New Results

Genomic dissection of Systemic Lupus Erythematosus: Distinct Susceptibility, Activity and Severity Signatures

Nikolaos I. Panousis, George Bertsias, Halit Ongen, Irini Gergianaki, Maria Tektonidou, Maria Trachana, Luciana Romano-Palumbo, Deborah Bielser, Cedric Howald, Cristina Pamfil, Antonis Fanouriakis, Despoina Kosmara, Argyro Repa, Prodromos Sidiropoulos, Emmanouil T. Dermitzakis, Dimitrios T. Boumpas
doi: https://doi.org/10.1101/255109

Abstract

Recent genetic and genomics approaches have yielded novel insights in the pathogenesis of Systemic Lupus Erythematosus (SLE) but the diagnosis, monitoring and treatment still remain largely empirical1,2. We reasoned that molecular characterization of SLE by whole blood transcriptomics may facilitate early diagnosis and personalized therapy. To this end, we analyzed genotypes and RNA-seq in 142 patients and 58 matched healthy individuals to define the global transcriptional signature of SLE. By controlling for the estimated proportions of circulating immune cell types, we show that the Interferon (IFN) and p53 pathways are robustly expressed. We also report cell-specific, disease-dependent regulation of gene expression and define a core/susceptibility and a flare/activity disease expression signature, with oxidative phosphorylation, ribosome regulation and cell cycle pathways being enriched in lupus flares. Using these data, we define a novel index of disease activity/severity by combining the validated Systemic Lupus Erythematosus Disease Activity Index (SLEDAI)1 with a new variable derived from principal component analysis (PCA) of RNA-seq data. We also delineate unique signatures across disease endo-phenotypes whereby active nephritis exhibits the most extensive changes in transcriptome, including prominent drugable signatures such as granulocyte and plasmablast/plasma cell activation. The substantial differences in gene expression between SLE and healthy individuals enables the classification of disease versus healthy status with median sensitivity and specificity of 83% and 100%, respectively. We explored the genetic regulation of blood transcriptome in SLE and found 3142 cis-expression quantitative trait loci (eQTLs). By integration of SLE genome-wide association study (GWAS) signals and eQTLs from 44 tissues from the Genotype-Tissue Expression (GTEx) consortium, we demonstrate that the genetic causality of SLE arises from multiple tissues with the top causal tissue being the liver, followed by brain basal ganglia, adrenal gland and whole blood. Collectively, our study defines distinct susceptibility and activity/severity signatures in SLE that may facilitate diagnosis, monitoring, and personalized therapy.

Systemic Lupus Erythematosus (SLE) is the prototypic systemic autoimmune disease that manifests a wide range of clinical and molecular abnormalities2,3. Despite advances in the pathogenesis and treatment, several unmet needs exist. To cite a few, the molecular events explaining the variability of SLE and the interspersing periods of inactivity and activity remain unspecified. The disease is often challenging to diagnose, especially at early stages, and there is lack of robust diagnostic criteria4. Existing instruments to monitor SLE suffer from inherent limitations and there are no accurate biomarkers of disease activity and severity. Moreover, the extent to which clinically-defined therapeutic targets correlate with changes in transcriptome is not known. Notwithstanding the successful introduction of the first targeted biological agent in SLE5, a considerable proportion of patients will be unresponsive to existing treatments, highlighting the need for novel, targeted therapies based on the underlying immune aberrancies. To examine these, we performed transcriptome profiling by RNA-seq (Supplementary Figure 1) in 142 SLE patients (Supplementary Table 1) with varying levels of disease activity, and compared it to matched healthy individuals.

Our data show widespread transcriptome perturbations in SLE with 6730 differentially expressed genes (DEGs) (false discovery rate [FDR] 5%) (Supplementary Figure 2A and Supplementary Table 2); 3977 genes were upregulated (59.1%) and 2753 genes were downregulated (40.9%) in SLE versus healthy individuals. The enriched KEGG pathways6 (5% FDR) for DEGs are shown in Supplementary Figures 2B-D with novel and previously identified pathways implicated in SLE being confirmed, such as the IFN signature represented in the Herpes simplex virus (HSV) and the NOD-like receptor signaling pathways7. A broad view of the biological regulation of SLE is provided by a network enrichment map of significantly enriched GO terms derived from DEGs (Supplementary Figure 2E). These results reveal marked aberrancies and specific signatures in SLE whole blood transcriptome.

Whole blood assays, while relevant to define complex inflammatory signatures8,9, do not inform on cell-specific mechanisms. To address this, we estimated the proportions of different immune cell types for each individual by using blood transcriptome deconvolution implemented in CIBERSORT10. We report significant differences in blood cell composition including reduction of naïve B-cells11 and natural killer cells12,13, and increase of memory activated CD4+ T-cells14 and myeloid-linage cells15-18 in SLE versus healthy individuals (Figure 1A). Next, we defined a global gene expression signature independent of the differences in cell composition by controlling for the estimated proportions of cell types, and found 1613 DEGs (5% FDR) between SLE and healthy individuals, which represents a 76% decrease in DEGs from the previous analysis. Pathway and GO term enrichment analysis revealed that the IFN signature (Figure 1B-C) is independent of cell type composition. By interrogating the interferome database19, we found that the DEGs are indicative of both type I and type II IFN response, while they also showed enrichment in the p53 signaling pathway (Figure 1B), which is implicated in apoptotic cell clearance and maintenance of immune tolerance20,21. These data demonstrate a robust transcriptional signal – independent of blood cell type composition – that could facilitate the discovery of novel biomarkers.

Figure 1:
Figure 1:

(A) Estimated proportions of different immune cell counts in healthy and SLE individuals. For every cell type the Mann-Whitney Wilcoxon test p-value comparing healthy and SLE patients is displayed on top. (B) Pathway enrichment analysis with all the DEGs after correcting for cell type proportion estimates. (C) Mechanistic map of the biological regulation in SLE after correcting for cell type proportion estimates. There is a prevalent type I IFN regulation independent of immune cell type composition. (D) Histogram’s of p-values for the interaction term (disease X estimated proportions) revealing cell type specific effects for SLE. For every cell type the proportion of estimated true positives (π1) and the number of significant genes at 5% FDR is presented. (E) Disease by estimated Neutrophils proportion interaction for the gene GTPBP2. X-axis indicate the estimated proportion of Neutrophils while y-axis indicate the normalized expression. Red dots indicate SLE patients while blue dots indicate healthy individuals. (F) Disease by estimated T-cell CD4 memory resting proportion interaction for the gene CD1C. X-axis indicate the estimated proportion of Neutrophils while y-axis indicate the normalized expression. Red dots indicate SLE patients while blue dots indicate healthy individuals.

To unravel cell-specific gene perturbations, we examined whether cell type composition has a different effect on gene expression between the SLE and healthy state. We controlled for the estimated proportions of all other immune cell types and tested for the significance of the disease × estimated cell proportion interaction term. We quantified the proportion of true positives estimated from the enrichment of significant p-values (π1)22 (Figure 1D). For naive and memory B-cells, CD4+ memory resting T-cells, CD8+ T cells, and neutrophils, the corresponding π1 values were above zero, suggesting that varying proportion of these cells interacts with the effects that cell type proportions have on gene expression in SLE versus healthy individuals. Illustratively, increasing proportion of neutrophils correlated positively with GTPBP2 (GTP Binding Protein 2) in SLE but not in healthy individuals (Figure 1E). The same trend was observed for the correlation between CD4+ memory resting T-cells and CD1c (cluster differentiation 1c) (Figure 1F). GTPBP2 interacts with IRF5 to regulate type I IFN production23, while CD1c is a cell membrane glycoprotein24 that mediates the presentation of modified peptides to CD1c-restricted T cells25 and has been linked to autoimmunity26,27. Together, differences in whole blood transcriptome in SLE are driven by both altered abundances of circulating immune cell types and cell-specific regulation of gene expression depending on disease status.

Remission of disease activity has been introduced as therapeutic target in SLE28 but whether this is mirrored by transcriptome changes remains unknown. This has implications for predicting the risk for subsequent flare and assessing the need for continuing long-term immunosuppression. To this end, we determined a ‘core’ gene expression signature that persists in the absence of disease activity following treatment. First, we applied PCA on the transcriptome in clinically active and inactive patients and healthy individuals (Supplementary Table 1 and Figures 2A-C). Inactive SLE were clearly differentiated from healthy (Figure 2B) but not from active SLE individuals (Figure 2C), signifying persistently deregulated gene expression despite disease remission. We took the intersection of DEGs in healthy versus active SLE (4938 DEGs 5% FDR; Supplementary Table 3) and healthy versus inactive SLE (4658 DEGs; Supplementary Table 4) that are not DEGs in active versus inactive SLE (377 DEGs; Supplementary Table 5), to reach 2726 DEGs which comprise the ‘core’ disease gene signature. These genes were enriched in biological processes and GO terms related to the regulation and response of the immune system (Supplementary Figure 3) suggesting persistence of immune system activation and inflammation despite remission of clinical signs and symptoms. By comparing patients with clinically inactive SLE but evidence for serologic activity (high anti-dsDNA titers, low serum complement)29 against those who are both clinically and serologically inactive, we found no DEGs at 5% FDR, corroborating studies showing similar favorable prognosis for these two groups30.

Figure 2:
Figure 2:

(A) Principal Component Analysis (PCA) of whole transcriptome between healthy and Active SLE individuals. The two first Principal Components are plotted. PC2 is clearly differentiating the 2 groups implying differences in gene expression. (B) PCA analysis between healthy and Inactive SLE individuals. PC2 is clearly differentiating the 2 groups indicating that even in remission the transcriptome of SLE patients is different compared to healthy individuals. (C) PCA analysis between Active and Inactive SLE patients. The first 2 PCs do not differentiate the 2 groups suggesting that there are not large differences in gene expression between them.

Flares of disease activity are frequent in SLE and contribute to accrual of irreversible organ damage31. Defining the signature of increased SLE activity has pathogenic and clinical ramifications. To this end, we selected the DEGs from the comparison of inactive versus active SLE that are not included in the ‘core’ disease signature. A total 365 DEGs were identified, which were enriched for KEGG pathways such as oxidative phosphorylation, consistent with the described alterations in mitochondrial mass and membrane potential in lupus T cells32-34 and the enhanced oxidative stress35-37. Other enriched pathways included ribosomes and cell cycle. Thus, increases in SLE activity may be linked to perturbed expression of genes that regulate metabolism, protein synthesis and proliferation of peripheral blood immune cells. The flare gene signature could enable the earlier identification of an impending clinical flare and more effective use of pre-emptive treatment, an unmet need in SLE in view of the limitations of traditional serologic tests38,39.

Biomarkers that accurately reflect disease severity and the underlying molecular pathogenesis, represent an additional unmet need in SLE. We assessed transcriptome differences according to varying degrees of disease activity/severity by using the validated SLEDAI-2K index as a quantitative measure (0=inactive; 1–5=mild; 6–10=moderate; >10=severe disease)40, among patients and healthy individuals (fixed score of -1). We performed PCA (Figure 3A) on the identified 3690 DEGs (5% FDR) (Supplementary Table 6) and defined PC1 (explaining 23% of the variance) as a new variable summarizing the expression properties of genes that recapitulate SLE severity (Figure 3B). Notably, PC1 clusters closely patients with inactive and low disease activity, which is consistent with evidence that both these states have a favorable outcome41,42. Pathway analysis showed enrichment in the oxidative phosphorylation and cell cycle KEGG pathways, similar to the ‘flare’ signature, suggesting that these two biological processes may be implicated not only in disease activity but also in severity. Functionally grouped networks of GO terms revealed gene signatures related to protein ubiquitination, electron transport chain, protein phosphorylation, cell cycle, defense response and regulation of response to stress (Figure 3C), all of which have been linked to SLE37,43,44. These results point out the involvement of multiple molecular pathways and biological processes in determining SLE progression/severity and suggest that whole blood transcriptome may serve as a robust biomarker of SLE activity and severity.

Figure 3:
Figure 3:

(A) Principal Component Analysis (PCA) of 3690 DEGs found by using SLEDAI score as a quantitative measurement of SLE severity. The first two Principal Components are plotted. PC1 is capturing the progression of activity/severity of SLE since is separating the different groups of activity. (B) Jitter plot of PC1 weights. For each group, the median is plotted. The regression line is plotted with red (p-value5.86e-17). PC1 defines a new phenotype that captures SLE activity. (C) Pathway enrichment analysis of 3690 DEGs on the left side of the plot. Oxidative phosphorylation and cell cycle KEGG pathways, that were regulated by genes belonging to the “flare” gene signature is also enriched suggesting that these genes not only capture the inactive to active progression but also the different levels of severity. Functionally grouped networks of GO terms on the right. Multiple biological aberrations are captured by differences in gene expression based on different levels of SLE activity.

SLE can affect multiple organs but the molecular basis of this heterogeneity remains elusive. Accordingly, SLE patients were grouped according to predominant organ activity (Table 1) and multiple comparisons were carried out. By comparing patients with active renal disease (nephritis) (group 1) versus those with activity from other organs (combined groups 2,3,4), we found 136 DEGs (5% FDR). These genes were enriched in functionally grouped networks of granulocyte activation and antimicrobial humoral response (Supplementary Figure 4), consistent with the role of neutrophil activation45,46 and their death by formation of chromatin extracellular traps,47-49 and of plasmablasts/plasma-cells11,50-52 in lupus nephritis. To further discern the transcriptome basis for kidney involvement in SLE, we took the intersection of DEGs in active nephritis (group 1) versus inactive SLE (combined groups 5 and 6) (1375 DEGs, 5% FDR) with those in active versus inactive SLE (377 DEGs). A total 305 genes were common in the two comparisons, suggesting a step-wise progression of transcriptome alterations from inactive to active non-renal and active renal SLE status. By comparing patients with active versus inactive nephritis, we found global gene expression differences captured by PCA (Supplementary Figure 5), while differential gene expression analysis between these two groups revealed 1496 DEGs. Together, lupus nephritis displays marked and gradually enriched changes in whole blood gene expression compared to other SLE subsets, some of which may be drugable.

Table 1:
Table 1:

Groups of individuals based on the tissue/organ that the disease activity originates.

SLE exhibits a striking gender bias with females being affected 7 to 12 times more frequently than men, yet the latter suffering from more severe disease53. To gain insights into the molecular basis of this sexual dimorphism, we took the non-overlapping sets of DEGs in male versus female SLE (Bonferroni significant) and male versus female healthy (90% FDR) individuals. Six genes showed gender-biased expression in SLE (Supplementary Figure 6A), two of which (SMC1A, ARSD) are located on X chromosome and escape X-inactivation. SMC1A, encoding for the cohesin complex protein Structural Maintenance Of Chromosomes 1A54-56, demonstrated the strongest gender bias (Supplementary Figure 6A-B), and this was confirmed in purified CD4+ T-cells (Supplementary Figure 6C). Although preliminary, these results provide candidate genes for further studies on the sexual dimorphism in the disease.

SLE diagnosis can be challenging especially at early stages before an adequate number of features accumulate4. We asked whether we could accurately classify individuals based on their transcriptional profile by building multiple classifiers based on Linear Discriminant Analysis (LDA) by the use of DEGs as features (see Methods). We measured a median diagnostic accuracy of 90% in the validation set with 86% sensitivity and 100% specificity (Supplementary Figure 7). Patients who were most frequently misclassified as healthy individuals had lower frequency of renal involvement, ANA/anti-dsDNA autoantibodies, and were less frequently treated with immunosuppressive agents, suggesting an overrepresentation of milder disease. Thus, whole blood transcriptome might be used to assist the diagnosis of SLE, a finding that needs to be confirmed and validated in longitudinal studies of patients with both early and established disease.

Definition of regulatory genetic effects is crucial to our understanding of the molecular basis of complex diseases57. We explored the genetic regulation of blood transcriptome in SLE by assessing expression-Quantitative Trait Loci (eQTLs) in our cohort. We found 3142 cis-eQTLs (5% FDR), with highly significant cis-eQTLs clustering close to the transcription start site of the genes (Supplementary Figure 8A-B). These eQTLs replicated well (pi=0.89, Supplementary Figure 8C) with a blood RNA-seq study in healthy donors58, suggesting the lack of disease specificity. This could be due to lack of adequate power to detect such effects and/or the need to study specific immune cell types. Assessing eQTLs from multiple tissues could enhance our interpretation of GWAS signals, as most of them reside in non-coding genome areas59 and enable the identification of the causal genes and tissues. To estimate the tissue(s) that determine the genetic causality of SLE, we employed the Regulatory Trait Concordance (RTC) method57,60 that estimates the causal tissues by using eQTLs in 44 tissues from the GTEx consortium61 (Figure 4). We found that the top causal tissue is liver followed by brain basal ganglia, adrenal gland and whole blood, suggesting that the genetic causality of SLE arises from multiple tissues. The finding of liver being linked as the top causal tissue is in agreement with our result that SLE exacerbation is associated with changes in expression of genes that regulate metabolism62.

Figure 4:
Figure 4:

(A) Estimation of the genetic causality of SLE in 44 tissues from GTEx. On the primary y-axis, the enrichments over the null per tissue are plotted as bars and on the secondary y-axis number of GWAS variants that co-localized with eQTLs per tissue are plotted as a line. The horizontal black line indicates the null. On top of each of the bars are the -log10 Benjamini-Hochberg corrected p-values for the enrichments.

In summary, SLE is an autoimmune disease with marked clinical and immunological heterogeneity, waxing and waning course, and variable response to immunosuppressive and biological treatments. In a previous study, Banchereau et al 63 profiled the blood transcriptome by microarrays in a longitudinal cohort of pediatric patients. However, their analysis focused on detection of immune correlates of disease activity and treatment effects based on a priori defined modules of transcripts co-expressed in blood across various immunological conditions. Our RNA-seq analysis in a well-characterized cohort of adult SLE patients provides an unbiased characterization of whole-blood gene signatures associated not only with susceptibility but also, with clinically relevant disease progression such as increased activity and major organ involvement. We also defined cell-specific effects of the disease on gene expression, which may provide insights into the pathogenic role of specific immune cell types by the further studies in purified cells. DEGs are involved in biological networks and pathways that are pertinent to SLE pathogenesis and may represent putative therapeutic targets. Importantly, we show that whole blood transcriptome can classify SLE versus healthy individuals with excellent specificity putting forward the possibility of using blood gene profiling as a diagnostic tool. Finally, our studies on the genetic variation on gene expression suggest that in addition to whole blood, other tissues such as the liver may determine genetic causality in SLE.

Author contributions

NIP carried out the analyses. NIP and GB drafted the manuscript, with contributions from all authors. GB recruited and took care of patients, collected blood samples, extracted RNA and DNA. HO performed the RTC analysis. LRP and DB prepared the RNA-seq libraries and CH processed the RNA-seq and genotyping data. IG contributed in recruitment of patients and healthy individuals, and extracted clinical data from the medical charts. DK isolated CD4+ T-cells from peripheral blood samples and performed RT-PCR studies. MT, MT, CP, AF, AR and PS contributed patient samples and participated in the analyses of data. DTB and ETD conceived the study. All authors read and approved the final manuscript for publication.

Competing financial interests

The authors declare no competing financial interests.

Methods

Patients, samples collection and clinical assessment

SLE patients, diagnosed according to board-certified rheumatologist’s judgment and/or the 1997 revised ACR classification criteria64, were enrolled from the rheumatology clinics at the University Hospital of Heraklion, University Hospital C.F.R Cluj Napoca, General Hospital of Athens ‘Laikon’, ‘Attikon’ University Hospital in Athens and ‘Hippokration’ Hospital of Thessaloniki. Demographic and clinical characteristics including history of biopsy-proven nephritis, ACR classification criteria, serum autoantibodies, assessment of disease activity (physician’s global assessment [PGA], SLEDAI-2K40), definitions of Lupus Low Disease Activity State and remission41,42, and use of medications, were evaluated by a standardized protocol. Informed consent was obtained from adults and the parents those younger than 18 years of age and all procedures were followed in accordance with protocols approved by the local institutional review boards. Patients were asked to withdraw all lupus medications for 12 hours prior to blood collection. Venipuncture was performed to collect blood in PaxGene RNA tubes (Qiagen) for mRNA extraction and in EDTA tubes for DNA extraction.

Study design

We performed RNA-seq to measure gene expression in whole blood from SLE patients and matched for age and sex healthy individuals. After quality control (Supplementary Figure 1), we obtained gene quantifications for 142 SLE patients and 58 healthy volunteers, measuring the expression of ~21.000 genes. All individuals were genotyped and imputed towards the 1000 genomes phase III reference panel65,66, yielding a set of ~6.9 millions of variants. For all SLE individuals we collected a variety of clinical phenotypes such as the 1997 American College of Rheumatology revised criteria for classification of SLE, presence of serum autoantibodies, physician-rated disease activity (PGA), an index measuring the activity of the disease (SLEDAI-2K)1,67, and the medical treatment at the time of sample collection.

Genotyping

All the individuals were genotyped with the Illumina HumanCoreExome-24 array. The individuals were phased with SHAPEIT68 and imputed to the 1000 Genomes Project Phase III66 using IMPUTE269. After imputation, we filtered out SNPs with MAF <=0.05, IMPUTE score < 0.4 and Hardy-Weinberg equilibrium p-value <5e-7. After filtering we ended up with ~7M variants for association testing.

RNA sequencing, mapping and quantifications

RNA libraries were prepared with the Illumina TruSeq sample preparation kit and were sequenced on Illumina HiSeq2000 according the manufacturer’s instructions. 49 bp paired-end reads were mapped to the GRCh37 reference human genome using the GEM mapper70. We kept reads with mapping quality >150 to quantify exons and genes corresponding to reads that are uniquely mapped to the genome, with the correct orientation between pairs, allowing for 5 mismatches for both reads of a pair. Exon quantification was performed using the GENCODE annotation v1571. The overlapping exons of a gene were merged into a meta-exon unit with starting coordinates the start position of the first exon that is merged and end position the end position of the last exon that is merged. We counted reads as mapped to this meta-exon unit if there was an overlap between either the start or end position of the read with the meta-exon. Gene level quantifications were then obtained by summing the meta-exons counts for each gene.

Quality control of RNA sequencing data

Several metrics were used to assess the quality of the RNA-seq data (Supplementary Figure 1A). We measured the proportion of exonic/total reads in quantified exons per sample. Samples with a ratio less than 0.2 were excluded from further analysis as technical outliers. We further excluded samples being outliers in the PCA plot (Supplementary Figure 1B). These samples were also outliers based on the proportion of total reads mapped to hemoglobin genes (Supplementary Figure 1C). Two samples were also excluded from further analysis based on discrepancy of the sex identified by the RNA-seq data and our clinical records (Supplementary Figure 1D). Finally, MBV72 was used to correct for any sample label swaps or cross contamination for the RNA-seq and the genotyped data without identifying any swaps in our dataset.

Differential gene expression analysis

We included 21851 genes in the analysis and used this set of genes in all downstream analyses for consistency. A gene was included if it had at least 5 reads in 10% of either the SLE or the healthy individuals. DESeq273 was used to call differentially expressed genes by including GC content, RNA integrity (RIN), center of collection, insert size mode, age, gender, amount of RNA to construct the library and plate number as technical covariates. These covariates were found to influence gene expression by performing step wise regression between gene expression and all the technical covariates that were obtained from the experimental procedure. Benjamini-Hochberg was used to assess significance at 5% or 1% false discovery rate (FDR). To identify genes that were differentially expressed despite the immune cell composition we used the same covariates as before by including the estimated cell proportions as covariates in the model. To assess cell specific differentially expression we used the latter model by including a disease × estimated proportion status interaction term and obtained p-values for every gene for the interaction term. For this analysis, we estimated the proportion of true positives from the enrichment of significant p-values by using the π1 statistics. To specify the DEGs based on disease severity/activity the SLEDAI-2K index was used as a quantitative measurement by fixing the score to -1 for healthy individuals. To call up and down regulated DEGs we used Spearman rank correlation between gene expression and the SLEDAI index.

Pathway and GO term enrichment analysis

To functionally characterize the identified DEGs a comprehensive functional enrichment gene set analysis for KEGG pathways and GO terms was performed by using the ClueGO74 plugin of Cytoscape75. Briefly, ClueGO visualize the functionally grouped networks of enriched pathways and GO terms. The enrichment is based on a two-sided hypergeometric test and corrected for multiple testing with either Benjamini-Hochberg for KEGG pathways or Bonferroni for GO terms by setting the threshold of the adjusted p-values in each case at 0.05.

We used a more stringent approach with GO terms by Bonferroni correction and fusion criteria implemented in the ClueGO plugin to reduce the plethora of the terms that have the same enriched genes allowing only for the most representative “parent” or “child” terms in the networks. Each node represents a pathway or a GO term and the size of each node represents the enrichment significance. The connection between nodes is based on Cohen’s kappa statistic score (≥ 0.4) which depends on the gene sharing between nodes. For each network, only the most significant node is labeled.

Cell type estimation

We used CIBERSORT10 to estimate the proportion of different immune cell types in whole blood in our dataset. CIBERSORT uses a well-defined gene expression signature (LM22) that consists of 547 genes that precisely differentiate mature human hematopoietic cells. To perform the analysis, we used the LM22 gene matrix as reference and normalized values for the library size per sample in our dataset obtained from the estimateSizeFactors function in DESeq2.

CD4+ T-cell purification and RT-PCR

Peripheral blood mononuclear cells (PBMCs) from SLE patients and age-matched healthy blood volunteers were isolated by Ficoll-Histopaque (Sigma-Aldrich) density-gradient centrifugation of heparinized venous blood. CD4+ T lymphocytes (98% purity) were isolated by negative immunomagnetic selection (Miltenyi Biotec, Bergisch Gladbach, Germany). Total RNA was extracted using the TRIzol™ extraction method and the Turbo DNAse kit (Ambion) was used to eliminate genomic DNA contamination. cDNA was prepared using Perfect Real time cDNA Synthesis Kit (Takara) according to manufacturer’s protocol. 200ng of RNA were used as a template for every reaction. RNAse H (2U/reaction) was added to clean the resulting cDNA from any RNA and incomplete cDNA products. PCR amplification of the resulting cDNA samples was performed using appropriate volumes of KAPA SYBR® FAST Universal 2x qPCR Master Mix and specific for each gene primers at a CFX Connect™, Real-Time System. Total volume of each PCR reaction was 20μl. Expression was normalized to GAPDH and calculated by the change-in-threshold method [2^(-ΔΔCT)]. The following primer sets were used: (5’ → 3’): SMC1A forward human: CAT CAA AGC TCG TAA CTT CCT CG; SMC1A reverse human: CCC CAG AAC GAC TAA TCT CTT CA; GAPDH forward human: CAT GTT CCA ATA TGA TTC CAC C; GAPDH reverse human: GAT GGG ATT TCC ATT GAT GAC.

Classification

we performed Linear Discriminant Analysis (LDA) by using DEGs as features. We divided our dataset to training (90%, corresponding to 128 SLE patients and 52 healthy individuals) and validation set (10%, corresponding to 14 SLE patients and 6 healthy individuals) and run 1000 iterations. For each iteration, we performed differential gene expression analysis between SLE and healthy using the training set and built the LDA classifier based on the identified DEGs in order to measure the accuracy, sensitivity (the probability of calling an SLE individual as patient) and specificity (the probability of calling a healthy individual as healthy) of the classifier. We found a median of 6126 DEGs ranging from 4671 to 7222 DEGs. We chose the strategy of building multiple classifiers to account for the SLE heterogeneity by inserting perturbations in the models by sampling in each iteration different individuals. In doing so, we achieved high levels of accurate classification.

eQTL mapping

eQTL analysis was performed in a 1MB window upstream or downstream the transcription start site of the gene using QTLtools76. We scaled the total number of raw reads per sample to the median number of all samples. We removed genes that were not expressed in 90% of the individuals and then we applied PCA to account for hidden technical variation. Significance was assessed by using the qvalue R package22 on beta approximated empirical p-values from 1000 permutations. Replication was assessed by using the π1 statistic on the p-value distribution obtained from significant cis-eQTLs.

Regulatory Trait Concordance (RTC)

We have previously described RTC score to assess whether a GWAS variant and an eQTL are tagging the same causal variant57. Briefly, we expect that if a GWAS variant and an eQTL do tag the same causal variant, by removing the genetic effect of the GWAS variant will have a significant consequence on the eQTL association. Following up, we can now estimate the causal tissues for complex traits and diseases by measuring the tissue sharing probabilities of eQTLs and calculating probabilities that a GWAS variant and the eQTL do tag the same functional effect. By normalizing the GWAS-eQTL probabilities with the tissue sharing estimates of the eQTLs, we can estimate the tissues from which GWAS genetic causality arises60.

Acknowledgments

This work was funded by the Foundation for Research in Rheumatology (FOREUM) and in part by the Greek General Secretariat of Research and Technology ‘Aristeia’ action of the Operational Program ‘Education and Lifelong Learning’ (co-funded by the European Social Fund and National Resources, Aristeia I 2344 to DB) and the European Research Council (ERC) Advanced Grant (“Lupuscare” to DB). GB was supported by the Greek State Scholarships Foundation-ΙKY (Postdoc Fellowships of Excellence-Siemens) and the European Union Seventh Framework Programme (project TransPOT; FP7-REGPOT-2011-1).

References

  1. 1.↵
    Bombardier, C., Gladman, D.D., Urowitz, M.B., Caron, D. & Chang, C.H. Derivation of the SLEDAI. A disease activity index for lupus patients. The Committee on Prognosis Studies in SLE. Arthritis Rheum 35, 630–640 (1992).
    OpenUrlCrossRefPubMedWeb of Science
  2. 2.↵
    Bertsias, G.K., Salmon, J.E. & Boumpas, D.T. Therapeutic opportunities in systemic lupus erythematosus: state of the art and prospects for the new decade. Ann Rheum Dis 69, 1603–1611 (2010).
    OpenUrlAbstract/FREE Full Text
  3. 3.↵
    Tsokos, G.C. Systemic lupus erythematosus. N Engl J Med 365, 2110–2121 (2011).
    OpenUrlCrossRefPubMedWeb of Science
  4. 4.↵
    Bertsias, G.K., Pamfil, C., Fanouriakis, A. & Boumpas, D.T. Diagnostic criteria for systemic lupus erythematosus: has the time come? Nat Rev Rheumatol 9, 687–694 (2013).
    OpenUrlCrossRefPubMed
  5. 5.↵
    Manzi, S., et al. Effects of belimumab, a B lymphocyte stimulator-specific inhibitor, on disease activity across multiple organ domains in patients with systemic lupus erythematosus: combined results from two phase III trials. Ann Rheum Dis 71, 1833–1838 (2012).
    OpenUrlAbstract/FREE Full Text
  6. 6.↵
    Kanehisa, M. & Goto, S. KEGG: kyoto encyclopedia of genes and genomes. Nucleic Acids Res 28, 27–30 (2000).
    OpenUrlCrossRefPubMedWeb of Science
  7. 7.↵
    Sabbah, A., et al. Activation of innate immune antiviral responses by Nod2. Nat Immunol 10, 1073–1080 (2009).
    OpenUrlCrossRefPubMedWeb of Science
  8. 8.↵
    Duffy, D., et al. Functional analysis via standardized whole-blood stimulation systems defines the boundaries of a healthy immune response to complex stimuli. Immunity 40, 436–450 (2014).
    OpenUrlCrossRefPubMed
  9. 9.↵
    Urrutia, A., et al. Standardized Whole-Blood Transcriptional Profiling Enables the Deconvolution of Complex Induced Immune Responses. Cell Rep 16, 2777–2791 (2016).
    OpenUrl
  10. 10.↵
    Newman, A.M., et al. Robust enumeration of cell subsets from tissue expression profiles. Nat Methods 12, 453–457 (2015).
    OpenUrlCrossRefPubMed
  11. 11.↵
    Odendahl, M., et al. Disturbed peripheral B lymphocyte homeostasis in systemic lupus erythematosus. J Immunol 165, 5970–5979 (2000).
    OpenUrlAbstract/FREE Full Text
  12. 12.↵
    Hervier, B., et al. Phenotype and function of natural killer cells in systemic lupus erythematosus: excess interferon-gamma production in patients with active disease. Arthritis Rheum 63, 1698–1706 (2011).
    OpenUrlCrossRefPubMedWeb of Science
  13. 13.↵
    Huang, Z., et al. Involvement of CD226+ NK cells in immunopathogenesis of systemic lupus erythematosus. J Immunol 186, 3421–3431 (2011).
    OpenUrlAbstract/FREE Full Text
  14. 14.↵
    Fritsch, R.D., et al. Abnormal differentiation of memory T cells in systemic lupus erythematosus. Arthritis Rheum 54, 2184–2197 (2006).
    OpenUrlCrossRefPubMedWeb of Science
  15. 15.↵
    Blanco, P., Palucka, A.K., Gill, M., Pascual, V. & Banchereau, J. Induction of dendritic cell differentiation by IFN-alpha in systemic lupus erythematosus. Science 294, 1540–1543 (2001).
    OpenUrlAbstract/FREE Full Text
  16. 16.
    Ding, D., Mehta, H., McCune, W.J. & Kaplan, M.J. Aberrant phenotype and function of myeloid dendritic cells in systemic lupus erythematosus. J Immunol 177, 5878–5889 (2006).
    OpenUrlAbstract/FREE Full Text
  17. 17.
    Gerl, V., et al. Blood dendritic cells in systemic lupus erythematosus exhibit altered activation state and chemokine receptor function. Ann Rheum Dis 69, 1370–1377 (2010).
    OpenUrlAbstract/FREE Full Text
  18. 18.↵
    Gupta, S. & Kaplan, M.J. The role of neutrophils and NETosis in autoimmune and renal diseases. Nat Rev Nephrol 12, 402–413 (2016).
    OpenUrlCrossRefPubMed
  19. 19.↵
    Rusinova, I., et al. Interferome v2.0: an updated database of annotated interferon-regulated genes. Nucleic Acids Res 41, D1040–1046 (2013).
    OpenUrlCrossRefPubMedWeb of Science
  20. 20.↵
    Watanabe, M., Moon, K.D., Vacchio, M.S., Hathcock, K.S. & Hodes, R.J. Downmodulation of tumor suppressor p53 by T cell receptor signaling is critical for antigen-specific CD4(+) T cell responses. Immunity 40, 681–691 (2014).
    OpenUrl
  21. 21.↵
    Yoon, K.W., et al. Control of signaling-mediated clearance of apoptotic cells by the tumor suppressor p53. Science 349, 1261669 (2015).
    OpenUrlAbstract/FREE Full Text
  22. 22.↵
    Storey, J.D. & Tibshirani, R. Statistical significance for genomewide studies. Proc Natl Acad Sci U S A 100, 9440–9445 (2003).
    OpenUrlAbstract/FREE Full Text
  23. 23.↵
    Li, S., Wang, L., Berman, M., Kong, Y.Y. & Dorf, M.E. Mapping a dynamic innate immunity protein interaction network regulating type I interferon production. Immunity 35, 426–440 (2011).
    OpenUrlCrossRefPubMedWeb of Science
  24. 24.↵
    del, C.S.M., Mendiguren, A.K., Salamone, G.V. & Fainboim, L. Membrane trafficking of CD1c on activated T cells. J Leukoc Biol 70, 567–577 (2001).
    OpenUrlPubMed
  25. 25.↵
    Salio, M., Silk, J.D., Jones, E.Y. & Cerundolo, V. Biology of CD1- and MR1-restricted T cells. Annu Rev Immunol 32, 323–366 (2014).
    OpenUrlCrossRefPubMedWeb of Science
  26. 26.↵
    Roura-Mir, C., et al. CD1a and CD1c activate intrathyroidal T cells during Graves’ disease and Hashimoto’s thyroiditis. J Immunol 174, 3773–3780 (2005).
    OpenUrlAbstract/FREE Full Text
  27. 27.↵
    Sieling, P.A., et al. Human double-negative T cells in systemic lupus erythematosus provide help for IgG and are restricted by CD1c. J Immunol 165, 5338–5344 (2000).
    OpenUrlAbstract/FREE Full Text
  28. 28.↵
    van Vollenhoven, R., et al. A framework for remission in SLE: consensus findings from a large international task force on definitions of remission in SLE (DORIS). Ann Rheum Dis 76, 554–561 (2017).
    OpenUrlAbstract/FREE Full Text
  29. 29.↵
    Gladman, D.D., Urowitz, M.B. & Keystone, E.C. Serologically active clinically quiescent systemic lupus erythematosus: a discordance between clinical and serologic features. Am J Med 66, 210–215 (1979).
    OpenUrlCrossRefPubMedWeb of Science
  30. 30.↵
    Steiman, A.J., Gladman, D.D., Ibanez, D. & Urowitz, M.B. Prolonged serologically active clinically quiescent systemic lupus erythematosus: frequency and outcome. J Rheumatol 37, 1822–1827 (2010).
    OpenUrlAbstract/FREE Full Text
  31. 31.↵
    Ugarte-Gil, M.F., et al. The number of flares patients experience impacts on damage accrual in systemic lupus erythematosus: data from a multiethnic Latin American cohort. Ann Rheum Dis 74, 1019–1023 (2015).
    OpenUrlAbstract/FREE Full Text
  32. 32.↵
    Caza, T.N., et al. HRES-1/Rab4-mediated depletion of Drp1 impairs mitochondrial homeostasis and represents a target for treatment in SLE. Ann Rheum Dis 73, 1888–1897 (2014).
    OpenUrlAbstract/FREE Full Text
  33. 33.
    Fernandez, D.R., et al. Activation of mammalian target of rapamycin controls the loss of TCRzeta in lupus T cells through HRES-1/Rab4-regulated lysosomal degradation. J Immunol 182, 2063–2073 (2009).
    OpenUrlAbstract/FREE Full Text
  34. 34.↵
    Perry, D.J., et al. Murine lupus susceptibility locus Sle1c2 mediates CD4+ T cell activation and maps to estrogen-related receptor gamma. J Immunol 189, 793–803 (2012).
    OpenUrlAbstract/FREE Full Text
  35. 35.↵
    Campbell, A.M., Kashgarian, M. & Shlomchik, M.J. NADPH oxidase inhibits the pathogenesis of systemic lupus erythematosus. Sci Transl Med 4, 157ra141 (2012).
    OpenUrlAbstract/FREE Full Text
  36. 36.
    Oates, J.C., Mashmoushi, A.K., Shaftman, S.R. & Gilkeson, G.S. NADPH oxidase and nitric oxide synthase-dependent superoxide production is increased in proliferative lupus nephritis. Lupus 22, 1361–1370 (2013).
    OpenUrlCrossRefPubMed
  37. 37.↵
    Perl, A. Oxidative stress in the pathology and treatment of systemic lupus erythematosus. Nat Rev Rheumatol 9, 674–686 (2013).
    OpenUrlCrossRefPubMed
  38. 38.↵
    Ho, A., Magder, L.S., Barr, S.G. & Petri, M. Decreases in anti-double-stranded DNA levels are associated with concurrent flares in patients with systemic lupus erythematosus. Arthritis Rheum 44, 2342–2349 (2001).
    OpenUrlCrossRefPubMedWeb of Science
  39. 39.↵
    Moroni, G., et al. Are laboratory tests useful for monitoring the activity of lupus nephritis? A 6-year prospective study in a cohort of 228 patients with lupus nephritis. Ann Rheum Dis 68, 234–237 (2009).
    OpenUrlAbstract/FREE Full Text
  40. 40.↵
    Lam, G.K. & Petri, M. Assessment of systemic lupus erythematosus. Clin Exp Rheumatol 23, S120–132 (2005).
    OpenUrlPubMedWeb of Science
  41. 41.↵
    Franklyn, K., et al. Definition and initial validation of a Lupus Low Disease Activity State (LLDAS). Ann Rheum Dis 75, 1615–1621 (2016).
    OpenUrlAbstract/FREE Full Text
  42. 42.↵
    Zen, M., et al. Prolonged remission in Caucasian patients with SLE: prevalence and outcomes. Ann Rheum Dis 74, 2117–2122 (2015).
    OpenUrlAbstract/FREE Full Text
  43. 43.↵
    Balomenos, D., et al. The cell cycle inhibitor p21 controls T-cell proliferation and sex-linked lupus development. Nat Med 6, 171–176 (2000).
    OpenUrlCrossRefPubMedWeb of Science
  44. 44.↵
    Popovic, D., Vucic, D. & Dikic, I. Ubiquitination in disease pathogenesis and treatment. Nat Med 20, 1242–1253 (2014).
    OpenUrlCrossRefPubMed
  45. 45.↵
    Cochrane, C.G., Unanue, E.R. & Dixon, F.J. A Role of Polymorphonuclear Leukocytes and Complement in Nephrotoxic Nephritis. J Exp Med 122, 99–116 (1965).
    OpenUrlAbstract
  46. 46.↵
    Kaplan, M.J. Neutrophils in the pathogenesis and manifestations of SLE. Nat Rev Rheumatol 7, 691–699 (2011).
    OpenUrlCrossRefPubMed
  47. 47.↵
    Hakkim, A., et al. Impairment of neutrophil extracellular trap degradation is associated with lupus nephritis. Proc Natl Acad Sci U S A 107, 9813–9818 (2010).
    OpenUrlAbstract/FREE Full Text
  48. 48.
    Jorch, S.K. & Kubes, P. An emerging role for neutrophil extracellular traps in noninfectious disease. Nat Med 23, 279–287 (2017).
    OpenUrlCrossRefPubMed
  49. 49.↵
    Villanueva, E., et al. Netting neutrophils induce endothelial damage, infiltrate tissues, and expose immunostimulatory molecules in systemic lupus erythematosus. J Immunol 187, 538–552 (2011).
    OpenUrlAbstract/FREE Full Text
  50. 50.↵
    Jacobi, A.M., et al. HLA-DRhigh/CD27high plasmablasts indicate active disease in patients with systemic lupus erythematosus. Ann Rheum Dis 69, 305–308 (2010).
    OpenUrlAbstract/FREE Full Text
  51. 51.
    Cheng, Q., et al. Autoantibodies from long-lived ‘memory’ plasma cells of NZB/W mice drive immune complex nephritis. Ann Rheum Dis 72, 2011–2017 (2013).
    OpenUrlAbstract/FREE Full Text
  52. 52.↵
    Starke, C., et al. High frequency of autoantibody-secreting cells and long-lived plasma cells within inflamed kidneys of NZB/W F1 lupus mice. Eur J Immunol 41, 2107–2112 (2011).
    OpenUrlCrossRefPubMed
  53. 53.↵
    Gergianaki, I., et al. Epidemiology and burden of systemic lupus erythematosus in a Southern European population: data from the community-based lupus registry of Crete, Greece. Ann Rheum Dis 76, 1992–2000 (2017).
    OpenUrlAbstract/FREE Full Text
  54. 54.↵
    Watrin, E., Kaiser, F.J. & Wendt, K.S. Gene regulation and chromatin organization: relevance of cohesin mutations to human disease. Curr Opin Genet Dev 37, 59–66 (2016).
    OpenUrlCrossRefPubMed
  55. 55.
    Busslinger, G.A., et al. Cohesin is positioned in mammalian genomes by transcription, CTCF and Wapl. Nature 544, 503–507 (2017).
    OpenUrlCrossRefPubMed
  56. 56.↵
    Richterova, J., Huraiova, B. & Gregan, J. Genome Organization: Cohesin on the Move. Mol Cell 66, 444–445 (2017).
    OpenUrl
  57. 57.↵
    Nica, A.C., et al. Candidate causal regulatory effects by integration of expression QTLs with complex trait genetic associations. PLoS Genet 6, e1000895 (2010).
    OpenUrlCrossRefPubMed
  58. 58.↵
    Buil, A., et al. Gene-gene and gene-environment interactions detected by transcriptome sequence analysis in twins. Nat Genet 47, 88–91 (2015).
    OpenUrlCrossRefPubMed
  59. 59.↵
    Hindorff, L.A., et al. Potential etiologic and functional implications of genome-wide association loci for human diseases and traits. Proc Natl Acad Sci U S A 106, 9362–9367 (2009).
    OpenUrlAbstract/FREE Full Text
  60. 60.↵
    Ongen, H., et al. Estimating the causal tissues for complex traits and diseases. Nat Genet (2017).
  61. 61.↵
    Consortium, G., et al. Genetic effects on gene expression across human tissues. Nature 550, 204–213 (2017).
    OpenUrlCrossRefPubMed
  62. 62.↵
    Oaks, Z., et al. Mitochondrial Dysfunction in the Liver and Antiphospholipid Antibody Production Precede Disease Onset and Respond to Rapamycin in Lupus-Prone Mice. Arthritis Rheumatol 68, 2728–2739 (2016).
    OpenUrl
  63. 63.↵
    Banchereau, R., et al. Personalized Immunomonitoring Uncovers Molecular Networks that Stratify Lupus Patients. Cell 165, 1548–1550 (2016).
    OpenUrl
  64. 64.↵
    Hochberg, M.C. Updating the American College of Rheumatology revised criteria for the classification of systemic lupus erythematosus. Arthritis Rheum 40, 1725 (1997).
    OpenUrlCrossRefPubMedWeb of Science
  65. 65.↵
    Genomes Project, C., et al. An integrated map of genetic variation from 1,092 human genomes. Nature 491, 56–65 (2012).
    OpenUrlCrossRefPubMedWeb of Science
  66. 66.↵
    Sudmant, P.H., et al. An integrated map of structural variation in 2,504 human genomes. Nature 526, 75–81 (2015).
    OpenUrlCrossRefPubMed
  67. 67.↵
    Gladman, D.D., Ibanez, D. & Urowitz, M.B. Systemic lupus erythematosus disease activity index 2000. J Rheumatol 29, 288–291 (2002).
    OpenUrlAbstract/FREE Full Text
  68. 68.↵
    Delaneau, O., Marchini, J. & Zagury, J.F. A linear complexity phasing method for thousands of genomes. Nat Methods 9, 179–181 (2011).
    OpenUrlCrossRefPubMed
  69. 69.↵
    Howie, B.N., Donnelly, P. & Marchini, J. A flexible and accurate genotype imputation method for the next generation of genome-wide association studies. PLoS Genet 5, e1000529 (2009).
    OpenUrlCrossRefPubMed
  70. 70.↵
    Marco-Sola, S., Sammeth, M., Guigo, R. & Ribeca, P. The GEM mapper: fast, accurate and versatile alignment by filtration. Nat Methods 9, 1185–1188 (2012).
    OpenUrlCrossRefPubMedWeb of Science
  71. 71.↵
    Harrow, J., et al. GENCODE: the reference human genome annotation for The ENCODE Project. Genome Res 22, 1760–1774 (2012).
    OpenUrlAbstract/FREE Full Text
  72. 72.↵
    Fort, A., et al. MBV: a method to solve sample mislabeling and detect technical bias in large combined genotype and sequencing assay data sets. Bioinformatics (2017).
  73. 73.↵
    Love, M.I., Huber, W. & Anders, S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol 15, 550 (2014).
    OpenUrlCrossRefPubMed
  74. 74.↵
    Bindea, G., et al. ClueGO: a Cytoscape plug-in to decipher functionally grouped gene ontology and pathway annotation networks. Bioinformatics 25, 1091–1093 (2009).
    OpenUrlCrossRefPubMedWeb of Science
  75. 75.↵
    Shannon, P., et al. Cytoscape: a software environment for integrated models of biomolecular interaction networks. Genome Res 13, 2498–2504 (2003).
    OpenUrlAbstract/FREE Full Text
  76. 76.↵
    Delaneau, O., et al. A complete tool set for molecular QTL discovery and analysis. Nat Commun 8, 15452 (2017).
    OpenUrlCrossRefPubMed
View Abstract
Back to top
Posted January 27, 2018.
Download PDF

Supplementary Material

Email
Genomic dissection of Systemic Lupus Erythematosus: Distinct Susceptibility, Activity and Severity Signatures
Nikolaos I. Panousis, George Bertsias, Halit Ongen, Irini Gergianaki, Maria Tektonidou, Maria Trachana, Luciana Romano-Palumbo, Deborah Bielser, Cedric Howald, Cristina Pamfil, Antonis Fanouriakis, Despoina Kosmara, Argyro Repa, Prodromos Sidiropoulos, Emmanouil T. Dermitzakis, Dimitrios T. Boumpas
bioRxiv 255109; doi: https://doi.org/10.1101/255109
Digg logo Reddit logo Twitter logo CiteULike logo Facebook logo Google logo Mendeley logo
Citation Tools

Subject Area