RNA-binding proteins in immune regulation: a focus on CCCH zinc finger proteins

CCCH zinc finger proteins

We previously identified 58 CCCH zinc finger proteins in mice and 56 CCCH zinc finger proteins in humans, through genome-wide surveys⁵. Based on updated information in GenBank, we identify here three new CCCH zinc finger proteins: CNOT4, HELZ2 and PAN3. In addition, two previously identified cDNAs that potentially encoded CCCH zinc finger proteins (BC003883 and BC019429) have now been deleted in the database. Thus, there are currently 59 and 57 CCCH zinc finger proteins in mice and humans, respectively (TABLE 1). Zfp36l3 is a rodent-specific gene⁶ and does not exist in humans. Human ZC3H21 is a pseudogene and not listed in TABLE 1. Furthermore, a recent genomic analysis revealed 68 and 67 CCCH zinc finger proteins in Arabidopsis thaliana and rice, respectively⁷, suggesting that CCCH zinc finger proteins are evolutionarily conserved.

Most CCCH zinc finger proteins with known functions act as regulators of RNA metabolism, including mRNA splicing, polyadenylation, export, translation and decay, and a number of CCCH zinc finger proteins also act as transcriptional repressors or signalling modulators (Supplementary information S2 (figure)). The molecular functions of 18 human CCCH zinc finger proteins have not yet been determined, and the biological functions of many CCCH zinc finger proteins remain unknown. However, most of the characterized CCCH zinc finger proteins have been shown to be crucial regulators of immune responses, through the regulation of cytokine production, immune cell activation, immune homeostasis and antiviral responses.

Emerging evidence suggests that several CCCH zinc finger proteins, such as TTP, roquin 1 and MCPIP1, constitute a novel regulatory network that promotes the resolution of inflammation, controls the magnitude and duration of adaptive immune responses, and maintains immune homeostasis by targeting mRNA decay and modulation of signalling pathways^8–11. TTP is the best-studied member of a small family of three proteins in humans that is characterized by a specific tandem CCCH zinc finger domain (FIG. 1). TTP was initially discovered as a gene that could be rapidly and transiently induced by the stimulation of fibroblasts with growth factors and mitogens¹². It is now known that this protein can bind to AU-rich elements (AREs) in mRNA, leading to the removal of the poly(A) tail from the mRNA and increased rates of mRNA decay¹³. In addition to the tandem CCCH zinc finger domains, TTP contains three proline-rich domains and a conserved carboxy-terminal sequence that can bind the NOT1 scaffolding protein. A nuclear export sequence (NES) is located at the amino terminus of TTP (FIG. 1). The other two human family members, encoded by ZFP36L1 and ZFP36L2, share very similar tandem CCCH zinc finger domains, as well as the putative C-terminal NOT1 binding sequences (FIG. 1).

Roquin 1 was identified through an N-ethyl-N-nitrosourea (ENU)-induced mutagenesis screen in mice⁹. In addition to a single CCCH zinc finger domain, roquin 1 contains a RING finger domain, a ROQ domain and a proline-rich domain (FIG. 1). Roquin 1 recognizes stem–loop motifs in the 3′ untranslated region (UTR) of its target mRNAs, through its ROQ domain and adjacent CCCH zinc finger domain, and promotes mRNA decay by recruiting the helicase RCK (also known as DDX6) and enhancer of mRNA-decapping protein 4 (EDC4)¹⁴. A second family member, roquin 2, shares a similar structure and has some overlapping functions with roquin 1 (FIG. 1).

MCPIP1 was identified as a novel protein containing a CCCH zinc finger domain and a PIN domain-like RNase domain; it also contains an ubiquitin-associated domain at its N terminus and a serine-rich domain at its C terminus (FIG. 1). MCPIP1 targets mRNA degradation via its intrinsic RNase activity¹¹. Three other related proteins (termed MCPIP2, MCPIP3 and MCPIP4) share similar domain structures with MCPIP1 (REF. 10) (FIG. 1). Interestingly, phylogenetic analysis⁵ showed that TTP, roquin 1 and MCPIP1 are evolutionarily closely related.

CCCH zinc finger proteins share many common features: the first is that each protein contains one or more CCCH zinc fingers, which defines this superfamily. Second, although other types of zinc fingers are well known as DNA-binding motifs, CCCH zinc fingers more commonly bind RNA by recognizing specific sequences or secondary structures in their mRNA targets⁴. For example, the tandem CCCH zinc finger motifs of the TTP protein family recognize specific ARE elements such as UUAUUUAUU in the 3′ UTR of their target mRNA¹³, which is further confirmed by structural analysis¹⁵. Of note, a single CCCH zinc finger can bind only weakly (albeit specifically) to AU-rich RNA¹⁶, suggesting that proteins that contain a single CCCH zinc finger domain may bind RNA as dimers. Biochemical analysis indicates that both the ROQ domain and the adjacent CCCH zinc finger of roquin 1 contribute to the recognition of RNA¹⁷. However, structural analysis suggests that the ROQ domain of roquin 1 is sufficient for binding to the conserved decay element (CDE) in the 3′ UTR of tumour necrosis factor (Tnf) mRNA^18,19. Although the CCCH zinc finger domain of MCPIP1 has been shown to play a part in the control of inter-leukin-6 (Il6) mRNA decay¹¹ (see later), the functional contributions of the CCCH zinc fingers in roquin 1 and MCPIP1 are not fully understood. Why CCCH zinc finger domains preferentially bind to RNA over DNA is also not well understood.

A third common feature of CCCH zinc finger proteins is that they commonly shuttle between different cellular compartments, such as from the nucleus to the cytoplasm, and between different RNA compartments, such as polysomes, stress granules and P-bodies, where they are thought to perform their various roles in RNA metabolism (BOX 1). A fourth common feature of this protein superfamily is that its members commonly interact with microRNA (miRNA) processing and effector pathways (BOX 2). Last, in addition to CCCH zinc finger domains, these family members contain other important functional domains. For example, MCPIP1 contains an RNase domain, and the roquin family proteins contain RING-type E3 ubiquitin ligase domains (FIG. 1). These domains greatly contribute to the functional variety within this protein superfamily.

Regulation of cytokine production

Regulation of immune cell activation

In addition to regulating immune responses through the control of cytokine production, CCCH zinc finger proteins can regulate immune cell activation via intrinsic pathways.

Regulation of immune homeostasis

The regulation of costimulatory signalling in CD4⁺ T cells is crucial for maintaining peripheral T cell tolerance and immune homeostasis. Failure to control costimulatory signalling may break immunological tolerance and contribute to the development of autoimmune diseases such as systemic lupus erythematosus (SLE). The expression of ICOS, which is a member of the CD28 superfamily of costimulatory molecules, is upregulated by T cells upon activation. ICOS signalling is required for the development and proliferation of T follicular helper (T_FH) cells⁶⁷. TFH cells promote the maturation of B cells into memory B cells or antibody-secreting plasma cells in germinal centres⁶⁸. Dysfunctional T_FH cells can lead to the aberrant selection of autoreactive germinal centre B cells and the production of autoantibodies that cause diseases such as SLE⁶⁹. Roquin 1 is a crucial factor that controls Icos mRNA levels and maintains activation-induced expression of ICOS by CD4⁺ T cells⁷⁰. Recognition of the 3′ UTR of this mRNA by the RNA-binding domain of roquin 1 facilitates the degradation of the transcript through interactions with the decapping enzyme EDC4 and the helicase RCK¹⁴.

Roquin 1^san/san (also called sanroque), a single point mutation in the ROQ domain of the gene encoding roquin 1, leads to a lupus-like autoimmune phenotype in mice, which is marked by enhanced numbers of T_FH cells and spontaneous germinal centre formation⁹. These mice had high-level ICOS expression. Partial correction of ICOS overexpression in roquin 1^san/sanIcos^+/− mice was accompanied by a reduction in lymphadenopathy, splenomegaly, T_FH cell population expansion and germinal centre B cell numbers, suggesting that ICOS overexpression is an essential contributor to the lupus phenotype in roquin 1^san/san mice⁷⁰. Interestingly, unlike roquin 1^san/san mice, the cell-specific ablation of roquin 1 in T cells or B cells did not cause autoimmunity⁷¹, and similarly to roquin 1, T cell-specific deletion of roquin 2 does not affect immune cell homeostasis⁷². However, mice deficient in both roquin 1 and roquin 2 developed an autoimmune phenotype with increased T_FH cell numbers and germinal centre expansion, which phenotypically resembles roquin 1^san/san mice⁷². These results suggest redundant functions of roquin 1 and roquin 2 in regulating ICOS expression in T_FH cells (FIG. 4).

The paradoxical observations between the autoimmune-prone roquin 1^san/san mice and the healthy roquin 1-deficient mice are not fully understood. A recent elegant study showed that a RING-less roquin 1-mutant protein failed to localize to stress granules and resulted in a compensatory role for roquin 2 in the repression of ICOS and T_FH cells²⁹, thus roquin 1^RINGless mice did not show any phenotypes of autoimmunity²⁹. By contrast, in roquin 1^san/san mice, the ROQ-mutant roquin 1 still localized to mRNA-regulating stress granules and prevented a compensationary response by roquin 2; thus roquin 1^san/san mice showed severe phenotypes of autoimmunity²⁹. Nevertheless, more investigation into this paradoxical observation is required for a full understanding.

In view of the important role of RING domains in driving protein ubiquitylation, determining whether roquin 1 and roquin 2 contain E3 ubiquitin ligase activity is crucial for understanding their complicated roles. Roquin 2 was identified as an E3 ubiquitin ligase required for reactive oxygen species (ROS)-induced ubiquitylation and degradation of apoptosis signal- regulating kinase 1 (ASK1; also known as MAP3K5)⁷³. As evident from research using Caenorhabditis elegans, mutation of the gene encoding the roquin 2 orthologue RLE1 results in the accumulation of the activated form of the ASK1 orthologue NSY1, which conferred resistance to Pseudomonas aeruginosa infection⁷³. Moreover, Zhang et al. recently demonstrated that both roquin 1 and roquin 2 are functional E3 ubiquitin ligases that drive the assembly of polyubiquitin chains of different linkages⁷⁴. Interestingly, a recent study showed that the RING domain of roquin 1 directly binds to the catalytic a1 subunit of adenosine monophosphate-activated kinase (AMPK) and represses its enzymatic activity⁷⁵. As AMPK is a central regulator of cellular metabolism in response to cellular stress and an inhibitor of mechanistic target of rapamycin (mTOR) signalling, roquin 1 may positively regulate T_FH cell formation by inhibition of AMPK activity (FIG. 4). Although this surprising observation⁷⁵ on the role of roquin 1 on T_FH cell formation needs more investigation, these results suggest that roquin 1 may fine-tune the regulation of T_FH cell differentiation through two different domains (RING domain and ROQ domain) and two different mechanisms (inhibition of AMPK and degradation of ICOS) (FIG. 4).

Taken together, these studies revealed the importance of roquin 1- and roquin 2-mediated post-transcriptional networks in T_FH cell differentiation and the maintenance of immune homeostasis. Investigations into the mechanisms through which roquin proteins and MCPIP1 repress their target mRNA molecules have highlighted the complex and intricate feedback loops these RNA-binding proteins use to augment or control adaptive immune responses.

Regulation of the regulators

The expression and function of CCCH zinc finger proteins are precisely regulated by multiple mechanisms, which add another layer of control on the regulation of innate and adaptive immune responses (FIG. 5). For example, Mcpip1 mRNA is induced by TLR ligands and inflammatory cytokines through the NF-κB signalling pathway⁷⁶. However, during LPS- and IL-1β-induced macrophage activation, MCPIP1 protein is rapidly degraded by the ubiquitin–proteasome system, which is dependent on IKK-mediated phosphorylation⁷⁷ (FIG. 5). Consistent with this observation, MG132, which is a potent proteasome inhibitor, has been shown to significantly increase MCPIP1 protein levels in HepG2 or HeLa cells⁷⁸. In addition, MCPIP1 protein can target the 3′ UTR of its own mRNA and promote its degradation⁷⁷. In activated T cells, MCPIP1 is regulated through TCR-induced MALT1 (mucosa-associated lymphoid tissue lymphoma translocation 1)-mediated cleavage. MALT1 is an arginine-specific protease that cleaves MCPIP1 at arginine 111; the resulting fragments are rapidly degraded by other proteases⁵⁷. Similarly to MCPIP1, both roquin 1 and roquin 2 are also cleaved by MALT1 upon T cell activation⁶² (FIG. 5).

Importantly, the function of MCPIP1 on mRNA decay is regulated by other RNA-binding proteins. For instance, AT-rich interactive domain-containing protein 5α (ARID5α) can compete with MCPIP1 for binding to the 3′ UTR of Il6 mRNA and protect it from MCPIP1-mediated degradation. Furthermore, ARID5a deficiency in mice attenuates acute inflammation and autoimmunity⁷⁹. Together, these results suggest that both the expression and function of MCPIP1 are dynamically regulated during the course of an immune response: MCPIP1 expression is suppressed by TLR signalling during the early phase of an infection, allowing the immune system to be activated and fully respond to the microbial challenge, whereas, in the later phases of the response, MCPIP1 expression is restored by increased transcription⁷⁶ and helps to promote the resolution of inflammation by promoting the degradation of inflammatory cytokines and immune molecules.

Similarly to MCPIP1, the transcription of Zfp36 is also regulated by NF-κB signalling in LPS-stimulated macrophages⁸⁰. TTP can also promote the degradation of its own mRNA by binding to the ARE on its 3′ UTR⁸¹. Phosphorylation of TTP by p38–MK2 has been reported to promote its sequestration by 14-3-3 proteins, resulting in diminished mRNA decay activity⁸² (FIG. 5). Additionally, phosphorylated TTP may be impaired in its ability to recruit mRNA decay enzymes, resulting in decreased deadenylation of target transcripts^82–84. In addition, the function of TTP in the regulation of cytokine production may be antagonized by another RNA-binding protein. For example, as an ARE-binding protein, HuR can compete with TTP for ARE-binding sites on the Il6 3′ UTR, thereby stabilizing Il6 mRNAs⁸⁵. TTP can also be ubiquitylated by the adaptor molecule TRAF2 in a phosphorylation-dependent manner^86,87. Ubiquitylated TTP no longer affects NF-κB activity but promotes the activation of JNK⁸⁶. A further study suggests that TTP degradation by proteasomes is ubiquitin independent but instead is dependent on intrinsically disordered N- and C-termini of the protein⁸⁸. These studies suggest that the expression and function of TTP are regulated by multiple mechanisms.

Taken together, these studies suggest that the expression and function of TTP, roquin 1 and MCPIP1 can be governed by the same signalling pathways that control immune cell activation in response to TCR, TLR and cytokine receptor signalling. It is likely that some of CCCH zinc finger proteins may act as a point of convergence for the activity of numerous kinases, including IKK, ERK1, ERK2, JNK and p38 MAP kinase. For example, TTP can be modified by different kinases in response to different environmental cues, and the different modified states of TTP have different roles in the regulation of target decay or inhibition of signalling pathways. Precise regulation of these regulators provides further layers of control to fine tune immune responses (FIG. 5).

Associated diseases and therapeutic potential

Although emerging evidence from studies of genetically engineered animals indicates that TTP, roquin 1 and MCPIP1 are crucial regulators of inflammation and immune homeostasis, mutations of these genes in humans have not been clearly implicated in disease development. Suzuki et al.⁸⁹ reported that a single nucleotide polymorphism (SNP) within the human ZFP36 promoter resulted in twofold greater promoter activity. These authors reported that patients with rheumatoid arthritis with the GG genotype might be prone to higher disease activity than those with the AG or AA genotypes. This suggests that this SNP mildly affects ZFP36 promoter activity, and may thus influence the activity of rheumatoid arthritis and perhaps other inflammatory diseases⁸⁹. Another polymorphism in the protein coding domain of ZFP36 has also been shown to be significantly associated with rheumatoid arthritis in African–Americans⁹⁰. Associations between human disease and mutations of RC3H1 (encoding roquin 1) and MCPIP1 have not been the subject of published investigations but certainly warrant future study.

Given their importance in immune regulation and RNA metabolism, CCCH zinc finger proteins are possible targets for the treatment of autoimmunity, viral infections and cancer, although the effects of targeting CCCH zinc finger proteins on an organismal level are currently unknown. Based on their complex function, both increasing and decreasing CCCH zinc finger protein expression and activity could have therapeutic benefits for specific indications. For instance, increasing MCPIP1 expression, by the selective suppression of degradative enzymatic activity within the cells, such as that exhibited by MALT1 and proteasomes, could potentially improve outcomes in diseases in which excessive inflammatory responses are considered detrimental, such as septic shock, atherosclerosis and certain viral infections. MI-2, a specific inhibitor of MALT1, was recently reported to selectively enhance MCPIP1 expression in CD4⁺ T cells and consequently promote apoptosis of T cell lines that were latently infected with HIV-1 in the presence of cell stimuli in vitro⁹¹. Given that MI-2 is a highly specific and non-toxic small molecular inhibitor of MALT1 (REF. 92), it would be interesting to explore its therapeutic potential for inflammatory diseases and viral infections in vivo. By contrast, suppressing MCPIP1 and roquin 1 activity may be a novel strategy for enhancing cancer immunotherapy or responses to vaccines. Recent studies with a mouse model of regulated TTP overexpression demonstrated a protective effect against several models of immune and inflammatory disease, supporting the potential role of TTP as a therapeutic target^93,94.

Conclusions and perspectives

Regulation of RNA metabolism is an important component of gene expression that facilitates the fine-tuning of transcript levels during physiological conditions and during the rapid and profound switch in global gene expression associated with inflammation and immune responses. Long-term dysregulation of RNA metabolism can often result in disease states, including inflammatory and autoimmune diseases⁹⁵. CCCH zinc finger proteins have emerged as important regulators of multiple facets of RNA metabolism and immune responses, with promising therapeutic potential. During the innate immune response, several CCCH zinc finger proteins, such as TTP, roquin 1 and MCPIP1, promote the resolution of inflammation by helping to eliminate the mRNAs of certain pro-inflammatory cytokines. During the activation of B cells and T cells, these CCCH zinc finger proteins intrinsically control the magnitude and duration of adaptive immune responses and maintain immune homeostasis via multiple mechanisms. Failure of these important regulatory controls (for example, in mice that are genetically deficient for TTP, roquin 1 or MCPIP1) often leads to the development of systemic inflammatory responses and autoimmune syndromes. In addition to the regulation of immune responses, other CCCH zinc finger proteins such as ZAP (encoded by ZC3HAV1), PARP12 (encoded by ZC3HDC1) and TOE1 are important regulators of viral replication (Supplementary information S3 (box)).

In the future, searching for associations between mutations in the genes for CCCH zinc finger proteins and various inflammatory disorders and cancer should help to assess the biological roles of members of this protein superfamily in humans. Furthermore, elucidating how these CCCH zinc finger proteins are regulated and exploiting this knowledge for therapeutic application, by enhancing or repressing their expression and/or function, is a crucial next step in understanding these multifaceted proteins. At a more basic level, it would be extremely interesting to determine how cross-regulation of two different domains occurs in a single protein. For example, does the E3 ubiquitin ligase activity of roquin 1 RING domain influence the RNA decay process mediated by its ROQ domain and CCCH zinc finger? Alternatively, does ROQ and CCCH zinc finger-mediated RNA binding affect its RING domain- mediated ubiquitylation function? Based on structural studies, it seems that the RING domain of roquin 1 can interact with its CCCH zinc finger domain and affect its RNA-binding activity⁷⁴. A recent study also showed that the N-terminal domain of MCPIP1 was associated with its PIN domain and significantly enhanced its RNase activity⁹⁶. Lastly, as most CCCH zinc finger proteins have not been well-characterized, further studies on the function and mechanisms of these less- characterized proteins will provide valuable information for understanding this fascinating group of proteins.

Supplementary Material