Logo of mcpAbout MCPASBMBMCPContactSubscriptionsSubmissionsThis Article
Mol Cell Proteomics. 2011 May; 10(5): M110.005033.
Published online 2011 Feb 24. doi:  10.1074/mcp.M110.005033
PMCID: PMC3098590

Serum Protein Profiling of Systemic Lupus Erythematosus and Systemic Sclerosis Using Recombinant Antibody Microarrays*An external file that holds a picture, illustration, etc.
Object name is sbox.jpg


Systemic lupus erythematosus (SLE) and systemic sclerosis (SSc) are two severe autoimmune connective tissue diseases. The fundamental knowledge about their etiology is limited and the conditions display complex pathogenesis, multifaceted presentations, and unpredictable courses. Despite significant efforts, the lack of fully validated biomarkers enabling diagnosis, classification, and monitoring of disease activity represents significant unmet clinical needs. In this discovery study, we have for the first time used recombinant antibody microarrays for miniaturized, multiplexed serum protein profiling of SLE and SSc, targeting mainly immunoregulatory proteins. The data showed that several candidate SLE-associated multiplexed serum biomarker signatures were delineated, reflecting disease (diagnosis), disease severity (phenotypic subsets), and disease activity. Selected differentially expressed markers were validated using orthogonal assays and a second, independent patient cohort. Further, biomarker signatures differentiating SLE versus SSc were demonstrated, and the observed differences increased with severity of SLE. In contrast, the data showed that the serum profiles of SSc versus healthy controls were more similar. Hence, we have shown that affinity proteomics could be used to de-convolute crude, nonfractionated serum proteomes, extracting molecular portraits of SLE and SSc, further enhancing our fundamental understanding of these complex autoimmune conditions.

Systemic lupus erythematosus (SLE)1 (1, 2) and systemic sclerosis (SSc), or scleroderma, (3, 4) are two severe, chronic autoimmune connective tissue diseases with still unknown etiology, complex pathogenesis, heterogeneous presentation, and unpredictable course. As a consequence, the difficulties in diagnosing, classifying, and treating both SLE (1, 5, 6) and SSc (3, 4, 7, 8) are significant. Thus, further studies delineating SLE and SSc, and revealing the underlying disease biology at the molecular level are highly warranted.

SLE is a multifaceted disease, with a prevalence of 40 to 200 cases per 100,000 persons (2), for which the lack of specific biomarkers is critical and impairs the clinical management of these patients (6, 912). First, the clinical symptoms vary so much that it often mimics or is mistaken for other conditions (1, 2). Because no single diagnostic test is at hand, SLE is currently diagnosed when at least 4 of 11 complex, clinical criteria, as defined by the American College of Rheumatology (13, 14), are fulfilled. Second, the course of the disease is characterized by alternating periods of flares and remissions (1, 2). There are no biomarkers at hand for predicting and/or identifying the start and end of a flare, which would be a key feature for optimizing treatment (1, 2, 5). Third, the therapeutic regime could be even further optimized if validated biomarkers for stratifying the patients into clinical phenotypic subsets, reflecting disease severity (15), were available. Fourth, the absence of markers has significantly hampered the efforts to monitor and evaluate the effects of (novel) therapeutics (6, 16). Considering the complexity of SLE, it is reasonable to argue that more than one biomarker signature will be required in order to reflect all aspects of SLE (6). Hence, the need to define molecular portraits associated with SLE is significant.

Compared with SLE (inflammatory phenotype) (1, 2), SSc displays a less anti-inflammatory and more fibrotic phenotype (4, 7, 17). This disorder, which has a prevalence of about 3 to 24 cases per million persons (18), is as SLE, diagnosed by evaluating an intricate pattern of clinical features. Based on the pattern of skin involvement (19), SSc is commonly classified into two subsets, limited cutaneous SSc (lcSSc) and diffuse cutaneous SSc (dcSSc). As for SLE, the need for specific biomarkers of SSc for diagnosis, classification, prognosis, and for monitoring the response to therapy is significant (8, 20).

Considering the nature of SLE (1, 2) and SSc (3, 8, 17), deciphering the serum, plasma, and/or urine proteomes, would shed further light on these diseases, and could provide the candidate biomarker signatures much longed for (6, 8, 1012, 20). Despite major efforts, using a plethora of methods, including conventional proteomic technologies, such as two-dimensional gels and mass spectrometry, our knowledge about the serum, plasma, and urine signatures reflecting SLE (6, 1012) and SSc (3, 7, 8, 20) is still very limited, and mainly restricted to single laboratory variables displaying inadequate specificity and sensitivity. Targeting crude proteomes, such as serum, has proven challenging using standard proteomic approaches because of sample complexity and methodological shortcomings (2123).

In recent years, affinity proteomics, mainly represented by antibody-based microarrays, have been established as a technology capable of performing multiplex profiling of complex proteomes in a sensitive manner (2426). In this context, we have developed a state-of-the-art recombinant antibody microarray technology platform (24, 27, 28) and validated its use within disease proteomics (24, 2932). Focusing on various cancers (3033) and inflammatory conditions (31) (Wingren et al., unpublished observations), we have deciphered candidate serum/plasma and tissue biomarker signatures for e.g. disease diagnosis, prognosis, and classification, as well as for monitoring the molecular effects of therapy and for selecting patients eligible for therapy.

In this proof-of-concept study, we have explored the potential of our recombinant antibody microarray set-up for profiling the serum proteome of SLE and SSc, targeting high- and low-abundant immunoregulatory proteins in crude, directly biotinylated sera. The data showed that several SLE-associated candidate serum protein signatures could for the first time be identified reflecting disease, disease severity (phenotypic subsets), and disease activity. Although SLE and SSc could be differentiated, the data implied that the serum profiles of SSc versus controls were more similar. Hence, this study demonstrated that a minimally invasive blood sample harbored disease-specific information reflecting autoimmune connective tissue diseases, further enhancing our fundamental understanding of SLE and SSc.


Clinical Samples

This study was approved by the regional ethics review board in Lund, Sweden. In total, 65 serum samples were collected at Lund University Hospital (Lund, Sweden), including 20 SSc patients (Table I), 30 SLE patients (Table II), and 15 healthy volunteers (mean age 44 (range 20–69), gender female:male 14:1). The SLE and SSc patients were all recruited from the Department of Rheumatology (Lund University Hospital) (34, 35). Serum samples of the SSc patients were collected at the first enrolment for inquiry of the SSc diagnosis, and the demographics are described in Table I. The SSc patients fulfilled the inclusions criteria of (a) a definitive diagnosis of SSc according to the ACR (36); (b) a disease duration of less than 5 years from the onset of skin involvement; and (c) had not been previously treated with any of the following drugs: azathioprine, chlorambucil, colchicine, cyclophosphamide, cyclosporine S, d-penicillamine, methotrexate, or mycophenolate mofetil. The SSc patients were grouped according to whether they had limited cutaneous SSc (lcSSc, n = 10) or diffuse cutaneous SSc (dcSSc, n = 10) (19), and the clinical data were obtained as previously described (37). The SLE patients had clinical SLE diagnosis and displayed four or more ACR classification criteria (13, 14). The serum samples were collected during follow-up when the patients were presented with a flare. The demographics are described in Table II. The SLE patients were grouped according to disease severity as previously described (15): (1) skin and musculoskeletal involvement (SLE1, n = 10); (2) serositis, systemic vasculitis but not kidney involvement (SLE2, n = 10); and (3) presence of SLE glomerulonephritis (SLE3). The clinical disease activity was defined as SLEDAI-2K score (38). All samples were aliquoted and stored at −80 °C until further use. A second, independent patient cohort, containing 20 SLE3 patients and 20 healthy volunteers, collected at Lund University Hospital (Lund, Sweden) (the demographics are described in supplemental Table 1) were used to validate selected differentially expressed analytes, including ≤10 cytokines (low-abundant markers) and one complement protein (high-abundant marker).

Table I
Demographic data of the SSc patients included in the study
Table II
Demographic data of the SLE patients included in the study

Labeling of Serum Samples

The serum samples were labeled using previously optimized labeling protocols for serum proteomes (27, 28, 39). Briefly, the serum samples were diluted 1:45 in phosphate-buffered saline (PBS), resulting in a concentration of about 2 mg/ml, and biotinylated at a molar ratio of biotin:protein of 15:1 using EZ-Link Sulfo-NHS-LC-Biotin (Pierce, Rockford, IL). Unreacted biotin was removed by dialysis against PBS for 72 h, using a 3.5 kDa molecular weight (MW) dialysis membrane (Spectrum Laboratories, Rancho Dominguez, CA). The samples were aliquoted and stored at −20 °C.

Production and Purification of Single-chain Fragment Variable (scFv)

One hundred thirty-five human recombinant single-chain fragment variable (scFv) antibody fragments directed against 60 different analytes mainly involved in immunoregulation, anticipated to reflect the events taking place in SLE and SSc (supplemental Table 1), were selected from the n-CoDeR library (40) and kindly provided by BioInvent International AB (Lund, Sweden). The specificity, affinity (normally in the nm range), and on-chip functionality of these phage display derived scFv antibodies (40) was ensured by using (i) stringent phage-display selection protocols (40), (ii) multiple clones (≤4) per target analyte, and (iii) a molecular design, adapted for microarray applications (24, 41) (Wingren et al., unpublished observations). The specificity of several of the antibodies have previously also been validated using well-characterized, standardized serum samples, and orthogonal methods, such as mass spectrometry, ELISA, MSD, CBA, and MS, as well as using spiking and blocking experiments (supplemental Table 2). All scFv antibodies were produced in 100 ml E. coli cultures and purified from expression supernatants, using affinity chromatography on Ni-NTA agarose (Qiagen, Hilden, Germany). Bound molecules were eluted with 250 mm imidazole, extensively dialyzed against PBS, and stored at 4 °C, until further use. The protein concentration was determined by measuring the absorbance at 280 nm (average concentration 480 μg/ml, range 100 to 800 μg/ml). The degree of purity and integrity of the scFv antibodies were evaluated by 10% SDS-PAGE (Invitrogen, Carlsbad, CA).

Fabrication and Processing of Antibody Microarrays

For production of the antibody microarrays, we used a set-up previously optimized and validated (27, 28, 30, 32). Briefly, the scFv microarrays were fabricated, using a noncontact printer (Biochip Arrayer, Perkin Elmer Life and Analytical Sciences). The antibodies were spotted onto black polymer MaxiSorb microarray slides (NUNC A/S, Roskilde, Denmark), resulting in an average of 11 fmol scFv per spot (range 2–19 fmol). Eight replicates of each scFv clone were arrayed to ensure adequate statistics. The on-chip performance of the antibodies has been previously validated (27, 3032, 41), for review see (4144). In total, 160 antibodies and controls were printed per slide orientated in two columns with 8 × 80 spots per column.

For handling of the arrays, we used a protocol recently optimized (33). Briefly, the slides were manually blocked in 5% (w/v) fat-free milk powder (Semper AB, Sundbyberg, Sweden) in PBS, and then placed in a Protein Array Work station (Perkin Ellmer Life and Analytical Sciences) for automated handling. The slides were washed with 0.5% (v/v) Tween-20 in PBS. The biotinylated serum sample was diluted 1:2 (resulting in a total serum dilution of 1:90) in 1% (w/v) fat-free milk powder and 1% (v/v) Tween in PBS (PBS-MT) prior to incubation on the array. The arrays were visualized with 1 μg/ml Alexa-647 conjugated streptavidin diluted in PBS-MT. Finally, the arrays were dried under a stream of nitrogen gas and scanned with a confocal microarray scanner (ScanArray Express, Perkin Elmer Life and Analytical Sciences) at 5-μm resolution, using three different scanner settings. The ScanArray Express software V3.0 (Perkin Elmer Life and Analytical Sciences) was used to quantify the intensity of each spot, using the fixed circle method. The local background was subtracted to compensate for possible local defects, the two highest and the two lowest replicates were automatically excluded and each data point represents the mean value of the remaining four replicates. The coefficient of correlation for intra-assays was >0.99 and for interassays >0.96, respectively.

Data Normalization

Only nonsaturated spots were used for further analysis of the data. Chip-to-chip normalization of the data sets was performed, using a semiglobal normalization approach (24, 30, 32), conceptually similar to the normalization developed for DNA microarrays. Thus, the coefficient of variation was first calculated for each analyte and ranked. Fifteen percent of the analytes that displayed the lowest coefficient of variation values over all samples were identified, corresponding to 21 analytes, and used to calculate a chip-to-chip normalization factor. The normalization factor Ni was calculated by the formula Ni = Si/μ, where Si is the sum of the signal intensities for the 21 analytes for each sample and μ is the sum of the signal intensities for the 21 analytes averaged over all samples. Each data set generated from one sample was divided with the normalization factor Ni. For the intensities, log2 values were used in the analysis.

Data Analysis

The support vector machine (SVM) is a supervised learning method in R (4547) that we used to classify the samples. The supervised classification was performed using a linear kernel, and the cost of constraints was set to 1, which is the default value in the R function SVM, and no attempt was performed to tune it. This absence of parameter tuning was chosen to avoid overfitting. The SVM was trained using a leave-one-out cross-validation procedure. Briefly, the training sets were generated in an iterative process in which the samples were excluded one by one. The SVM was then asked to blindly classify the left out samples as either disease or healthy, and to assign a SVM decision value, which is the signed distance to the hyperplane. No filtration on the data was done before training the SVM, i.e. all antibodies used on the microarray were included in the analysis. Further, a receiver operating characteristics (ROC) curve, as constructed using the SVM decision values and the area under the curve (AUC), was calculated. AUC values were in general interpreted as 0.5–0.6 = poor; 0.6–0.7 = fair; 0.7–0.8 = medium; 0.8–0.9 = good; 0.9–1.0 = excellent. The discriminatory power of clinical laboratory parameters (anti-DNA antibodies, anti-nuclear antibodies (ANA), complement protein C1q, complement protein C3, complement protein C4, C-reactive protein (CRP), and/or SLEDAI-2K) were evaluated single predictors or in combination by using the same leave-one-out SVM procedure as for the array data.

Significantly up- or down-regulated plasma proteins (p < 0.05) were defined based on the relative protein levels and identified using Wilcoxon signed-rank test, log transformed and mean centered. It should be noted that the approach did not differentiate whether the observed up- or down-regulated levels of an analyte was because of an increased or decreased production or an increased or decreased consumption. The samples were visualized using Qlucore (Lund, Sweden) or R (heat maps).

Finally, we investigated the impact of the small sample groups at the end results of the statistical analysis. To this end, we randomly changed the group labels across the samples, and re-ran the classification using the randomized (permutated) groups. The randomization process was repeated 20 times for each classification. The analysis showed that the even though the sample groups were small, the classifier performed among the best when the groups were made up of their original, correct members (when the original group displayed an adequate ROC AUC value) (supplemental Fig. 1A). Thus, the data showed that the successful classification adequately reflected the original stratification, and was not the results of a confounding factor derived from the small sample groups. In the cases where SVM models also were built using data subsets composed of only significant analytes, a corresponding analysis was performed (supplemental Fig. 1B). Here, each permutated model was constructed using only the analytes found to be significant for each individual permutation, i.e. the analytes selected for training and testing each permutation was identified using the same criterion (wilcoxon p value < 0.05) as for the original stratification.

Validation of Array Data

A human Th1/Th2 10-plex MSD (Meso Scale Discovery, Gaithersburg, MD) assay was run in an attempt to validate the antibody microarray results (differentially expressed analytes), focusing on SLE. Both patient cohorts of SLE were profiled using MSD (supplemental Figs. 2–6). Each well of the MSD 96-plate had been prefunctionalized with antibodies against IFN-γ, IL-1β, IL-2, IL-4, IL-5, IL-8, IL-10, IL-12p70, IL-13, and TNF-α in spatially distinct electrode spots. The assay was run according to the protocol provided by the manufacturer and the electrochemiluminiscence-based readout was performed in an MSD SECTOR® instrument. The limit of detection was defined as 2.5 times the standard deviation of the zero point in the standard curve. The MSD assay was applied to both patient cohorts.

Further, the levels of three complement proteins, including C1q, C3, and C4, were determined by an electroimmuno-, turbidometric-, and/or nephelometric-based assay in the first patient cohort (supplemental Figs. 2 and 6). Finally, the levels of complement protein C3 was determined in both patient cohorts using our array set-up (supplemental Figs. 3 and 5).


Profiling of SLE and SSc Serum Proteomes

First, we determined a focused serum proteome profiles of SLE and SSc, and compared them to that of the healthy controls. The top ≤25 significantly (p < 0.05) expressed nonredundant analytes are shown as heat maps in Fig. 1. Although only four analytes (e.g. IFN-γ, and IL-4) were found to be differentially expressed in SSc (Fig. 1A), the data showed that the expression patterns of 40 analytes differed between SLE versus controls, including both down-regulated (e.g. C1q, C3, and C5) and up-regulated analytes (e.g. IL-6, IL-10, IFN-γ, and TNF-α) (Fig. 1B). It should be noted that we could not differentiate whether the observed up- and down-regulated levels of an analyte was because of an increased or decreased production or increased or decreased consumption.

Fig. 1.
Expression profiling of serum proteomes from SSc and SLE using a 135-recombinant-antibody microarray. The samples were classified using a leave-one-out cross validation approach with a SVM, and illustrated as ROC curves. The patients are ordered by decreasing ...

In order to evaluate the ability of the set-up to differentiate SSc and SLE versus controls based on the observed protein expression profiles, we ran a leave-one-out cross-validation based on all antibodies with a SVM, and collected the decision values for all samples. These prediction values were then used to construct a ROC curve, and the AUC value was calculated (Fig. 1). The results showed that SLE versus controls could be reasonably differentiated (AUC = 0.76) (Fig. 1B), with a specificity and sensitivity of 0.80% and 0.67%, respectively. Using orthogonal assays, targeting low-abundant markers (≤10 cytokines; MSD-based assay) (supplemental Fig. 2A) and high-abundant markers (≤3 complement proteins; electroimmuno-, turbidometric-, and/or nephelometric-based assays) (supplemental Fig. 2B), differentially expressed analytes were successfully validated. In contrast, SSc versus controls could not be distinguished (AUC = 0.47) (Fig. 1A).

Next, we examined whether a multiplexed serum profile differentiating SLE versus SSc could be determined (Fig. 1C). The data showed that 42 analytes were significantly differentially expressed of which all, but 3 (C1q, C3, and mucin-1), were up-regulated in SLE (e.g. IL-6, IL-12, IFN-γ, and TNF-α). Although the discriminatory power was poor (AUC = 0.59) using unfiltered data, i.e. all antibodies, the AUC value was found to be 0.69 when only the differentially expressed analytes were used.

Serum Protein Profiles of SSc Phenotypic Subsets

In an attempt to further refine the serological portrait of SSc, we determined the expression profiles of the two commonly recognized phenotypic subsets, dcSSc and lcSSc (Fig. 2). The data showed that seven significantly differentially expressed analytes were observed for dcSSc versus controls (the most prominent being IFN-γ and IL-4), five for lcSSc versus controls (e.g. IL-4, MCP-4, and CD-40), whereas none could be observed for dcSSc versus lcSSc. Using unfiltered data, the phenotypic subsets could neither be differentiated from the controls nor from each other (AUC = 0.50). In contrast, based on the differentially expressed analytes only, both dcSSc (AUC = 0.71) and lcSSc (AUC = 0.90) could be distinguished from the controls.

Fig. 2.
Expression profiling of serum proteomes from clinical subsets of SSc, dcSSc, and lcSSc, using a 135-recombinant-antibody microarray. A, Classification of dcSSc and lcSSc versus healthy controls based on all 135 antibodies, using a SVM-based leave-one-out ...

Classification of SLE

We then determined the serum protein profile of three phenotypic subsets of SLE grouped according to increased disease severity, SLE1 < SLE2 < SLE3, and compared them with that of the controls (Fig. 3). The results showed that the number of differentially expressed analytes (15 < 28 < 44) and AUC values (0.55 < 0.67 < 0.99) increased with the severity of the symptoms.

Fig. 3.
Expression profiling of phenotypic subsets of SLE reflecting increased disease severity, including SLE1 (least symptoms), SLE2, and SLE3 (most severe symptoms), using a 135-recombinant-antibody microarray. The samples were classified using a leave-one-out ...

In more detail, in the case of SLE1, complement proteins were found to be down-regulated (i.e. found at lower levels) (e.g. C1q, C3, C5, and Factor B), whereas several cytokines were found to be up-regulated (e.g. IL-4, IL-6, IL-10, TNF-α) (Fig. 3A and 3D). Despite the observed differences in serum protein expression patterns, SLE1 versus controls could not be discriminated (AUC = 0.55). Although complement proteins were down-regulated in SLE2 as well (e.g. C1q), the pattern of up-regulated cytokines contained a mixture of both Th1 cytokines (e.g. IL-2, IL-12, IFN-γ, and TNF-α) and Th2 cytokines (e.g. IL-6, IL-10) (Figs. 3B and and33D). The discriminatory power of this serum biomarker profile was moderate (AUC = 0.67), corresponding to a specificity of 60% and sensitivity of 80%. In contrast, SLE3 versus controls could readily be discriminated (p = 6×10−7; AUC = 0.99) with a sensitivity and specificity of 100 and 93%, respectively (Figs. 3C). Of note, in the training of the SVM, no filtration of data was made, i.e. all the measurements were included. Apart from the complement proteins (e.g. C1q, C3, and C5) and IL-18 (down-regulated), the pattern of 44 differentially expressed analytes contained mainly up-regulated cytokines, including both Th1 and Th2 cytokines (Figs. 3C and and33D). For all three comparisons, a set of differentially expressed markers (cytokines) were successfully validated using a 10-plexed MSD-based assay (supplemental Figs. 3A–3C). Furthermore, the differential expression pattern observed for both a high-abundant marker (complement protein C3) and a set of low-abundant markers (cytokines) were also validated on an independent, second patient cohort (SLE3 versus healthy controls) using both array technology and a MSD-based assay (supplemental Figs. 3D and 3E). Hence, the data showed that first candidate SLE serological biomarker signatures reflecting disease severity had been delineated using recombinant antibody microarrays.

Refined Classification of SLE versus SSc

Considering the fact that the phenotypes of SLE were reflected in the serum profiles, we compared the molecular pattern of the phenotypic subsets of SLE with that of SSc to refine the observed serological differences between these conditions (Fig. 4). As the subsets of SSc could not be differentiated (Fig. 2), SSc was kept as one cohort. The data showed that the number of significantly differentially expressed analytes (3 < 36 < 49) and AUC values (0.40 < 0.69 < 0.98) for SLE phenotypic subsets versus SSc increased with the severity of SLE (Fig. 4).

Fig. 4.
Expression profiling of phenotypic subsets of SLE reflecting increased disease severity and SSc, using a 135-recombinant-antibody microarray. The samples were classified using a leave-one-out cross validation approach with a SVM, and illustrated as ROC ...

In the case of SLE versus SSc, only three differentially expressed analytes (C1q, C3 and TNF-α) were detected, and the cohorts could not be differentiated (AUC = 0.40) (Figs. 4A and and44D). In contrast, SLE2 versus SSc could be differentiated (AUC = 0.69) with a 75% specificity and 60% sensitivity, and 36 differentially expressed analytes were observed (Figs. 4B and and44D). Except for mucin-1 and Ku 70/80 (down-regulated), a pattern of mainly up-regulated analytes was observed for SLE2, including both Th1 and Th2 cytokines. Further, SLE3 versus SSc was readily classified (AUC = 0.98; p = 6×10−7), displaying a sensitivity and specificity of 100 and 85%, respectively (Fig. 4C). Although complement proteins (e.g. C1q, C3, and C5) and TNF-β were found to be down-regulated (i.e. found at lower levels), a majority of the 49 significantly differentially expressed analytes were found to be up-regulated in SLE3 and composed of a panel of both Th1 and Th2 cytokines (Figs. 4C and and44D). Hence, the serological portraits differentiating SLE and SSc had been further deciphered and refined (cfs. Figs. 1 and and44).

Refined Classification of SLE Reflecting Disease Severity

Next, we further investigated whether the SLE samples could be classified according to disease severity, determined á priori based on an intricate pattern of clinical parameters (Fig. 5). To this end, we compared the serum protein profiles of the three SLE phenotypic subsets with each other. The results showed that SLE1 (least symptoms) versus SLE3 (most symptoms) could be distinguished (AUC = 0.80), corresponding to a specificity and sensitivity of 70% and 80%, respectively (Fig. 5A). Moreover, SLE1 versus SLE2 could also be separated (AUC = 0.73), whereas SLE2 versus SLE3 (AUC = 0.58) appeared to be more similar. Notably, these comparisons were all based on unfiltered data. When only the differentially expressed analytes were used, SLE1 versus SLE2 and SLE2 versus SLE3 could also be classified, displaying AUC values of 0.79 and 0.81, respectively. The relationship between the three clinical phenotypes is further highlighted in Fig. 5B, where a principle component analysis, based on the 13 most significantly differentially expressed analytes is shown. Again, the data outlined that the three cohorts could be classified according to disease severity (phenotype), indicating that serological profiles had been delineated that could be used for phenotype classification. In addition, the array data was also found to display a better classification than single or combined conventional clinical parameters, including, ANA, anti-DNA, SLEDAI-2k, C1q, C3, C4, and CRP, in a majority of the cases (cfs. Fig. 5 and Table III).

Fig. 5.
Expression profiling of phenotypic subsets of SLE reflecting increased disease severity, using a 135-recombinant-antibody microarray. A, Classification of SLE1, SLE2 and SLE3 based on all 135 antibodies, using a SVM-based leave-one-out cross validation ...
Table III
The discriminatory power of clinical laboratory parameters for classification and prognosis of SLE evaluated as single predictors or in combination by using a leave-one-out cross-validation SVM procedure as for the antibody array data

In more detail, 23 significantly differentially expressed analytes were observed for SLE1 versus SLE2 (Fig. 5C). Apart from five up-regulated analytes (e.g. IL-4 and TNF-β), the markers were down-regulated (e.g. IL-10 and IL-12) in SLE1. Although only seven differentially expressed (down-regulated) analytes (e.g. IFN-γ, IL-6, IL-10, and IL-13) were detected for SLE2 versus SLE3, a total of 32 analytes were detected for SLE1 versus SLE3. In the latter case, all but two analytes (IL-3 and IL-1β), were found to be down-regulated in SLE1. Notably, the Th1 cytokines IL-12, IFN-γ, and IL-2 increased as the disease progressed from SLE1 through SLE2 to SLE3, whereas central Th2 cytokines either decreased (IL-4) or showed moderate increase (IL-5 and IL-10) (Fig. 5D), indicating a potentially Th1 skewed immune response as the severity of SLE progressed. Finally, a set differentially expressed markers (cytokines) were successfully validated for all three comparisons using a 10-plexed MSD-based assay (supplemental Fig. 4), further highlighting the findings.

Patients Can be Classified According to Disease Activity

Finally, we investigated whether serological biomarker signatures reflecting SLE disease activity could de-convoluted. To this end, the SLE samples were grouped into three groups based solely on their disease activity, as reflected by their SLEDAI-2K values; low (36), mid (919) and high (2234).

The antibody microarray data showed that low versus mid SLE activity samples could not be distinguish using unfiltered data (AUC = 0.53), whereas the AUC value was found to reach 0.82 when only the 12 differentially expressed analytes were used (Fig. 6A). Except for IL-18, only down-regulated analytes (e.g. IL-8, IL-10, IL-12, and IFN-γ) were observed in low versus mid SLE activity (Fig. 6B). A set of differentially expressed markers (cytokines) were validated using an MSD-based assay (supplemental Fig. 5A). In contrast, mid versus high SLE activity could be differentiated with very high confidence based on nonfiltered data, as reflected by an AUC of 0.93, corresponding to a specificity and sensitivity of 91% and 100%, respectively (Fig. 6A). Among the eight differentially expressed analytes, five were up-regulated (e.g. C1q, C4, IL-8, and IL-12) and three down-regulated (IL-1-ra, IL-18, and TNF-α) in mid versus high SLE activity (Fig. 6B). A set of differentially expressed analytes (cytokines) were successfully validated using a MSD-based assay (supplemental Fig. 5B). Moreover, low versus high SLE activity could be moderately differentiated (AUC = 0.70) with a specificity of 82% and sensitivity of 67%, showing a panel of 10 differentially expressed analytes, including five up-regulated complement proteins (C1q, C3, C4, C5 and Factor B) and five down-regulated analytes (e.g. CD40, IL-6, and IFN-γ) (Figs. 6A and and66B). A set of differentially expressed markers (cytokines) were again validated using a MSD assay (supplemental Fig. 5C). Focusing on the complement proteins, the trends showed that the levels of complement proteins decreased as the activity of the disease increased, as reflected by C1q, C3, C4, and C5 (Figs. 6C and and66D). In addition, the pattern of differentially expressed C1q, C3, and C4 was successfully validated using electroimmuno-, turbidometric-, and/or nephelometric-based assays. Hence, the data indicated that the first generation of multiplexed serum profiles reflecting SLE disease activity could be identified. Of note, the array data was also found to display a better classification than single or combined conventional clinical parameters, including, ANA, anti-DNA, SLEDAI-2k, C1q, C3, C4, and CRP, in a majority of the cases (cfs. Fig. 6 and Table III).

Fig. 6.
Expression profiling of subsets of SLE reflecting disease activity, using a 135 recombinant antibody microarray. A, Classification of SLE, grouped based solely on disease activity (SLEDAI values); low (3 to 6), mid (9 to 19) and high (22 to 34), based ...

In order to eliminate any influence of the disease severity on the observed serum profiles (Fig. 5), we re-ran the analysis above focusing on the SLE3 samples only (Fig. 7). The SLE3 samples were then divided into two groups, based on their SLEDAI-2K values; low (mean 13, range 10–16) and high (mean 24, range 17–32). The data showed that the SLE3 samples could be correctly classified according to disease activity, as reflected by an AUC value of 1.00, corresponding to a 100% specificity and 100% sensitivity (Fig. 7). The 20 significantly differentially expressed analytes are shown as a heatmap. Except for IL-4 and IL-1β, a pattern of mainly up-regulated analytes (e.g. CD40, IL-2, and IL-10) was observed as the disease activity increased. Notably, a set of differentially expressed analytes (cytokines) were successfully validated using a 10-plexed MSD-based assay (supplemental Fig. 6A). Furthermore, the expression pattern of selected differentially expressed markers (cytokines) was also validated on an independent, second patient cohort using an MSD-based assay (supplemental Fig. 6B). Noteworthy, the array data was also found to display a better classification than single or combined conventional clinical parameters in all cases (cfs. Fig. 7 and Table III).

Fig. 7.
Expression profiling of subsets of SLE3 reflecting disease activity, using a 135 recombinant antibody microarray. Classification of SLE3 subsets, grouped based on disease activity (SLEDAI values); SLE3 low (blue) (mean 13, range 10–16), SLE high ...

In conclusion, we have shown that protein expression profiling of serum samples could be performed using recombinant antibody microarrays, enabling us to extract several unique biomarker signatures reflecting SLE and SSc and clinical subsets thereof.


Based on the notion that immunoregulation is a particular phenomenon in SLE and SSc, we have in this pilot study taken on an affinity proteomics approach in order to harness the diagnostic, prognostic, and classification power of the immune system to address these conditions. To this end, we designed our antibody microarray to target predominantly some of the key regulatory serum molecules. With this limited range of specificities (n = 60), we showed that affinity proteomics could still be used to extract disease-associated serum portraits for SSc and in particular SLE reflecting disease, disease severity, disease activity, and phenotypic subsets. In the case of SLE, selected differentially expressed markers were validated using orthogonal assays and a second, independent patient cohort. Although the data needs to be further validated in follow-up studies, targeting larger, independent cohorts of patient samples before it could be transferred into clinical practice (29), the results clearly outlined the biomarker signatures.

SSc is a multisystem fibrotic disorder often described with a Th2 skewed, anti-inflammatory immune response, although the Th1/Th2 balance in SSc is still a matter of controversy (4, 48). Considering the fibrotic pathology of SSc, it would be natural to expect increased levels of IL-4 (Th2 cytokine) and decreased expression of IFN-γ (48). In accordance, we found IL-4 to be up-regulated and IFN-γ to be down-regulated in SSc versus controls. Although this feature was prominent also for the clinical subset dcSSc, only IL-4 was up-regulated in lcSSc. To some extent this might reflect the fact that dcSSc is considered to be dominated by rapidly progressive fibrosis of the skin, lungs, and other internal organs, whereas lcSSc is dominated by vascular manifestations, and skin and organ fibrosis is generally limited and slow to progress (4).

We found very few serological markers to be differentially expressed in SSc versus healthy controls, indicating on small differences in this particular data set. As far as we know, the association of mucin-1 and MCP-4 with SSc is a novel finding. In previous studies, serological biomarkers, exemplified by auto-antibodies, not targeted here, have been found to be an important marker of SSc (8). Other circulating serum biomarkers that have recently been indicated to be central and/or potentially differentially expressed in SSc include e.g. TGF-β, IL-1α, IL-4, IL-6, IL-10, and IL-13 (7, 8, 4951). Apart from IL-4, we were unable to replicate these previous reports. The reason for these discrepancies could include significant biological heterogeneity of SSc (3, 20, 49), the size of the patient cohorts, and/or inherent difficulties in reproducing exploratory findings generated at different sites using different proteomic approaches (29, 5254). This clearly implicated the importance of launching large, multicenter studies using parallel complementary technologies for deciphering complex, heterogeneous diseases (8, 29, 5254).

Diagnosing SLE and SSc is challenging, which could result in patients suffering unexplained symptoms and untreated disease (13, 6, 8, 20). The situation is further complicated by patients suffering from overlap syndromes in which the autoimmune disease could start out as SLE and then transforms into SSc, and vice versa (55, 56). In this proof-of-concept study, distinct features and candidate serum signatures differentiating SLE and SSc were delineated. Notably, we showed that the magnitude of the observed differences coincided with the severity of SLE. Although only three analytes were found to be differentially expressed in SLE1 (least symptoms) versus SSc, the up-regulation of TNF-α and to some extent, the decreased levels of complement proteins in SLE1 agreed well with previous reports (4, 6, 8, 12, 15, 20). As the severity of SLE increased, several other analytes that were anticipated to be over expressed in SLE (6, 12) were also detected, e.g. IL-2, IL-12, and IFN-γ. However, some analytes anticipated to be more associated with SSc than SLE (3, 4, 6, 12, 20), were also found to be up-regulated in SLE versus SSc, e.g. IL-4, IL-5, IL-10, and TGF-β, indicating that the view of describing SLE versus SSc simply as an inflammatory phenotype versus a fibrotic phenotype needs to be refined. In general, our data indicated a much more vigorous and broader (systemic) immune response in SLE than in SSc.

Focusing on SLE, the absence of serological biomarkers for diagnosis is significant (6). Panel(s) of auto-antibodies, erythrocyte-bound complement protein product C4d, complement receptor 1, and platelet-bound-C4d have been proposed, but their potential remains to be validated (6). Notably, we found 40 differentially expressed, nonredundant, circulating serum biomarkers, reflecting the disease, again indicating the potential of our approach. This list of variables was composed of novel markers, as well as some markers that had previously been reported to be associated with SLE, such as IL-4, IL-6, IL-10, IL-12, IFN-γ, C3 and C4 (6, 12, 57, 58), further supporting our observations. It should, however, be noted that these known variables had not previously been reported in the context of a candidate signature for SLE diagnosis, but mainly as single or a few combined markers potentially reflecting disease activity (6, 12, 57, 58). Further studies will be required to validate our candidate signature(s) and to extract and interpret the disease biology reflected by these serological predictors.

Subsequently, these findings were extended, and we report here a candidate serum-based signature for classifying SLE according to disease severity (15). When we compared the serum portraits of the phenotypic subsets with those of the controls, the data showed, as could be expected, that the number of differentially expressed analytes and the discriminatory power of the candidate signature increased with the severity of SLE. As for example, our results implied that SLE3, lupus nephritis, versus controls could be differentiated with a sensitivity and specificity of 100% and 93%, respectively. Previous attempts to pin-point lupus nephritis has mainly relied on single or a few markers, although attempts have been made to identify multiplex signatures using e.g. mass spectrometry based approaches, but so far with limited success (6, 10, 11, 59). Further, our data implied a somewhat skewed Th2 response in SLE1, as IL-4, IL-6, and IL-10 were found to be up-regulated, although a mixture of Th1 and Th2 cytokines were up-regulated in SLE2 and SLE3, indicating a complex pattern of cytokine expression. This could, at least to some extent, reflect the complex and intricate disease processes known to take place as the disease progresses, as mirrored by its unpredictable course and heterogeneous presentation (13, 7).

A refined view emerged when the molecular fingerprints of the SLE phenotypic subsets were compared with each other. Then, the data implied a potentially Th1 skewed immune response as the severity of SLE progressed, which is more in line with the common view of a strong inflammatory component in SLE (13, 7). The impact of being able to classify SLE into phenotypes reflecting disease severity based on a minimally invasive predictor will be central, because this is vital for therapy selection (1, 5, 11). Currently, no blood-based tests exist for such classification, further indicating the potential of using affinity proteomics for extracting key biological information from complex proteomes. Furthermore, the potentially clinically added value was illustrated by the fact that the candidate biomarker signatures provided better classification than single/combined conventional clinical laboratory variables, e.g. ANA, anti-DNA, SLEDAI-2K, C1q, C3, C4, and CRP, in a majority of the cases.

Alternating periods of flare and remission is a typical feature of SLE (13). The active phase of the disease often has serious consequences, like tissue and organ damages, and means to predict and monitor a flare would be instrumental for improving patient management (13, 11). From a clinical perspective, monitoring the activity of renal involvement in SLE is today mainly dependent on microscopic evaluation of urine, which has been shown to be associated with large methodological shortcomings (10). Other clinical laboratory variables that have been evaluated are e.g. auto-antibodies, C1q, C3, and C4, although the performance of these single predictors for monitoring disease activity has varied (6062). Despite the fact that major efforts have been made to reveal SLE-activity associated markers, ranging from traditional efforts, targeting single analytes, e.g. CD40L, to proteomic approaches (57, 58, 60, 6365), the need for additional information and a refined scenario is significant (6). Here, we have outlined candidate serological biomarker signatures for staging SLE patients based on disease activity that provided better staging than single/combined conventional clinical laboratory variables in a majority of the cases. This identified list of variables was composed of novel markers and some markers that had previously been reported to be indirectly (gene-level) or directly (protein-level) associated with disease activity, such as C3, C4, IL-6, IL-8, IL-10, IL-12, and RANTES (6, 57, 58, 66). In particular interferon-inducible chemokines, of which some were detected here, such as RANTES, IL-6 and IL-8, have frequently been indicated as central players to be associated with disease activity (57, 58, 66, 67). Notably, the first pre-validated three-plexed signature, based on the serum levels of IFN-γ-inducible 10-kDa protein, monocyte chemotactic protein 1, and macrophage inflammatory protein 3β, was recently published (58), outlining the potential of delineating multiplexed serum predictors for staging SLE disease activity.

In order to eliminate the influence of disease severity on our ability to reflect disease activity, we focused on lupus nephritis, the most severe form of SLE, and split this group into 2 cohorts based on disease activity; SLE3-high versus SLE3-low. The data showed that we could differentiate the SLE3 samples based on disease activity with high confidence, and the array data was found to provide a better differentiation than single/combined conventional clinical variables in a majority of cases. Again, a mixture of novel and previously reported candidate markers (6, 57, 58, 66) were delineated, further outlining our capability of reflecting disease severity. As compared with when all SLE samples were used, including patients with a broad range of disease severity, the candidate serological biomarker signatures were refined, although several variables were retained, including interferon-inducible chemokines, such as IL-8 and RANTES.

In this pilot study, we have shown that a simple blood sample harbored significant disease-specific information reflecting complex autoimmune connective tissue diseases that could be extracted and interpreted using affinity proteomics. The results could provide new perspective on the nature of SLE, and have outlined several candidate serological biomarker signatures reflecting disease, disease severity and disease activity. The potential of these disease predictors will be validated in larger prospective studies, where also extended array layouts targeting an even wider range of prospective markers will be explored, but suggest novel means for improved clinical management of patients.


* This study was supported by grants from the Alfred Österlund Foundation, Greta and Johan Kock Foundation, Foundation of Strategic Research (Strategic Center for Translational Cancer Research - CREATE Health (www.createhealth.lth.se), the Swedish Research council (VR-M), the Medical Faculty at Lund University, King Gustaf V′s 80th birthday foundation, Skanes University Hospital, the foundation of the National Board of Health and Well-fare, Crafoord Research foundation, the Swedish Rheumatism Association, and the Swedish Heart and Lung Foundation.

An external file that holds a picture, illustration, etc.
Object name is sbox.jpg This article contains supplemental Figs. S1 to S6 and Tables S1 and S2.

1 The abbreviations used are:

systemic lupus erythematosus
anti-nuclear antibodies
area under the curve
C-reactive protein
diffuse cutaneous SSc
limited cutaneous SSc
receiver operating characteristics
single-chain fragment variable
SLE disease activity index
systemic sclerosis
support vector machine.


1. D'Cruz D. P., Khamashta M. A., Hughes G. R. (2007) Systemic lupus erythematosus. Lancet 369, 587–596 [PubMed]
2. Rahman A., Isenberg D. A. (2008) Systemic lupus erythematosus. New Engl. J. Med. 358, 929–939 [PubMed]
3. Varga J. (2008) Systemic sclerosis: an update. Bull. NYU Hosp. Joint Diseases 66, 198–202 [PubMed]
4. Varga J., Abraham D. (2007) Systemic sclerosis: a prototypic multisystem fibrotic disorder. J. Clin. Invest. 117, 557–567 [PMC free article] [PubMed]
5. Eisenberg R. (2009) Why can't we find a new treatment for SLE?. J. Autoimmun. 32, 223–230 [PMC free article] [PubMed]
6. Liu C. C., Ahearn J. M. (2009) The search for lupus biomarkers. Best Practice Res. 23, 507–523 [PMC free article] [PubMed]
7. Abraham D. J., Krieg T., Distler J., Distler O. (2009) Overview of pathogenesis of systemic sclerosis. Rheumatology 48 Suppl 3, iii3–7 [PubMed]
8. Matucci-Cerinic M., Steen V., Nash P., Hachulla E. (2009) The complexity of managing systemic sclerosis: screening and diagnosis. Rheumatology 48 Suppl 3, iii8–13 [PubMed]
9. Liu C. C., Manzi S., Ahearn J. M. (2005) Biomarkers for systemic lupus erythematosus: a review and perspective. Curr. Opin. Rheumatol. 17, 543–549 [PubMed]
10. Rovin B. H., Birmingham D. J., Nagaraja H. N., Yu C. Y., Hebert L. A. (2007) Biomarker discovery in human SLE nephritis. Bull. NYU Hosp. Joint Diseases 65, 187–193 [PubMed]
11. Rovin B. H., Zhang X. (2009) Biomarkers for lupus nephritis: the quest continues. Clin. J. Am. Soc. Nephrol. 4, 1858–1865 [PubMed]
12. Suh C. H., Kim H. A. (2008) Cytokines and their receptors as biomarkers of systemic lupus erythematosus. Expert Rev. Mol. Diagnostics 8, 189–198 [PubMed]
13. Hochberg M. C. (1997) Updating the American College of Rheumatology revised criteria for the classification of systemic lupus erythematosus. Arthritis Rheumatism 40, 1725. [PubMed]
14. Tan E. M., Cohen A. S., Fries J. F., Masi A. T., McShane D. J., Rothfield N. F., Schaller J. G., Talal N., Winchester R. J. (1982) The 1982 revised criteria for the classification of systemic lupus erythematosus. Arthritis Rheumatism 25, 1271–1277 [PubMed]
15. Sturfelt G., Sjöholm A. G. (1984) Complement components, complement activation, and acute phase response in systemic lupus erythematosus. Int. Arch. Allergy Appl. Immunol. 75, 75–83 [PubMed]
16. Eisen M. B., Spellman P. T., Brown P. O., Botstein D. (1998) Cluster analysis and display of genome-wide expression patterns. Proc. Natl. Acad. Sci. U.S.A. 95, 14863–14868 [PMC free article] [PubMed]
17. Yamamoto T. (2009) Scleroderma—pathophysiology. Eur. J. Dermatol. 19, 14–24 [PubMed]
18. Mayes M. D., Lacey J. V., Jr., Beebe-Dimmer J., Gillespie B. W., Cooper B., Laing T. J., Schottenfeld D. (2003) Prevalence, incidence, survival, and disease characteristics of systemic sclerosis in a large US population. Arthritis Rheumatism 48, 2246–2255 [PubMed]
19. LeRoy E. C., Black C., Fleischmajer R., Jablonska S., Krieg T., Medsger T. A., Jr., Rowell N., Wollheim F. (1988) Scleroderma (systemic sclerosis): classification, subsets and pathogenesis. J. Rheumatol. 15, 202–205 [PubMed]
20. Doran J. P., Veale D. J. (2008) Biomarkers in systemic sclerosis. Rheumatology 47 Suppl 5, 36–38 [PubMed]
21. Chen C. H. (2008) Review of a current role of mass spectrometry for proteome research. Analyt. Chim. Acta 624, 16–36 [PubMed]
22. Cravatt B. F., Simon G. M., Yates J. R., 3rd (2007) The biological impact of mass-spectrometry-based proteomics. Nature 450, 991–1000 [PubMed]
23. Hanash S. M., Pitteri S. J., Faca V. M. (2008) Mining the plasma proteome for cancer biomarkers. Nature 452, 571–579 [PubMed]
24. Borrebaeck C. A., Wingren C. (2009) Design of high-density antibody microarrays for disease proteomics: Key technological issues. J. Proteomics 72, 928–935 [PubMed]
25. Haab B. B. (2006) Applications of antibody array platforms. Curr Opin Biotechnol 17, 415–421 [PubMed]
26. Kingsmore S. F. (2006) Multiplexed protein measurement: technologies and applications of protein and antibody arrays. Nat. Rev. Drug Discov. 5, 310–320 [PMC free article] [PubMed]
27. Ingvarsson J., Larsson A., Sjöholm A. G., Truedsson L., Jansson B., Borrebaeck C. A., Wingren C. (2007) Design of recombinant antibody microarrays for serum protein profiling: targeting of complement proteins. J. Proteome Res. 6, 3527–3536 [PubMed]
28. Wingren C., Ingvarsson J., Dexlin L., Szul D., Borrebaeck C. A. (2007) Design of recombinant antibody microarrays for complex proteome analysis: choice of sample labeling-tag and solid support. Proteomics 7, 3055–3065 [PubMed]
29. Borrebaeck C. A., Wingren C. (2009) Transferring proteomic discoveries into clinical practice. Expert Rev. Proteomics 6, 11–13 [PubMed]
30. Carlsson A., Wingren C., Ingvarsson J., Ellmark P., Baldertorp B., Fernö M., Olsson H., Borrebaeck C. A. (2008) Serum proteome profiling of metastatic breast cancer using recombinant antibody microarrays. Eur. J. Cancer 44, 472–480 [PubMed]
31. Ellmark P., Ingvarsson J., Carlsson A., Lundin B. S., Wingren C., Borrebaeck C. A. (2006) Identification of protein expression signatures associated with Helicobacter pylori infection and gastric adenocarcinoma using recombinant antibody microarrays. Mol. Cell Proteomics 5, 1638–1646 [PubMed]
32. Ingvarsson J., Wingren C., Carlsson A., Ellmark P., Wahren B., Engström G., Harmenberg U., Krogh M., Peterson C., Borrebaeck C. A. (2008) Detection of pancreatic cancer using antibody microarray-based serum protein profiling. Proteomics 8, 2211–2219 [PubMed]
33. Carlsson A., Persson O., Ingvarsson J., Widegren B., Salford L., Borrebaeck C. A., Wingren C. (2010) Plasma protein profiling reveals biomarker patterns associated with prognosis and therapy selection in glioblastoma multiforme patients. Proteomics – Clin. Appl. In press [PubMed]
34. Hesselstrand R., Scheja A., Akesson A. (1998) Mortality and causes of death in a Swedish series of systemic sclerosis patients. Ann. Rheumatic Dis. 57, 682–686 [PMC free article] [PubMed]
35. Ståhl-Hallengren C., Jönsen A., Nived O., Sturfelt G. (2000) Incidence studies of systemic lupus erythematosus in Southern Sweden: increasing age, decreasing frequency of renal manifestations and good prognosis. J. Rheumatol. 27, 685–691 [PubMed]
36. Masi A. T. (1988) Classification of systemic sclerosis (scleroderma): relationship of cutaneous subgroups in early disease to outcome and serologic reactivity. J. Rheumatol. 15, 894–898 [PubMed]
37. Wuttge D. M., Wildt M., Geborek P., Wollheim F. A., Scheja A., Akesson A. (2007) Serum IL-15 in patients with early systemic sclerosis: a potential novel marker of lung disease. Arthritis Res. Ther. 9, R85. [PMC free article] [PubMed]
38. Gladman D. D., Ibañez D., Urowitz M. B. (2002) Systemic lupus erythematosus disease activity index 2000. J. Rheumatol. 29, 288–291 [PubMed]
39. Wingren C., Borrebaeck C. A. (2008) Antibody microarray analysis of directly labelled complex proteomes. Curr. Opin. Biotechnol. 19, 55–61 [PubMed]
40. Söderlind E., Strandberg L., Jirholt P., Kobayashi N., Alexeiva V., Aberg A. M., Nilsson A., Jansson B., Ohlin M., Wingren C., Danielsson L., Carlsson R., Borrebaeck C. A. (2000) Recombining germline-derived CDR sequences for creating diverse single-framework antibody libraries. Nat. Biotechnol. 18, 852–856 [PubMed]
41. Borrebaeck C. A., Wingren C. (2007) High-throughput proteomics using antibody microarrays: an update. Expert Rev. Mol. Diagn. 7, 673–686 [PubMed]
42. Wingren C., Borrebaeck C. A. (2006) Antibody microarrays: current status and key technological advances. Omics 10, 411–427 [PubMed]
43. Wingren C., Borrebaeck C. A. (2007) Progress in miniaturization of protein arrays–a step closer to high-density nanoarrays. Drug Discov. Today 12, 813–819 [PubMed]
44. Wingren C., Ingvarsson J., Lindstedt M., Borrebaeck C. A. (2003) Recombinant antibody microarrays–a viable option?. Nat. Biotechnol. 21, 223. [PubMed]
45. Chih-chung C., Chih-Jen L. (2007) LIBSVM: a library for support vector machines., http/::www.csie.ntu.edu.tw/cjlin/libsvm
46. Cristianini N., Shawe-Taylor J. (2000) An Introduction to Support Vector Machines and Other Kernel-based Learning Methods. Cambridge Univeristy Press
47. Ihaka R., Gentleman R. (1996) A language for data analysis and graphics. J. Comp. Graph Stat. 5, 299–314
48. Atamas S. P., White B. (2003) Cytokine regulation of pulmonary fibrosis in scleroderma. Cytokine Growth Factor Rev. 14, 537–550 [PubMed]
49. Beirne P., Pantelidis P., Charles P., Wells A. U., Abraham D. J., Denton C. P., Welsh K. I., Shah P. L., du Bois R. M., Kelleher P. (2009) Multiplex immune serum biomarker profiling in sarcoidosis and systemic sclerosis. Eur. Respir. J. 34, 1376–1382 [PubMed]
50. Duan H., Fleming J., Pritchard D. K., Amon L. M., Xue J., Arnett H. A., Chen G., Breen P., Buckner J. H., Molitor J. A., Elkon K. B., Schwartz S. M. (2008) Combined analysis of monocyte and lymphocyte messenger RNA expression with serum protein profiles in patients with scleroderma. Arthritis Rheumatism 58, 1465–1474 [PubMed]
51. Radstake T. R., van Bon L., Broen J., Hussiani A., Hesselstrand R., Wuttge D. M., Deng Y., Simms R., Lubberts E., Lafyatis R. (2009) The pronounced Th17 profile in systemic sclerosis (SSc) together with intracellular expression of TGFbeta and IFNgamma distinguishes SSc phenotypes. PloS one 4, e5903. [PMC free article] [PubMed]
52. Gerszten R. E., Accurso F., Bernard G. R., Caprioli R. M., Klee E. W., Klee G. G., Kullo I., Laguna T. A., Roth F. P., Sabatine M., Srinivas P., Wang T. J., Ware L. B. (2008) Challenges in translating plasma proteomics from bench to bedside: update from the NHLBI Clinical Proteomics Programs. Am. J. Physiol. 295, L16–22 [PMC free article] [PubMed]
53. Hanash S. (2003) Disease proteomics. Nature 422, 226–232 [PubMed]
54. Omenn G. S., States D. J., Adamski M., Blackwell T. W., Menon R., Hermjakob H., Apweiler R., Haab B. B., Simpson R. J., Eddes J. S., Kapp E. A., Moritz R. L., Chan D. W., Rai A. J., Admon A., Aebersold R., Eng J., Hancock W. S., Hefta S. A., Meyer H., Paik Y. K., Yoo J. S., Ping P., Pounds J., Adkins J., Qian X., Wang R., Wasinger V., Wu C. Y., Zhao X., Zeng R., Archakov A., Tsugita A., Beer I., Pandey A., Pisano M., Andrews P., Tammen H., Speicher D. W., Hanash S. M. (2005) Overview of the HUPO Plasma Proteome Project: results from the pilot phase with 35 collaborating laboratories and multiple analytical groups, generating a core dataset of 3020 proteins and a publicly-available database. Proteomics 5, 3226–3245 [PubMed]
55. Lorber M., Gershwin M. E., Shoenfeld Y. (1994) The coexistence of systemic lupus erythematosus with other autoimmune diseases: the kaleidoscope of autoimmunity. Sem. Arthritis Rheumatism 24, 105–113 [PubMed]
56. Pope J. E. (2002) Scleroderma overlap syndromes. Curr. Opin. Rheumatol. 14, 704–710 [PubMed]
57. Bauer J. W., Baechler E. C., Petri M., Batliwalla F. M., Crawford D., Ortmann W. A., Espe K. J., Li W., Patel D. D., Gregersen P. K., Behrens T. W. (2006) Elevated serum levels of interferon-regulated chemokines are biomarkers for active human systemic lupus erythematosus. PLoS Med. 3, e491. [PMC free article] [PubMed]
58. Bauer J. W., Petri M., Batliwalla F. M., Koeuth T., Wilson J., Slattery C., Panoskaltsis-Mortari A., Gregersen P. K., Behrens T. W., Baechler E. C. (2009) Interferon-regulated chemokines as biomarkers of systemic lupus erythematosus disease activity: a validation study. Arthritis Rheumatism 60, 3098–3107 [PMC free article] [PubMed]
59. Suzuki M., Ross G. F., Wiers K., Nelson S., Bennett M., Passo M. H., Devarajan P., Brunner H. I. (2007) Identification of a urinary proteomic signature for lupus nephritis in children. Ped. Nephrol. 22, 2047–2057 [PubMed]
60. Villegas-Zambrano N., Martínez-Taboada V. M., Bolívar A., San Martín M., Alvarez L., Marín M. J., López-Hoyos M. (2009) Correlation between clinical activity and serological markers in a wide cohort of patients with systemic lupus erythematosus: an eight-year prospective study. Ann. N.Y. Acad. Sci. 1173, 60–66 [PubMed]
61. Sjoholm A. G., Martensson U., Sturfelt G. (1997) Serial analysis of autoantibody responses to the collagen-like region of Clq, collagen type II, and double stranded DNA in patients with systemic lupus erythematosus. J. Rheumatol. 24, 871–878 [PubMed]
62. Sturfelt G., Sjöholm A. G., Svensson B. (1983) Complement components, C1 activation and disease activity in SLE. Int. Arch. Allergy Appl. Immunol. 70, 12–18 [PubMed]
63. Mosley K., Tam F. W., Edwards R. J., Crozier J., Pusey C. D., Lightstone L. (2006) Urinary proteomic profiles distinguish between active and inactive lupus nephritis. Rheumatology 45, 1497–1504 [PubMed]
64. Pitashny M., Schwartz N., Qing X., Hojaili B., Aranow C., Mackay M., Putterman C. (2007) Urinary lipocalin-2 is associated with renal disease activity in human lupus nephritis. Arthritis Rheumatism 56, 1894–1903 [PubMed]
65. Urquizu-Padilla M., Balada E., Cortés F., Pérez E. H., Vilardell-Tarrés M., Ordi-Ros J. (2009) Serum levels of soluble CD40 ligand at flare and at remission in patients with systemic lupus erythematosus. J. Rheumatol. 36, 953–960 [PubMed]
66. Fu Q., Chen X., Cui H., Guo Y., Chen J., Shen N., Bao C. (2008) Association of elevated transcript levels of interferon-inducible chemokines with disease activity and organ damage in systemic lupus erythematosus patients. Arthritis Res. Ther. 10, R112. [PMC free article] [PubMed]
67. Pascual V., Farkas L., Banchereau J. (2006) Systemic lupus erythematosus: all roads lead to type I interferons. Curr. Opin. Immunol. 18, 676–682 [PubMed]

Articles from Molecular & Cellular Proteomics : MCP are provided here courtesy of American Society for Biochemistry and Molecular Biology

Save items

Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...


  • MedGen
    Related information in MedGen
  • PubMed
    PubMed citations for these articles
  • Substance
    PubChem chemical substance records that cite the current articles. These references are taken from those provided on submitted PubChem chemical substance records.
  • Taxonomy
    Taxonomy records associated with the current articles through taxonomic information on related molecular database records (Nucleotide, Protein, Gene, SNP, Structure).
  • Taxonomy Tree
    Taxonomy Tree

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...