Since genome-wide patterns of DNA methylation are known to differ between cell-types 
, and different cell-types in the immune system are implicated in the pathogenesis of psoriasis, we isolated and studied CD4+
cells to overcome the issue of epigenetic heterogeneity in whole blood. Both cell-types are relatively abundant in blood and have important functions in the immune system, thus they are good targets for epigenetic alterations which might contribute to the development of psoriasis. Comparisons of CD4+
cells revealed many significant differences for both DNA methylation (1,288 of the 26,690 CpG sites, 4,8%, n
12, Table S1
) and gene expression (2,126 of the 37,846 transcripts, 5,6%, n
13, Table S2
) in unaffected individual twins (), which clearly demonstrate the importance of isolating specific cell-types.
Volcano plots of cell-type specific differences in DNA methylation and gene expression.
MZ co-twins are highly correlated for DNA methylation in both CD4+
17 pairs, mean ρ
0.96–0.99) and CD8+
13 pairs, mean ρ
0.95–0.98) (). Both analyses of individual CpG sites and mean methylation per gene did not reveal any differentially methylated sites between unaffected and affected co-twins. Individual scatter plots of DNA methylation clearly demonstrate greater similarity among the MZ twins than among unrelated individuals (Figure S1
). To ensure that the observed differences in DNA methylation between co-twins were genuine rather than technical artifacts, we ran internal replicates on a subset of the twins. Specifically, we replicated 7 pairs and calculated technical and biological differences between replicated samples (self-self comparisons) and between co-twins, respectively. The overall distributions of the technical differences in DNA methylation per CpG site were significantly smaller than the biological differences (Kolmogorov-Smirnov test, two-sided, P
, Figure S2
). This clearly shows that the observed biological differences between unaffected and affected co-twins are genuine, although they are small. Similarly, differences in gene expression between MZ co-twins were small in both CD4+
17 pairs, mean r
0.97–0.99) and CD8+
14 pairs, mean r
0.98–0.99). Although there are many small differences, we did not detect any genome-wide significant differences in DNA methylation or gene expression between co-twins discordant for psoriasis when analyzed separately ().
Volcano plots of differences in DNA methylation and gene expression in discordant MZ twins.
DNA methylation is essential for the regulation of gene expression. We reasoned that a combined analysis of DNA methylation and gene expression could select functional methylation sites involved in regulating gene expression. We therefore investigated if co-twin differences in DNA methylation and gene expression were correlated. To do this, we compared the differences (between co-twins) in mean β-values for the CpGs associated with each gene, with the log fold changes of the gene expression. Using Spearman's rho as a measure for correlation, we then ranked the genes according to the significance of the correlation coefficients. This combined analysis of DNA methylation and gene expression revealed cell-type specific differences, identifying genes known to be involved in immune response and associated with psoriasis, only in CD4+
cells. shows the top 50 genes ranked according to the significance of the correlation in CD4+
cells. Genes associated with psoriasis (shown in bold) are overrepresented in this list (Fisher's exact test, P
, Table S3
). IL13 have been identified in GWAS 
, and ALOX5AP 
, PTHLH 
and TNFSF11 
have all been linked to different aspects of the disease. The entire list of the 11,933 genes studied ranked according to the significance of the correlation in CD4+
can be found as Table S4
in supporting information. Scatter plots depicting the relationship of MZ co-twin differences in DNA methylation and gene expression of TNFSF11 demonstrate the strong correlation in CD4+
cells compared to a non-significant correlation in CD8+
Subset of genes with a correlated difference in DNA methylation and gene expression.
Scatter plots of differences in DNA methylation against the differences in gene expression of TNFSF11.
We used DAVID 
to explore the potential of shared biological pathways among the genes in the list we generated from the combined analysis of DNA methylation and gene expression in CD4+
cells (Table S4
). GO analyses identified significant enrichment of GO terms in CD4+
cells (), whereas the analysis did not detect any enriched terms in CD8+
cells. A significant portion of the top 1% of the genes ranked at the top of the list were found to be involved in the immune response (GO: 0006955, 12.5%, P
0.042), positive regulation of response to stimulus (GO: 0048584, 7.5%, P
0.037), immune system process (GO: 0002376, 15%, P
0.043) and regulation of response to stimulus (GO: 0048583, 10.8%, P
0.043). All of these categories contain genes involved in the immune response, which are potentially important in autoimmune diseases. Interestingly, a subset of the genes identified in the GO analysis (IL13, IL23R, CCL1, CCL5, CSF2, TNFSF11, LTB and SF9) comprises part of the cytokine-cytokine receptor interaction pathway, which is essential in communication between cells in the immune system. Skewed cytokine levels of pro-inflammatory and anti-inflammatory cytokines characterize the pathogenesis of psoriasis. Cytokines and chemokines are essential in the communication between cells in the immune system. Whereas cytokines generally influence proliferation, differentiation and secretion of pro- or anti-inflammatory factors, chemokines primarily have an effect on the movement of cells 
. Thus, pathways like the cytokine-cytokine receptor interaction are indeed relevant in the etiology of psoriasis. In this context, our results suggest that DNA methylation is important in regulation of the cytokine cascade and signaling pathways involved in psoriasis.
Gene ontology results of significantly enriched GO terms identified in a combined analysis of DNA methylation and gene expression in CD4+ cells.
cells are known to be present in psoriatic plaques, and current evidence indicates that CD4+
cells play a more critical role than CD8+
. Our data strongly suggest that CD4+
cells are important in the pathogenesis of psoriasis. However, in this context it is important to recognize the complexity of CD4+
cells, which consists of several subpopulations with specific roles in the immune system (i.e. upon activation, naïve CD4+
cells develop into several lineages; Th
1 cells, Th
2 cells, Th
22 and T regulatory cells). Recently, much attention has been drawn towards Th
17 cells and the role in the pathogenesis of psoriasis 
. It has also been speculated that CD4+
cells at different differentiation states may be present, which complicates the picture even further 
. Distinctive compositions of these subpopulations can potentially contribute to the observed intra-pair differences. In addition, the complexity of the CD4+
cells could explain the small intra-pair differences by averaging out the level of DNA methylation and gene expression. Thus, a disease-associated change in DNA methylation and gene expression in a subset of cells can ultimately appear as an overall small difference.
Recently, several genes and pathways associated with psoriasis have been identified in GWAS 
. Many of these have an essential role in the immune system and this clearly demonstrates the importance of immune response regulation in the disease. The molecular mechanisms driving the inflammation in skin causing psoriasis are complex. Our findings identify new potential susceptibility genes and point to different plausible biological pathways in psoriasis that are under epigenetic regulation and suggest an epigenetic dysregulation of biological pathways implicated in immune function. It will be important to expand on these findings in larger twin and other non-twin cohorts.