A20/Tumor Necrosis Factor α-Induced Protein 3 in Immune Cells Controls Development of Autoinflammation and Autoimmunity: Lessons from Mouse Models

Immune cell activation is a stringently regulated process, as exaggerated innate and adaptive immune responses can lead to autoinflammatory and autoimmune diseases. Perhaps the best-characterized molecular pathway promoting cell activation is the nuclear factor-κB (NF-κB) signaling pathway. Stimulation of this pathway leads to transcription of numerous pro-inflammatory and cell-survival genes. Several mechanisms tightly control NF-κB activity, including the key regulatory zinc finger (de)ubiquitinating enzyme A20/tumor necrosis factor α-induced protein 3 (TNFAIP3). Single nucleotide polymorphisms (SNPs) in the vicinity of the TNFAIP3 gene are associated with a spectrum of chronic systemic inflammatory diseases, indicative of its clinical relevance. Mice harboring targeted cell-specific deletions of the Tnfaip3 gene in innate immune cells such as macrophages spontaneously develop autoinflammatory disease. When immune cells involved in the adaptive immune response, such as dendritic cells or B-cells, are targeted for A20/TNFAIP3 deletion, mice develop spontaneous inflammation that resembles human autoimmune disease. Therefore, more knowledge on A20/TNFAIP3 function in cells of the immune system is beneficial in our understanding of autoinflammation and autoimmunity. Using the aforementioned mouse models, novel A20/TNFAIP3 functions have recently been described including control of necroptosis and inflammasome activity. In this review, we discuss the function of the A20/TNFAIP3 enzyme and its critical role in various innate and adaptive immune cells. Finally, we discuss the latest findings on TNFAIP3 SNPs in human autoinflammatory and autoimmune diseases and address that genotyping of TNFAIP3 SNPs may guide treatment decisions.

Immune cell activation is a stringently regulated process, as exaggerated innate and adaptive immune responses can lead to autoinflammatory and autoimmune diseases. Perhaps the best-characterized molecular pathway promoting cell activation is the nuclear factor-κB (NF-κB) signaling pathway. Stimulation of this pathway leads to transcription of numerous pro-inflammatory and cell-survival genes. Several mechanisms tightly control NF-κB activity, including the key regulatory zinc finger (de)ubiquitinating enzyme A20/tumor necrosis factor α-induced protein 3 (TNFAIP3). Single nucleotide polymorphisms (SNPs) in the vicinity of the TNFAIP3 gene are associated with a spectrum of chronic systemic inflammatory diseases, indicative of its clinical relevance. Mice harboring targeted cell-specific deletions of the Tnfaip3 gene in innate immune cells such as macrophages spontaneously develop autoinflammatory disease. When immune cells involved in the adaptive immune response, such as dendritic cells or B-cells, are targeted for A20/TNFAIP3 deletion, mice develop spontaneous inflammation that resembles human autoimmune disease. Therefore, more knowledge on A20/TNFAIP3 function in cells of the immune system is beneficial in our understanding of autoinflammation and autoimmunity. Using the aforementioned mouse models, novel A20/TNFAIP3 functions have recently been described including control of necroptosis and inflammasome activity. In this review, we discuss the function of the A20/TNFAIP3 enzyme and its critical role in various innate and adaptive immune cells. Finally, we discuss the latest findings on TNFAIP3 SNPs in human autoinflammatory and autoimmune diseases and address that genotyping of TNFAIP3 SNPs may guide treatment decisions.
Keywords: A20, tumor necrosis factor α-induced protein 3, nF-κB, ubiquitination, autoinflammation, autoimmune disease, mouse models, single nucleotide polymorphisms inTRODUCTiOn Autoinflammatory and autoimmune diseases share a spectrum of chronic immune system disorders (1). Autoinflammatory diseases are rare and occur due to innate immune cell dysfunction with increased cytokines such as interleukin (IL)-1β and tumor necrosis factor (TNF) α (2,3). In contrast, autoimmune diseases are caused by adaptive immune cell dysfunction and affect millions of people A20/TNFAIP3 in Immune Cells in Autoinflammation and Autoimmunity Frontiers in Immunology | www.frontiersin.org February 2018 | Volume 9 | Article 104 worldwide (4). Self-reactive T-cells and/or autoreactive antibodies facilitate responses against harmless tissue (5). Essential for development of these diseases is the activation status of immune cells, wherein nuclear factor-κB (NF-κB) plays a key role. NF-κB activation is tightly controlled by several mechanisms, including the key regulatory (de)ubiquitinating enzyme A20 or tumor necrosis factor α-induced protein 3 (TNFAIP3) (6). Genetic studies have demonstrated the association of TNFAIP3 single nucleotide polymorphisms (SNPs) with multiple human diseases (7), such as systemic lupus erythematosus (SLE) (8-10), rheumatoid arthritis (RA) (9), and Crohn's disease (CD) (11,12). A20/ TNFAIP3 regulates crucial stages in immune cell homeostasis, such as NF-κB activation and apoptosis. Recently, new functions have become apparent, including the control of necroptosis and inflammasome activity (13)(14)(15). Here, we review the latest understanding of A20/TNFAIP3 as a key regulator of immune signaling and its cell-specific role in the pathogenesis of autoinflammation and autoimmunity as demonstrated in murine models.

nF-κB PATHwAY nF-κB Activation
An important and well-characterized signaling pathway of immune cell activation is the NF-κB pathway (7), which is activated through canonical or non-canonical cascades (16). The canonical pathway is triggered by several pattern recognition receptors (PRRs), such as toll-like receptors (TLRs) and nucleotide oligomerization domain (NOD)-like receptors (NLRs) and cytokine receptors, such as TNF receptor (TNFR) and IL-1 receptor (16). PRRs are essential within the innate immune response in defense against invading pathogens. In addition, T-cell receptor (TCR) or B-cell receptor (BCR) triggering, crucial in the adaptive immune response, also leads to NF-κB activation (17). In total, five NF-κB family members have been identified thus far, termed p65 (RelA), RelB, c-Rel, NF-κB1, and NF-κB2 (18). These five members can form homoor heterodimers and distinctive NF-κB dimers bind different DNA-binding sites, resulting in cytokine release, enhanced cell survival, proliferation, differentiation, and changes in metabolism (18,19).

Regulation of nF-κB Activity
Several regulatory mechanisms control NF-κB signaling to maintain tissue homeostasis. One of the proteins that terminate NF-κB signaling is A20/TNFAIP3 (6). A20/TNFAIP3 regulates protein ubiquitination, an important post-translational modification (6). Ubiquitination is reversible and tightly controlled by opposing actions of ubiquitin ligases and deubiquitinases (DUBs) (20). Several ubiquitin chains are known, each having specific functions. Lysine (K)48-linked polyubiquitin chains target a protein for proteasomal degradation, whereas K63linked or linear polyubiquitin chains stabilize protein-protein interactions important for downstream signaling molecules (16). Interestingly, A20/TNFAIP3 has both ligase and DUB activity to perform both K48 ubiquitination and K63 deubiquitination (6).
In short, in macrophages, A20/TNFAIP3 regulates IL-1β/ IL-18 release by controlling NLRP3 inflammasome activity and CXCL9/CXCL10 production through STAT1 signaling. Both pathways are essential in controlling the autoinflammatory arthritis phenotype. However, a role for neutrophils and/or DCs in the pathogenesis of arthritis cannot be excluded.
Summarizing, the expression of co-stimulatory molecules, pro-inflammatory cytokines such as IL-6, and anti-apoptotic proteins in DCs is controlled by A20/TNFAIP3. A20/TNFAIP3 in DCs functions to maintain in vivo T-cell and B-cell homeostasis, thereby preventing spontaneous autoinflammation.
Differences in T-cell-specific Tnfaip3 deletion between the two mouse strains could indicate that either CD8 + T-cells drive inflammation in Tnfaip3 maT-KO mice or CD4 + T-cells have increased regulatory function in Tnfaip3 CD4-KO mice. Indeed, regulatory T cell (Treg) proportions were increased in Tnfaip3 CD4-KO mice, because of a reduced IL-2 dependence for their development (93). In vitro activated CD4 + T-cells from Tnfaip3 CD4-KO mice died quicker than wild-type T-cells (14,15), due to A20/TNFAIP3's control on necroptosis (14) and autophagy (15). Necroptosis is RIPK3-dependent programmed cell death (94). Increased necroptosis in A20/Tnfaip3-deficient CD4 + T-cells impaired Th1 and Th17-cell differentiation in vitro (14). Interestingly, perinatal death of Tnfaip3 KO mice was greatly delayed by RIPK3 deficiency, implying that A20/TNFAIP3 may control necroptosis in other cell types (14), such as CD8 + T-cells (95). Preventing necroptosis did not fully restore survival of A20/Tnfaip3-deficient CD4 + T-cells (14), which could be attributed to autophagy, a lysosomal degradation pathway necessary for survival after TCR stimulation (96). Autophagy is regulated by mechanistic target of rapamycin (mTOR), which is increased in Tnfaip3-deficient CD4 + T-cells after TCR stimulation (15). Consequently, treatment with an mTOR inhibitor improves survival by enhancing autophagy (15). mTOR inhibitors are effective in murine SLE and RA (97), but should not be used in patients with A20/TNFAIP3 alterations, as it may improve pathogenic T-cell survival.
In conclusion, in CD4 + T-cells, A20/TNFAIP3 regulates necroptosis and autophagy. In contrast to conventional Th-cells, Treg development is restricted by A20/TNFAIP3. In CD8 + T-cells, A20/TNFAIP3 regulates necroptosis, IL-2, and IFNγ release, of which IFNγ might have contributed to a further undefined lung and liver inflammatory phenotype in Tnfaip3 maT-KO mice.

A20/TnFAiP3 in AUTOinFLAMMATORY AnD AUTOiMMUne PATienTS
TNFAIP3 is one of the few genes that has been linked by genomewide association studies (GWAS) to multiple immune diseases (106,107). The list of common coding and non-coding variants (SNPs) in the vicinity of the TNFAIP3 gene region associated with autoimmune conditions keeps expanding, with recently reported associations with autoimmune hepatitis (AIH) (108,109), primary biliary cirrhosis (110) and colitis ulcerosa (111). Since a comprehensive overview of SNPs within and around the TNFAIP3 gene has been provided elsewhere (7), we focus on a selection of SNPs with known different functional, clinical, and therapeutical consequences (Figure 2). We also discuss a recently described monogenic disease "Haplo-insufficiency of A20 (HA20)" (112), which clearly illustrates the importance of functional A20/TNFAIP3 protein expression levels (Figure 2).

TNFAIP3 SnPs and novel Mutations Affecting A20/TnFAiP3 expression and Function
Reduced TNFAIP3 mRNA expression was observed in peri pheral blood mononuclear cells (PBMCs) in SLE and RA patients (115)(116)(117) and in disease-affected organs, e.g., in colon or skin biopsies from CD and psoriasis patients compared to healthy tissues (118)(119)(120). In RA synovium, reduced A20/TNFAIP3 protein expression was detected compared to non-autoimmune osteoarthritic synovium (121). SNPs near the TNFAIP3 gene can result in reduced A20/TNFAIP3 mRNA expression and consequently protein concentrations. For instance, specific SNPs associated with SLE ("TT>A", Figure 2H) are situated in an enhancer region of the TNFAIP3 gene and hamper DNA looping, resulting in reduced TNFAIP3 mRNA expression (122) and reduced A20/ TNFAIP3 protein expression in B-cells (8).
Recently, novel rare familial TNFAIP3 mutations (Figures 2B,G) causing HA20 have been described (112). These mutations lead to severely reduced functional A20/TNFAIP3 protein expression (112,123). HA20 is a dominantly inherited disease caused by high-penetrance heterozygous germ line (mostly nonsense or frameshift) mutations in TNFAIP3 (112). Previously, A20/ TNFAIP3 loss-of-function mutations were only identified as somatic variants in lymphomas (105) [reviewed in Ref. (124)]. HA20-associated mutations were first reported in seven unrelated families with an early-onset inflammatory disease resembling the common polygenic Behçet disease (112). Some patients diagnosed with Behçet-like disease were found to have similar HA20 mutations (125,126). Recently, in a Japanese cohort the majority (59%) of HA20 patients did not fulfill the criteria of Behçet disease (127). In this study, a genotype-phenotype correlation was not observed (127). However, careful evaluation of clinical characteristics can aid diagnosing patients with HA20 or Behçet disease (128). Autoimmune diseases such as autoimmune lymphoproliferative syndrome (ALPS) and SLE were additionally recognized in HA20 patients (113,123,127). Excess Th17-cell differentiation was also observed in HA20 patients (127). All HA20 patients identified thus far have a strong inflammatory signature as demonstrated by elevated levels of many pro-inflammatory cytokines (e.g., IL-1β, IL-6, TNFα, IL-17, and IFNγ) and most patients respond to treatment with cytokine inhibitors (anti-TNF and anti-IL-1) (112,127,128). Interestingly, Tnfaip3 +/− mice do not have an overt inflammatory phenotype despite elevated inflammatory cytokines (e.g., IL-1β and IL-6) in serum (129) and brain (130). Nevertheless, Tnfaip3 +/− mice are more susceptible to experimental psoriasis (120) and atherosclerosis (129), but these specific symptoms are not commonly reported for HA20. Increased NLRP3 activity was detected in PBMCs of HA20 patients after LPS stimulation, leading to elevated IL-1β (112). Transfection of mutant-truncated A20/TNFAIP3 prolonged NF-κB activation due to reduced deubiquitinating function (112) (Figure 2B). PBMCs of a patient with HA20 also demonstrated prolonged NF-κB activation (112,123). Mutant-truncated A20/ TNFAIP3 proteins do not exert a dominant-negative effect on protein function, and this indicates that sustained NF-κB activation in HA20 is due to haploinsufficiency rather than an aberrant protein function (112). It remains unclear whether missense high penetrance mutations may have a different impact on A20/ TNFAIP3 function.
Two SNPs, rs5029941 (A125V) and rs2230926 (F127C), are located in close proximity of each other near the C103 catalytic site in the OTU domain and result in non-synonymous coding changes in the A20/TNFAIP3 protein (Figures 2D,E). The rs2230926 (F127C) SNP, associated with multiple autoimmune diseases (Figure 2E), hampers A20/TNFAIP3 function after TNFα stimulation (10). The SNP location within the OTU domain ( Figure 2E) suggests that the K63-deubiquitinating efficacy is decreased, although this was not evaluated. The A125V mutation ( Figure 2D) results in reduced DUB activity and was shown to impair A20-mediated degradation and deubiquitination of TRAF2 (131). Although the A125V mutation was associated with protection from SLE, surprisingly the same allele was associated with increased risk of IBD (131).
In conclusion, specific SNPs functionally alter A20/TNFAIP3 expression or function, and HA20 is a disease with generalized inflammation due to severely reduced functional A20/TNFAIP3 protein expression.

TNFAIP3 SnPs Affecting Disease Progression and Treatment in Patients
Common, presumably hypomorphic, variants in TNFAIP3 can have clinical consequences. For instance, lower TNFAIP3 mRNA expression in PBMCs correlates with SLE disease activity as susceptibility to lupus nephritis is increased (115). SLE or SSc patients with an intron SNP ( Figure 2C) predisposes for increased risk for either renal involvement (132) or aggravated FiGURe 2 | Overview of single nucleotide polymorphisms (SNPs) and novel haplo-insufficiency of A20 (HA20) mutations in the proximity of TNFAIP3 which are highlighted in this review. TNFAIP3 gene SNPs, adapted with permission from: Springer Nature, Nature Reviews Immunology, A. Ma & B.A. Malynn © 2012 (7). Exons contributing to the OTU domain are depicted in green, and exons forming the zinc finger (ZnF) domains are blue. Non-coding exons are gray. The catalytic C103 site, ZnF4, and ZnF7 are highlighted. Black triangles indicate all known SNPs in the TNFAIP3 gene with associations with autoimmune diseases. Among the various documented SNPs/novel mutations, several lead to (1) reduced A20/TNFAIP3 protein level, (2) reduced A20/TNFAIP3 efficiency, (3) altered disease prognosis, or (4) therapeutic implications and are thus highlighted in this figure (A-H). Known associations with (autoinflammatory and autoimmune) diseases for SNPs are indicated in the top gray bar. Multiple novel mutations causing "HA20" and two SNPs termed "TT>A" (associated with SLE) are listed in the box in the lower left corner. The reported p.Gln415fs mutation (113) should be reported as p.Lys417Serfs*4 to stay consistent with Human Genome Variation Society nomenclature (114). Abbreviations: OTU, ovarian tumor; ZnF, zinc finger; TF, transcription factors; TNFAIP3, tumor necrosis factor α-induced protein 3; HA20, haploinsufficiency of A20; AIH, autoimmune hepatitis; SLE, systemic lupus erythematosus; SSc, systemic sclerosis; RA, rheumatoid arthritis; T1D, type 1 Diabetes; JIA, juvenile idiopathic arthritis; CD, Crohn's disease; Pso, psoriasis; SS, Sjögren syndrome. disease with fibrosing alveolitis and pulmonary hypertension (133). Similarly, RA patients with a previously described functional SNP (Figure 2E) had more swollen joints and increased disease activity scores (DAS28) compared to RA patients without this SNP, indicating worse clinical prognosis (9,117). Finally, AIH patients with an upstream SNP (Figure 2A) harbored increased liver enzymes and more cirrhosis at disease presentation compared to patients without this SNP (109). These findings illustrate that within autoimmune patients certain SNPs around the TNFAIP3 gene predispose a worse clinical prognosis. Analysis of TNFAIP3 SNPs might guide treatment choices, e.g., with TNF-blocking therapy. For RA and CD patients, reduced TNFAIP3 mRNA in PBMCs or colonic biopsies, A20/TNFAIP3 in Immune Cells in Autoinflammation and Autoimmunity Frontiers in Immunology | www.frontiersin.org February 2018 | Volume 9 | Article 104 respectively, is correlated with effective TNF-blocking therapy (118,134). Psoriasis patients harboring specific TNFAIP3 SNPs (Figures 2E,F) respond more effectively to TNF blockade (135). This indicates that TNFAIP3 SNP analysis before TNFblocking therapy initiation is worthwhile to perform in several autoimmune diseases and may be more practical than evaluating TNFAIP3 mRNA expression.

Treatment of Autoinflammation and Autoimmunity
Knowledge from cell-specific targeting studies in mice illustrate that loss of A20/TNFAIP3 results in either autoinflammation or autoimmunity. The pathophysiologic distinction between these conditions has therapeutic implications. Autoinflammatory diseases such as Still's disease, Behçet's disease, and most cases of HA20 are well treated with IL-1 blockade, which has only marginal effect in autoimmune diseases including RA (136). Autoinflammation may also underlie other chronic disorders such as atherosclerosis, as these patients benefit from anti-IL-1 therapy (137,138). In contrast, autoimmune disorders (e.g., SLE) have a strong contribution of IL-6 highlighted by successful anti-IL-6 treatment (139). This is in line with mouse studies in which innate cell activation (e.g., Tnfaip3 LysM-KO mice) leads to increased IL-1β (13) and adaptive immune cell activation (e.g., Tnfaip3 CD19-KO mice) leads to enhanced IL-6 (70,71,100,102). In line with the adaptive nature of the disease, several autoimmune diseases also improve after treatments targeting adaptive immune cells [e.g., T-cell suppression using cyclosporine (140,141) or B-cell depletion using Rituximab] (142).

COnCLUSiOn
Control of immune system activation is crucial to prevent both autoinflammation and autoimmunity. A20/TNFAIP3 hereby plays an important role in several innate and adaptive immune cells. Through analysis of cell-specific deletion of A20/Tnfaip3 in mice, it became apparent that innate myeloid cells require A20/ TNFAIP3 to suppress autoinflammation, while the development of autoimmunity is primarily controlled by A20/TNFAIP3 in DCs and B-cells. In addition, novel functions of A20/TNFAIP3 on inflammasome activity and necroptosis are uncovered. It would be of great value to examine in patient material cell-specific profiles of A20/TNFAIP3 and its effector function. The direct consequence of many SNPs on A20/TNFAIP3 is yet unknown. However, it is becoming increasingly clear that specific TNFAIP3 SNPs can alter A20/TNFAIP3 function, can affect its expression level, or are associated with poor clinical outcomes. Finally, future studies on TNFAIP3 SNPs to predict therapeutic effectivity would greatly benefit patient health care to obtain personalized therapy.

AUTHOR COnTRiBUTiOnS
All the authors listed have made a substantial, direct, and intellectual contribution to the work and approved it for publication.

FUnDinG
This project was supported by grants from the Dutch Arthritis Association (12-2-410) and European Framework program 7 (FP7-MC-CIG grant 304221).