Whole-genome transcription and DNA methylation analysis of peripheral blood mononuclear cells identified aberrant gene regulation pathways in systemic lupus erythematosus

Multiple studies on a variety of cell types have demonstrated altered gene expression and DNA methylation patterns in autoimmune diseases, providing important new directions toward understanding the pathogenesis of immune-mediated diseases [27, 28]. A number of studies have compared gene expression in PBMC from patients with SLE versus healthy individuals or other autoimmune conditions, which provide substantial advantages in the search for diagnostic and prognostic biomarkers in autoimmune disease [29, 30]. But LN remains a major challenge and one of the most severe manifestations of SLE. Early diagnosis and intervention are essential for favorable patient outcomes [31]. It would be of great help to identify biomarkers so that we could assess LN activity or kidney damage, and predict an imminent exacerbation of renal disease accurately [2].

In this study, we performed an unbiased genome-wide gene transcription and DNA methylation study in PBMC from SLE patients with LN and without LN and compared with matched normal controls. By cross comparison of different SLE groups with NC, we identified a group of genes (552 up- and 550 downregulated) which are consistently differentially expressed in SLE as compared with controls. DNA methylation analysis on the whole-genome level distinguished a large number of CpG sites which were differentially methylated (1813 hypermethylated and 3785 hypomethylated sites) in SLE compared to NC. With integrated analysis of the gene transcription and DNA methylation data, we found that 60.5 % of the upregulated genes contain hypomethylated CpG sites and 87.1 % downregulated genes harbor hypermethylated CpG sites in their regulatory regions of the genome, suggesting that DNA methylation may have regulatory effect on most of the differentially expressed genes in SLE. Profiling mRNA expression and DNA methylation in the same set of SLE and NC samples enable us to investigate the complexity of gene dysregulation in the context of this disease.

Further analysis was performed to elucidate the potential molecular pathways modulated by the aberrantly expressed and methylated genes in SLE. The current analysis was mainly focused on the common upregulated genes with DNA hypomethylation. To our expectation, the upregulated genes in SLE were significantly enriched in immune regulatory pathways such as IFN signaling pathway, TLR signaling pathway, and proinflammatory signaling pathways, which have been reported to be related with pathogenesis in SLE and other autoimmune disorders [32–34]. It is well known that the overexpression of type I IFN-related genes in peripheral blood immune cells is a major SLE signature [35]. However, the pathogenic role of IFNs and their exact regulation in the development of SLE is still under debate. By exploring the globe gene transcription and DNA methylation in PBMC of SLE with different clinical manifestations (with or without LN), our study provided further evidence that the overexpression of a cluster of IFN genes was involved in the activation of the IFN signaling pathway. In addition, we noticed a stepwise increased expression of nine IFN genes from NC to SLE LN^? to SLE LN⁺, suggesting a pathogenic role of the IFNs in the development of SLE and LN, even though the difference between SLE LN^? and SLE LN⁺ was not statistically significant for all IFNs except for MX1. Interestingly, almost all the overexpressed IFN genes in SLE were associated with multiple hypomethylated CpG sites in the regulatory region of the genes, indicating epigenetic modification may play a major role in the regulation of IFN gene expression. Previous studies have reported the hypomethylation of IFN genes in the CD4+ T cells in SLE [13]. Our study detected the hypomethylated CpG sites in association with upregulation of IFN genes in PBMC of SLE, suggesting the hypomethylation may be a general regulating mechanism in all immune cells in PBMC, including T, B, macrophages, and dendritic cells. Further studies need to be performed on different cell types and different stages of the diseases to elucidate the exact molecular mechanisms of IFN gene regulation.

Toll-like receptor pathway is another molecular pathway identified in our study to be enriched with dysregulated genes in SLE. Seven upregulated genes in both SLE LN^? and SLE LN⁺, including TLR1, TLR4, TLR7, TLR8, CD14, EIF2AK2 and IRAK3, were involved in the regulation of TLR pathways. TLR regulation plays an essential role in the pathogenesis of SLE [36]. It is believed that abnormal activation of TLR7/TLR8 or TLR9 by self RNA or DNA in antigen-presenting cells may cause breach of immune tolerance and results in cytokine, IFN-I, and autoantibody production in T, B, and myeloid cells [37–39]. TLR4 have also been shown to induce TNF-? production in monocytes of SLE patients [40, 41]. Substantial studies have been focused on the TLR signaling pathway for exploration of therapeutic target using adjuvants or inhibitors [42, 43]. Our data revealed the aberrant DNA methylation of TLR and its signaling pathway in SLE, implying a novel therapeutic strategy by intervention of methylation in SLE.

Serum cytokines/chemokines are important biomarkers reflecting immune cell activation and inflammatory reaction during the development of SLE [44]. In this study, we measured the serum levels of 27 pre-inflammatory cytokines/chemokines in SLE patients with or without lupus nephritis and compared them with healthy controls. Our analysis identified 15 cytokines/chemokines which significantly increased in the sera of both SLE LN^? and SLE LN⁺ patients, as compared with NC. The increased levels of Th1 cytokines (IL-2, IL-12, IL-15, TNF-a), and Th2 cytokines (IL-4, IL-5, IL-6, IL-10,) reflecting the activation of both Th1 and Th2 T helper cell subsets, which is in agreement with previous studies showing that IL-12, IL-10, IL-6 and TNF-? were correlated with active SLE [45–48]. We also observed the increased level of IL-17, which is secreted by Th17 cells and involved in T cell activation and autoantibodies in SLE [49–51]. It is not surprising that we could not find significant difference on any of the serum cytokines between SLE LN⁺ and SLE LN^? in our sample set. Although several cytokines, including MCP-1, IP-10 and IL-6, have been reported to have increased expression in the kidney and urine of SLE patients with active LN compared with no renal SLE, the serum level were not different between SLE patients with or without LN [52], implying that the serum levels of inflammatory cytokines/chemokines are not reflecting the LN activity in SLE patients.

Our data also showed that RANTES (CCL5) and platelet-derived growth factor-BB (PDGF-BB) were significantly lower in both SLE LN^? and SLE LN⁺ compared with NC. RANTES, as a proinflammatory chemokine, has been extensively studied and reported to be increased in SLE patients [53, 54]. The discrepancy of our data with previous studies may be due to sample variation and different method used. PDGF-BB is a platelet-derived factor involved in cell proliferation. Its role in SLE has not been reported before.

The regulation of the serum cytokines/chemokines in SLE on the genome level was not fully understood. By analyzing the mRNA expression and the DNA methylation of the cytokine/chemokine genes, we were able to find out if the increased serum level of cytokines in SLE is associated with their mRNA transcription and DNA methylation in the PBMC of the patients. To our surprise, we only found two out of the 17 differentially expressed cytokines, the IL-15 and RANTES, in which the mRNA expression in PBMC correspond with their serum levels. The discrepancy between serum protein level and mRNA for most of the cytokines/chemokines may be due to the cytokine gene expression regulation is cell-type specific and the differential expression may not be detected in PBMC, which is a mixture of different cell types. On the other hand, the increased mRNA transcription on IL-15 in SLE PBMC suggesting the transcription of this cytokine may be activated in multiple cell types. Indeed, IL-15 can be produced by macrophage/monocytes, activated NK cells, CD4+ T cells, and B cells, and involved in the pathogenesis of SLE and other autoimmune diseases [55–57]. Furthermore, the hypomethylation on several CpG sites in the regulatory region of this gene suggested a strong epigenetic regulation of IL-15 in SLE. In addition, our study showed that the lower level of RANTES in serum is correlated with lower mRNA expression in PBMC in SLE patients. Interestingly, the hypermethylated CpG sites at the 5?UTR region of CCL5 (RANTES) gene in SLE patients also indicated a suppressive effect of epigenetic regulation on this gene. This data is currently under validation in a larger sample cohort.

LN is one of the most devastating complications of SLE, but the molecular basis for this disease is not yet clear. Gene expression profiling on PBMC and different subsets of blood cells have identified candidate genes that might contribute to the pathogenesis of SLE and its complications [7, 58, 59]. However, the role of the differentially expressed genes identified from peripheral blood of SLE patients in the pathogenesis of LN is still under debate. By comparing the gene expression profiles in PBMC from SLE LN^? and SLE LN⁺ patients, we identified three genes which are significantly elevated in SLE LN⁺ patients, namely, MX1, GPR84 and E2F2. MX1 is one of the IFN-I-inducible genes which have been extensively studied in SLE. Increased MX1 gene expression has been detected in both PBMC and renal intrinsic cells of lupus nephritis patients suggesting its role in the pathogenesis of LN [60, 61]. Our data consolidated previous finding by detecting the upregulation of MX1 on transcription in PBMC and in serum in SLE LN⁺ patients. Furthermore, by analyzing DNA methylation of this gene, we postulated that the higher expression of MX1 in SLE may be regulated by hypomethylation of the CpGs in the regulatory region.

E2F2 and GPR84 are novel biomarkers which have not been previously reported to be associated with SLE or LN. E2F2 is a member of E2F transcription factor family. As a cell cycle activator, E2F2 can regulate the gene expression that govern the cell cycle and DNA synthesis, and thus cell proliferation [62, 63]. There was a study that showed that E2F2 can promote the expression of microRNA let-7a by binding to its promoter and let-7a has been shown to contribute to hyperplasia and the proinflammatory response in SLE [64]. Therefore, the increased expression of E2F2 may associate with the inflammation in SLE and LN. GPR84 is a G protein-coupled receptor for sensing free fatty acid and active inflammatory reaction. Activation of GPR84 by its ligand could elicit chemotaxis of polymorphonuclear leukocytes (PMNs) and macrophages and amplified lipopolysaccharide (LPS)-stimulated production of the proinflammatory cytokine IL-8 from PMNs and TNF-? from macrophages [65]. Further analysis on the expression of these two genes in different cell types and the molecular pathways is warranted to elucidate their role in the pathogenesis of SLE and LN.

Although our current analysis was primarily focused on the genes which were upregulated and hypomethylated in SLE, the identification of a large amount of downregulated genes which were hypermethylated in the SLE genome by globe gene expression and DNA methylation analysis implying that downregulation or silencing of gene expression in PBMC may play an equally important role in SLE and LN [66–70]. Previous studies have demonstrated that the downregulation of IL-2 [35, 67, 71], IL-17 [66], and Notch-1 [69] in SLE T lymphocytes was due to the binding cAMP-responsive element modulator (CREM)-? to the promoter of the genes followed by enhanced CpG DNA methylation. We therefore examined methylation status on the CpG sites of these genes in our study. By comparing the mean ? value between NC and SLE (both with LN and without LN involvement), we found that four out of 70 sites on Notch-1 gene, two out of ten sites on IL-17A gene and one out of two sites on IL-2 gene were hypermethylated in SLE (p??0.05) (data not shown). Further analysis on the SLE downregulation gene data set is going on with the expectation to find out more dysregulated pathways modulated by DNA hypermethylation.

There are some limitations in the current study. First, the gene expression and DNA methylation profiling were only performed in PBMC. Given the highly complex nature of this disease, different cell subsets in peripheral blood may have different transcription and methylation profiles during the development of the diseases. However, previous studies using different cell types have identified specific gene signatures from different cell types. These cell type-specific genes allowed a correct classification of PBMC independent from their heterogenic cellular composition. Therefore, based on the expression of cell type-specific signature genes, we should be able to normalize the expression profile of PBMC from all individuals [72]. On the other hand, the advantage of using PBMC includes easy accessibility and less blood volume requirement. The result from PBMC has reflected a general picture of profiling alteration in whole blood. It is therefore ideal for biomarker screening [73]. The second limitation is that the sample size in this study is relatively small. The statistical power may not be strong enough to draw conclusions on some of the differential genes and pathways. The current data will need to be further validated in a larger sample cohort. Third, functional analysis will need to be conducted to confirm the genetic and epigenetic interactions on the dysregulation of genes and pathways in SLE.