The Multi-Omics Architecture of Juvenile Idiopathic Arthritis

Juvenile idiopathic arthritis (JIA) is highly heterogeneous in terms of etiology and clinical presentation with ambiguity in JIA classification. The advance of high-throughput omics technologies in recent years has gained us significant knowledge about the molecular mechanisms of JIA. Besides a minor proportion of JIA cases as monogenic, most JIA cases are polygenic disease caused by autoimmune mechanisms. A number of HLA alleles (including both HLA class I and class II genes), and 23 non-HLA genetic loci have been identified of association with different JIA subtypes. Omics technologies, i.e., transcriptome profiling and epigenomic analysis, contributed significant knowledge on the molecular mechanisms of JIA in addition to the genetic approach. New molecular knowledge on different JIA subtypes enables us to reconsider the JIA classification, but also highlights novel therapeutic targets to develop a cure for the devastating JIA.


Introduction
Juvenile idiopathic arthritis (JIA) is a common type of chronic rheumatic diseases affecting children with the age of onset under 16 years, and an important cause of disability. Epidemiological studies showed that incidence rate of JIA ranges from 1.6 to 23 per 100,000 children annually and the prevalence of JIA was about 3.8-400 per 100,000 children in Europe [1]. Girls exhibit a higher incidence rate than boys (10.0/100,000 compared to 5.7/100,000) [1]. Its negative effects on children's physical development, as well as psychiatric development, cause serious damage to the quality of life of affected children, causing pain of active joints, physical disability, anxiety, and depression. There is no cure for JIA. JIA needs aggressive treatment to control its symptoms. For this difficult disease, it is important to understand the underlying molecular mechanisms of the development of JIA systematically through unbiased approaches. In recent years, people have started to gain knowledge about the multiomics architecture of JIA, thanks to the advance of high-throughput omics technologies, which is the focus of this review article.

Genetic Studies of Monogenic Forms of JIA
Considering the poorly known etiologies of JIA and ambiguities of JIA classification, identification of monogenic forms of JIA has gained us significant knowledge on some critical molecular mechanisms (Table 1).
In a recent in vivo functional study on LACC1, Skon-Hegg et al. demonstrated the important role of LACC1 in inflammatory responses [18]. As shown in this study, LACC1 knockout mice showed worse disease in the Citrobacter rodentium model of colitis, and the collagen-induced arthritis models, compared to wild type (WT) mice, while there were no differences identified between LACC1 KO and WT in the dextran sulfate sodium (DSS) model of colitis and the K/BxN model of arthritis [18]. Besides JIA, the pathogenic effects of LACC1 mutation I254V have been correlated with other diseases, e.g., inflammatory bowel disease [19], leprosy [20], and Behçet disease [21]. C284R was also identified of correlation with severe pediatric Crohn's disease in a Saudi family [15].

LRBA
The LPS responsive beige-like anchor protein gene (LRBA) encodes a member of the WDL-BEACH-WD (WBW) gene family, with its expression induced in B cells and macrophages by bacterial LPS [22]. LRBA protein helps to keep intracellular storage of cytotoxic T-lymphocyte-associated antigen 4 (CTLA-4) protein and prevents its degradation [23]. For activation of T lymphocytes, the T cell receptor (TCR) complex requires the costimulation by CD28 after antigen recognition [24]. CTLA-4 transmits inhibitory signals to attenuate T cell activation by competing for the B7 ligands with its homologue CD28 [25,26]. Reduced CTLA-4 activity is likely associated with JIA [27]. Meanwhile the expression of CTLA-4 in CD4+CD28-T-cells in JIA patients is increased in JIA patients and CD28-T cells are not susceptible to inhibition by CTLA-4 [28]. LRBA deficiency is a cause of common variable immunodeficiency (CVID), a group of congenital dysregulatory disorders in immune system. According to the study by Azizi et al., 10.1% of 227 CVID patients show rheumatologic disorders, and JIA is the most common manifestation (3.1%) [29]. Lopez-Herrera et al. reported for the first time that homozygous mutations of LRBA correlated with common variable immunodeficiency-8 with autoimmunity (CVID8) in four consanguineous families with CVID8 [30]. Semo Oz et al. reported a case of LRBA deficiency who initially presented as polyarthritis and was diagnosed with JIA [31].
Gámez-Díaz et al. observed that Lrba knockout (Lrba −/− ) mice showed reduced CTLA4 expression by Tregs and activated conventional CD4+ and CD8+ T lymphocytes, decreased frequency of peritoneal B-1a cells with reduced IL-10 production [32]. Lymphatic tissue showed no obvious signs of autoimmunity. The Lrba −/− mice can produce a normal amount of serum IgM and IgG, generate specific antibody responses after immunization, and produce elevated serum and secretory basal IgA levels [32]. These findings expanded the roles of LRBA protein in dysfunction of immune system, and suggested that LRBA might be important for maintaining autoantigen tolerance.

NFIL3
The nuclear factor, interleukin 3 regulated gene (NFIL3) encodes an important transcription factor in human immune system, regulating cytokine production of type 2 T helper (T(H)2) cells [33]. Research found that the expression of NFIL3 is impaired in Crohn's disease and ulcerative colitis patients, which may shift tolerogenic functions of macrophages to proinflammatory [34]. Schlenner et al. studied two monozygotic twin girls with JIA, using whole exome sequencing, single cell sequencing, and flow cytometry [35]. A novel homozygous mutation, methionine to isoleucine mutation at residue 170 (M170I) in NFIL3, was identified in the patients [35]. The mutation reduced stability of NFIL3 protein. Nfil3-knockout mice had upregulated IL-1β production, and increased susceptibility to arthritis induction [35].

UNC13D
The unc-13 homolog D gene (UNC13D) encodes a protein essential for cytolytic granules secretion at the immunologic synapse [36]. Partial knockdown expression of UNC13D resulted in impaired NK cell degranulation [37]. Schulert et al. reported an intronic mutation in UNC13D, c.117 + 143A > G in a patient with sJIA and recurrent MAS [37]. This mutation is located in a region regulating lymphocyte-specific UNC13D expression, disrupted the binding of NF-κB with a transcriptional enhancer, and is associated with downregulated transcription of UNC13D in PBMCs. A previous study also reported genetic association between the MUNC13-4 (UNC13D) sequence polymorphisms and sJIA/macrophage activation syndrome (MAS) [38]. Hazen et al. reported compound heterozygous mutations of UNC13D correlated with systemic JIA without MAS in an 8-year-old girl with decreased cytotoxic function of NK cells [39].

Genetic Studies of Polygenic Forms of JIA
Besides the above monogenic forms of JIAs, JIA is commonly considered as an immune disorder caused by complex interactions between environmental factors and multiple genetic risk factors [40]. Early studies revealed that siblings or twins of JIA patients have an increased risk up to 15-30%, which is close to that of type 1 diabetes [41]. The age of onset and disease duration among JIA twins are highly consistent, suggesting JIA as a polygenic genetic disease [40].
In recent years, numerous susceptibility loci have been identified through genome-wide association studies (GWAS) of autoimmune diseases. With the application of molecular genetic and genomic technologies in JIA research, some disease-causing genetic variants have been identified in JIA families [42]. Although some risk variants are located in coding regions of genes affecting protein structure and leading to prominent clinical significance, a larger number of risk variants are mapped to noncoding regions containing regulatory elements, which may have an impact on chromatin structure or noncoding RNA binding, and affect gene expression in a tissue-specific or disease state-related manner [42]. The functional genes in these loci have significantly increased the knowledge about the pathogenesis underlying this complex disease.

The Association with HLA
Among the identified JIA genetic loci, the human leukocyte antigen locus (HLA) confers the strongest effect in JIA genetic susceptibility because of its major roles in autoimmune destruction ( Table 2) [43]. The HLA genes locate at Chr6p21.3, including HLA class I and class II genes. Antigens presented by HLA class I molecules are of intracellular origin, while antigens presented by HLA class II molecules are from extracellular proteins. The HLA proteins are by far the most polymorphic products encoded by the human genome. Multiple amino acid variations have been driven by the evolutionary advantage of heterozygosity, allowing the presentation of diverse antigens from emerging pathogens, despite increased risk of many polygenic diseases [44]. In the past few decades, studies have provided multiple lines of evidence that polyarthritis RF-negative subtype has a strong association with HLA Class II alleles DRB1*08 and DPB1*03, and HLA Class I allele A2 [45,46]. Oligoarthritis is also associated with DPB1*02, DQB1*04, DRB1*01, DRB1*08, DRB1*11, and DRB1*13 [47,48]. DRB1*11:03/11:04 and DRB1*08:01 confer susceptibility to both oligoarthritis and polyarthritis RF-negative JIA patients [49]. Interestingly, DRB1*04 and DRB1*07 confer a very strong protective effect to oligoarthritis and DRB1*15:01 is protective against both oligoarthritis and polyarthritis RF-negative [49,50]. Fine-mapping the HLA locus in JIA reveals that extended oligoarthritis and the RF-negative polyarthritis are genetically similar in their HLA associations [51]. HLA-DRB1 amino acid position 13 showed the strongest connection with oligoarthritis and both RF-positive and RF-negative polyarthritis [51].

Polyarthritis RF-Positive
There have been many earlier studies focusing on the connection between polyarticular RF-positive JIA and HLA [52][53][54][55]. Examination of the genotypes of JIA patients by PCR-sequence-specific oligonucleotide probe methodology has shown that HLA-DRB1*04:05 is associated with polyarthritis RF-positive [56]. Within the HLA region, the most significant locus is rs3129769. This locus is near HLA-DRB1, and in a strong linkage disequilibrium (r 2 = 0.88) with the SNP rs660895 reported in rheumatoid arthritis (RA) [57,58]. These SNPs map to the HLA-DRB1*0401 allele.

sJIA
As the one of the common subtypes, accounting for 10-20% of all JIA, knowledge on the pathogenesis of sJIA is still lacking. As the patients with sJIA present with prominent systemic symptoms, such as fever, rash, and synovitis, activation of the autoimmune system is suggested. However, any association between autoantibodies and HLA in sJIA is still lacking [59]. Research to date shows that HLA class II alleles are significantly associated with sJIA [60]. Meta-analysis of six populations with Western European ancestries showed that HLA-DRB1*11 had strong association [60]. The SNP rs151043342 had the strongest association with sJIA among 482 sJIA-associated SNPs in the class II HLA region [60].

Association with Non-HLA Loci
To date, many studies have been performed to test the association between JIA and functional candidate genes with known biological functions or associations with other autoimmune diseases, such as rheumatoid arthritis (RA) and type 1 diabetes (T1D). In addition, a number of JIA-associated non-HLA loci have been identified through the unbiased approach of genome-wide association studies (GWAS). Most of these loci are located at noncoding regions, about a half of which overlap with adult-onset arthritis [42,66].
Due to the low prevalence of JIA, sample size has been a major limitation for GWASs on JIA. To increase the sample size, several GWASs have been conducted with a cohort of mixed JIA subtypes. In the first GWAS on JIA, 279 cases (including 48% persistent oligoarthritis, 29% extended oligoarthritis, 20% RF-negative polyarthritis, and 3% RF-positive polyarthritis) and 184 controls were studied. Besides the most strongly associated SNP (rs2187684) mapped to the HLA region, Hinks et al. identified a significantly associated SNP (rs2358820) located within the gene VTCN1, which encodes a negative regulator of T-cell mediated immune responses [67]. By fine-mapping of the VTCN1 locus, Hinks et al. identified 10 SNPs associated with JIA [67]. Consequently, a GWAS including 388 children with JIA cases and 2500 controls with European ancestry replicated three previously reported loci mapped to the genes PTPN22, IL2RA, and ANTXR2. Besides, they identified a novel locus with the peak SNP (rs953387), an expression quantitative trait locus (eQTL) associated with the expression level of CXCR4 in T-cells (p = 0.0054) and lymphoblastoid cells (p = 0.014) [68]. They also observed four rare variants existed only in JIA patients, not in controls.
The protein tyrosine phosphatase non-receptor type 22 gene (PTPN22) variants have been identified of association with JIA and many other autoimmune diseases [69]. PTPN22 encodes a protein tyrosine phosphatase, which is mainly expressed in cytoplasm of hematopoietic origin cells. The proteins PTPN22, SHP-1 (PTPN6), and PTPN2 may limit TCR proximal signaling by dephosphorylation of Lck Y394 (a key tyrosine residues of Src family) in T-cells [70,71]. Previous studies showed that the counts of effector/memory T-cells increased in Ptpn22 deficient mice, compared with wild-type mice during rest state, but had no influence on naïve T cells [72,73]. Consistently, PTPN22 expression has been shown elevated in effector and memory T-cells, relative to naïve T-cells [74,75]. PTPN22 may thus be important for the regulation of effector but not naïve T-cell activation.
To date, 23 JIA loci, harboring 33 genes, outside of the HLA region have been identified with genome-wide significance (Table 3). To systematically examine potential pleiotropic effects of genes associated with different autoimmune diseases, Li et al. conducted a meta-analysis on GWASs of 10 different pediatric-age-of-onset autoimmune diseases [76]. SNPs at seven loci of ADGRL2, PTPN22, TENM3, ANKRD55, IL2RA, IL21, and ANKRD30A, were identified of association with both JIA and other autoimmune diseases, such as common variable immunodeficiency (CVID), autoimmune thyroiditis (THY), ankylosing spondylitis (AS), celiac disease (CEL), ulcerative colitis (UC), Crohn's disease (CD), and systemic lupus erythematosus (SLE), although some loci with opposite direction of effect [76]. The genes at JIA GWAS loci are enriched in signaling pathways playing key roles in both innate immune responses and adaptive immunity, showing interactions with each other (Figures 1 and 2). Table 3. Genome-wide significant loci (non-HLA) associated with JIA summarized in GWAS Catalog (https://www.ebi.ac.uk/gwas/). JIA associated SNPs listed in GWAS catalog were clumped into independent loci based on the linkage disequilibrium and distance between SNPs.
In the largest GWAS on JIA to date, Hinks et al. identified 14 novel JIA susceptibility loci with genome-wide significance (p < 5 × 10 −8 ) by genotyping 2816 JIA cases with European ancestry (including both oligoarthritis and polyarthritis RF-negative patients) and 13,056 controls with the Illumina Infinium Immunochip [42]. The Immunochip provides cost-effective genotyping solution for both common and rare variants as a helpful tool for immunogenetic study. The Immunochip is suitable to conduct replication of autoimmune and inflammatory diseases, and to fine-map established GWAS loci [79]. Another GWAS study on JIA with a large sample size, including 2751 oligoarthritis and polyarthritis RF-negative patients and 15,886 controls, replicated nine previously reported GWAS loci, and suggested five additional loci with p < 1 × 10 −6 , i.e., PRR9_LOR, ILDR1_CD86, RNF215, LINC00951, and HBP1, but none of these loci reached genome-wide significance [80].
Among the GWAS loci, CD86 and JAK1 are important for signaling pathways of T cell differentiation and proliferation [81]. Many of the oligoarthritis and polyarthritis RF-negative GWAS loci were shared with other autoimmune diseases, suggesting pleiotropic effects of these loci as well as shared molecular mechanisms of autoimmune diseases, such as T-cell receptor activation and signaling. For example, these JIA associated loci have been identified of association with other autoimmune diseases: JAK1 with CEL and multiple sclerosis (MS); PTH1R with CEL; AHI1_LINC00271 with CEL, MS, THY, and T1D; WDFY4 with SLE.
Zervou et al. investigated the nonsynonymous SNP rs34536443 on TYK2 [82], which has been shown of association with various rheumatic diseases. The SNP results in amino acid substitution of Pro1104 to Ala. The study found that the SNP led to the extension of the α-helical segment and alternation of the protein 3D structure, which is likely to affect the function of the TYK2 protein [82]. Couturier et al. performed a study on the association between TYK2 polymorphism and multiple sclerosis in 1366 French patients and 1802 matched controls [83]. The study found that the protective genotype of the TYK2 polymorphism rs34536443 may reduce the activity of TYK2, which may promote the secretion of Th2 cytokines and induce T lymphocyte differentiation toward a Th2 phenotype [83]. These functional effects of TYK2 polymorphism may explain its association with oligoarthritis and polyarthritis RF-negative JIA [42].

Polyarthritis RF-Positive
Some children with polyarthritis RF-positive JIA have the phenotypes of chronic inflammatory arthritis and one or more positive tests of RF and/or anti-citrullinated peptide antibodies (ACPA) [84]. These phenotypes are similar to adult-onset RA. Polyarthritis RF-positive JIA is likely to share clinical features, pathogenesis, and genetic/environmental factors with RA in adults. Previous candidate gene association studies have shown that polyarthritis RF-positive JIA is associated with adult RA-associated loci, i.e., rs10499194 (TNFAIP3), rs2476601 (PTPN22), and rs7574865 (STAT4) [85]. Furthermore, large sample studies on polyarthritis RF-positive provide solid evidence that this infrequent subtype of JIA not only has a lot in common with adult seropositive RA in phenotype, but also has more in common with adult RA in genetics than the other subtypes of JIA.
Hinks et al. recruited 340 polyarthritis RF-positive JIA patients and 14,412 controls from US, UK, Germany, Canada, and Norway, and genotyped these participants using the Immunochip array [57]. This study assessed 44 previously reported non-HLA loci associated with RA, and 27 oligoarthritis/polyarthritis RF-negative JIA loci, to see if these loci were associated with polyarthritis RF-positive JIA in this population. In total, 19 of 44 RA risk loci showed significant association with polyarthritis RF-positive (23 polyarthritis RF-positive associated loci in total) [57,66]. On the other hand, only three of 44 RA risk loci showed significant association with oligoarthritis and polyarthritis RF-negative (27 oligoarthritis/polyarthritis RF-negative associated loci in total) [57,66]. This study provided convincing evidence that polyarthritis RF-positive JIA is more similar to adult RA in genetic architecture than oligoarthritis and polyarthritis RF-negative JIA. In contrast, six loci were shared by polyarthritis RF-positive (out of 23 loci) and oligoarthritis/polyarthritis RF-negative JIA (out of 27 loci) [57].

sJIA
No reported genetic risk factors have been shared by sJIA and other JIA subtypes, suggesting that sJIA may have a special disease process which distinguishes it from other types of JIA. Many candidate gene studies reported sJIA association of a number of SNPs at different genetic loci, such as IL6 [86,87], MIF [88,89], IL10/20 [90,91], IL1 [92,93], MVK, TNFRSF1A [93], CCR5 [94], SLC26A2 [95,96], and TPSN [96], etc. The first GWAS on sJIA included 982 sJIA cases and 431 controls from nine countries [97]. Besides the HLA locus, a novel locus at chr1p36.32 was reported of sJIA association with genome-wide significance (GWS). This locus contains 14 SNPs with GWS, while the index SNP rs72632736 is mapped 263.5 kb upstream of the adherens junction-associated protein 1 gene (AJAP1) [97], which plays a role in cell adhesion and cell migration [98]. The sJIA GWAS also observed 23 novel loci with suggestive evidence (p < 5×10 −6 ), and found no overlapped genetic loci between sJIA and other common subtypes of JIA [97]. More notably, they found treatment-related susceptibility loci in sJIA, i.e., the HLA class II locus. The HLA class II molecules may result in activation of CD4+ T cells. Abatacept has shown promising therapeutic effect particularly in refractory cases with sJIA [99,100] by reducing T-cell activation through costimulatory inhibition.
According to the study by Arthur et al. [101], a SNP in the promoter region of IL1RN was associated with sJIA. In addition, many of the top sJIA-associated SNPs were correlated with IL1RN expression in lymphoblastoid cell lines. It has been observed in sJIA patients that homozygotes of alleles with higher expression of IL1RN have strong association with non-response to anakinra therapy. This discovery has important implications in precision treatment of sJIA.

Transcriptome Study of JIA
Transcriptome study has demonstrated the important roles of specific cell types in JIA pathogenesis of different subtypes, clinical remission, and response to treatment. Differential gene expression analysis uncovered a gradient of order among the JIA subtypes, from healthy controls, to oligoarticular, polyarticular, and sJIA, to Crohn's disease [102]. Approximately 246 genes were identified as significantly upregulated in patients with active sJIA [103], including IL-6 expressed in monocytes and B cells, IL-10 in monocytes, and suppressor of cytokine signaling 3 (SOCS3) in monocytes and T cells. Cell type analyses allow further probe into the pathogenic mechanisms. Hierarchical clustering suggested the division of polyarticular JIA into three distinct subsets [104]. The three subgroups were associated with different monocyte markers, transforming growth factor β (TGF-β)-inducible genes, and immediate early genes, respectively.

Transcriptome Profiling of Neutrophils in JIA
Ter Haar et al. showed that neutrophils played an important role in the early inflammatory phase of sJIA, and the number and inflammatory activity of neutrophils were highly sensitive to the blockade of IL-1 signaling [105]. To examine activated neutrophil subsets, Brown et al. isolated neutrophils from children with active sJIA or clinically inactive disease (CID) [106]. In both groups, neutrophils from sJIA patients exhibited increased S100A8/A9 release upon PMA stimulation compared to control neutrophils. The study also showed that the number of suppressive neutrophils CD16 + CD62L lo increased with nuclear hypersegmentation in the children with sJIA. Transcriptome analysis of purified neutrophils showed that the genes AIM2, IL18RAP, and NLRC4, involved in critical immune system processes were significantly upregulated [106]. These studies increased our understanding on the roles of neutrophils in sJIA.

Transcriptome Profiling of Macrophages in JIA
CD163, a hallmark of macrophage activation [107], is highly expressed in macrophages of patients with active sJIA or macrophage activation syndrome (MAS) [108]. Two microRNAs, miR-125a-5p and miR-181c, which were previously found to be elevated in active sJIA [109], may reduce CD163 expression of macrophages. Both microRNAs were not elevated in macrophages with stable CD163 expression, but were induced in macrophages with increased CD163 mRNA expression. The miR-181s may target directly at CD163 mRNA for degradation. Overexpression of miR-181c inhibits the anti-inflammatory response of CD163 to hemoglobin or the high-mobility box 1 complex [110]. In contrast, transcriptome analysis of miR-125a-5p overexpression identified "cytokine-cytokine receptor interactions" as the most significantly inhibited gene pathway with reduced interleukin 10 Receptor Subunit Alpha (IL10RA), while IL-10 mediation is required for CD163 expression [108].

Transcriptome Profiling of Monocytes in JIA
Comparing JIA patients with controls, CXCL8 (IL-8) was the most significantly expressed gene transcript [111]. The difference in clinical response to methotrexate in patients with JIA was related to the difference in a number of gene transcripts regulated in monocytes [111]. These gene expression profiles could provide the basis for predicting biomarkers of response to methotrexate treatment [111].

Transcriptome Profiling of T Cells in JIA
CD27, CD276, CTLA4, IL2RG, and SLAMF7 are significantly upregulated in patients with sJIA-related lung diseases, suggesting the activation of T cells [112]. By the transcriptome analysis of 579 immune genes of T cells, UBE2L3, IL-6, STAT4, TYK2, TNFAIP3, and PTPN2 were significantly dysregulated in the relapse and remission of JIA cases [113]. Upregulation of HLA class II and another less-appreciated marker of T-cell activation, CD86, was found in the transcriptome profiling of 92% of JIA samples tested for DNA methylation [114]. Super-enhancers related to autoimmune diseases exist in T cells. Pro-inflammatory signals may inhibit the expression of disease-related genes by inhibiting the activity of super-enhancers [115]. Compared with that in sJIA patients, the number of HLA-DR + T cells and T cells coexpressing CD57 and CD16/56 increased in oligoarthritis JIA and polyarthritis JIA patients [116]. In general, transcriptome analysis of T cells is important for understanding the pathogenesis of JIA as well as its symptomatic treatment.

Epigenomics Study of JIA
It is well known that only about 2% of sequence in the human genome encodes proteins [117]. What roles do the remaining 98% noncoding regions play? Studies by genome annotations provided evidence that noncoding sequences can actually be transcribed into functional RNA molecules or bind transcription factors to fine-tune gene transcription in the both physiological and pathological processes [118,119]. In the era of GWAS, investigators discovered that the vast majority of SNPs associated with diseases were mapped to noncoding regions of the human genome, including intronic regions and intergenic regions [120].
Jiang et al. studied whether activation histone marks, i.e., enhancers H3K4me1 and H3K27ac, were enriched in the linkage disequilibrium (LD) blocks, containing 22 SNPs reported in the previous GWAS studies on JIA [121]. Chromatin immunoprecipitation (ChIP) followed by sequencing was used in human neutrophils and CD4+ T cells. The study showed that H3K4me1 and/or H3K27ac were enriched in 15 of 22 known JIA-associated loci. In addition, human CD4+ T cells had 18 regions with H3K4me1 and/or H3K27ac histone marks. The study also investigated noncoding RNAs (ncRNAs) within 5 kb of JIA-associated SNPs in human neutrophils, and found ncRNA transcripts in the rs4705862 and rs6894249 loci as determined by chromatin interaction analysis by paired-end tag sequencing (ChIA-PET) [121]. Consequently, the same research group showed that the JIA-associated LD blocks were enriched with H3K27ac and/or H3K4me1 marks, and these blocks bound with transcription factors in neutrophils [122]. In addition, the genes with critical immune functions, including HLA-DQA1, HLA-DQB2, TRAF1, and IRF1 have long-distance interactions within the LD blocks in human CD4+ T cells as determined by ChIA-PET [122].
The genome-wide analysis by Ellis et al. identified differential DNA methylation of promoter regions in peripheral blood CD4+ T cells from methotrexate-naïve JIA patients, compared to healthy controls, and found that reduced methylation at IL32 was associated with JIA [123]. Ai et al. collected fibroblast-like synoviocytes (FLS) from four early RA and three JIA patients, and used the previously reported 11 longstanding RA and 11 osteoarthritis as controls, to study DNA methylome signature. The study showed that early RA and JIA clustered with longstanding RA, but were different from osteoarthritis. JIA segregated with the longstanding RA group, and formed a subset in the RA super-group. These findings suggested that the inflammatory injuries in JIA and RA could be caused by abnormal methylation, and levels of methylation are related to the types and stages of disease [124].
Histone deacetylases (HDACs) are acetylation 'erasers' that can deacetylate both histone and non-histone proteins. HDACs are involved in cellular signaling, epigenetic regulation, and important in regulating the function of Teff cells and Treg cells [125]. Besides the above mentioned HLA class II molecules related sJIA that may be suitable to use abatacept, studies showed the epigenetic effects of HDAC9 in regulating critical innate immune processes through deacetylation of histone proteins [126,127]. HDAC inhibitors (HDACi) have been extensively studied for the application to the treatment of autoimmune diseases, such as Givinostat, a pan-class I/II HDACi, which is currently being investigated in JIA [128,129], raising the possibility that HDACi represents plausible targeted therapeutic strategies in JIA.

Conclusions
JIA is a group of diseases, highly heterogeneous in terms of etiology and clinical presentation. Multiomics studies make significant contribution to the understanding of the genetic basis and molecular mechanism of JIA pathogenesis. Similar to other complex human diseases, only a small proportion of JIA familial cases can be attributable to single-gene mutations. For the majority of sporadic JIA cases, the development of the disease is shaped by genetic, epigenetic elements, and environmental factors. Though JIA was classified into seven subtypes, phenotypic overlap was observed between subtypes, suggesting shared genetic/epigenetic basis. The relative low prevalence of JIA makes it hard to acquire enough samples to carry out GWAS on each individual subtype. Proper statistical methods for meta-analysis are needed to identify genetic loci shared by JIA subtypes by taking into account of pleotropic effects and potentially different effect directions between subtypes, as well as small sample sizes. In addition to genetic components, epigenetic modifications also contribute to JIA pathogenesis and development. The shared and distinct epigenetic regulations between JIA subtypes are even less understood, and the interactions between genetic and epigenetic mechanisms warrant further investigation. Integrative analyses of genetic, epigenetic, and transcriptome data are needed to elucidate the full picture of underlying molecular mechanisms of JIA subtypes and to classify JIA cases by molecular markers, and to further facilitate drug development and drug repositioning. New molecular knowledge on different JIA subtypes encourages us to reconsider the JIA classification, but also highlights novel therapeutic targets to develop a cure for the devastating JIA.

Conflicts of Interest:
The authors declare no conflict of interest.