Drs. Watanabe, Charles-Schoeman, and Miao contributed equally to this work.
Proteomic profiling following immunoaffinity capture of high-density lipoprotein: Association of acute-phase proteins and complement factors with proinflammatory high-density lipoprotein in rheumatoid arthritis
Article first published online: 25 MAY 2012
Copyright © 2012 by the American College of Rheumatology
Arthritis & Rheumatism
Volume 64, Issue 6, pages 1828–1837, June 2012
How to Cite
Watanabe, J., Charles-Schoeman, C., Miao, Y., Elashoff, D., Lee, Y. Y., Katselis, G., Lee, T. D. and Reddy, S. T. (2012), Proteomic profiling following immunoaffinity capture of high-density lipoprotein: Association of acute-phase proteins and complement factors with proinflammatory high-density lipoprotein in rheumatoid arthritis. Arthritis & Rheumatism, 64: 1828–1837. doi: 10.1002/art.34363
- Issue published online: 25 MAY 2012
- Article first published online: 25 MAY 2012
- Accepted manuscript online: 9 JAN 2012 11:56AM EST
- Manuscript Accepted: 22 DEC 2011
- Manuscript Received: 9 MAR 2011
- NIH (National Heart, Lung, and Blood Institute). Grant Numbers: 5K23-HL-094834, 5R01-HL-082823
- Arthritis Foundation
- Southern California Chapter
To identify protein biomarkers associated with proinflammatory high-density lipoprotein (HDL) in patients with active rheumatoid arthritis (RA) by proteomic analysis.
Liquid chromatography tandem mass spectrometry (LC-MS/MS) was used to analyze proteins associated with immunoaffinity-purified HDL from plasma obtained from 2 sets of RA patients, 1 with antiinflammatory HDL and 1 with proinflammatory HDL. Proteins were fractionated by Offgel electrophoresis and analyzed using an LC-MS/MS system equipped with a high-capacity high-performance liquid chromatography chip incorporating C18 reverse-phase trapping and analytical columns. Sandwich enzyme-linked immunosorbent assays were used to validate the association between select proteins and proinflammatory HDL in a second cohort of RA patients.
Seventy-eight proteins were identified in the HDL complexes. The levels of 12 proteins were significantly increased in RA patients with proinflammatory HDL compared to RA patients with antiinflammatory HDL. These proteins included the acute-phase proteins apolipoprotein J, fibrinogen, haptoglobin, serum amyloid A, and complement factors (B, C3, and C9). The associations between proinflammatory HDL and 4 of the proteins were validated in a second RA cohort.
Our findings indicate that proinflammatory HDL in patients with RA contains a significantly altered proteome, including increased amounts of acute-phase proteins and proteins involved in the complement cascade. These findings suggest that HDL is significantly altered in the setting of chronic inflammation in active RA, with resultant loss of its antiinflammatory function. The characterization of the biomarkers described herein may identify novel molecular connections that contribute to the higher risk of cardiovascular disease in RA patients.
Patients with rheumatoid arthritis (RA) have a significantly increased risk of cardiovascular disease (CVD), including myocardial infarction and sudden cardiac death, which is not explained by traditional CVD risk factors (1). Previous work has suggested that systemic inflammation contributes to CVD in RA, and patients with active disease and high inflammatory burden are at significantly increased risk (2–4). Although some studies have suggested that RA mortality rates may be falling in response to new therapies (5), survival trends are not keeping pace with the general population, and CVD remains the leading cause of death (6). The investigation of novel mechanisms of accelerated atherosclerosis in patients with RA is therefore important both for appropriate treatment and for aggressive primary prevention.
The inverse relationship between high-density lipoprotein (HDL) cholesterol and the risk of CVD is well established (7, 8). However, a significant number of cardiovascular events occur in patients with normal HDL and low-density lipoprotein (LDL) cholesterol levels (9, 10). Previous work has suggested that the inflammatory nature of HDL may be a more sensitive marker of CVD than HDL cholesterol levels. Thus, there is a need for further investigation of biomarkers with better predictive value based on HDL function (11–13).
HDL plays numerous antiinflammatory and atheroprotective roles by promoting reverse cholesterol transport and preventing the oxidation of LDL (14, 15). However, the protective function of HDL is impaired and HDL becomes proinflammatory during pathologic processes that accelerate cardiovascular events (16–18). The molecular changes and mechanisms that promote the conversion of antiinflammatory HDL to proinflammatory HDL are currently unknown. Knowledge of the molecular profiles that distinguish proinflammatory HDL from antiinflammatory HDL may facilitate better understanding of the alterations in the protein cargo of HDL, which adversely affect its normal antioxidant and antiinflammatory functions.
Proinflammatory HDL levels are increased in RA patients compared to healthy controls (19). We recently reported that levels of proinflammatory HDL are positively correlated with disease activity in RA patients (20). However, how proinflammatory HDL is involved in RA disease activity and its relationship to CVD is unknown. In this study, we evaluated HDL with distinct inflammation properties (either antiinflammatory or proinflammatory) from the plasma of RA patients. HDL was isolated using immunocapture columns and further subjected to our newly developed Offgel electrophoresis and liquid chromatography tandem mass spectrometry (LC-MS/MS) method to identify protein markers that distinguish proinflammatory HDL from antiinflammatory HDL. Multiple proteins were identified in HDL, including several proteins significantly associated with proinflammatory HDL. The associations between proinflammatory HDL and 4 of these proteins were validated in a second RA cohort by HDL-capturing enzyme-linked immunosorbent assay (ELISA). Since the inflammatory nature of HDL has previously been directly linked to CVD, and is also significantly correlated with disease activity in RA, characterization of the biomarkers we identified in the present study may lead to the detection of novel molecular connections that play a role in the higher risk of CVD in RA patients.
MATERIALS AND METHODS
Reagents for proteomic analysis.
The following high-performance liquid chromatography (HPLC)–grade and proteomic reagents were used: acetonitrile (Fisher), 2,2,2-trifluoroethanol and formic acid (both from Fluka), dithiothreitol and iodoacetamide (both from Sigma), sequencing-grade modified trypsin (Promega), iTRAQ 8-Plex reagents (Applied Biosystems), and Immobiline DryStrip and immobilized pH gradient buffer, pH3–10 (both from GE Healthcare) for Offgel fractionation.
RA patients were recruited through flyers posted in the University of California, Los Angeles (UCLA) rheumatology offices and in the UCLA Medical Center. Patients inquiring about the study from flyers posted in their rheumatologist's office or elsewhere were referred to one of the study physicians if the treating rheumatologist was not involved in the protocol. All patients met the American College of Rheumatology 1987 revised criteria for the classification of RA (21), which was verified by chart review. Written informed consent for the study was obtained from all subjects under a protocol approved by the Human Research Subject Protection Committee at UCLA.
Clinical evaluation of RA patients.
The Disease Activity Score in 28 joints (DAS28) (22), levels of inflammation markers (high-sensitivity C-reactive protein and erythrocyte sedimentation rate), and fasting lipid profiles were obtained in individual patients as previously described (20). Disease-related disability was assessed using the disability index of the Health Assessment Questionnaire (23).
Determination of the inflammation properties of HDL.
The antioxidative status of HDL was determined by cell-free assay with 2,7,7′-dichlorofluorescein diacetate as previously described (24). Briefly, plasma samples were isolated from RA patients after an overnight fast, cryopreserved in 10% sucrose, and freshly frozen at −80°C until used. HDL-containing supernatants were isolated by LipiDirect HDL reagent (Polymedco) and assayed for cholesterol content (Thermo DMA). Fifty microliters of HDL (100 μg HDL cholesterol/ml) was tested for each sample. Oxidized dichlorofluorescein (DCF) was used as an indicator of reactive oxygen species measured by fluorescence intensity at 485 nm/530 nm. Readings with DCF and LDL were normalized to 1.0 (HDL inflammatory index). After the addition of HDL to LDL, samples with an HDL inflammatory index of <1.0 were considered antiinflammatory, while samples with an HDL inflammatory index of ≥1.0 were considered proinflammatory.
Paraoxonase and arylesterase activity assays.
Both assays were performed as previously described, with minor modifications (14, 25). For the arylesterase assay, plasma was diluted 1:50 with assay buffer and incubated with phenyl acetate. Arylesterase activity was measured by kinetic reading with absorbance at 270 nm (A270 nm) for 2 minutes. The kinetic rate at A270 nm was multiplied by the molecular extinction coefficient of 0.7633 and dilution factor of 2.5 to represent a unit of arylesterase activity per ml of plasma.
Western blot analysis of paraoxonase 1.
Equal amounts of plasma protein (50 μg) from all samples (determined by measuring A280 nm using a Thermo Scientific NanoDrop 2000 spectrometer) were loaded onto 10% mini-Protean TGX gels (Bio-Rad) for separation and transferred onto nitrocellulose membranes (Bio-Rad). After blocking with 5% skim milk in Tris buffered saline–Tween 20 for 60 minutes at room temperature, the membranes were incubated overnight at 4°C with a goat anti-human paraoxonase 1 antibody (1:2,000; R&D Systems). After washing, the membranes were further incubated with horseradish peroxidase (HRP)–conjugated anti-goat IgG antibody (1:5,000; Santa Cruz Biotechnology), and developed with ECL Plus (GE Healthcare).
HDL isolation with anti-HDL IgY spin columns for proteomic analysis.
Plasma protein (6.5 mg) was loaded onto anti-HDL chicken IgY immunocapture spin columns according to the recommendations of the manufacturer (GenWay Biotech). The protein concentrations in plasma were measured by the BCA Protein Assay. The first wash was combined with the initial sample solution to make the flowthrough sample. Eluted HDL was desalted using a YM-3 centrifugal filter unit (Millipore). The protein concentration of HDL captured by the IgY column was determined using a NanoDrop spectrophotometer (Thermo).
Separation of HDL proteins by sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE).
HDL (30 μg of proteins) was loaded onto a 4–12% Bis-Tris NuPAGE precast gel (1.5 mm thick). Proteins were visualized by staining gels overnight in SimplyBlue SafeStain (Invitrogen). The gel lanes were cut into 12 sections for trypsin digestion and analysis by LC-MS/MS.
Separation of HDL proteins by Offgel electrophoresis.
Proteins were separated according to their isoelectric point (pH 3–10) by Offgel electrophoresis according to the recommendations of the manufacturer (Agilent). Briefly, a total of 100 mg of HDL protein was evenly distributed across 12 wells of an Agilent 3100 Offgel fractionator and separated at pH 3–10. The proteins were focused with a maximum current of 50 mA until a volt hour limit of 50 kVh was reached. Fractions were collected by first pipetting the liquid from each well. Next, 100 μl of water/methanol/formic acid (49%/50%/1% by volume) was added to each well and incubated for 90 minutes without voltage. The liquid was then removed, combined with the first portion removed from the well, acidified with formic acid, and concentrated by vacuum centrifugation prior to LC-MS/MS analysis. LC-MS/MS analysis was performed using 25% of each fraction.
Analysis of HDL in RA patient samples by 8-Plex iTRAQ.
The iTRAQ 8-Plex reagent kit was used to test all 8 HDL samples from the RA patients (4 with proinflammatory HDL and 4 with antiinflammatory HDL) in a single experiment. Equivalent amounts of each isolated HDL protein were digested with trypsin and derivatized with one of the iTRAQ reagents.
Since glycine in the IgY elution buffer is incompatible with the iTRAQ reagent, eluted HDL was desalted with a reverse-phase C18 HPLC column (4.6 mm inner diameter and 50 mm long, 5 μm particle size; Agilent) at a temperature of 80°C using a trifluoroacetic acid (TFA) buffer system (0.1% aqueous TFA in solvent A and 0.08% TFA in acetonitrile in solvent B). Samples were loaded onto 10% solvent B, eluted with a 10–70% gradient of solvent B over 2 minutes, ramped to 100% over 1 minute, and held at 100% solvent B for 2 minutes. The entire protein peak was collected and concentrated using a vacuum centrifuge. The protein concentration was determined using NanoDrop spectrophotometry.
The desalted HDL sample (200 μg of proteins) was reduced, alkylated, and digested with trypsin using trifluoroethanol as a detergent. A 1-dimensional LC-MS/MS analysis was performed on each digest mixture to ensure the quality of the digestion reaction. A 60-μg aliquot of each sample was labeled with one of the iTRAQ reagents according to the recommendations of the manufacturer (Applied Biosystems). Briefly, 1 vial of iTRAQ reagent was used for each digested HDL sample. The iTRAQ reagent was solubilized in 100% isopropanol and added to the peptide sample at a final concentration of 60% (volume/volume). The solution pH was adjusted to 7.5–8.5 to optimize the labeling efficiency (>99%). After labeling for 2 hours at room temperature, the samples were combined and concentrated by vacuum centrifugation. The labeled samples (200 μg of proteins) were separated by Offgel electrophoresis. Each fraction was analyzed in triplicate by LC-MS/MS.
All MS analyses were performed on an Agilent 6520 quadrupole time-of-flight mass spectrometer equipped with an Agilent 1200 series liquid chromatography instrument and an Agilent Chip Cube LC-MS interface. LC separations used a high-capacity HPLC chip consisting of a 160-nl enrichment column and a 75 μm × 150 mm analytical column, both packed with Zorbax 300 SB-C18, 5-μm reverse-phase support. Peptides were loaded onto the enrichment column with 97% solvent A (water with 0.1% formic acid) and 3% solvent B (90% acetonitrile, 10% water with 0.1% formic acid) at a flow rate of 4 μl/minute, and eluted with a linear gradient of 8–30% solvent B for 45 minutes and then 30–80% solvent B for 1 minute at a flow rate of 0.3 μl/minute. Positive-ion electrospray mass spectra were acquired using a capillary voltage set at 1,850V, the ion fragmentor set at 175V, and the drying gas set at 300°C and 4 liters/minute. MS spectra were collected in centroid mode over a mass range of 375–2,500 mass/charge (m/z) at a scan rate of 4 spectra/second. MS/MS spectra were collected in centroid mode over a range of 50–3,000 m/z and an isolation width of 4 atomic mass units. The collision energy was ramped at a slope of 2.5 and an offset of 3.7. The top 6 most intense precursor ions for each MS scan were selected for tandem MS with active exclusion for 0.33 minutes.
Peptide and protein identification.
The iTRAQ labels are designed such that the same peptide from each sample has the same molecular weight and chromatographic retention time, but yields a different fragment or reporter ion when the MS/MS spectrum is acquired. The relative intensity of the reporter ion signals in an individual MS/MS spectrum is a direct measure of the relative levels of that peptide in each of the original samples.
The spectral data were converted to mass/charge data format using Agilent MassHunter Qualitative Analysis Software (B.03.01) and processed through the Computational Proteomics Integrated Environment (COPINE) hosted at the City of Hope National Medical Center. COPINE is a collection of comprehensive proteomics data analysis tools, which comprise the Global Proteome Machine (from The Global Proteome Machine Organization). X!Tandem Tornado (2009.04.01.1) was used as the database search engine. Spectra were searched against the UniProt Human database together with the reverse UniProt Human database and a custom database of common contaminants. Search parameters included a fragment mass error of 50 parts per million, a parent mass error of 50 ppm, trypsin cleavage specificity, and carbamidomethyl as a fixed modification of cysteine. Deamidation of asparagine and glutamine, oxidation of methionine to methionine sulfoxide and sulfone, and acetylation of the N-terminus were specified as variable modifications. For the X!Tandem search of the iTRAQ-derivatized samples, the iTRAQ 8-Plex was specified for peptide N-terminus and lysine residues. A merged, nonredundant output file was generated for protein identifications with loge values <−1. A concatenated decoy database consisting of the reverse of the sequences in the UniProt database was used for all searches.
Results of the database search were analyzed by Scaffold (version 3.00.01) running on COPINE. Peptide identifications were accepted if they could be established at >90% probability as specified by the Peptide Prophet algorithm (26). Protein identifications were accepted if they could be established at >99% probability and contained ≥2 identified peptides. Protein probabilities were assigned by the Protein Prophet algorithm (27). Proteins that contained similar peptides and could not be differentiated based on MS/MS analysis alone were grouped to satisfy the principle of parsimony.
Reporter ion intensity data were collected from the iTRAQ LC-MS/MS runs using the Agilent Spectrum Mill database search program. Peptides of <7 amino acid residues were excluded because of the high probability that they would match >1 protein. Only proteins with ≥2 unique iTRAQ peptides in each of the triplicate runs were included in the results. Values reported are for the sum of the reporter ion intensities for all of the peptides assigned to a protein. Reporter ion values for each of the replicated runs were averaged to yield a total reporter ion intensity for each sample. The average coefficient of variation was 0.37 for the proteins in the antiinflammatory data set and 0.34 for the proteins in the proinflammatory data set.
Analysis of HDL-associated proteins by ELISA.
Individual plasma samples from randomly selected RA patients with antiinflammatory or proinflammatory HDL were assayed by HDL-capturing sandwich ELISAs as previously described (16). The following antibodies were used in the ELISA experiments: chicken antibody for human HDL (GenWay Biotech); mouse antibodies for human apolipoprotein J (Apo J), serum amyloid A (SAA), HRP-conjugated Apo A-I, fibrinogen, and haptoglobin (all from Abcam); and HRP-conjugated secondary antibodies for anti-mouse antibody (GE Healthcare).
Briefly, 96-well polyvinyl chloride microtiter plates were precoated with 2 μg/ml of chicken anti-human HDL antibodies overnight at 4°C. Following incubation of the precoated plates with individual plasma samples diluted 1:10 with 1× phosphate buffered saline (PBS), the plates were washed thoroughly, blocked with 5% nonfat milk in PBS, and incubated with corresponding primary antibodies to Apo J or SAA, or HRP-conjugated antibodies to Apo A-I, fibrinogen, or haptoglobin at a 1:2,500 dilution. The unconjugated primary antibodies were detected by HRP-conjugated secondary antibodies for mouse immunoglobulin at a 1:2,500 dilution. Following incubation with tetramethylbenzidine solution, HRP activity was measured at an optical density of 450 nm (OD450 nm). The HRP-conjugated antibody for each target protein was coated in empty wells in a series of different concentrations as a standard to convert the OD450 nm of each sample to the concentration of the HRP-conjugated antibody of each target protein. The value for each protein in a given sample was represented as a relative concentration compared to the average value for that protein in plasma samples of RA patients with antiinflammatory HDL, which was set at 1.
The distributions of all data were carefully examined to determine the appropriate parametric or nonparametric test for analysis. For the proteomics data, a 2-sample t-test with unequal variance (with Satterthwaite adjustment) was used. P values were not corrected for multiple comparisons given the exploratory nature of the proteomics analyses. For the clinical measures and HDL-associated protein ELISAs, patient groups were compared using a t-test for continuous variables and the chi-square test of association for categorical variables, along with Fisher's exact test in cases where there were small sample sizes in individual categories. When needed, nonparametric Wilcoxon rank sum tests were used to compare groups for skewed continuous variables. DAS28 is known to be normally distributed and was treated as such even with a small sample size (Table 1). P values less than 0.05 (2-tailed) were considered significant.
|Patients with antiinflammatory HDL (n = 4)||Patients with proinflammatory HDL (n = 4)||P|
|Age, years||53.2 ± 9.7||55.3 ± 9.2||0.76|
|Sex, % female||100||100||1.0|
|DAS28||2.12 ± 0.84||7.84 ± 0.71||<0.0001|
|HAQ DI||0.0 (0.2)||2.5 (1.7)||0.027|
|hsCRP, mg/liter||0.3 (0.6)||39.9 (64.9)||0.029|
|ESR, mm/hour||7.25 ± 7.09||84.3 ± 30.1||0.012|
|Total cholesterol, mg/dl||185.3 ± 7.3||184.8 ± 31.4||0.98|
|LDL cholesterol, mg/dl||91.5 ± 16.3||100.0 ± 20.4||0.54|
|HDL cholesterol, mg/dl||68 (40)||52 (10)||0.06|
|Triglycerides, mg/dl||90 (90)||173 (170)||0.31|
|HDL inflammatory index†||0.22 ± 0.11||1.34 ± 0.29||0.002|
|Paraoxonase, units/ml (by paraoxonase assay)||243 (233)||59 (18)||0.19|
|Paraoxonase, units/ml (by arylesterase assay)||163 ± 144||238 ± 127||0.47|
Findings of HDL isolation with immunocapture spin columns.
HDL complexes were isolated using immunocapture spin columns with IgY antibody for human HDL. For these particular patient samples, the amount of protein obtained from the IgY capture was significantly higher for the proinflammatory group. This suggests that there was either a significant difference in the amount of Apo A-I or Apo A-II present in the samples, or that the ratio of the amount of Apo A-I or Apo A-II relative to the total amount of other proteins was different. When the results were corrected for the fraction of the sample used, the MS analysis indicated that there was no significant difference in the amounts of Apo A-I and Apo A-II, but that the levels of many other proteins were significantly higher in the patients with proinflammatory HDL, as described below. The ability of this method to enrich for HDL-associated proteins is evident in the SDS-PAGE analysis of the starting plasma (input), the fraction that does not bind to the column (flowthrough), and the fraction that comes off with the elution buffer (bound) (Figure 1).
Results of proteomic analysis of proinflammatory HDL in RA patients.
We recently demonstrated that the proinflammatory properties of HDL are positively correlated with systemic inflammation in patients with CVD (28) and patients with RA (20). In the present study, we used plasma samples from RA patients with proinflammatory HDL (HDL inflammatory index ≥1.0) and those with antiinflammatory HDL (HDL inflammatory index <1.0). There was no difference between RA patients with proinflammatory HDL and those with antiinflammatory HDL with regard to age, sex, cholesterol levels, or triglyceride levels (Table 1). RA patients with proinflammatory HDL had significantly higher disease activity (Table 1).
Proteins associated with HDL complexes were labeled with iTRAQ reagents as previously described (29) and subjected to LC-MS/MS following Offgel electrophoresis. A total of 78 proteins were identified in HDL (Figure 2), including lipid metabolism factors and transporters, protease inhibitors, acute-phase proteins, complement proteins, and coagulation factors. Twelve of these proteins were significantly increased in proinflammatory HDL compared to antiinflammatory HDL in RA patients, including fibrinogen, complement factors, α1-antitrypsin, haptoglobin, Apo J, immunoglobulin heavy chain, serpin D1 (heparin cofactor II), and SAA (Figure 2). The total reporter ion intensity assigned to each protein provided a rough measure of the relative amount of each protein in the sample. The value of the total reporter ion intensity for the 2 patient groups was within 5% (P = 0.71), providing confirmation that the protein amounts analyzed were well matched (data not shown).
Association of proinflammatory mediators with proinflammatory HDL in RA patients.
To validate the discovery of protein biomarkers of proinflammatory HDL in RA patients (Figure 2), we tested plasma samples from additional RA patients with antiinflammatory HDL and additional RA patients with proinflammatory HDL. Sandwich ELISA was performed to capture HDL and validate the association of proinflammatory HDL with the following 4 markers involved in the acute phase of inflammation: Apo J, fibrinogen, haptoglobin, and SAA. We chose these 4 markers because their levels were highly significantly increased in proinflammatory HDL samples in the first cohort of RA patients, their levels are increased during vascular inflammation, and they have been linked specifically to RA, a disease associated with increased CVD risk secondary to systemic inflammation. All 4 markers were significantly associated with proinflammatory HDL (Figure 3). No significant differences in HDL-associated Apo A-1 levels were noted between the 2 groups (Figure 3).
A trend toward decreased paraoxonase 1 protein levels in the proinflammatory HDL samples was noted in the initial proteomic analysis (Figure 2). Followup assessment of paraoxonase 1 protein levels by Western blotting followed by densitometry in the validation cohort (n = 15 samples per group) revealed a significant decrease (of a mean ± SD of 21% ± 16.4%) in paraoxonase 1 protein levels in proinflammatory HDL samples compared to antiinflammatory HDL samples. Paraoxonase 1 activity was also assessed by means of the arylesterase and paraoxonase assays in both cohorts of patients. A nonsignificant trend toward higher paraoxonase 1 activity (as determined by paraoxonase assay) in the antiinflammatory HDL samples compared to the proinflammatory HDL samples was noted in the proteomics cohort (P = 0.19) (Table 1). A similar trend was noted in the larger validation cohort (Table 2). These data are consistent with previous work suggesting a link between paraoxonase 1 activity and inflammation in patients with RA (30).
|Patients with antiinflammatory HDL (n = 16)||Patients with proinflammatory HDL (n = 15)||P|
|Age, years||56.3 ± 10.7||58.3 ± 13.3||0.66|
|Sex, % female||94||93||1.0|
|Ethnicity, % Caucasian||69||67||1.0|
|DAS28||3.1 ± 0.9||6.0 ± 1.4||<0.0001|
|HAQ DI||0.13 (0.84)||1.31 (1.46)||0.002|
|hsCRP, mg/liter||1.3 (1.9)||4.7 (16.2)||0.005|
|ESR, mm/hour||10 (8)||45 (35)||0.0001|
|Total cholesterol, mg/dl||184.2 ± 38.2||208.7 ± 47.2||0.12|
|LDL cholesterol, mg/dl||101.3 ± 31.8||114.9 ± 29.7||0.23|
|HDL cholesterol, mg/dl||59 (23)||71 (42)||0.28|
|Triglycerides, mg/dl||107 (75)||102 (64)||0.84|
|HDL inflammatory index†||0.32 (0.09)||1.38 (0.46)||<0.0001|
|Paraoxonase, units/ml (by paraoxonase assay)||201 (233)||133 (124)||0.12|
|Paraoxonase, units/ml (by arylesterase assay)||221 (180)||195 (87)||0.38|
Oxidation of LDL is one of the major factors in the development of human atherosclerosis (31, 32). Entrapment and oxidation of LDL in the subendothelial space and the subsequent interactions between endothelial cells and monocytes is a key process in the initiation of atherosclerotic lesion development (33, 34). The inverse relationship between HDL cholesterol and the risk of atherosclerosis is well established (8). Although there does not appear to be a single explanation for the antiatherogenic role of HDL, it has become clear that the functional status of HDL, which is largely dependent on its protein components, is probably an important determinant of CVD (18).
Over a decade ago, it was shown that the antiinflammatory properties of HDL are impaired in rabbits (14), mice (15), and humans (13) during inflammatory processes. This impaired HDL is proinflammatory in nature, as characterized by decreased levels and activity of antiinflammatory, antioxidant factors (17); a gain of proinflammatory proteins (14); increased lipid hydroperoxide content (18); reduced potential to efflux cholesterol (35); and diminished ability to prevent LDL oxidation (36). Previous work has also suggested that HDL inflammatory properties may be a more sensitive marker of CVD than HDL cholesterol levels (13). However, the molecular changes and mechanisms that promote the conversion of antiinflammatory HDL to proinflammatory HDL are currently not well understood.
RA is a systemic inflammatory disease associated with high cardiovascular risk (1). Previous work by our group has shown that HDL in patients with RA has abnormal antiinflammatory properties compared to that in healthy controls, and that the antiinflammatory function of HDL is significantly correlated with disease activity and systemic inflammation; higher disease activity was associated with worsened ability of HDL to inhibit LDL oxidation (20). Previous studies also suggested alterations in the levels of select proteins associated with HDL (16, 28).
In the present study, we developed an Offgel electrophoresis and LC-MS/MS method to identify proteins that indicate whether HDL is proinflammatory or antiinflammatory. Among the 78 proteins identified in HDL, the following 12 were significantly associated with proinflammatory HDL: fibrinogen (α, β, and γ chains), complement factors (C3, C9, and B), α1-antitrypsin, haptoglobin, Apo J, immunoglobulin heavy chain, serpin D1 (heparin cofactor II), and SAA (Figure 2). These results suggest that alternative proteins, particularly acute-phase proteins, associate with HDL that might exert proinflammatory and proatherogenic functions.
Interestingly, 11 of the 12 proteins (all except fibrinogen) were also identified in HDL from CVD patients (37). Fibrinogen is a plasma glycoprotein that is converted to fibrin by thrombin during blood coagulation. Fibrinogen is considered to be a thrombosis marker that contributes to the high rate of mortality from CVD in RA patients (38, 39). How HDL-associated fibrinogen is involved in thrombosis needs to be further investigated in order to understand the pathologic mechanisms involved in cardiovascular events in RA patients.
We previously identified hemoglobin (Hgb) and its accessory proteins, haptoglobin and hemopexin, as biomarkers of proinflammatory HDL in CVD patients (16, 28). Haptoglobin was also significantly associated with proinflammatory HDL in RA patients in the present study (Figures 2 and 3). In contrast, Hgb and hemopexin were associated with HDL, but were found to be independent of HDL inflammation properties in this study (Figure 2), which may be attributable to the small sample size. Heme regulatory factors associated with proinflammatory HDL regulate the redox reaction and vascular tone in the circulation (16). The oxidative properties of heme produce reactive oxygen species, causing oxidative stress (40). Thus, based on our previous findings and the findings of this study, haptoglobin appears to be the most predictive biomarker of proinflammatory HDL to evaluate the protective role on oxidative status.
Proinflammatory HDL can be converted to antiinflammatory HDL by diet (37) or therapeutic interventions such as Apo A-I mimetic peptides (41). Previous studies showed that HDL function in RA patients may be modestly improved with high-dose statin therapy (42), methotrexate (20), and infliximab (30). The present study identified significant alterations in the proteome of proinflammatory HDL in RA patients, which may aid in the development of additional therapeutic strategies to improve HDL antiinflammatory properties. Large, prospective controlled studies are necessary to confirm the direct cause-and-effect relationship between HDL function and disease activity in RA and its relationship to CVD.
It should be noted, however, that the protein changes demonstrated in this study were limited by the fact that we analyzed the entire HDL population through the immunoaffinity method. Given the fact that HDL is highly heterogeneous in particle composition, this approach cannot identify the specific particles of HDL with which the protein cargo is associated. The method can be further improved by adding a size exclusion step following the immunoaffinity capture of HDL to further identify specific changes in protein cargo that are associated with specific HDL particles in RA patients. Furthermore, it should also be noted that for most of the highly abundant proteins, including fibrinogen, we suspect that the HDL-associated component is quite small compared to the corresponding serum concentrations.
In summary, this is the first study to use immunoaffinity capture of HDL with subsequent proteomic analysis to describe the protein cargo of HDL in patients with RA, a chronic inflammatory disease associated with significantly increased cardiovascular morbidity and mortality. Our findings implicate HDL as an active participant in the inflammatory response, carrying multiple complement proteins, serine protease inhibitors, and other proteins involved in cell signaling and the coagulation cascade. In addition, marked differences were seen in the HDL cargo of patients with active disease and proinflammatory HDL, compared to patients with low disease activity and antiinflammatory HDL. These findings are consistent with a known link between inflammation and accelerated CVD in RA, and suggest that HDL-associated biomarkers may warrant further investigation, including network and pathway analyses. Future work on the proteins identified in these studies will determine whether any single one of these markers or any combination of these markers is predictive of cardiovascular events in RA patients, and can be reversed with antiinflammatory therapies.
All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Reddy had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study conception and design. Watanabe, Charles-Schoeman, Miao, Katselis, T. D. Lee, Reddy.
Acquisition of data. Watanabe, Charles-Schoeman, Miao, Y. Y. Lee, Katselis, T. D. Lee.
Analysis and interpretation of data. Watanabe, Charles-Schoeman, Miao, Elashoff, Y. Y. Lee, Katselis, T. D. Lee, Reddy.
- 13Inflammatory/antiinflammatory properties of high- density lipoprotein distinguish patients from control subjects better than high-density lipoprotein cholesterol levels and are favorably affected by simvastatin treatment. Circulation 2003; 108: 2751–6., , , , , , et al.
- 39Elevated levels of small, low-density lipoprotein with high affinity for arterial matrix components in patients with rheumatoid arthritis: possible contribution of phospholipase A2 to this atherogenic profile. Arthritis Rheum 2001; 44: 2761–7., , , , , , et al.