Genomewide Analyses of Pathogenic and Regulatory T Cells of NOD Mice Reveal a Signiﬁcant Difference in DNA Methylation on Chromosome X

c Indian Academy of Sciences

RESEARCH NOTE

Genomewide analyses of pathogenic and regulatory T cells of NOD mice reveal a signiﬁcant difference in DNA methylation on chromosome X

DANG SUN1, QINGSHENG YU2,3,PINGLI3 and JIANYING SHEN4,5∗

1School of Life Sciences, Tsinghua University, Qinghuayuan Road, Beijing 100084, People’s Republic of China 2Alliance PKU Management Consultants, Central North Fourth Ring Road, Beijing 100101, People’s Republic of China 3Beijing Key Lab for Immune-Mediated Inﬂammatory Diseases, Beijing 100029, People’s Republic of China 4Institute of Chinese Materia Medica, China Academy of Chinese Medical Sciences, Beijing 100700, People’s Republic of China 5Department of Diabetes and Metabolic Disease Research, Beckman Research Institute, City of Hope, Duarte, CA 91010, USA

[Sun D., Yu Q., and Shen J. 2016 Genomewide analyses of pathogenic and regulatory T cells of NOD mice reveal a signiﬁcant difference in DNA methylation on chromosome X. J. Genet. 95, 1023–1029]

Introduction suggests that DNA methylation is an important nongenetic component in regulating T1D development (Disanto et al. Reestablishing a well-balanced population of regulatory T 2013). cells (Tregs) and pathogenic T cells (Tpaths) is necessary To uncover the pathological mechanism of T1D, we first for diabetic patients to regain glucose control. However, the performed methylated CpG island recovery assay (MIRA) molecular mechanisms modulating functional differentiation (Rauch and Pfeifer 2010) to analyse the genomewide of Tpaths and Tregs remain unclear. In this study, we anal- + methylation differences among four different CD4 T cell ysed the gene expression and DNA methylation profiling in lines from female nonobese diabetic (NOD) mice, including one Tpath, two Tregs and one control T cell line (Tctrl). one Tpath cell line, two different Treg cell lines and one Dramatic distinct methylation patterns of chromosome X control T cell line (Tctrl). Tpaths were originally isolated from and autosomes were found in Tregs and Tpaths. Combining BDC2.5 T cell receptor (TCR) transgenic NOD mice, cul- with cDNA microarray data, eight differentially expressed tured and activated by a mimotope p79 peptide in vitro genes were identified on chromosome X in Tpaths. Such and maintained as a cell line. These Tpaths could induce methylation patterns and eight identified genes may play an aggressive form of T1D when transferred to NOD/SCID important roles in regulating functional differentiation and/or mice (You et al. 2003). Treg1 is a clone derived from the pre- maturation of these T cells. viously described N206 Treg line, whose function does not Type 1 diabetes (T1D) is an autoimmune disease resulting require cell contact (Chen et al. 2003). Treg2 is another clone from selective destruction of insulin-producing pancreatic derived from the NR206 Treg line, whose function is depen- islet β cells by autoreactive Tpaths. Tregs, including both + + − + dent on both IFN-γ production and cell contact with target Foxp3 CD4 and antigen-expanded Foxp3 CD4 Tregs, are important for preventing autoimmune diseases by sup- cells (Chen et al. 2006). Both of these Tregs inhibit Tpath pressing Tpaths (Kleinewietfeld and Hafler 2014). Hence, proliferation in vitro and prevent T1D in adoptive transfer reestablishing a well-balanced population of Tregs and experiments. In contrast, Tctrls that function neither as Tregs Tpaths is necessary for diabetic patients to regain glu- nor as Tpaths were used as neutral controls (You et al. 2003). cose control. However, the molecular mechanisms modulating functional differentiation of Tpaths and Tregs remain unclear. Epigenetic regulation of gene expression is a Materials and methods dynamic process in which DNA methylation is a primary Cells mechanism (Jaenisch and Bird 2003). Emerging evidence Two regulatory T cell clones (Tregs) were used in this study. ∗ For correspondence. E-mail: [email protected]. Treg1 cells were clone-derived from the previously described Keywords. methylation; cDNA microarray; type 1 diabetes; pathogenic T cells; regulatory T cells; chromosome X; ubiquitin; epigenetics; autoimmune disease; autosome; T cell lines.

Journal of Genetics, DOI 10.1007/s12041-016-0729-8, Vol. 95, No. 4, December 2016 1023 Dang Sun et al.

N206 Treg line (Chen et al. 2003). Treg2 cells were dialysis in buffer containing 150 mM NaCl, 50 mM Hepes, clone-derived from the NR206 Treg line (Chen et al. 2006). pH 7.4, 50% (v/v) glycerol and 5 mM 2-mercaptoethanol. Tpath cells were originally isolated from BDC2.5 T-cell GST-tagged MBD2b protein was kept at −20◦C. Genomic receptor (TCR) transgenic NOD mice. The functionally DNA fragments digested by MseI were ligated with linker. neutral control cells (Tctrl), N79 cells (Chen et al. 2003; Then, after overnight incubation with GST-MBD2b and His- You et al. 2003), function neither as Tregs nor as Tpaths. MBD3L1 proteins, MagneGST beads (2.5 μL) (Promega, These cells were all from female mice and were cultured and Madison, USA) which were preblocked by JM110 bacterial maintained in vitro. Ten days after activation through specific DNA, were added into the reaction mixture and incubated peptides, the cells were used in the study. at 4◦C for 45 min. Beads were washed thrice with wash- ing buffer which contained 700 mM NaCl,10 mM Tris·HCl, pH 7.5, 0.1% Triton X-1001 mM EDTA and 3 mM MgCl . cDNA microarray analysis 2 Qiaquick PCR purification kits (Qiagen, Valencia, USA) Data in the Gene Expression Omnibus (GEO) database were used to elute the methylated CpG-enriched fraction. (accession no. GSE26456) were used for cDNA microarray The number of replicated used for MIRA microarray was analysis. Using Partek Genomics Suite program, the differ- one. entially expressed genes in Tpaths versus Tregs and Tctrls were identified by comparing Tpaths with grouped Tregs and Definition of CpG methylation peaks: Log2 ratios between Tctrls, and the differentially expressed genes in Tregs ver- MIRA-enriched and input DNA samples were generated sus Tctrls and Tpaths were identified by comparing Tregs using NimbleScan software (Roche NimbleGen) and further with grouped Tpaths and Tregs. Statistically significant genes processed by Quantile normalization. Probes with log2 ratios were identified using mixed model analysis with cut-offs of greater than 1 (2-fold enriched) were selected as positive. A a false discovery rate (Benjamini–Hochberg test) P < 0.05 methylation peak was defined as a region with at least two and a fold-change values ≥| ± 1.5|. positive probes covering a minimum length of 250 bp allow- ing one probe gap, and then mapped relative to known tran- Genomic DNA purification and enrichment of methylated CpG scripts defined in the UCSC genome browser mm8 RefSeq sequences database. Methylation peaks falling into 1500 bp relative to transcription start sites were defined as ‘promoter’; methy- Genomic DNA was purified using proteinase K digestion. lation peaks falling within 1500 bp of RefSeq transcript end Methylated DNA enrichment was performed as described sites were defined as ‘downstream’, and those falling within previously with minor modifications (Rauch and Pfeifer gene bodies (from 1500 bp downstream of transcription start 2010). Briefly, a glutathione-Sepharose CL-4B matrix to 1500 bp upstream of the transcript end) were defined as (Amersham Biosciences, Piscataway) was used to elute puri- ‘intragenic’ peaks. fied GST-tagged MBD2b protein with elution buffer, which contained 0.1% Triton X-100, 150 mM NaCl, 50 mM Tris·HCl, pH 8.5 and 20 mM glutathione, at 4◦C. Phos- Functional identification of gene networks and protein signalling phate buffered saline (PBS) was used for dialysis of the pathways: Functional identification of the gene networks eluted GST-MBD2b fraction for 5 h, followed by overnight and canonical signalling pathways was performed using

Figure 1. General methylation patterns of four cell lines. Hierarchical clustering of methylation data. The heat map shows relative methylation differences depicted in red (hypomethylated) and green (hypermethylated) log2 (MI) scores.

1024 Journal of Genetics, Vol. 95, No. 4, December 2016 Different methylation pattern in pathogenic and regulatory T cells

Ingenuity Pathway Analysis program (Ingenuity Systems, Then we defined a methylation peak as a region with Redwood, USA). The eight Tpath-specific hypermethylated at least two positive probes covering a minimum length of and differentially expressed genes on chromosome X were 250 bp, and then compared the methylation peaks among used with the Core Analysis tool. Analysis was performed the four cell lines. On autosomes, fewer methylation peaks using Ingenuity Knowledge Base (genes only) as the refer- appeared in Tpaths compared with Tregs. However, methy- ence set and was limited to direct or indirect relationships. lation peaks of Tpaths were significantly more than those of Tregs on chromosome X (data not shown). Further, methylation peaks detected in only Tpaths or Tregs were Results and discussion defined as either Tpath-specific or Treg-specific hypermethylation peaks, while peaks detected in the other cells but The heat map of the methylation clusters is shown in figure 1, not in Tpaths or Tregs were respectively defined as Tpath- with the two Tregs as one group and the Tpath and Tctrl as specific or Treg-specific hypomethylation peaks. Among the other for the analyses. A total of 2204 methylation peaks the 65 identified Tpath-specific hypermethylation peaks, 23 were positively detected from the CpG methylation array, peaks (35.4%) were on autosomes and 42 peaks (64.6%) with 1421, 1493, 1254, and 1159 methylation peaks identi- were on chromosome X. In contrast, 267 of the 268 Tpath- fied in Treg1, Treg2, Tctrl, and Tpath, respectively. To ver- specific hypomethylation peaks (99.63%) were detected on ify the methylation microarray data, we analysed the DNA the autosomes, and only one hypomethylation peak appeared methylation status of four genes, S1pr1, Hexim1, Dusp18 on chromosome X (0.37%) (figure 3a). Also, 322 of the 323 and Barx2, which were hypermethylated in Tregs but not in identified Treg-specific hypermethylation peaks (99.69%) Tpath by combined bisulphate restriction analysis (COBRA) were located on autosomes. Only one methylation peak was assays and found this data to be consistent with the array data detected on chromosome X (0.31%). Significantly, more (figure 2). Treg-specific hypomethylation peaks (7 of 30, 23.3%) were located on chromosome X (figure 3b). Thus, distinct methylation patterns of chromosome X and autosomes in Tregs versus Tpaths might be important for regulating functional differentiation and/or maturation of these T cells. Most of our knowledge about the pathogenesis of T1D are derived from studies in the NOD mouse, which shares many features with human T1D (Roep et al. 2004). Both female and male NOD mice spontaneously developed T1D; however, females had more invasive and destructive insuli- tis, leading to an earlier onset (12 weeks of age) and greater incidence (80–90%) of diabetes compared with males (20 weeks of age, 10–30%). Gender-specific differences in susceptibility to T cell-mediated autoimmune diabetes are poorly understood. It has been suggested that sex hormones are associated with the sexual dimorphism in the onset of autoimmune diabetes in NOD mice, because long-term administration of androgen or its derivatives to young female NOD mice decreased the incidence of diabetes (Fox 1992; Toyoda et al. 1996). Estrogen increased IL-12-induced activation of STAT4 which may modulate the Th1/Th2 immune balance at early stages of the T cell-mediated autoimmune process and affect the development of diabetes in NOD mice (Bao et al. 2002). DNA methylation was proved to be involved in the initiation and maintenance of chromosome X inactivation (Kaslow and Migeon 1987), which contributed to disorders in self-recognition and autoimmunity (Libert et al. 2010). Many gene promoters have been found to be hypomethylated in autoimmune disorders, such as MET in rheumatoid arthritis (Neidhart et al. 2000), PAD2 in multiple sclerosis (Mastronardi et al. 2007), and p16, SHP1, p15,andp21 Figure 2. Verification of methylation difference in the four cell − in psoriasis (Zhang et al. 2009). Global DNA hypomethy- lines by bisulphite-based COBRA assays. , control digestion with + no BstUI; +, BstUI-digested samples. Digestion by BstUI indicates ation of CD4 T cells has been reported in systemic lupus methylation of the sequence that was tested. erythematosus (SLE). Hypomethylation of specific genes

Journal of Genetics, Vol. 95, No. 4, December 2016 1025 Dang Sun et al. overexpressed in lupus T cells such as ITGAL (CD11a), Therefore, to better understand the unique methylation CD40LG (CD40L), TNFSF7 (CD70), KIR2DL4, IFNGR2, status of the chromosome X and its potential effect on and PRF1 (perforin) was believed to contribute to the devel- gene expression pattern in Tpaths and Tregs, we first iden- opment of autoreactivity and overstimulation of autoanti- tified the location of methylation peaks in genes on chro- bodies (Jeffries and Sawalha 2011). A genomewide DNA mosome X in these cell populations. On chromosome X methylation study of CD4+ T cells of SLE patients revealed in Tpaths, 42 genes were specifically hypermethylated: 38 236 hypomethylated and 105 hypermethylated CG sites, had methylation peaks in their promoter region, two had and genes involved in autoimmunity, such as MMP9 and peaks in intragenic region, and two had peaks in down- PDGFRA, were hypomethylated, whereas genes involved stream region; only one gene (2900008C10Rik) was specifi- in DNA methylation were hypermethylated (Jeffries et al. cally hypomethylated, with its methylation peaks located in 2011). In a streptozotocin-induced T1D rat model, dia- intragenic region. In contrast, on chromosome X in Tregs, betes led to functional methyl deficiency which resulted just one gene (Ophn1) was specifically hypermethylated with in the hypomethylation of DNA in a tissue-specific fash- it methylation peaks located in intragenic region; seven genes ion (Williams et al. 2008). A genomewide DNA methyl- were specifically hypomethylated: six had methylation peaks ation study of monocytes from T1D-discordant monozygous in promoter region, and one (Zfp92) had methylation peaks in twin pairs has identified 132 T1D-associated methylation intragenic region (table 1). Subsequently, we analysed the variable positions (T1D-MVPs), which preceded disease expression of these 51 genes (43 from Tpaths and eight diagnosis (Rakyan et al. 2011). from Tregs) using cDNA microarray data of the four CD4+

Figure 3. Chromosomal distribution of Tpath-specific or Treg-specific methylation peaks and identified gene network. (a) Chromosomal distribution of Tpath-specific hypermethylation (blue) or hypomethylation (red) peaks. Tpath-specific hypermethylation peaks were mainly located on chromosome X, while Tpath-specific hypomethylation peaks were mainly distributed on autosomes. (b) Chromosomal distribution of Treg-specific hypermethylation (blue) or hypomethylation (red) peaks. Treg-specific hypermethylation peaks were mainly located on autosomes, while Treg-specific hypomethylation peaks were mainly distributed on chromosome X. (c) A gene network involved in the eight identified genes. IPA was used to identify biologically related networks from the eight identified genes. Only one network was identified, including seven of the eight genes. Solid and dotted lines denote direct and indirect interactions, respectively. Network compo- nents highlighted in red were upregulated, and those highlighted in green were downregulated (Tpaths versus Tregs and Tctrls, FC ≥1.5 or ≤−1.5, and P < 0.05). The associated network functions included cancer and haematological and immunological diseases.

1026 Journal of Genetics, Vol. 95, No. 4, December 2016 Different methylation pattern in pathogenic and regulatory T cells

Table 1. Differentially expressed genes with hypermethylation or hypomethylation peaks detected on chromosome X in Tpaths and Tregs.

Fold change Gene Methylated region (Tpaths vs Tctrls and Tregs) P value

Tpath-specific hypermethylated genes Pim2 Promoter 2.92222 0.04092 Mcart6 Promoter 1.58576 0.046091 Wdr45 Downstream 1.44907 0.39275 Thoc2 Intragenic 1.41012 0.344131 Utp14a Promoter 1.20748 0.277816 Renbp Downstream 1.13631 0.164487 Taf1 Promoter 1.12545 0.21392 Las1l Promoter 1.11593 0.514542 Uxt Promoter 1.09026 0.742053 Prps1 Promoter 1.04131 0.758049 Upf3b Promoter 1.03925 0.655838 Nkap Promoter 1.01149 0.935704 Araf Promoter −1.07498 0.552556 Htatsf1 Promoter −1.08158 0.188832 Ebp Intragenic −1.11256 0.582235 Flna Promoter −1.13324 0.313217 Rbm3 Promoter −1.17744 0.53547 Prrg1 Promoter −1.19775 0.660164 Rps4x Promoter −1.21376 0.304607 Ndufb11 Promoter −1.24874 0.204381 Nono Promoter −1.33402 0.063087 Idh3g Promoter −1.34627 0.008265 Rpl10 Promoter −1.3923 0.102753 Rpl36a Promoter −1.4012 0.143272 Mpp1 Promoter −1.41681 0.036609 Apex2 Promoter −1.43221 0.049442 Abcb7 Promoter −1.47244 0.376535 Siah1b Promoter −1.56637 0.212923 Aifm1 Promoter −1.57782 0.02524 Pdzd11 Promoter −1.67702 0.049239 Rpl39 Promoter −1.79079 0.149878 Rragb Promoter −1.79391 0.122815 Ercc6l Promoter −1.85092 0.250508 Hmgn5 Promoter −1.85702 0.107097 Naa10 Promoter −1.9306 0.066697 Rab39b Promoter −1.97062 0.151809 Praf2 Promoter −2.14887 0.1464 Magt1 Promoter −2.17832 0.002495 Uprt Promoter −2.56146 0.089979 Phka2 Promoter −3.26021 0.038513 Pola1 Promoter −3.36815 0.028419 Hmgb3 Promoter −5.00736 0.047592 Tpath-specific hypomethylated genes 2900008C10Rik Intragenic Treg-specific hypermethylated genes Ophn1 Intragenic 1.37795 0.161557 Treg-specific hypomethylated genes Gnl3l Promoter 2.14253 0.267149 Rpl10 Promoter 1.29028 0.280126 Ribc1 Promoter 1.21916 0.161692 Zfp92 Intragenic 1.13146 0.819171 Rbm3 Promoter 1.04154 0.696003 Gpkow Promoter −1.26625 0.231716 Snord61 Promoter −1.95121 0.131149

T cell lines, which can be accessed in the Gene Expres- Tpaths versus Tregs and Tctrls (fold-change, i.e., FC ≥ 1.5 sion Omnibus (GEO) database (accession no. GSE26456). or ≤−1.5, and P < 0.05), all of which were Tpath-speciﬁc Among these 51 genes, eight were differentially expressed in hypermethylated genes with methylation peaks detected in

Journal of Genetics, Vol. 95, No. 4, December 2016 1027 Dang Sun et al.

their promoter regions. Among them, the expression level of Mcart6 and Pim2 was upregulated, while the expression of the other six genes (Aifm1, Pdzd11, Magt11, Phka2, Pola1 and Hmgb3) was downregulated in Tpaths. Regard- ing Treg-specific hypermethylated or hypomethylated genes, none was differentially expressed in Tregs versus Tpaths and Tctrls (FC > 1.5 or < −1.5, and P < 0.05; table 1). The functions and the locations on chromosome X of these eight genes were summarized in table 2. Thus, the eight Tpath- specific hypermethylated and differentially expressed genes on chromosome X may play important roles in the differ- ential regulation and functional maturation of Tpaths and Tregs. Function To understand potential roles of these eight genes in T cells, virtual functional pathway analyses using ingenuity pathway analysis (IPA) were performed, and only one gene network was identified. This gene network included seven

exonuclease activity, chromatin binding, etc. of the eight genes identiﬁed, as the function of Mcart6 is -5

currently unknown. The identified network includes functions involved in cancer and haematological and immunological diseases. Notably, six of the seven genes in this network were directly linked to the ubiquitin C (Ubc) gene, suggesting that these genes may play an important role in ubiquitin-mediated protein degradation or posttranscrip- tional modification in critical cellular activities including cell b strand Unknown strand Magnesium ion transmembrane transporter activity strand Calmodulin binding, catalytic activity, etc. strand DNA binding strand Protein C-terminus binding strand DNA binding, electron-transferring-flavoprotein dehydrogenase activity, etc. strand 3 division, proliferation, differentiation, and death (figure 3c). strand ATP binding, kinase activity, etc. − − + + − − −

+ How these eight genes work in T cell function differentiation requires further studies.

Acknowledgements

We thank Wen-Hui Lee for preparing cell samples and Xiwei Wu for help in data analysis. We apologize to colleagues who are not 7455388-7460558 bp, 68809093-68813851 bp, 45828121-45866740 bp, 97818222-97821907 bp, 90550106-90877494 bp, cited here because of space restrictions. 133515655-133572841 bp, 103163423-103207245 bp, 156940098-157036810 bp,

References Pim2 Aifm1 Pola1 Pdzd11 Mcart6 Phka2 Magt1 Hmgb3

Gene symbol Location on chromosome X Bao M., Yang Y., Jun H. S. and Yoon J. W. 2002 Molecular mechanisms for gender differences in susceptibility to T cell-mediated autoimmune diabetes in nonobese diabetic mice. J. Immunol. 168, 5369–5375. Chen C., Lee W. H., Yun P., Snow P. and Liu C. P. 2003 Induction of autoantigen-speciﬁc Th2 and Tr1 regulatory T cells and modulation of autoimmune diabetes. J. Immunol. 171, 733–744. Chen C., Lee W. H., Zhong L. W. and Liu C. P. 2006 Regula- tory T cells can mediate their function through the stimulation of APCs to produce immunosuppressive nitric oxide. J. Immunol. 176, 3449–3460. Disanto G., Vcelakova J., Pakpoor J., Elangovan R. I., Sumnik Z., Ulmannova T. et al. 2013 DNA methylation in monozygotic quadruplets affected by type 1 diabetes. Diabetologia 56, 2093– 2095.

a Fox H. S. 1992 Androgen treatment prevents diabetes in nonobese Chromosome X locations of eight differentially expressed hypermethylated genes in Tpaths and their functions. diabetic mice. J. Exp. Med. 175, 1409–1412. Jaenisch R. and Bird A. 2003 Epigenetic regulation of gene expression: how the genome integrates intrinsic and environmental From VEGA annotation of NCBI Build 37. From Mouse Genome Informatics, http://www.informatics.jax.org. Mitochondrial carrier triple repeat 6 Table 2. Gene name Proviral integration site 2 b Apoptosis-inducing factor, mitochondrion-associated, 1 High mobility group box 3 PDZ domain containing 11 Magnesium transporter 1: Phosphorylase kinase, alpha 2: Polymerase (DNA directed), alpha 1 a signals. Nat. Genet. 33 suppl, 245–254.

1028 Journal of Genetics, Vol. 95, No. 4, December 2016 Different methylation pattern in pathogenic and regulatory T cells

Jeffries M. A., Dozmorov M., Tang Y. H., Merrill J. T., Wren Rakyan V. K., Beyan H., Down T. A., Hawa M. I., Maslau S., Aden J. D. and Sawalha A. H. 2011 Genome-wide DNA methyla- D. et al. 2011 Identification of type 1 diabetes-associated DNA tion patterns in CD4+ T cells from patients with systemic lupus methylation variable positions that precede disease diagnosis. erythematosus. Epigenetics 6, 593–601. PLoS Genet. 7, e1002300. Jeffries M. A. and Sawalha A. H. 2011 Epigenetics in systemic Rauch T. A. and Pfeifer G. P. 2010 DNA methylation profiling using lupus erythematosus: leading the way for specific therapeutic the methylated-CpG island recovery assay (MIRA). Methods 52, agents. Int. J. Clin. Rheumtol. 6, 423–439. 213–217. Kaslow D. C. and Migeon B. R. 1987 DNA methylation stabilizes Roep B. O., Atkinson M. and von Herrath M. 2004 Satisfaction X chromosome inactivation in eutherians but not in marsupi- (not) guaranteed: re-evaluating the use of animal models of type als: evidence for multistep maintenance of mammalian X dosage 1 diabetes. Nat. Rev. Immunol. 4, 989–997. compensation. Proc. Natl. Acad. Sci. USA 84, 6210–6214. Toyoda H., Takei S. and Formby B. 1996 Effect of 5- Kleinewietfeld M. and Hafler D. A. 2014 Regulatory T cells in alpha dihydrotestosterone on T-cell proliferation of the female autoimmune neuroinflammation. Immunol. Rev. 259, 231–244. nonobese diabetic mouse. Proc. Soc. Exp. Biol. Med. 213, 287– Libert C., Dejager L. and Pinheiro I. 2010 The X chromosome in 293. immune functions: when a chromosome makes the difference. Williams K. T., Garrow T. A. and Schalinske K. L. 2008 Type I Nat. Rev. Immunol. 10, 594–604. diabetes leads to tissue-specific DNA hypomethylation in male Mastronardi F. G., Noor A., Wood D. D., Paton T. and Moscarello rats. J. Nutr. 138, 2064–2069. M. A. 2007 Peptidyl argininedeiminase 2 CpG island in multiple You S., Chen C., Lee W. H., Wu C. H., Judkowski V., Pinilla C. sclerosis white matter is hypomethylated. J. Neurosci. Res. 85, et al. 2003 Detection and characterization of T cells specific 2006–2016. for BDC2.5 T cell-stimulating peptides. J. Immunol. 170, 4011– Neidhart M., Rethage J., Kuchen S., Kunzler P., Crowl R. M., 4020. Billingham M. E. et al. 2000 Retrotransposable L1 elements Zhang K. M., Zhang R. L., Li X. H., Yin G. H. and Niu X. P. 2009 expressed in rheumatoid arthritis synovial tissue: association Promoter methylation status of p15 and p21 genes in HPP-CFCs with genomic DNA hypomethylation and influence on gene of bone marrow of patients with psoriasis. Eur. J. Dermatol. 19, expression. Arthritis. Rheum. 43, 2634–2647. 141–146.

Received 5 January 2016, in revised form 5 March 2016; accepted 11 April 2016 Unedited version published online: 15 April 2016 Final version published online: 29 November 2016

Corresponding editor: S. GANESH

Journal of Genetics, Vol. 95, No. 4, December 2016 1029