiovascular events in SLE patients, and the inhibition of the LOX-1 pathway may provide protection from the development of cardiovascular disease.

Patients samples and clinical assessments
SLE patient blood and serum samples were obtained from the Warren G. Magnuson Clinical Center Blood Bank (Bethesda, MD) as approved by the National Institute of Arthritis and Musculoskeletal and Skin Diseases/National Institute of Health between 2013 and 2018. Clinical and demographic characteristics, SLEDAI-2K [71] and Framingham Risk Scores (FRS) were calculated at each visit. Laboratory parameters including fasting blood glucose and lipid panel, white blood count, platelet count, C3, C4 complement levels, and systemic inflammatory markers such as hsCRP and ESR were quantified in the clinical laboratory at the NIH. Laboratory tests for anti-nuclear antibodies (ANA), extractable nuclear antigens (ENA) and double-stranded DNA (dsDNA) antibodies were all performed and reported for the cohort. Medication usage for all patients was also noted. Blood and serum from healthy donors were obtained from individuals enrolled at the MedImmune Research Specimen Collection Program. SLE and healthy cohorts are described in S1 Table.

Measurement of sLOX-1 levels in SLE serum and supernatants
An in-house sandwich ELISA was developed using the Mesoscale diagnostics (Mesoscale Diagnostics, MD, USA) platform to measure soluble LOX-1 levels in human serum and supernatants. MSD high bind plates were coated with 5 μg/ml in-house generated anti-LOX-1 antibody overnight, blocked for 1 hour and samples (25 μl/well, no dilution) were added along with recombinant human LOX-1 as standard for 2h. Plates were washed using MSD Tris wash buffer 3 times after each incubation step. Human LOX-1/OLR1 antibody (AF1798 from R&D systems, MN, USA) was sulfotagged using MSD conjugation kit (R31AA-2) to generate detection antibody. Sulfo-tagged detection antibody was added and incubated for 1 hr. 2X MSD read buffer was used to read plates on MSD machine. sLOX-1 levels from the samples were interpolated from standard curve values using MSD workbench software (Mesoscale Diagnostics, MD, USA).

HDL assays
HDL was isolated from human serum by polyethylene glycol precipitation (PEG) of LDL using 20% w/v PEG in PBS as described before [72].
Proinflammatory HDL (piHDL) index. piHDL was measured using a cell-free assay has been developed and reported previously [9, 73,74]. 20 μL of the normal LDL (Cell Biolabs, CA, USA) at a concentration of 50 μg/ml and 90 μL of test HDL from healthy of SLE individuals at a final concentration of 10 μg/mL cholesterol were incubated in quadruplicate in 96-well plates for 1h. 10 μL of DCFH-DA solution (0.2 mg/mL) was then incubated in each well for 2 hours. Presence of oxidized form of LDL leads to the conversion of normally non-fluorescent dichlorofluorescein diacetate (DCFH-DA) into a fluorescent form (DCFH). DCFH is then measured on a plate reader (SpectraMax, Molecular Devices, CA, USA) at an excitation wavelength of 485 nm and an emission wavelength of 530 nm. Fluorescence units were then compared with absence of test HDL in the mixture which was set at a value of 1.

Cholesterol Efflux Capacity (CEC) assays
CEC assays were performed based on previously reported technique on J774 cells [75]. Briefly, cells were plated and radiolabeled with 2 μCi of 3 H-cholesterol/ml. 0.3 mmol/l 8-(4-chlorophenylthio)-cAMP was added to the cells for 16h to upregulate ATP-binding cassette transporter A1 (ABCA1). Serum HDL obtained after apoB depletion, as described above, from healthy and SLE individuals was added for 4h. Liquid scintillation counting was performed to count effluxed radioactive cholesterol by HDL from cells. CEC was then calculated by using the following formula: (μCi of 3 H-cholesterol in media containing 2.8% apoB-depleted subject plasma-μCi of 3 H-cholesterol in serum-free media/μCi of 3 H-cholesterol in media containing 2.8% apoB-depleted pooled control serum-μCi of 3 H-cholesterol in pooled control plasmafree media). % efflux was then divided by total cholesterol content from HDL serum that was added to obtain % efflux per μg HDL cholesterol. The pooled healthy serum was obtained from 3 healthy volunteers. All assays were performed in triplicate.

oxLDL measurements
Oxidized LDL (oxLDL) was measured in human serum samples by an ELISA kit (Mercodia Inc, Uppsala, Sweden) which is a solid phase two-site enzyme immunoassay based on the direct sandwich technique in which two monoclonal antibodies were directed against separate antigenic determinants on the oxidized apolipoprotein-B molecule. During incubation, oxLDL in the sample reacts with anti-oxLDL antibodies bound to microtitration well. After washing, which removes non-reactive plasma components, a peroxidase conjugated antihuman apolipoprotein B antibody recognizes the oxidized LDL bound to the solid phase. After a second incubation and a simple washing step that removes unbound enzyme labeled antibody, the bound conjugate is detected by reaction with 3,3', 5,5'-tetramethylbenzidine (TMB). The reaction is stopped by adding acid to give a colorimetric endpoint, then read spectrophotometrically.

Measurement of TACE activity in SLE serum and supernatants
TACE activity was measured in SLE patient serum as well as from supernatants of stimulated monocytes with oxLDL and DNA-IC. Where indicated, anti-LOX-1 antibody (50 μg/mL) or TAPI-1 inhibitor (100 μM) were used to pre-treat cells for 30m. TACE activity was measured by using a synthetic peptide substrate containing the cleavage site (Mca-Pro-Leu-Ala-Gln-Ala-Val-Dpa-Arg-Ser-Ser-Ser-Arg-NH2) primarily for ADAM17 and related enzymes such as ADAM8, ADAM9 and ADAM10 (R&D Systems, Inc., MN, USA). The cleavage site by ADAM17 and ADAM10 is the peptide bond between Ala and Val. 10 μL of each sample (serum or supernatant) was added to 90 μL of 25 mM Tris buffer, pH 8.0. The substrate was then added at a final concentration of 10 μM in a total of 100 μL reaction mixture for 1h. Fluorescence units (FU) were measured using a fluorescent microplate reader, Spectramax (Molecular Devices, CA, USA), with an excitation wave length at 320 nm and an emission wavelength at 405 nm. TACE activity results were normalized to total protein concentration of serum samples or supernatant. Enzymatic activity of TACE is represented as FU/min/μg.

In vitro determination of NET formation by LDGs
Isolated LDGs at 2x10 4 cells per well were plated in 96-well flat-bottom plates in HBSS (Ca 2+ and Mg2+ free, ThermoFisher) with 2% heat-inactivated FBS. After incubation with anti-LOX-1 antibody (50 μg/mL) or diphenyleneiodonium (DPI; 10 μM) treatment at 37˚C for 30 min, cells were primed with oxLDL where indicated. PMA (25 nM) or RNP immune complexes (RNP 1 μg/mL + 2% RNP + SLE serum) were used to treat LDGs for 4 h in the presence of anti-MPO-FITC at 5 μg/mL and NucRed at 4 μL/100 μl/well. Without washing, cells were imaged using high content imaging system, Incucyte (Essen BioScience, MI, USA) with a 20x objective. Cells from 9 fields were quantified in duplicates for each treatment. The number of MPO positive NETs were quantified by the Incucyte software (Essen BioScience, MI, USA).

Statistical analysis
Summary statistics are presented as mean ± SEM or median as indicated for continuous variables and categorical variables. For group comparisons, two-tailed Student's t test was used for parametric data analysis as seen in in vitro data, and Welch's test was used for nonparametric data analysis for comparing sample means with unequal variances and unequal sample sizes in low sLOX-1 versus high sLOX-1 SLE groups, when normal distribution of the data was not guaranteed. For correlation analysis, Spearman's correlation analysis was used and the coefficient reported as r s . For certain values in tables, standardized multivariate regression analysis was performed, and standardized β-coefficients and p values were reported. hsCRP, TACE activity, triglycerides, HDL efflux, oxLDL levels represent dependent variables, adjusting for cardiovascular and cardiometabolic risk factors. Statistical analysis was performed using R. P values � 0.05 are considered statistically significant.
Supporting information S1