Contributions of mass spectrometry-based proteomics to defining cellular mechanisms and diagnostic markers for systemic lupus erythematosus

Systematic lupus erythematosus (SLE) is a complex disease for which molecular diagnostics are limited and pathogenesis is not clearly understood. Important information is provided in this regard by identification and characterization of more specific molecular and cellular targets in SLE immune cells and target tissue and markers of early-onset and effective response to treatment of SLE complications. In recent years, advances in proteomic technologies and applications have facilitated such discoveries. Here we provide a review of insights into SLE pathogenesis, diagnosis and treatment that have been provided by mass spectrometry-based proteomic approaches.


Introduction
Immune-mediated reactions in systemic lupus erythe mato sus (SLE) aff ect many diff erent body systems, including the central nervous system (CNS), blood cells, joints, skin, heart, lungs, and kidneys [1]. Although immunesuppressive treatment is eff ective, there are undesirable side eff ects and reduced response to treatment in some cases [1][2][3]. Th us, an important objective in SLE research is the identifi cation of more specifi c therapeutic targets by defi ning novel cellular mechanisms of disease. Another objective is the identifi cation of reliable, non-invasive, and quantifi able markers for detection of the early-onset of specifi c complications. Such markers would enable treatment to be administered most eff ectively along with assessment of positive response to therapy such that the treatment can be modifi ed or stopped in a timely manner to best manage adverse side eff ects. Th ese markers also provide important pathogenic insight and tools for testing new or improved therapeutics. In addition to clinical features (rash, arthritis), diagnostic information for SLE is also provided by measurement of immune cell profi les and activity, identifi cation of specifi c autoantibodies, and identifi cation of changes in protein expression profi les in bodily fl uid (urine, blood, cerebral spinal fl uid).
Mass spectrometry-based proteomic technologies have played an important role in each area of clinical diagnosis as well as the development of a more comprehensive understanding of the underlying disease process using a myriad of diverse sample types and techniques. One of the preferred mass spectrometry methods in proteomics combines one-or two-dimensional liquid chromato graphy (LC) peptide separation with electro-spray ionization (ESI) tandem mass spectrometry (MS/MS) [4]. Th is LC-MS/MS methodology allows for a direct and highly sensitive identifi cation of hundreds of individual proteins from virtually any type of biomedical sample [5]. Th e decision to use one-dimensional versus two-dimensional LC is based on the complexity of the protein sample. Typically, SDS-PAGE bands or spots (<30 proteins) are analyzed by one-dimensional and much more complex samples, such as affi nity-purifi cations and whole cell, tissue, or bodily fl uid extracts are analyzed with twodimensional LC-MS/MS. Another type of mass spectrometry that is employed is matrix-assisted laser desorption/ ionization (MALDI)-time of fl ight (TOF) [6]. With this approach up to 96 individual protein samples are spotted onto a stationary target for analysis. Although the sensitivity of MALDI-TOF is limited to characterizing the 10 to 15 most abundant proteins in each sample, it has the benefi t of being a higher throughput platform for lower complexity, pre-fractionated protein mixtures (after gel separation) because each sample is analyzed in minutes whereas a typical one-dimensional LC run requires an hour and two-dimensional LC-MS/MS takes 10 to 12 hours. A third mass spectrometry approach is surface-enhanced laser desorption/ionization (SELDI)-TOF, a modifi cation of MALDI-TOF [7]. With SELDI-TOF, diff erent surface components, such as strong anion exchangers or hydrophobic characteristics, allow binding of proteins with certain characteristics while dissimilar proteins are washed away. Th is allows analysis of targeted subsets of structurally related proteins and reducing the complexity of the sample improves the overall sensitivity or dynamic range of detection. SELDI-TOF results diff er from LC-MS/MS and MALDI-TOF-TOF in that the results are given in mass to charge ratios (m/z) rather than peptide sequence, so positive protein identifi cation is not possible. It is, however, useful for rapid analysis of the protein m/z profi les of semi-complex samples by reducing upfront separation while preserving the fast analysis time of a MALDI platform. Although not as desirable for discovery, these attributes of relative ease of sample preparation and speed of analysis and data output, as well as lower startup and operation costs, present SELDI-TOF as a more suitable mass spectrometry platform for a clinical test.
A key component of these studies is the method used to prepare selectively targeted protein samples for mass spectrometry interpretation. Auto-antigen identifi cation studies usually involve separation of the tissue of interest by one-dimensional or two-dimensional electrophoresis SDS-PAGE, followed by immuno-blotting with anti-sera from SLE patients, and identifi cation of proteins in immune-reactive bands by MALDI-TOF-TOF or LC-MS/MS analysis [8]. Another approach isolates autoantibodies from SLE patient serum using immobilized protein G, which is then used to affi nity enrich antigens in targeted tissue extracts for direct characterization with two-dimensional LC-MS/MS analysis [9]. Studies aimed at defi ning diagnostic markers for SLE complications by comparing protein expression in normal and diseased bodily fl uids have used two basic approaches: assessment of protein expression diff erences from stained twodimensional gels followed by MALDI-TOF-TOF identi fication of the proteins in diff ering spots; or direct assessment and comparison of protein expression from intrinsic LC-MS/MS data [8,10]. Th e work fl ow of these appli cations is described in Figure 1.
In this review we seek to explain prior contributions of mass spectrometry techniques to the biological under standing, diagnosis and treatment of SLE and to lay out a framework for future work in the fi eld. It is our hope that the subsequent information will allow a greater under standing of the mass spectrometry-based techniques and illustrate their potential in translational and clinical research related to SLE and other autoimmune diseases.

Cardiovascular systemic lupus erythematosus
A great deal of eff ort has been put into understanding the changes to the cardiovascular system in SLE. Cardiac involvement in SLE has been reported with a prevalence up to 50% [11], but on average only 9% will develop true myocarditis [12,13]. SLE has long been known to result in immune deposits in the pericardium and/or myocardium of SLE patients, resulting in both histopathologic and gross anatomic changes [14]. Arrhythmias, including sinus tachycardia, atrial fi brillation, long QT, atrio ventricular block and bundle branch block, are rare complications, but can have signifi cant eff ects when combined with other factors such as antimalarial therapy [13] and studies have reported up to 16% of children with SLE to have asymptomatic cardiovascular disease [15]. Th e risk of developing coronary artery disease is four to eight times higher in SLE patients than in controls, and this risk increases to 50-fold in young women [16]. Treatment can be tailored to specifi c patients at risk for the development of such complications, but for this to be possible, early diagnosis of SLE is required followed by the accurate gauging of cardiovascular risk. Mass spectro metry fi ndings have been used to characterize the proteomic changes in the SLE disease state to better understand the pathogenic mechanisms behind cardiovascular disease to allow earlier diagnosis and improved risk assessment [10,[17][18][19][20].
To better understand pathogenesis of cardiovascular SLE and provide biomarkers for prompt diagnosis, autoantigens involved in congenital heart block were subjected to proteolytic cleavage with and without sera of SLE patients, followed by MALDI-TOF-TOF identifi cation of antibody-protected sites (epitopes) [19]. Th e resulting peptides were found to correspond to the 200 to 239 amino acid portion of the Ro52 protein, supporting the use of autoantibodies against Ro52 as a biomarker. Subsequent studies have supported the role of anti-Ro antibodies (both Ro52 and Ro60) in congenital heart block and suggested potential mechanisms for the develop ment of the condition [21,22].
Haptoglobin α2 polypeptide chain diff erences have been noted between SLE and non-SLE populations via MALDI-TOF mass spectrometry [23]. Patients expressing the haptoglobin α2 isoform have been shown to have shorter survival times after coronary artery bypass graft [24] and more severe complications after myocardial infarc tion [25], possibly due to decreased ability to reduce oxidative stress. Th e association of haptoglobin α2 peptides with SLE fl are may off er insight into the cardiac complications seen in SLE and higher levels of haptoglobin α2 have been reported in SLE patients versus controls [23].
To identify target autoantigens and explore the pathogenesis of dilated cardiomyopathy, Seko and colleagues [10] applied a two-dimensional electrophoresis separation approach to plasma membrane proteins isolated from rat cardiac myocyte cultures. Th e protein blot was then probed with serum of healthy controls as well as a 17-year-old patient with an unusual presentation of delayed cardiomyopathy subsequent to neonatal lupus erythe matosus. While the control showed no signifi cant reactivity, many immune-reactive spots appeared on the sample probed with SLE serum, which were excised and analyzed on a LC-MS/MS system. Th e analysis identifi ed two high-probability proteins: annexin A6, known for calcium signaling and membrane traffi cking, and vimentin, known for maintenance of cellular structure (Table 1). Th e authors downplayed the likelihood of vimentin play ing a role in the pathogenesis of myocardial contractile dysfunction. However, mice with overexpressed annexin A6 had decreased contraction and relaxation of cardiac myocytes as well as dilated cardiomyopathy [26], suggesting that if autoantibodies to annexin A6 stimulate its function, then the result would be suppression of myocardial activity.

Immune cell involvement in systemic lupus erythematosus
SLE is, at its core, a disease of the immune system, so a deeper understanding of immune cell function in both normal and diseased states is critical for a comprehensive understanding of SLE. Complementary studies by Dai Starting from sample, which could be any preparation of proteins, a work fl ow could lead to a number of separation/purifi cation techniques or proceed directly to tryptic digestion. Both one-and two-dimensional electrophoresis (1-DE and 2-DE) are used to immunoblot for novel autoantigens followed by analysis, frequently matrix-assisted laser desorption/ionization (MALDI) or surfaceenhanced laser desorption/ionization (SELDI), but also including one-dimensional and two-dimensional liquid chromatography (1D-LC and 2D-LC)-tandem mass spectrometry (MS/MS). Complex samples may be separated by affi nity chromatography, immunoprecipitation or magnetic bead separation. Techniques may be employed to remove primarily high-abundance proteins, separated based on protein chemistry or interaction kinetics involving immunoglobulins or other proteins. The digested peptides could be subjected to analysis by MALDI, SELDI, and one-or twodimensional electrospray ionization (ESI)-LC/MS/MS. Gel separated and digested proteins are frequently analyzed with MALDI or SELDI platforms, but are appropriate for ESI as well. More complex sample types, in particular whole cell lysates, which have not been separated, are frequently subjected to one-or two-dimensional LC peptide separation coupled with ESI-MS/MS analysis. The two-dimensional aspect of LC separation allows a greater number of proteins to be identifi ed from very complex samples when compared to the other methods, with the tradeoff being the greatly increased time (10 to 12 hours) to analyze one sample. SELDI-time-of-fl ight (TOF) and MALDI-TOF are used for profi ling while MALDI-TOF-TOF and ESI-MS/MS can sequence the peptides in the sample to positively identify their parent proteins.   [24,25] Ro52 [19] MALDI-TOF-TOF Antibody protected sites found with SLE sera in portion of the Ro52 protein Both comprising amino acids 200 to 239 Annexin A6 [10] LC-MS/MS Autoantibodies identifi ed against annexin A6 Both Activation of annexin A6 via autoantibody resulting in impaired heart function

Immune cell involvement in SLE
Immunoglobulin J [27] MALDI-TOF Upregulated in SLE 2.46-fold Both Early diff erentiation marker for B cells Calprotectin L1 [27] MALDI-TOF Upregulated in SLE 2.02-fold Both Known marker of disease activity in RA Released from granulocytes and monocytes during their activation Signifi cantly tied to SLEDAI and higher levels of anti-double-stranded DNA antibodies and SLE arthritis May also inhibit immunoglobulin production Glutathione S-transferase MALDI-TOF and Downregulated in SLE 0.15-fold (P = 0.001) in one study and versus diseased Both [18,27] iTRAQ 2D LC-MS/MS and healthy controls Lower GST results in higher ROS levels and increased oxidative stress Polymorphisms with altered enzyme activity of GST have been shown to lead to SLE Apolipoprotein A-IV MALDI-TOF Upregulated in SLE 2.67-fold (P = 0.010) Biomarker precursor [27] Similar results seen in RA [70] and Alzheimer's disease [71] Zinc fi nger protein MALDI-TOF Upregulated in SLE 2.24-fold (P = 0.041) Both subfamily 1A [27] Zinc fi nger protein -iTRAQ 2D LC-MS/MS Upregulated in active SLE versus stable SLE (2.7160), RA (2.3824) and healthy Both isoform 2 protein 549 [18] controls ( and colleagues [27] and Wang and colleagues [18] show the evolution of a mass spectrometry-based inter pretation of peripheral blood mononuclear cells (PBMCs) in SLE. Dai and colleagues published their work in 2008 comparing the relative changes in the proteome of PBMCs of both healthy controls and SLE patients. Th ey employed the conventional approach of two-dimensional electrophoresis and silver stain followed by MALDI-TOF-TOF to identify protein spots expressed at diff erent levels. A total of 20 spots diff ered signifi cantly between the SLE and control group, with 11 being upregulated and 9 downregulated. Th e fi ndings of the study are summarized in Table 1. Briefl y, immunoglobulin J chain, an early diff erentiation marker in B cells, was up regulated, as was calprotectin L1, a protein released by granulocytes and monocytes during activation. Interestingly, calprotectin L1 levels were associated with the severity of disease and at the same time were linked signifi cantly to anti-double-stranded DNA antibodies. Glutathione S-transferase (GST) was downregulated in SLE, which would  reduce the cell's ability to degrade reactive oxygen species, which may contribute to infl ammation and tissue damage.
Wang and colleagues [18] followed Dai two years later with an expanded study that employed the use of isobaric tagging for relative and absolute protein quantifi cation (iTRAQ) as a means of studying PBMC proteome changes. Th e iTRAQ method was used with LC-MS/MS to allow multiplexing of samples (up to eight per run), which not only saves time but also reduces technical variability. Using six samples each for active SLE, inactive SLE, rheumatoid arthritis (RA) as a diseased control and healthy patient samples, 452 proteins with a variety of cellular locations and biological functions were identifi ed and quantifi ed. Of these, 126 were either upregulated or downregulated across all the groups. When compared to the 2008 data, this study showed a similar downregulation of glutathione S-transferase and upregulation of zinc fi nger proteins with similar fold changes in both experiments, but also identifi ed several other known SLE regulators and novel functional candidates. In total, 17 proteins were upregulated and 13 were downregulated in active versus stable SLE. Th e authors' choices for future study are listed in Table 1. Known SLE and infl ammatory proteins were identifi ed, such as resistin and calciumbinding family of S100 proteins. Additionally, several novel candidate proteins were identifi ed, such as brain acid soluble protein 1 and ras-related C3 botulism toxin substrate 2, which have roles in apoptotic mechanisms in other diseases [28,29], but this was their fi rst link to SLE disease activity.

Lupus nephritis
Lupus nephritis (LN) must be considered in a discussion on the use of proteomic methods in lupus, because LN is one of the most common and most severe complications that can arise from SLE (aff ecting approximately 50% of patients), leading to signifi cant mor bidity and mortality [30]. As with other renal disorders, a major diagnostic source for LN is urine. Th e advantage of using urine as a screening method for the presence or severity of disease is the non-invasive nature of sample collection. Urine studies also allow long-term observation of patients, allowing tracking of changes in the urine proteome associated with changing disease states. SLE and LN biomarkers have been a topic of discussion dating back to the 1960s and continue to be of interest today [20,[31][32][33]. Recent proteomic studies have provided in sight into novel urinary biomarkers for LN.
Oates and colleagues [34] built on the ideas of urine protein quantifi cation using two-dimensional electrophoresis to train an artifi cial neural network capable of identifying potentially signifi cant changes to urine proteomes immediately before validation by renal biopsy.
Briefl y, 200 μg of protein were collected from 20 diff erent patients with SLE at the time of biopsy. Th e recovered protein was denatured and separated using two-dimensional electrophoresis. A total of 213 protein spots across the 20 patient gels were used to calculate protein abundance and an artifi cial neural network was trained to identify the class of LN based on the protein abundance profi le. Th e resulting neural network was able to successfully identify >86% of all positives and >92% of all negatives, with the exception of class V, where each false positive was present in a sample also positive for class III or IV LN, which were correctly identifi ed. Th e group then identifi ed the most critical spots for determining sensitivity and subjected each to MALDI-TOF mass spectrometry, which identifi ed six proteins: α-1 acid glycoprotein, α 1 microglobulin (two spots), zinc α-2 glycoprotein (two spots), and IgG κ light chain (Table 1).
Early work successfully identifi ed proteomic profi les of renal fl are in SLE through the use of SELDI-TOF to identify protein signatures without the need to positively identify proteins. Th is allowed the identifi cation of novel biomarkers, but the protein identifi cation limitation of the SELDI-TOF technique prevented mechanistic inference and predictions [35]. Zhang and colleagues [36] provided both diagnostic and mechanistic understanding of LN using longitudinal urine proteomics to assess the changes in the urine proteome before, during and after SLE renal fl are by combining SELDI-TOF with LC-MS/ MS sequencing for protein identifi cation. Using a sample set of 145 urine specimens from SLE class III, IV and V patients, they identifi ed 27 low molecular weight proteins that showed signifi cantly diff erent levels of expression before, during or after SLE fl are. While no single peptide was present in all the fl are or non-fl are samples, combina tions of two peptides were found that successfully diff erentiated 100% of the remission and pre-fl are samples from fl are and post-fl are samples. LC-MS/MS analysis identifi ed multiple isoforms of α-1 antitrypsin, albumin, and hepcidin as proteins of interest (Table 1). Interestingly, hepcidin-20 levels were only altered prefl are, which means it may be a useful indicator of impend ing fl are, allowing modulation of clinical management to reduce severity and duration of the fl are. A follow up study two years later supported this idea and further presented interleukin-6-and interferon-α-dependent upregulation of hepcidin transcription as a possible mechanism of hepcidin increase in LN [37].
A study using LC-MS/MS to characterize immunereactive bands from two-dimensional electrophoresis separated endothelial cell lysate with sera of SLE patients, other autoimmune conditions and healthy controls identi fi ed the glycolytic enzyme aldolase A as a candidate target antigen [20] (Table 1). Moreover, this was found to be more common for LN patients than in SLE without LN (43.4% with LN versus 11.1% without LN). However, reactivity to aldolase A was not appropriately signifi cant between LN and other infl ammatory disease controls such as RA and polymyositis to be used as a stand-alone biomarker for LN. Th e authors did, however, note specifi city of LN patient anti-sera to the region of aldolase A comprising amino acids 94 to 183 over other autoimmune conditions.
Suzuki and colleagues used a combination of SELDI-TOF [38] and MALDI-TOF [39] for evaluation of the urine proteome in pediatric LN. Four potentially significant biomarkers were identifi ed, including transferrin (Tf ), ceruloplasmin, α1-acid-glycoprotein (AGP), and lipocalin-type prostaglandin-D synthetase (L-PGDS), each with great signifi cance (P < 0.005; Table 1). Signifi cant increases of Tf, L-PGDS, and AGP were seen as early as three months before clinical diagnosis or worsening of LN. Both Tf and AGP were also seen in adult forms of the disease [40], while L-PGDS is novel to pediatric LN diagnosis. Ceruloplasmin has been implicated in global SLE, but was not found to be an eff ective metric for defi ning the course of LN in pediatric patients.

Neuropsychiatric systemic lupus erythematosus
Neuropsychiatric SLE (NPSLE) is a potentially severe complication that varies in prevalence between 9.5% and 95% based on the literature and the diagnostic criteria [41]. Findings suggest that NPSLE can be present in a mild form in as many as 95% of SLE patients [42], making it more common but typically less severe than renal or cardiovascular complications. Mood disorders and cogni tive disorders are common symptoms and can be particularly challenging to diagnose in pediatric and adoles cent patients due to the rapid emotional and physical changes that occur at those ages. While acquisition of cerebral spinal fl uid (CSF) is invasive, it has become routine in the diagnosis of many neurological diseases and, in addition to serum, is therefore a useful medium for proteomic study of NPSLE. NPSLE research has focused on both biomarker identifi cation for its diagnostic value as well as autoantigen identifi cation for improved mechanistic understanding of the disease state and potential therapeutic development.
Lu and colleagues [8] found CSF and sera obtained from NPSLE patients to be reactive to rat, mouse and monkey brain homogenate when separated by twodimen sional electrophoresis. MALDI-TOF-TOF was used to identify immune-reactive proteins. A total of 275 serum samples and 177 CSF samples were included, giving a great deal of depth and statistical power to the study. Briefl y, four antinuclear antibody (ANA)-positive serum samples, which did not react to Hep2 cells, were used as pilot samples. Initial analysis of the reactive spots revealed an interaction in three of four samples with α-internexin (INA), a highly conserved protein between rodents and humans (Table 1). Recombi nant human INA was then used in a western blot to confi rm reactivity of sera as well as immunofl uorescence with sera and anti-INA polyclonal antibody. Both the sera and the polyclonal antibody were found to react to cultured rat cortical neurons and the reactivity of sera was reduced by incubation with recombinant INA. Th ese fi ndings opened the door for analysis of the entire cohort of CSF and serum samples. Using two standard devia tions as a cutoff for positivity, 41.7% of the NPSLE samples of CSF were found to be positive for anti-INA antibodies (compared to 0% of SLE with cerebral infarc tion, 7.1% of normal SLE controls, and 10% of other neurological diseases). In total, the data showed reactivity in more than 40% of NPSLE patients in a dose-dependent manner toward INA. Longitudinal studies found that antibody levels were inversely proportional to cognitive level and declined after initiation of therapy and clinical remission. Furthermore, mice immunized with INA plus lipopolysaccharide developed cognitive impairments similar to NPSLE and exhibited hippocampal and cortical neuron apoptosis, suggesting that the presence of anti-INA antibodies in the serum of patients was pathogenic.
Ndhlovu and colleagues [43] identifi ed α-and β-tubulin as target autoantigens for NPSLE using patient sera to probe isolated mitochondria from bovine brain followed by MALDI-TOF analysis. Autoantibodies to β-tubulin have been identifi ed in other CNS targeting autoimmune diseases such as Guillain-Barre syndrome, multiple sclerosis, and Sydenham's chorea, but targeting of αtubulin is potentially specifi c to NPSLE. α-Tubulin autoantibodies were present in 36% of NPSLE patient sera, only 4% of SLE patients without neurological symptoms and none of the multiple sclerosis, epileptic or healthy control samples (Table 1). Possibly due to the small size of the NPSLE population in the study (n = 37), no correlation was seen between specifi c symptoms of NPSLE and the presence of α-tubulin autoantibodies in sera. Larger patient population studies are needed to further explore the involvement of anti-α-tubulin antibodies in NPSLE.
LC-MS/MS incorporates a direct front-end separation of complex samples to greatly increase the sensitivity of the procedure for low abundance protein detection. LC-MS/MS may be superior to MALDI-based techniques when dealing with complex samples. Katsumata and colleagues found this to be true while attempting to uncover novel serum autoantibodies for SLE, in parti cular those involved in CNS syndromes [9]. Whole protein extracts from human neuroblastoma and glioblastoma cell lines were separated using two-dimensional electrophoresis and reacted to patient sera. It was determined that MALDI-TOF sensitivity was too low for eff ective protein identifi cation in this case. To address this concern, anti-sera from SLE patients with and without CNS syndromes was bound to protein Gsepharose beads and incubated with the same cell line extracts. Proteins pulled down with the sepharose beads were analyzed with a LC-MS/MS method using a Q-TOF instrument. Most of the 154 proteins identifi ed by the LC-MS/MS approach were not identifi ed in the twodimensional electrophoresis/MALDI-TOF analysis and 50 were found in the CNS group but not the non-CNS group. After exclusion of previously described autoantigens, 11 recombinant candidates showed reactivity against pooled serum of 28 patients with active CNS SLE by western blot analysis (5 of which were only detected by LC-MS/MS). Th ese 11 candidates were then probed with sera of 106 SLE patients (42 with active CNS syndromes) as well as 100 healthy controls, resulting in a list of 5 confi rmed autoantigens, 3 of which were novel (crystalline αB, esterase D and APEX nuclease 1; Table 1). When compared to other diseased sera (RA, systemic sclerosis, Sjögren's syndrome, and multiple sclerosis), levels of anti-crystalline αB were not found to be signifi cantly diff erent. Anti-APEX nuclease 1 and esterase D, on the other hand, had signifi cantly higher levels in SLE than in these disease controls and between CNS and non-CNS SLE sera. APEX nuclease 1 is the major apurinic/apyrimidinic endonuclease in human cells and is found to be upregulated in tissues under oxidative stress and has a signifi cant eff ect on apoptotic regulation in certain tissues [44]. Th e eff ects of esterase D are less well known.
In a similar study, Kimura and colleagues [45] used a combination two-dimensional electrophoresis and LC-MS/MS with an ESI-LCQ instrument to obtain higher sensitivity for low abundance proteins. Patient sera from a NPSLE patient with white matter hyperintensities on T2-weighted MRI scan was reacted to a two-dimensional electrophoresis rat cerebral lysate blot. In total, nine reactive spots were seen and fi ve were analyzed by LC-MS/MS. Th e two spots of greatest intensity corresponded to 60 kDa heat shock protein (Hsp60) [45], which functions to refold proteins that are misfolded and/or facilitate degradation of proteins that cannot be recovered [46]. Enhanced expression of Hsp60 can induce endothelial cell apoptosis and Hsp60 expression on the surface of endothelial cells has been reported to be directly related to coronary artery disease [47], opening the possibility of a similar relationship in the cerebral vasculature. In a follow-up study of 180 patients with neurological diseases (including 15 with SLE) and 23 healthy controls, Kimura and colleagues showed that the titer levels of anti-Hsp60 antibodies were directly correlated to severity of white matter hyperintensities in the brain, suggesting the presence of anti-Hsp60 antibodies as both a diag nostic tool as well as a mechanistic agent in the develop ment of NPSLE [32]. Although the authors chose to focus on Hsp60, the study also found SLE-specifi c antibodies to α-internexin, β-actin, and glial fi brillary acidic protein (Table 1).

Other uses for proteomics in the study of SLE
Aside from assessing diff erences in disease states, pharma cologic reactions can be measured using proteomic techniques. Kao and colleagues [48] used two-dimensional electrophoresis and MALDI-TOF to investigate the changes to apoptotic-prone macrophages (APMs) from lupus-prone NZB/W-F1 mice before and after treat ment with cystamine, a known inhibitor of transglutaminase 2 activities. A number of the fi ndings indicated that cystamine has an anti-apoptotic aff ect in APMs and suggests a possible mechanism of this activity. Th e protein expression and enzyme activity of caspase 3 were downregulated and anti-apoptotic proteins APIP, PEBP4, and Bcl-2 were upregulated. ERK1/2 showed reduced phosphorylation after treatment in APMs, which also would result in survival of APMs and fewer apoptotic bodies to serve as potential targets for autoantibody produc tion. Previous studies have shown that ERK1/2 phosphorylation is increased with cystamine treatment in endothelial cells [49] and neuronal tissue [50]; therefore, the fi ndings of decreased phosphorylation in APMs may help further explain the development of SLE in specifi c tissues and lead to a more complete model of pathogenesis.
Mass spectrometry can also be useful as a supplemental technique of a study. In work primarily focused on genetics, D'souza and colleagues [51] utilized MALDI-TOF mass spectrometry to support the hypothesis that catalase polymorphisms aff ect the risk and severity of SLE. Oxidative processes resulted in the 4-hydroxy 2-nonenol adducts on membrane catalase of red blood cells. MALDI-TOF was of great use in this situation because it allowed the positive identifi cation of catalase from red blood cell membranes (an uncommon location) while migrating at a higher molecular weight, possibly due to 4-hydroxy 2-nonenol adducts. Ye and colleagues [52] expanded on previous microarray data by selecting a gene that is upregulated in SLE, IFIT1 (interferoninduced with tetratricopeptide repeats 1), creating a IFIT1 fusion protein with GST, and then using bead immobilized GST-IFIT1 as bait to capture interacting proteins in white blood cell lysate. MALDI-TOF was performed on the proteins pulled down in the procedure and found Rho/Rac guanine nucleo tide exchange factor to be a potential target for IFIT1 interaction in SLE. Follow-up studies have supported the role of the interferon pathway in SLE and other autoimmune diseases [53][54][55].

Conclusion
Many of the studies presented in this review were aimed at using proteomic applications to defi ne events in specifi c SLE complications, but it is our view some of these discoveries could also provide insight into other areas of SLE or that approaches applied to similar diseases could be modifi ed for use in SLE. Recent studies have shown the use of MALDI-TOF techniques for synovial fl uid analysis in autoimmune arthritis [56], which could be used to assess changes in SLE arthritis. Kimura and colleagues [45] identifi ed a 60 kDa heat shock protein as an autoantigen in NPSLE, a protein with known ties to coronary artery disease and endothelial cell apoptosis [47,57,58], that may play a role in the development of cardiovascular SLE. Several studies have shown the benefi ts of proteomics using bronchoalveolar lavage fl uid as a sample for the study of sarcoidosis [59], systemic sclerosis [60,61], and other pulmonary diseases [62,63]. Th e use of bronchoalveolar lavage fl uid for the study and diagnosis of pulmonary complications in SLE would be less invasive and safer than biopsy. Tamburro and colleagues [64] recently published a proteomic review of the vitreous humor of the eye. Contrasting these fi ndings to the proteome of SLE patients' vitreous humor could shed more light on the pathogenesis of retino pathy in SLE. Dai and colleagues [27] found immuno globulin J to be upregulated in peripheral blood mono nuclear cells. Other mass spectrometry-based techniques could be used, such as multiple reaction monitoring (MRM) and selected reaction monitoring (SRM) LC-MS/MS, which allows focused study on only m/z ratios of interest, reducing sample processing time and greatly increasing throughput. After the initial biomarker discovery stages, MRM/SRM has been used as a tool in biomarker validation in blood [65] and synovial fl uid, specifi cally in RA patients [66]. Additionally, the speed and potential for multiplexing means MRM/SRM could have clinical utility in patient diagnosis, prognosis, and monitoring [67]. We used a LC-MS/MS approach to identify the M-type phospholipase A 2 receptor (PLA 2 R) as the target antigen in idiopathic membranous nephropathy [68]; antibodies recognizing the PLA 2 R were present in 70 to 80% of these patients. However, they are not present in any patients with membranous lupus nephritis, which develops in up to 40% of patients with LN [69]. Th is suggests that there are unique auto antibodies directed against specifi c kidney proteins other than the PLA 2 R in membranous lupus nephritis; therefore, proteomic studies could be designed to identify such target antigens as well as specifi c epitopes important in this disease.
Even with the strides we have made in the understanding of SLE, it is still a debilitating condition aff ecting hundreds of thousands of patients every year. While treatments are available, they bring with them signifi cant side eff ects and frequent failure rates. Recent advances in mass spectrometry technology have allowed an even more thorough analysis of complex samples to better understand the signifi cance of minute changes in the pathogenesis of SLE, while accommodating large sample sizes through multiplexing and high-throughput methods. Furthermore, the options researchers or clinicians have in mass spectrometry techniques are rapidly expanding and have the ability to be tailored to a specifi c study. It is our hope that this review has aff orded the reader a greater knowledge of the potential uses of mass spectrometry and other proteomic technologies in the study of SLE. While a proteomic understanding of disease is by no means a panacea, it off ers the potential to help advance our knowledge and understanding of SLE to a point that the disease could be less of a burden for those affl icted.