Lipid Anti-Lipid Antibody Responses Correlate with Disease Activity in Systemic Lupus Erythematosus

Systemic Lupus Erythematosus (SLE) is a chronic autoimmune disorder characterized by broad clinical manifestations including cardiovascular and renal complications with periodic disease flares and significant morbidity and mortality. One of the main contributing factors to the pathology of SLE is the accumulation and impaired clearance of immune complexes of which the principle components are host auto-antigens and antibodies. The contribution of host lipids to the formation of these autoimmune complexes remains poorly defined. The aim of the present study was to identify and analyze candidate lipid autoantigens and their corresponding anti–lipid antibody responses in a well-defined SLE patient cohort using a combination of immunological and biophysical techniques. Disease monitoring in the SLE cohort was undertaken with serial British Isles Lupus Assessment Group (BILAG) scoring. Correlations between specific lipid/anti-lipid responses were investigated as disease activity developed from active flares to quiescent during a follow up period. We report a significant negative correlation between anti-lipid antibodies for 24S-hydroxycholesterol, cardiolipin and phosphatidylserine with SLE disease activity. Taken together, these data suggest that lipid autoantigens represent a new family of biomarkers that can be employed to monitor disease activity plus the efficacy of therapeutic intervention in SLE.


Introduction
Systemic Lupus Erythematosus (SLE) is a chronic inflammatory autoimmune disease found predominantly in women. Complex interactions amongst immune, genetic, environmental and hormonal factors have been implicated in SLE susceptibility and pathogenesis [1]. Numerous mouse and human studies have implicated dysfunctional cellular and immune components including autoimmune T and B lymphocytes [2,3,4]; elevated levels of pro-inflammatory cytokines [5]; formation of antinuclear antibodies [6]; accumulation and impaired clearance of postapoptotic cell remnants [7,8] or failure of FccR-mediated clearance of immune complexes [9] in the pathology of Systemic Lupus Erythematosus.
The role of lipids and anti-lipid responses in Systemic Lupus Erythematosus and other autoimmune diseases remains poorly defined in comparison to proteins and genetic factors based on the technical challenges inherent in their analysis. A summary of studies linking oxysterols, phospholipids and prostaglandin derivatives with autoimmune, degenerative and age-related diseases including SLE is provided in Table 1. Thus there is a requirement for a broader and more detailed analysis of the role of lipids in these diseases.
Oxysterols represent the family of host lipids most strongly implicated in autoimmune conditions (Table 1). These are oxygenated derivatives of cholesterol that are intermediates in the cholesterol excretion pathway [10]. Cholesterol oxidation is either through attack by reactive oxygen species (ROS) that oxygenate the sterol ring at the C7-position or by enzymatic hydroxylation of cholesterol side-chains that generate 24S-, 25and 27-hydroxycholesterol respectively [11]. 24S-hydroxycholesterol is specifically generated in the central nervous system [12][13] and plasma levels of this lipid have been implicated in diseases linked to CNS inflammation including Alzheimer's and Vascular dementia [14].
Elevated plasma levels of 24S-hydroxycholesterol was reported in Multiple Sclerosis (MS) patients with positive cranial MRI scans indicating an acute inflammatory episode of demyelination [15]. Oxidized phosphatidylcholine and their corresponding autoantibodies have also been implicated in MS [16]. Other lipid markers including F(2)-isoprostanes, 7-b-hydroxycholesterol, 7-ketocholesterol and 27-hydroxycholesterol have been linked to Parkinson's disease [17]. 7-ketocholesterol may also be involved in the pathophysiology of atherosclerosis where it is suspected of inducing apoptosis in the cells of the vascular wall including monocytes/macrophages [18]. This lipid is also known to be related to oxidized-LDL-mediated cytotoxicity [19]. 7-b hydroxycholesterol is proposed to promote human NK cell death and may also be involved in atherosclerosis [20]. This study focuses on the role of oxidized lipids and anti-lipid responses in Systemic Lupus Erythematosus (SLE).

Patients
The patient cohort employed was composed of individuals referred to Addenbrooke's Hospital, Cambridge, UK between 2004 and 2008. All patients provided written informed consent and ethical approval was obtained from the Cambridge Local Research Ethics Committee (Ref: 04/023). Blood was collected at two time points: the moment of disease [21] -flare; and during the follow-up period. Follow-up was defined as the period between 3months and 12-months post therapy. Disease monitoring was undertaken with serial BILAG scoring [22]. All patients were enrolled with active disease with an average BILAG score of 16.01 prior to treatment. Patients on treatment entered clinical remission and the average BILAG score in the follow up period was reduced to 2.4+/22.1. For all patients, a full haematological, biochemical and immunological profiling was done [21]. BILAG scores for 3 time-points and clinical data for patients with flare are included (Supplementary table). Thirteen paired SLE patients' samples were used to analyze changes in lipid and anti-lipid IgG levels between the flare and follow-up period (between 3-12 months post-therapy). Twenty patients with flare (including 13 previously mentioned patients) were used for the correlation analysis.

Blood Processing
Blood was collected in EDTA tubes and peripheral blood mononuclear cells (PBMC) separated on Ficoll-Paque PLUS gradient (GE Healthcare, Sweden). Plasma was stored at 280uC prior to use.

Enzyme-linked Immunosorbent Assay (ELISA) and Lipid Quantification
Maxi-sorp plates (NUNC, Denmark) were coated with 5 ug/ml of lipids in EtOH evaporated for 2 hrs at RT. Plates were blocked with 0.8% collagen in PBS (Sigma-Aldrich, USA). Anti-lipid IgG responses were detected using goat-anti human IgG antibody conjugated with horseradish peroxidase (HRP) (Thermo Scientific Pierce, USA). TMB substrate (BD OptEIA TM , BD Biosciences) was employed for 15 mins and colour development assayed at 450 nm using a Perkin-Elmer Victor 3 V plate reader.

Lipid Extraction and Gas Chromatography-mass Spectrometry (GC-MS)
Lipids were extracted using a modified Folch method and hydrolysed in methanolic KOH [23]. COPs and F 2 -Isoprostanes were then purified into 2 separate fractions by solid phase extraction on mixed mode anion exchange columns, dried under a stream of nitrogen, derivatised and injected onto an Agilent 5973/ 6890 GC-MS system as described previously [24]. Briefly, cholesterol oxidative products (COPs) were derivatised to generate their trimethyl silyl ether derivatives and analysed in the EI (electron ionization) mode. F 2 -isoprostanes were derivatised to form their pentafluorobenzyl (PFB) ester, trimethyl silyl ether derivatives and analysed in the NCI (negative chemical ionization) mode. Quantification was achieved by relating peak area of samples with respective deuterated internal standards according to the previously published method [25]. 8-Iso-PGF 2 a and 5-F 2t -IsoP F 2 -isoprostane isomers co-eluted and were measured together as a single peak.

Lipid Extraction and Liquid Chromatography -Mass Spectrometry (LC -MS)
Phospholipids were extracted according to a modified protocol of Bligh and Dyer, 1959 [26]. Synthetic lipids obtained from Avanti Polar Lipids (Alabaster, AL, USA) were spiked as internal standards. The extracted lipids were measured using ABI 3200QT (Applied Biosystems, Foster City, CA) interfaced to a HPLC system using multiple reaction-monitoring mode [27]. Phospholipids were separated on a Phenomenex Luna 3 m C18 column (150 mm62 mm)(Phenomenex, Torrance, CA, USA). Signal intensities obtained for each lipid class were normalized to the appropriate internal standard.

Statistical Analysis
Differential levels of lipids or anti-lipid IgGs between time points were assessed in Prism (GraphPad Software) using the Wilcoxon nonparametric matched pairs test with a p,0.05 value considered significant in all cases. A correlation between different clinical parameters and anti-lipid IgG levels was examined using a Spearman nonparametric correlation and Linear regression tests.

Total IgG Levels in SLE Patients Remain Unchanged during Therapy
A reduction in BILAG score was employed as the principal indicator for treatment success in our SLE patient cohort. We observed a decrease in this score as patients progressed from flare through follow-up period (Fig. 1a). Absolute levels of IgG in the patients' plasma remain unchanged at these time points (Fig. 1b).
In Table S1, we present BILAG scores for 8 systems: General (Gen), Mucocutaneous (Muc), Neurological (Cns), Musculoskeletal (Msk), Cardiovascular and Respiratory (Car), Vasculitis (Vas), Renal (Ren) and Haematological (Hae). As documented in this table, BILAG scores in SLE patients at 3-months and 12-months indicate that these patients have quiescent disease and can thus be combined as our follow-up cohort.

SLE Patients with Active Disease Exhibit Increased Levels of Lipid and Anti-lipid Responses
Gas chromatography-mass spectrometry (GC-MS) analyses indicate significantly higher levels of oxidized cholesterols in patients during flare versus follow-up. Specifically, 7-b-hydroxycholesterol, 7-ketocholesterol (Fig. 2bi, 2ci) were significantly increased. At the same time, 24S-hydroxycholesterol and 7-ahydroxycholesterol levels remained unchanged (Fig. 2ai, 2di). Only five of the patients had a form of therapy aimed at reducing cholesterol levels-a combination of fenofibrate, pravastatin or simvastatin (Table S2). In four out of five patients, administration of lipid-lowering drugs started from the moment of flare and lasted over next 12 months. Administration of statins will reduce cholesterol levels. However, we have not addressed to what extent statins will affect levels of oxidized cholesterols over time in these patients. Anti-7-a-hydroxycholesterol IgG responses against oxysterols were significantly higher in patients with flare in comparison with the results obtained during the follow-up period (Fig. 2aii). The same trend was seen in anti-7-b-hydroxycholesterol and anti-7-ketocholesterol IgGs though these differences were not significant ( Fig. 2bii and 2cii). Anti-24S-hydroxycholesterol IgG levels did not change over time (Fig. 2dii).

Levels of Oxidatively Modified Unsaturated Fatty Acids and Anti-lipid Antibodies Change during Therapy
Mass-spectroscopy analyses of phosphatidylserine during the flare and follow-up period suggest a reduction in this lipid over time (Fig. 3ai). Also, anti-phosphatidylserine IgG levels tend to increase during the follow-up period (Fig. 3aii).
Normalized values for isoprostanes (normalized against arachidonic acid (AA)) significantly decrease from the moment of flare ( Fig. 3bi and 3ci). We employed 15-F2t-IsoP as the antigen in this assay where we measured anti-isoprostane IgG response (Fig. 3bii). Anti-15-F2t-IsoP IgG exhibited a trend of higher levels in flare compared to follow-up.

Anti-lipid IgG Levels for Other Lipids also Change during Therapy
Lathosterol is a precursor of cholesterol and we detected significantly higher anti-lathosterol IgG responses in patients with flare (Fig. 4ai). Phospholipids and anti-phospholipid antibodies are often seen in autoimmune disorders and are associated with cerebrovascular disease or moderate-to-severe cognitive dysfunction. In our study we detected significantly higher levels of anticardiolipin (Fig. 4aii) and anti-oxidized phopsphatidylcholine in flare (Fig. 4bii). Anti-phopsphatidylcholine IgG in patients with flare was also increased (Fig. 4bi).

Levels of Anti-lipid Antibodies Negatively Correlate with BILAG Scores
We analyzed our lipid anti-lipid data for correlations with a score of disease activity in SLE. For those analyses we included an additional 7 SLE patients for which we had BILAG scores during the flare. In total we analyzed 20 patients. Spearman correlation and linear regression tests showed that anti-24S hydroxycholesterol IgG, anti-cardiolipin IgG and anti-phosphatidylserine IgG, negatively correlate with BILAG score -the disease activity index (Fig. 5ai, 5aii, 5bi respectively). A similar, although not significant trend was seen in anti-7-a hydroxycholesterol IgGs (Fig. 5bii). A higher level of these anti-lipid antibodies was observed in patients with lower BILAG scores suggesting that these may be markers for improved prognosis.

Discussion
In SLE, the heterogeneity of disease presentations in patient populations combined with a poor understanding of the underlying pathological mechanisms result in challenges for successful disease management. The American College of Rheumatology classification utilizes 11 criteria for diagnosis of lupus, of which a patient must meet four [28]. Several computerised indices for measuring clinical disease activity in Systemic Lupus Erythematosus are employed in clinical practice-BILAG, SLEDAI, LUMINA [29][30][31][32]. Despite advances in treatment protocols there remains a dearth of good diagnostic and prognostic biomarkers to facilitate improvements in disease characterization and management [33][34][35]. Moreover, the immune modulating treatment used for SLE remains a significant problem with many patients developing treatment-associated complications [36]. Thus, the identification of biomarkers that correlate with a good response to therapy will have an impact on treatment-associated side-effects where less-aggressive approaches can be employed [33].
In our lipidomic and immunologic approaches we have identified a new corpus of biomarkers relevant to this disease. This study focused on a clinically well-defined SLE cohort that was monitored over a period of 12 months. Based on clinical score and statistical tests, we grouped samples provided between 3 and 12 months post-therapy as our follow-up cohort. This approach enabled us to apply statistical tests for paired samples.
We were also interested in correlations between lipid/anti-lipid IgG levels and BILAG scores during disease flares. In this case we did not need paired samples and thus employed 20 samples in total.
We have identified new targets that correlate with disease score. Our approach employed gas chromatography-mass spectrometry (GC-MS) and liquid chromatography-mass spectrometry (LC-MS) to investigate the lipids of interest. The presence of anti-lipid autoantibodies was confirmed by an enzyme-linked immunosorbent assay where we focused on the immunoglobulin isotype G.
Our high-resolution analysis has identified a cohort of new antilipid autoantibodies. Whilst total IgG levels remain consistent during therapy, we detected differential levels of anti-lipid IgGs for four different lipid species: lathosterol; enzymatically produced oxysterols 7-a-hydroxycholesterol and oxidatively modified phospholipids -cardiolipin, and oxidized phosphatidylcholine.
The measurement of oxygenated cholesterol species remains technically challenging based on the proposed auto-oxidation that can occur during lipid derivation, processing or storage from biological samples. [37][38][39]. In our study plasma levels of oxysterols showed a general trend of decrease over time. Thus, if auto-oxidation occurs, it should happen at a similar level in all samples tested without a significant influence on absolute ratios. The main conclusion of this study concerns relationships between anti-lipid antibodies and BILAG score. Anti-lipid antibodies identified in lupus patients' plasma are part of an autoimmune response that targets lipid antigens. These develop as a result of pathogenic processes in the afflicted patients. Thus, antibodies from SLE patient's plasma should not target oxidized cholesterols that are result of in vitro auto-oxidation processes (e.g. during the ELISA plate coating with lipid-antigen). We found similar levels of anti-24S-hydroxycholesterol IgGs between flare and follow-up (Fig. 2dii). There was no significant difference in IgG levels against 27-hydroxycholesterol when analyzing the same time points (data not shown). Phosphatidylserine levels in plasma showed a trend of reduction over the time. Anti-cardiolipin antibodies are one of several anti-phospholipid antibodies that have been previously identified in SLE patients [40] where cardiolipin present on the surface of apoptotic cells acts as an immunologic trigger for the production of the autoantibodies [41].
Isoprostanes are generated by the free radical-mediated peroxidation of arachidonic acid (AA) [42]. 15-F2t-IsoP is a marker of free radical damage and lipid peroxidation in vivo that is formed by free radical catalysis of arachidonic acid [43]. Serum levels of 15-iso-PGF2alpha and 8-iso-PGF2alpha in SLE patients showed a significantly higher level at flare compared to the posttherapy period.
BILAG is currently accepted as the best disease activity score in SLE [22] and thus we analyzed which if any of our lipid/anti-lipid parameters correlate. We observed that anti-phosphatidylserine, anti-cardiolipin and anti-24S hydroxycholesterol IgG negatively correlate with the BILAG score. Anti-7-a-hydroxycholesterol IgGs also show trend of negative correlation with BILAG score. However, for this anti-lipid response we were not able to confirm a statistically significant correlation.
At the same time points we were not able to detect statistically significant correlations between BILAG scores and one traditional SLE biomarker -anti-DNA antibodies (data not shown). This can potentially be explained by the fact that anti-DNA antibodies are found in only 60% of SLE patients and those antibodies are particularly associated with lupus nephritis [44]. The negative correlations observed can be explained by two possible mechanisms: the presence of anti-lipid IgG during flare may be beneficial and aid immune complex clearance through IgG receptors expressed on phagocytic cells such as the Kuppfer cells in the liver [45][46]; or reduced levels of lipids and their corresponding anti-lipid antibodies in plasma are a consequence of deposition of immune complexes in the tissues. Both of these mechanisms are currently under investigation in our laboratory.
It is an interesting observation that levels of anti-lipid IgGs for 24S-hydroxycholesterol and phosphatidylserine remained very similar over the time period of study. Since levels of these antibodies negatively correlate with BILAG score at the time of flare, one explanation could be that their presence during the postflare period might have a protective role. For the actual levels of phosphatidylserine we saw a trend of reduction between flare and follow-up. This phenomenon suggests that either oxidation processes in unsaturated fatty acids were reduced or the development of phosphatidylserine immune complexes intensified over time.
IgM autoantibodies have been linked to apoptosis-associated antigen clearance [47][48]. IgM anti-phosphorylcholine was shown to be notably higher in patients with low SELENA-SLEDAI disease activity index [47]. Those patients encountered less cardiovascular and renal problems and high levels of IgM are attributed to the homeostatic and protective roles [47]. The panel of auto-antigens that appear during the apoptosis is large and further identification of protective IgG and IgM antibodies is required.
Clinical and laboratory markers currently used in SLE have moderate utility based on specificity and sensitivity. Based on our findings, we suggest that the measurement of anti-lipid IgGs for 24S-hydroxycholesterol, cardiolipin and phosphatidylserine may be used as a sensitive and non-invasive method of surveillance during treatment. These therefore represent improved biomarkers for the evaluation and development of better therapeutic strategies aimed at reducing treatment associated morbidity and mortality, a significant problem in SLE.  Figure  4ai anti-lathosterol IgG levels in plasma. Figure 4aii anti-cardiolipin IgG levels in plasma. Figure 4bi anti-phosphatidylcholine IgG levels in plasma. Figure 4bi anti-oxidized phosphatidylcholine IgG levels in plasma. p,0.05 value considered significant in Wilcoxon test. doi:10.1371/journal.pone.0055639.g004 Figure 5. Biomarker candidates that correlate with BILAG scores. Figure 5ai, 5bi, 5ci. Anti-24S hydroxycholesterol IgG, anti-cardiolipin IgG and anti-phosphatidylserine IgG levels negatively correlate with BILAG scores. Figure 5bii. Anti-7-a-hydroxycholesterol IgG levels also show trend of negative correlation with BILAG scores. p,0.05 value was considered as significant in Spearman correlation and Linear regression tests. doi:10.1371/journal.pone.0055639.g005 Table S1 BILAG scores summary for three