Post‐transcriptional control of immune responses and its potential application

Abstract

Inflammation is the host response against stresses such as infection. Although the inflammation process is required for the elimination of pathogens, uncontrolled inflammation leads to tissue destruction and inflammatory diseases. To avoid this, the inflammatory response is tightly controlled by multiple layers of regulation. Post‐transcriptional control of inflammatory mRNAs is increasingly understood to perform critical roles in this process. This is mediated primarily by a set of RNA binding proteins (RBPs) including tristetraprolin, Roquin and Regnase‐1, and RNA methylases. These key regulators coordinate the inflammatory response by modulating mRNA pools in both immune and local nonimmune cells. In this review, we provide an overview of the post‐transcriptional coordination of immune responses in various tissues and discuss how RBP‐mediated regulation of inflammation may be harnessed as a potential class of treatments for inflammatory diseases.

Introduction

The immune response against pathogens is typically initiated by the engagement of pattern recognition receptors such as Toll‐like receptors (TLRs) and their cognate molecules on macrophages and dendritic cells.1 Downstream immune signal transduction leads to the activation of pivotal transcription factors such as nuclear factor‐kappa B (NF‐κB), activator protein 1 (AP‐1) and interferon regulatory factors (IRFs). This induces a rapid and robust transcriptional activation and production of proinflammatory cytokines and interferons (IFNs) including interleukin‐6 (IL‐6) and tumor necrosis factor (TNF), thereby facilitating the activation and recruitment of other immune cells and the elimination of pathogens. The signal transduction machineries and modulators of signalling have been intensively studied over the past two decades, providing us with a near‐complete picture of transcriptional regulation in inflammation.1, 2

While robust transcriptional regulation plays a pivotal role in the initiation of inflammation, it is now known that post‐transcriptional mechanisms also operate to fine‐tune the final output of inflammatory cytokines and mediators in immune cells such as macrophages and T cells.3, 4, 5 Post‐transcriptional regulation is a type of gene expression control at the mRNA level, which includes the regulation of mRNA stability, translation, modification, localisation as well as alternative splicing. These systems, mediated by multiple RNA binding proteins (RBPs), coordinately control inflammatory gene expression in immune cells, providing an efficient way to cease immune responses during the resolution of inflammation. Moreover, recent studies have revealed that post‐transcriptional programmes also regulate the diverse functions of nonimmune cells, thereby mediating the interplay between immune cells and nonimmune cells to facilitate the elimination of pathogens in the local tissue. Here, we present an overview of our understanding of post‐transcriptional regulation of immune responses in local tissues and discuss the potential applications of the treatment of inflammatory and autoimmune diseases, especially focusing on recent advances.

Inflammation‐related mRNAs are regulated by a set of RNA binding proteins

Many mRNAs encoding inflammatory mediators (inflammatory mRNAs) are known to be relatively unstable, enabling the fine‐tuning of gene expression and inflammatory responses.6 The rapid turnover of such unstable mRNAs is achieved by a variety of sequences/structures present in their 3′ untranslated region (UTR) termed cis‐elements. These cis‐elements are recognised by a set of RBPs, which positively or negatively affect the stability of mRNAs (Figure 1).

Figure 1.

RNA binding proteins shape the fate of immune‐related mRNAs. Inflammatory mRNAs harbor multiple cis‐elements. AU‐rich element (ARE)‐containing mRNAs are destabilised by TTP, AUF1 and ZFP36L1, 2, while HuR is responsible for their stabilisation. Stem‐loop‐containing mRNAs are recognised and degraded by Regnase‐1 and Roquin‐1, 2. ARID5A is proposed to counteract the function of Regnase‐1, thereby stabilising stem‐loop‐containing mRNAs. RNA modification is another type of cis‐element. m⁶A RNA modification is deposited by the METTL3/14 complex in the nucleus. This modification may be erased by demethylases FTO and ALKBH5. Subsequently, m⁶A‐containing mRNAs are recognised by YTHDF1‐3, thereby controlling their translation and decay.

Cis‐elements in the 3′ UTR of inflammatory mRNAs

There are several canonical cis‐elements which are frequently found in unstable inflammatory mRNAs. Among these, the best‐characterised motifs are AU‐rich elements (AREs) and stem‐loop structures. AREs typically contain a stretch of adenine and uridine sequences, typically represented as a repetitive AUUUA motif.7 These motifs are found in the 3′ UTR of mRNAs of proinflammatory genes, including IL‐6, TNF, IL‐2, cyclooxygenase 2 (COX2), IFN‐γ and granulocyte–macrophage colony‐stimulating factor (GM‐CSF). AREs are bound and recognised by ARE‐binding proteins TTP (tristetraprolin, otherwise known as ZFP36), ZFP36L1, ZFP36L2, AU‐rich binding factor 1 (AUF1), human antigen R (HuR), and KH‐type splicing regulatory protein (KHSRP). Most of these ARE‐binding proteins promote the degradation of target mRNAs, while HuR is known to stabilise ARE‐containing mRNAs.8

The stem‐loop structure is a secondary structure of mRNA which forms a hairpin‐like motif. Compared to AU‐rich elements, stem‐loop structures are found in partly overlapping sets of inflammatory mRNAs. Well‐characterised stem‐loop‐containing mRNAs are as follows: IL6, TNF, IL2, inducible T‐cell costimulator (ICOS), NF‐κB inhibitor zeta (NFKBIZ) and NF‐κB inhibitor delta (NFKBID). This class of motifs includes constitutive decay elements (CDEs) which are targeted by Roquin‐1,2 (RC3H1, 2). Stem‐loop structures are recognised and degraded by a set of RBPs, Regnase‐1 and Roquin‐1,2, and this process is known to be counteracted by another RBP, AT‐rich interactive domain‐containing protein 5A (Arid5a). A recent study has demonstrated that Regnase‐1 and Roquin recognise an overlapping set of stem‐loop structures, suggesting a synergistic function of these proteins.9

The repertoire of cis‐elements is still expanding. For instance, via genome‐wide identification of Roquin target mRNAs, Roquin has been demonstrated to bind to CDEs, which canonically form triloop stem‐loop structures.10 In addition to this structure, recent reports using systematic evolution of ligands by exponential enrichment (SELEX) and photoactivatable ribonucleoside‐enhanced cross‐linking and immunoprecipitation (PAR‐CLIP) methods have shown that Roquin‐1 also recognises noncanonical hexaloop structures and linear binding elements (LBEs), respectively.11, 12 Thus, our understanding of cis‐elements remains incomplete and further studies are necessary to comprehensively identify and validate potential RBP binding sites.

Secondary structures which serve as cis‐elements may be dynamically altered during immune responses. Recently, a global profiling study of mRNA secondary structures found that viral infection leads to a change in the secondary structures of 3′ UTR of cellular mRNAs.13 These structurally dynamic regions are enriched in RBP binding sites and regulate RBP‐mediated mRNA stabilisation upon viral infection. Thus, cis‐elements may be dynamically generated or abolished during immune responses, allowing for the intricate regulation of immune‐related transcripts in a manner dependent on stimulation. It would be of interest to investigate the roles of RNA helicases in this context.

Canonical RBPs

Canonical RBPs harbor one or more RNA binding domains such as the KH domain and the CCCH‐type zinc finger domain. In a similar fashion, a set of immune‐related RBPs harbor CCCH‐type zinc finger domains.14 Key examples of such RBPs include members of the ZFP36 family,15 Roquin‐1,216 and Regnase family (ZC3H12A‐D, MCPIP1‐4 or Regnase‐1‐4).17

TTP recognises AREs in the 3′ UTR of target mRNAs via its two zinc finger domains and destabilises them by recruiting the carbon catabolite repression 4‐negative on TATA‐less (CCR4‐NOT) deadenylation complex via an evolutionarily conserved C‐terminal motif.18 TTP deficiency in mice leads to autoimmune symptoms, with cachexia and arthritis resulting from excess TNF production.19 The tandem zinc finger domain of TTP is conserved in other ZFP36 family members, ZFP36L1 and ZFP36L2, which also recognise ARE‐containing mRNAs and induce the deadenylation and/or translational repression of the latter. These RBPs are redundantly required for thymocyte development; the paucity of these proteins during T‐cell development leads to T‐cell acute lymphoblastic leukaemia via dysregulation of Notch1 expression.20 Additionally, ZFP36L1 and ZFP36L2 are reportedly critical for maintaining the quiescent state before the expression of the precursor B‐cell receptor during B‐cell development.21 Furthermore, ZFP36L1 and ZFP36L2 have unique roles in the regulation of lymphocytes. On the one hand, ZFP36L1 is required for the maintenance of marginal zone B cells by suppressing transcription factors including Krüppel‐like factor 2 (KLF2) and IRF8.22 On the other hand, ZFP36L2 suppresses production of IFN‐γ in memory T cells by binding and inhibiting the translation of pre‐existing Ifng mRNAs via AREs.23

Roquin‐1,2 harbors a RING finger domain and a unique RNA binding motif, ROQ domain, as well as a CCCH‐type zinc finger domain. Roquin promotes the decay of target mRNAs via the recruitment of CCR4‐NOT deadenylation complex.10 A point mutation in the ROQ domain (M199R) of Roquin‐1 induces a lupus‐like autoimmune disease in mice.24 This pathology is primarily mediated by upregulation of Icos mRNA in follicular helper T cells. Conditional deletion of both Roquin‐1,2, but not either protein individually, in T cells reveals a similar phenotype, suggesting a functional redundancy in these proteins.25

Regnase family members possess PIN‐like ribonuclease domains in addition to CCCH‐type zinc finger domains. Among these family members, Regnase‐1 (Zc3h12a) is the best‐characterised protein. Regnase‐1 recognises stem‐loop structures in the 3′ UTR of inflammation‐related mRNAs such as Il6, Ptgs2 and Regnase‐1 itself and destabilises these mRNAs via its endonucleolytic activity in a translation‐dependent manner.9, 26, 27 Thus, Regnase‐1 is critical for degrading inflammation‐related mRNAs. Regnase‐1 exerts its activity in the control of both innate and adaptive immune systems. Regnase‐1‐deficient macrophages produce increased amounts of cytokines in response to stimulation by TLR ligands.26 In CD4 T cells, Regnase‐1 downregulates the expression of Icos, Il2, Ox40 and c‐Rel, thereby inhibiting effector T‐cell function.28

Zc3h12d, otherwise known as transformed follicular lymphoma (TFL), is also reportedly involved in post‐transcriptional regulation of immune‐related transcripts.29 Zc3h12d was shown to promote the destabilisation of IL2, IL6, IL10, IL17 and TNF mRNA via their 3′ UTRs through unknown mechanisms. Although they did not develop spontaneous inflammatory disease unlike Regnase‐1‐deficient mice, Zc3h12d‐deficient mice showed exacerbated paralysis in an experimental autoimmune encephalomyelitis (EAE) model. Further studies are needed to determine the mRNA degradation mechanisms of this family member.

While Regnase‐1 and Roquin recognise an overlapping set of mRNAs via stem‐loop structures, they play a nonredundant role in the control of immune‐related mRNAs. This view is underscored by the observation that the deletion of both proteins in CD4 T cells in vivo leads to the exacerbation of inflammatory disease found in either Regnase‐1‐ or Roquin‐defective mice.30 Mechanistically, Regnase‐1 function requires up‐frameshift protein 1 (UPF1) and active translation, conditions which are not necessary for Roquin‐mediated mRNA decay.9 Moreover, Regnase‐1 colocalises with the endoplasmic reticulum (ER), while Roquin localises in P body/stress granules, where translation inactive mRNAs are stored and degraded. These findings indicate that Regnase‐1 and Roquin regulate the same set of target mRNAs in a spatiotemporally distinct manner.

Although the findings above provide mechanistic insights into the functions of Regnase family members, the roles of other members, particularly ZC3H12B and ZC3H12C, are not well understood. Very recently, it has been reported that ZC3H12B also regulates IL‐6 and Ier3 in vitro, which are targets of Regnase‐1.31 Also, ZC3H12C is associated with an increased risk of psoriasis,32 although the functional role remains to be clarified. Given conservation of the PIN‐like RNase domain, it is conceivable that these family members may also play a pivotal role in regulating immune responses and this warrants further investigation.

Noncanonical RBPs

Recently, attempts to globally identify RNA binding proteins reveal that more than 1500 genes are potentially associated with RNA.33, 34 While canonical RBPs often harbor a combination of several RNA binding domains, some of the identified proteins do not have an RNA binding domain. Instead, their enzymatic core or protein–protein interaction sites could be responsible for RNA interaction.35 Here, we focus on representative noncanonical RBPs that are reported to function in immune regulation.

It is known that IL‐17 signalling regulates the stability of chemokine mRNA.36 An adaptor for the IL‐17‐receptor, NF‐κB activator 1 (Act1), has been proposed to be an RBP involved in this regulation process.37 Although Act1 lacks canonical RNA binding domains, Act1 SEFIR [SEF (similar expression to fibroblast growth factor genes) and IL‐17R] domain, which is responsible for binding to IL‐17 receptor, is reported to serve as an RNA binding domain. The Act1 SEFIR domain binds to a stem‐loop structure called SEFIR‐binding element (SBE) in Cxcl1 mRNA, thereby increasing the expression of this chemokine. The abrogation of this interaction in vivo using RNA aptamers leads to the amelioration of skin inflammation or airway inflammation in respective disease models.37 Moreover, Act1 regulates Cxcl1 mRNA stability by recruiting other RBPs. It has been reported that Act1 forms a complex with TNF receptor‐associated factor 2 (TRAF2), TRAF5 and splicing factor 2 (SF2) upon the stimulation of IL‐17.38 This complex formation induces the disengagement of Cxcl1 mRNA from SF2, thereby stabilising this mRNA. In another line of study, Act1 mediates polyubiquitination of HuR, leading to the stabilisation of Cxcl1 mRNA.39 Thus, Act1 may regulate the stability of mRNA via multiple mechanisms.

Arid5a is another noncanonical RBP that binds to stem‐loop elements.40 Arid5a harbors an AT‐rich interaction domain which was previously annotated as a DNA binding domain.41 Arid5a expression is induced upon the stimulation of macrophages by LPS, IL‐6 and IL‐1β and stabilises Il6 mRNA by counteracting Regnase‐1‐mediated mRNA decay.42 It is also reported that Arid5a binds to T‐box‐containing protein expressed in T cells (T‐bet) and signal transducer and activator of transcription 3 (Stat3) mRNA to enhance the inflammation.43, 44 Moreover, Arid5a is reported to promote translation of Nfkbiz and Cebpb mRNA.45 Thus, these studies demonstrate that Arid5a is a positive regulator of immune responses which antagonises Regnase‐1 function. It would be interesting to globally identify Arid5a binding sites at a transcriptome‐wide level to gain insights into its binding motifs and to compare them with those of Regnase‐1 and Roquin.

As exemplified here, noncanonical RBPs also participate in the regulation of immune responses. Further studies will expand the list of noncanonical RBPs which are critical for immune regulation.

RNA modifications deposited by RBPs

It has long been recognised that the nucleoside and the base of mRNA can be modified46 to potentially serve as cis‐elements of mRNAs; however, the role of mRNA modification was relatively unclear until recently. Rapid improvements in the detection of RNA modifications allow us to examine mRNA modification more precisely, even at a single nucleotide resolution.47 One of the most abundant mRNA modifications is N⁶‐methyladenosine (m⁶A), which is decorated on transcripts of more than 7000 human genes.48 The functions of m⁶A modification are diverse and include the regulation of mRNA decay, translation and splicing.49 m⁶A modification is deposited by the methylation complex formed by METTL3 and METTL14, together with several auxiliary factors, and can be removed by the demethylases AlkB homolog 5 (ALKBH5) and fat mass and obesity‐associated (FTO). m⁶A‐modified mRNA is recognised by a set of m⁶A‐binding ‘reader’ proteins including YTH domain‐containing proteins (YTHDF1‐3, YTHDC1‐2). It is reported that this modification plays a critical role in the mRNA degradation of immune‐related transcripts.50 For example, m⁶A methylation promotes the destabilisation of Socs1, Socs3 and Cish mRNA in CD4 T cells, which are negative regulators of STAT signalling. The deletion of Mettl3 or Mettl14 in CD4 T cells leads to the downregulation of the IL‐7‐STAT5 pathway, thereby preventing their homeostatic proliferation and differentiation into effector cells.50 It is also reported that the deletion of Mettl3 in regulatory T (Treg) cells leads to severe autoimmune disease due to the loss of immunosuppressive functions.51 Therefore, the m⁶A modification is crucial for T‐cell homeostasis activation and regulatory functions.

Another report showed that METTL3 plays a critical role in the differentiation and proliferation of human hematopoietic stem/progenitor cells (HSPCs) and acute myeloid leukaemia cells.52, 53 The overexpression of METTL3 or METTL14 in HSPCs blocks cell differentiation and promotes cell growth. The authors also found that m⁶A marks are deposited on mRNAs of oncogenic proteins such as c‐MYC, MYB, PTEN and BCL2, leading to the upregulation of translation efficiency and/or stability of these mRNAs. Since the expression of METTL3 and METTL14 is downregulated during myeloid differentiation, the dynamic control mediated by m⁶A deposition is critical for proper differentiation of HSPCs into myeloid cells.

Recently, several groups have demonstrated that m⁶A deposition regulates the production of IFN‐β.54, 55 The functional abrogation of METTL3 or METTL14 leads to the overproduction of IFN‐β and IFN‐stimulated genes (ISGs) in response to double‐stranded DNA (dsDNA) exposure, and better control of human cytomegalovirus infection, while the knockdown of the m⁶A demethylase ALKBH5 has the opposite effect. Mechanistically, Roni and colleagues reported that m⁶A deposition mediates the destabilisation of IFNB mRNA by YTHDF2.55 In contrast, Rubio and colleagues described a reduction of IFN‐β production upon YTHDF1 or YTHDF2 knockdown.54 Detailed investigation will be necessary to reconcile this apparent discrepancy. Nevertheless, these findings demonstrate a pivotal role of m⁶A deposition in the control of infection. In another line of study, YTHDF3 inhibits the expression of ISGs via the translational activation of forkhead box O3 (FOXO3), a transcription corepressor, although YTHDF3 recognises its target mRNA in an m⁶A‐independent manner.56 Thus, further research is needed on the role of m⁶A readers in the setting of infection.

In summary, these findings suggest that m⁶A RNA modification is a multifunctional mark in controlling inflammation in various immune cell types. It would be interesting to investigate the role of this modification in other immune cells or in other settings such as chronic inflammation.

RBPs orchestrate immune responses via the regulation of immune and nonimmune cells

Immune cells infiltrate various tissues to eliminate and/or tolerate exogenous antigens. To efficiently mount immune responses, the milieu surrounding immune cells plays an important role. The expression of RBPs is also dynamically regulated in local nonimmune cells such as the epithelial lining, thereby facilitating the function of immune cells (Figure 2). However, the dysregulation of immune and nonimmune cells leads to a variety of pathological conditions such as autoimmune diseases. Here, we will summarise the physiological and pathophysiological contributions of RBPs in each tissue especially focusing on several very recent findings in this context.

Figure 2.

Regulation of the local immune response by RBPs. (a) RBP‐mediated pathogen control in the lung. The stimulation of airway epithelial cells leads to the degradation of Regnase‐1. CCL28 and pIgR mRNAs are then derepressed, thereby facilitating the recruitment of plasma cells and promoting the secretion of IgA into the airway. (b) RBP‐mediated systemic regulation of iron. During iron deficiency, Regnase‐1 in the duodenum promotes the uptake of dietary iron by indirectly activating HIF2α. Additionally, Regnase‐1 in macrophages decreases the expression of transferrin receptor (TfR1). Both of these actions by Regnase‐1 result in increased transferrin‐bound iron in the blood stream.

Lung

The airway is the place where inhaled pathogens or antigens and immune cells come into contact with each other. It is reported that airway epithelial cells are regulated by Regnase‐1 to control inhaled pathogens.57 The intratracheal administration of heat‐killed Pseudomonas aeruginosa, a clinically relevant pathogen, induces the reduction of Regnase‐1 expression in the lung and the sustained expression of Regnase‐1‐target genes such as Il6 and Il12b. The deletion of Regnase‐1 in airway epithelial cells confers protection against P. aeruginosa infection, suggesting that the downregulation of Regnase‐1 is beneficial to control pulmonary infection. Mechanistically, this protection could be attributed to the augmented expression of Regnase‐1 targets, Ccl28 and Pigr, which recruits plasma cells to the lung and promotes the secretion of IgA, respectively. These findings suggest that RBP‐mediated regulation controls the crosstalk between immune and nonimmune cells, facilitating the elimination of pathogens in the airway. Given that the downregulation of Regnase‐1 enhances the immune responses, it would also be interesting in the future to investigate its role in pathological conditions in the lung, such as lung cancer and interstitial pneumonia.

Intestine

The intestinal mucosa encounters various exogenous materials, including microbes and nutrients. Immune tolerance against nonharmful materials needs to be properly established; otherwise, unwanted immune responses induce inflammatory bowel diseases. Roquin is known to play a role in the regulation of colitis development.58 In the colitis model which develops independently of B cells, Roquin‐1 and Roquin‐2 in Treg cells confer protection against colitis. Mechanistically, Roquin in Treg cells inhibits the PI3K‐AKT‐mTOR pathway at several levels, which is an upstream suppressor of Foxo1, a factor promoting Treg functions. Therefore, Roquin‐deficient Treg cells show impaired suppressive activity on conventional T cells.

Regnase‐1 is also expressed in intestinal epithelial cells where it plays diverse roles. Regnase‐1 in colon epithelium is reported to regulate epithelial regeneration.59 The lack of Regnase‐1 specifically in intestinal epithelial cells shows resistance to the colitis induced by dextran sulphate sodium (DSS) presumably via the regulation of mTOR signalling and purine metabolism. In addition to the immune regulation, Regnase‐1 has a distinct function in the duodenum, the uppermost part in the intestine which absorbs most of the iron required for the body. The deletion of Regnase‐1 in the whole body or intestinal epithelial cells leads to iron deficiency anemia.60 Under the deficiency of Regnase‐1, the activity of a transcription factor hypoxia‐inducible factor 2 α (HIF2α), a critical facilitator of iron uptake in the duodenum, was not augmented, suggesting impaired iron uptake in Regnase‐1‐deficient mice. This could be attributable to the upregulation of prolyl hydroxylase domain‐containing protein 3 (PHD3) mRNA, a negative regulator of HIF2α. Mechanistically, Regnase‐1 has been shown to destabilise iron‐regulatory transcripts such as PHD3 and transferrin receptor in addition to immune‐related ones. Moreover, Regnase‐1 itself is transcriptionally upregulated by HIF2α, forming a positive feedback loop. These findings suggest that Regnase‐1 in the duodenal epithelial cells maintains iron homeostasis by regulating iron uptake. Since iron is an essential trace metal required for invading bacteria, it is tempting to speculate that the control of iron uptake into blood stream by Regnase‐1 participates in the immune responses against blood‐borne pathogens.

Taken together, RBPs in intestinal epithelial cells and immune cells regulate the local inflammation and the intestinal tissue‐specific functions.

Skin

Because of its exposure to the external environment, the skin is also a location where immune responses are mounted against pathogens and/or allergens. IL‐17 family cytokines are known to regulate psoriatic skin inflammation, as shown by the therapeutic success of the IL‐23/IL‐17A pathway blockade.61 The signal transduction of the IL‐17 receptor is reportedly regulated by Regnase‐1 in the disease model of psoriasis.62, 63 Regnase‐1 was shown to be upregulated in psoriatic skin lesions in the imiquimod‐treated mouse model and human patients. Furthermore, Regnase‐1 haploinsufficiency in mice shows exacerbated inflammation upon imiquimod treatment, which is attributable to nonhematopoietic cells. Mechanistically, Regnase‐1 degrades a set of mRNAs related to IL‐17‐signalling such as Il17ra and Il17rc in a manner independent of the 3′ UTR.62 The deletion of Il17ra in Regnase‐1 heterozygous mice almost completely reduced the severity of skin inflammation.63 Thus, RBP‐mediated control regulates the sensitivity to IL‐17 signalling in the skin.

Central nervous system (CNS)

It is well known that RBPs such as TDP‐43 and FUS play a critical role in the pathogenesis of neurodegenerative diseases.64 Here, we do not cover such RBPs which promote the degeneration of neurons themselves, but primarily focus on RBPs which impact disease progression by modulating immune‐related aspects of pathogenesis.

EAE is a mouse model of autoimmune disease which is similar to the human disease multiple sclerosis. Regnase‐1, which regulates the sensitivity to IL‐17 signalling as mentioned above, is reportedly critical for the inhibition of EAE development.62 Conversely, Arid5a, which counteracts the function of Regnase‐1, is critical for the exacerbation of EAE severity.42 In the EAE model, Arid5a‐deficient T cells failed to differentiate into Th17 cells due to downregulation of IL‐6 production, accounting for the pathogenic mechanism mediated by Arid5a.

Moreover, ARE‐binding proteins are also critical for EAE pathogenesis. Augmented expression of TTP by depleting its autoinhibitory AREs results in protection against EAE, presumably via the suppression of proinflammatory cytokine expression.65 Correspondingly, adoptive transfer of HuR‐deficient T cells leads to the amelioration of EAE disease severity.66 Mechanistically, HuR stabilises Il17 mRNA and promotes T‐cell proliferation under the conditions of EAE. In addition, HuR upregulates the expression of the chemokine receptor CCR6 by promoting its translation and/or mRNA stabilisation, further enhancing the infiltration of Th17 cells into the CNS.67

Inflammation in the central nervous system is also linked with iron metabolism. Members of the poly(rC) binding protein (PCBP) family are multifunctional proteins that serve as an iron chaperones and RNA binding proteins.68 It is reported that PCBP1 promotes EAE pathology in iron‐replete conditions.69 PCBP1 binds to UC‐rich elements in Csf2 mRNA, a pivotal proinflammatory cytokine in pathogenic T cells, thereby stabilising the target mRNA. PCBP1 can be degraded by caspase‐mediated proteolysis under iron‐depleted conditions. Although the function of PCBP1 in iron chaperoning may also contribute to this pathology, these findings imply a potential link between post‐transcriptional regulation and neuroinflammation.

Taken together, RBPs play diverse roles in the control of immune and nonimmune cells in a tissue‐specific manner. These findings will be also informative when considering the tissue‐specific modulation of the function of RBPs in order to regulate local inflammation.

Harnessing post‐transcriptional regulation to control the immune system

Treatment with monoclonal antibodies against key proinflammatory cytokines IL‐6 and TNF has been shown to be efficacious in the treatment of autoimmune diseases such as rheumatoid arthritis, psoriasis and ulcerative colitis.70 Although these compounds have provided great benefits, there are some limitations including the administration route and pharmacokinetics. Moreover, there is a fraction of autoimmune disease that cannot be cured by current antibody‐based treatments.71 Therefore, there is still a need to develop new treatments for autoimmune diseases. While antibody‐based compounds generally target produced and released cytokines or their receptors, the modulation of cytokine production before protein synthesis could be another useful therapeutic option.

RBPs play a critical role in the regulation of inflammation as exemplified in a number of loss‐of‐function studies which we have discussed above. Thus, these RBPs and their regulatory mechanisms serve as good therapeutic targets for treating inflammatory and autoimmune diseases. We will summarise attempts to exploit or modulate the RBP‐mediated control of inflammation (Figure 3).

Figure 3.

Potential therapeutic approaches targeting post‐transcriptional regulation of inflammatory mRNAs. Post‐transcriptional mechanisms may be exploited to treat autoimmune or autoinflammatory diseases in multiple ways. siRNA‐mediated gene silencing and decoy molecules may decrease expression or activity of RBPs that stabilise proinflammatory mRNAs, respectively (a, b). Disruption of secondary structures may inhibit the access of RBPs to mRNAs of anti‐inflammatory genes or RBPs that promote the degradation of proinflammatory mRNAs (e.g. Regnase‐1 and Roquin, c). Moreover, signal‐dependent activation of proteases [e.g. T‐cell receptor (TCR)‐mediated activation of MALT1] which degrade RBPs can be targeted (d). Additionally, m⁶A RNA modification may be stabilised by demethylase inhibitors (e).

siRNA‐mediated gene silencing

As we have discussed above, some RBPs, such as HuR and Arid5a, stabilise inflammatory mRNAs. These RBPs could be targeted by microRNA (miRNA) or small interfering RNA (siRNA).72 This approach is theoretically possible; however, compounds to treat diseases in this manner are facing setbacks because of poor performance. In particular, these compounds have pharmacokinetics issues, especially with regard to intestinal absorption, and drug delivery to the intended tissue. Additionally, these compounds may be recognised by RNA sensors which elicit the innate immune response. Thus, further optimisation and improvement of these compounds for efficient and specific gene silencing are necessary.

Decoy molecules

RNA‐RBP interactions may be inhibited by the administration of decoy molecules, which mimic the bona fide RNA targets of individual RBPs. The proof of concept of this approach has been demonstrated in the case of Act1, whereby the administration of an RNA aptamer mimicking Cxcl1 SBE ameliorated IL‐17‐mediated inflammation in a skin inflammation and asthma model, respectively.37 This finding is suggestive of the potential efficacy in the clinical setting. Further studies are needed to show the efficacy of this strategy for human diseases.

Disruption of mRNA secondary structure

Since secondary structures such as the stem‐loop element in the 3′ UTR are targeted by a set of RBPs such as Regnase‐1 or Roquin, the disruption of this structure could be one possible way to intervene in its degradation. Morpholino oligos are a class of nuclease‐resistant antisense oligos which show higher binding affinity to target RNA compared to conventional DNA. In vitro studies have shown that these oligos which target stem‐loop elements can disrupt their secondary structure, thereby inhibiting the recognition by RBPs and stabilising their target mRNA.9, 10 Since this disruption takes place in a sequence‐specific manner, this approach can selectively increase the expression of gene of interest. For instance, the Roquin binding site in the mRNA of A20,73 a negative regulator of NF‐κB signalling, could be targeted by this approach to ameliorate inflammation. Recently, antisense oligonucleotide‐based treatment has been approved by US Food and Drug Administration for the treatment of neuromuscular disease SMA (spinal muscular atrophy). This treatment showed acceptable safety and tolerability, suggesting further potential for this treatment modality.74 Although further studies are needed, this class of compounds represents a new and promising therapeutic option.

Protease inhibition

The expression of RBPs is dynamically regulated by external stimuli such as T‐cell receptor (TCR) engagement. For example, Regnase‐1 and Roquin undergo proteolysis by a paracaspase, mucosa‐associated lymphoid tissue protein 1 (MALT1), upon TCR stimulation.28, 75 Thus, it is plausible that the inhibitor of MALT1 could enhance the activity of these proteins, thereby suppressing the immune responses. A MALT1 inhibitor has been shown to be effective in aggressive activated B‐cell‐like diffuse large B‐cell lymphoma (ABC‐DLBCL) in in vitro and in vivo models.76 These compounds could be repurposed to treat autoimmune or inflammatory diseases. Further studies are required to provide a conceptual proof of this approach.

Another interesting application of this approach is antiviral and anticancer therapeutics. In another article of this Special Feature, the critical roles of the proprotein convertase Furin in the regulation of viral replication and cancer development are discussed. As it is known that Furin mRNA is targeted by Regnase‐1 and Roquin,30 there may be a therapeutic application for MALT1 inhibitors in these contexts.

However, one should be conscious about the side effects of these drugs. As we have discussed above, RBP functions are diverse and the abrogation of their functions in the whole body may lead to unwanted reactions. The deletion of Malt1 in mice leads to autoimmune symptoms presumably due to the paucity of Treg cells.77 Therefore, selective delivery of the compounds is necessary to prevent potential adverse effects.

Demethylation inhibitors

m⁶A mRNA methylation can be removed by demethylases such as ALKBH5 and FTO. As these proteins are iron‐dependent, α‐ketoglutarate‐dependent hydroxylases, antagonistic inhibition is possible using α‐ketoglutarate (α‐KG) mimetics such as dimethyloxalylglycine (DMOG) or R‐2‐hydroxyglutarate (R‐2HG). DMOG treatment increases the level of m⁶A abundance and alters the level of proteins which binds to mRNA in partial m⁶A‐dependent manner.78 R‐2HG has been shown to inhibit FTO‐mediated m⁶A demethylation.79 These compounds could be repurposed to modulate inflammatory responses. As mentioned above, IFN‐β production is reported to be regulated by m⁶A deposition. Since the role of IFN‐β upon dsDNA stimulation is implicated in cellular senescence as well as pathogen control, the regulation of m⁶A status may have an impact on not only infection but also aging.80

Although these compounds could be used for modulating above‐mentioned biological processes, they are relatively nonspecific, inherently inhibiting many other hydroxylases that require α‐KG. To circumvent this, several attempts to find demethylase‐specific inhibitors have been reported.81, 82 Of these studies, a natural demethylase‐specific inhibitor, rhein, which is not an α‐KG mimetic or an iron chelator, competitively binds FTO‐active sites.81 Another compound identified is meclofenamic acid. While its mechanism of action is independent of α‐KG and iron, it has also been shown to selectively inhibit FTO rather than ALKBH5.44 These and future studies will provide valuable compounds that specifically target individual demethylases, all of which should be tested in animal models in the future.

Concluding remarks

As we have discussed, we now know that RBPs regulate many aspects of the inflammatory response. RBPs bind to a variety of immune‐related mRNAs via cis‐elements, thereby post‐transcriptionally controlling the magnitude of inflammation. Nevertheless, we are just starting to understand the functions of only a fraction of RBPs and cis‐elements; the roles of a myriad of RBPs in immune regulation have yet to be revealed. Interestingly, a bioinformatics approach employing comparative genomics identified more than 200 families of highly confident RNA structures.83 This study identified several putative RNA structures with unknown regulatory functions in immune‐related transcripts, as well as already‐characterised motifs such as CDEs. It would be intriguing to investigate novel interactions between these newly identified cis‐elements and corresponding RBPs. In addition, mapping RNA binding sites at the transcriptome‐wide scale will provide us with a more complete picture of the cis‐element landscape.

Most of the above‐mentioned studies focus on the function of RBPs in immune cells. However, recent pioneering studies have revealed that these RBPs appear to be multifunctional in order to achieve the proper mounting of immune responses in specific milieu.12, 20 Thus, future studies will be necessary to integrate the role of RBPs in individual cell populations at the organismal level. These studies also provide valuable information regarding potential adverse effects of modulating RBP functions as a therapeutic approach.

RNA‐RBP interactions could be exploited to treat excessive responses observed in inflammatory and autoimmune diseases. Although we are still in the exploratory stages of determining drug candidates, compounds targeting RNA‐RBP interaction stand to be a novel class of anti‐inflammatory treatments given the importance of RBPs in regulating immune response. Thus, further intensive research will be undoubtedly important to advance these findings from bench to bedside.

Conflict of interest

The authors report no conflict of interest.