RNA-binding proteins in autoimmunity: From genetics to molecular biology

Autoimmune diseases (ADs) are chronic pathologies generated by the loss of immune tolerance to the body's own cells and tissues. There is growing recognition that RNA-binding proteins (RBPs) critically govern immunity in healthy and pathological conditions by modulating gene expression post-transcriptionally at all levels: nuclear mRNA splicing and modification, export to the cytoplasm, as well as cytoplasmic mRNA transport, storage, editing, stability, and translation. Despite enormous efforts to identify new therapies for ADs, definitive solutions are not yet available in many instances. Recognizing that many ADs have a strong genetic component, we have explored connections between the molecular biology and the genetics of RBPs in ADs. Here, we review the genetics and molecular biology of RBPs in four major ADs, multiple sclerosis (MS), type 1 diabetes mellitus (T1D), systemic lupus erythematosus (SLE), and rheumatoid arthritis (RA). We anticipate that gaining insights into the genetics and biology of ADs can facilitate the discovery of

RNA immunoprecipitation methods, where RBPs in native conditions are purified together with the bound RNAs, and the associated RNAs can then be identified using RT-qPCR or RNA-sequencing (RNA-seq) analysis (Keene et al., 2006;Martindale et al., 2020). Another powerful approach to study RNA-RBP interaction involves a crosslinking step before immunoprecipitation (CLIP) and its many variations (e.g., HITS-CLIP, iCLIP, PAR-CLIP, eCLIP), that permits the identification of the precise binding sites of an RBP on a target mRNA. Many other protocols have been developed over the years to study RBPs, as reviewed elsewhere (Singh et al., 2021;Wheeler et al., 2018).

| Multiple sclerosis
Multiple sclerosis (MS) is an autoimmune disease of the central nervous system (CNS) caused by an aberrant immune response leading to inflammation, damage of the blood-brain barrier, demyelination, and axonal loss, culminating in neurodegeneration (Martinsen & Kursula, 2022). A severely debilitating disease, MS causes physical and cognitive deficits leading to disability, and affects more than 2.5 million people worldwide (Kavaliunas et al., 2022). The etiology of MS is still largely unknown, but it is believed that environmental factors and genetic predisposition cooperate to trigger the autoimmune response against CNS antigens (Olsson et al., 2016). Among environmental factors, a key role for Epstein-Barr virus infection was recently demonstrated. Using data from 10 million US military personnel monitored over a 20-year period, Bjornevik et al. determined that the risk of MS increased 32-fold after infection with EBV, supporting its role in the pathogenesis of MS (Bjornevik et al., 2022). Sam68 F I G U R E 1 RNA binding proteins (RBPs) regulate many steps of mRNA biology. In the nucleus, RBPs interact with pre-mRNAs recruiting the spliceosome, and regulating splicing events. Mature mRNA export is mediated by different RBPs and once the mRNA arrives in the cytoplasm, RBPs control the localization, initiation of translation and mRNA decay. In the figure we show the RBPs described in the text in the context of each autoimmune disease (AD) discussed in the manuscript. Comprehensive descriptions of the functions of each RBP can be found elsewhere in literature.
MS appears to involve imbalanced interactions between the different immune cell populations, and the antibodies and cytokines that they produce (Jones et al., 2017;Steri et al., 2017). The ensuing inflammatory response leads to demyelination and early neuronal loss associated with the neurodegenerative process (Frischer et al., 2009). Unfortunately, at present there is no current cure or therapy for MS, and it is managed mainly by the use of anti-inflammatory and immunomodulatory drugs to ameliorate MS symptoms. Recently, RBPs have emerged as key players in different biological and physiological processes including inflammation and neurodegeneration. Here we highlight mechanisms of neurodegeneration and inflammation in MS with a focus on RBP dysfunctions (Table 1).

| Neurodegeneration
The aberrant localization of RBPs represents a critical step for altered regulation of RNAs and is a key player in MS progression and disability. Mislocalization of two RBPs-the heterogeneous nuclear ribonucleo-protein A1 (HNRNPA1) and TAR-DNA-binding protein-43 (TDP-43)-has been observed in the CNS of MS patients. Both HNRNPA1 and TDP-43 shuttle between the nucleus and cytoplasm to modulate mRNA splicing, stabilization, and trafficking. In MS pathology, HNRNPA1 and TDP-43 move from a homeostatic nuclear location to the cytoplasm, leading to the formation of protein aggregates (Salapa et al., 2017(Salapa et al., , 2018. HNRNPA1 is involved in several neuronal processes including the regulation of neuron morphology, viability, and stress granule formation (Low et al., 2021). Thus, dysfunction of HNRNPA1 induces neuronal deficits and its knock down was reported to reduce neurite outgrowth, increase in cell toxicity, and change in stress granule formation (Anees et al., 2021). Supporting a role of HNRNPA1 in neuronal processes, Lee and colleagues identified nine single-nucleotide polymorphisms (SNPs) in MS patients contributing to HNRNPA1 dysfunction. The SNPs are located in the nucleocytoplasmic binding domain of HNRNPA1, known as M9 (Lee & Levin, 2014), that modulates the nuclear export sequence/nuclear localization sequence (NES/NLS) responsible for the mobilization of HNRNPA1 between the nucleus and the cytosol (Iijima et al., 2006). The authors found that, transfection of plasmids containing HNRNPA1 M9 mutants in neuron-like cells resulted in HNRNPA1 mislocalization to the cytoplasm, stress granule formation, and apoptosis (Lee & Levin, 2014). Similarly, Clarke et al. showed that a somatic mutation in HNRNPA1 alters its function and promotes stress granule formation (Clarke et al., 2021). Interestingly, antibodies recognizing HNRNPA1 and HNRNPB1 were found to be more abundant in the cerebrospinal fluid (CSF) and serum of MS patients than in other neurological diseases, suggesting their possible utility as markers for MS (Lee et al., 2011;Yukitake et al., 2008). Mislocalization and decreased expression have also been reported for the RPBs TDP-43 and polypyrimidine tractbinding protein 1 and 2 (PTB1, PTB2) in MS lesions and in cultured primary human oligodendrocytes exposed to metabolic stress. PTB1 and PTB2 regulate mRNA stability and trans-splicing of genes containing C-rich polypyrimidine tracts (Stern et al., 2009), as well as neuronal differentiation, while TDP-43 plays a crucial role in oligodendrocyte function (Jo et al., 2020). Mislocalization and altered expression of these RBPs in MS lesions lead to variation in the regulation of target mRNAs contributing to neurodegeneration (Masaki et al., 2020). Interestingly, misregulation of HNRNPA1, TDP-43, PTB1, and PTB2 is also associated with other neurological disorders such as amyotrophic lateral sclerosis, Alzheimer's disease, and frontotemporal dementia (Bampton et al., 2020;Bekenstein & Soreq, 2013;Deshaies et al., 2018).

| Demyelination
The role of RBPs in myelin formation and damage has been studied by different groups. Jiayi et al. described the involvement of HuR in regulating, together with miRNA-29a, the production of the protein cystatin F, a papain-like lysosomal cysteine proteinase inhibitor. Cystatin F plays a key role in the processes of demyelination and remyelination. Both HuR and cystatin F were present in reduced levels in the core areas of MS plaques as compared to the border zone, suggesting that HuR downregulation may be linked to cystatin F mRNA instability resulting in higher demyelination in MS patients. A direct function of HuR in cystatin F regulation was uncovered in cultured cells .
Another RBP implicated in the myelination process is QUAKING (QKI), a protein that belongs to the heteronuclear ribonucleoprotein particle K (HNRNPK) homology (KH) domain family. QKI protein affects pre-mRNA splicing, mRNA turnover, and translation of several target mRNAs through interaction with QKI response element (QRE) localized in the 3 0 UTR of target mRNAs including the MBP mRNA, encoding myelin basic protein (MBP) (Chen Yin et al., 2021). Different isoforms of QKI (QKI-5, QKI-6, QKI-7) are generated via alternative splicing, and a balance between them in oligodendrocytes and Schwann cells was found to be essential in myelination process (Darbelli & Richard, 2016;Pilotte et al., 2001). Interestingly, human brains from MS patients had reduced microglia QKI expression compared to individuals with normal white matter content (Lee Villarreal et al., 2020). In QKI-deficient mice, microglia showed reduced CNS remyelination, suggesting an important role of QKI in the regulation of microglial activity in the brain (Lee Villarreal et al., 2020).

| Inflammation
In MS patients, deregulated lymphocytes are activated in the periphery and infiltrate the CNS, participating in local inflammation and demyelination (Haase & Linker, 2021). Pro-inflammatory factors secreted by immune cells may drive RBP dysfunctions (Libner et al., 2020). For example, TNFα and IFNγ, shown to be increased in CSF samples of MS patients (Khaibullin et al., 2017), cause RBP dysfunctions in neuronal cells (Salapa et al., 2018). IFNγ induces HNRNPA1 mislocalization and consequently changes in RNA metabolism (Salapa et al., 2018). The pro-inflammatory cytokines IFNγ and TNFα and anti-A1 antibodies induced HNRPA1 dysfunction and damage in primary cortical neurons, a physiologically relevant in vitro model system . Reciprocally, RBPs can bind to mRNAs that encode pro-and anti-inflammatory cytokines and control their expression post-transcriptionally. HuR binds to and regulates many target mRNAs including those that encode interleukin (IL) 1β, IL6, TNFα, and IL17. HuR binding to IL17 3 0 UTR stabilized IL17 mRNA, increasing in IL17 protein levels (Chen Cascio et al., 2013), and high IL17 levels were detected in plaques and CSF of MS patients, correlating with MS severity (Matusevicius et al., 1999). Still, a direct correlation between MS, HuR, and IL17 mRNA metabolism awaits to be reported. Aberrant levels of HuR were also found in PBMCs and T cells isolated from 52 MS patients compared to healthy controls, and MS patients with moderate-to-severe forms of MS showed reduced HuR protein levels compared to patients with mild disease, suggesting that HuR protein levels decline gradually with disease progression (Pistono et al., 2020).
Recently we studied the role of NF90 in regulating the cytokine BAFF (Idda et al., 2018), whose increased function is a disease-predisposing factor for MS and SLE. NF90 is a RBP that influences RNA metabolism at several levels, including pre-RNA splicing, mRNA turnover, and translation (Castella et al., 2015). Using GWAS, our group identified a variant in the BAFF 3 0 UTR associated with increased risk of MS, SLE and levels of the cytokine BAFF, B cells and immunoglobulins (Steri et al., 2017). The variant, an insertion-deletion located in the 3 0 UTR of the TNFSF13B gene, generated an alternative polyadenylation site leading to the formation of a shorter 3 0 UTR that escaped binding by NF90 and the microRNA miR-15a. NF90 cooperated with miRNA-15a to repress the translation of BAFF in normal conditions. In presence of the variant, the RBP binding site and the miRNA were lost, leading to higher production of soluble BAFF (Idda et al., 2018).
Overall, these data indicate that dysregulation in RBPs or binding site for RBPs in target RNAs can alter the inflammatory environment with negative consequence for MS onset.

| Systemic lupus erythematosus
SLE is a chronic debilitating multi-system autoimmune disease with significant morbidity and mortality (Yeneric & Singh, 2012). It is characterized by the breakdown of the immune system with production of autoantibodies, activation of the complement cascade, and deposition of immune complexes leading to systemic autoimmunity and organ damage (Dema & Charles, 2016). One of the main features of SLE is the production of antinuclear antibodies targeting doublestranded DNA (dsDNA) and other nuclear autoantigens originating from uncleared apoptotic cells (Satoh et al., 2009). The onset of SLE involves the entire immune system, with dysregulation of T cells, B cells, and dendritic cells, along with altered production of cytokines, which together cause immune and tissue dysfunction (Apostolidis et al., 2011). As a multifactorial disease, genetics and environmental factors interact in complex ways to drive dysregulation of the immune system. Due to the high morbidity and mortality or SLE, and the absence of effective therapies, understanding the role of RBPs and posttranscriptional regulation in SLE pathogenesis is an essential step toward the identification of new and more appropriate therapeutic targets (Table 1).

| RPBs and systemic lupus erythematosus
The Serine/arginine-rich splicing factor 1 (SRSF1) is an RBP that controls posttranscriptional gene expression via pre-mRNA alternative splicing, stability, and translation that has been associated with SLE pathogenesis (Katsuyama et al., 2019). Reduced expression of SRSF1 correlates with activation of T cells and increased risk of autoimmune disease. Low SRSF1 levels lead to the dysregulation of genes involved in apoptosis, and are linked to lymphopenia in SLE (Katsuyama et al., 2019), while reduced SRSF1 abundance in T cells from patients with SLE correlate with severe disease (Kono et al., 2018). However, the role of SRSF1 in T cell physiology and autoimmune disease is largely unknown. In mice, reduced SRSF1 levels triggered systemic autoimmunity and lupus nephritis (Katsuyama et al., 2019) associated with increased frequency of activated/effector T cells producing proinflammatory cytokines, elevated activation of mTORC1 pathway, and reduced expression of phosphatase and tensin homolog (PTEN). Aberrant mTOR and PTEN regulation were also implicated in T and B cells dysfunctions and autoimmunity (Perl, 2015).
The KH-type splicing regulatory protein (KSRP), an RBP that promotes the decay of mRNAs encoding cytokines, has also been studied in SLE (Briata et al., 2016). In line with the role of KSRP as a negative regulator of cytokine expression, lupus nephritis is augmented in KSRP À/À mice. Accordingly, KSRP was found to protect and ameliorate lupus nephritis by negatively regulating IFN production, and by reducing the migration of immune cells, associated with the modification of chemokines and the expression of adhesion molecules. Interestingly, the phenotype of the KSRP À/À mice, characterized by enhanced pro-inflammatory gene expression and exacerbation of kidney morphology, helped to discover that the posttranscriptional regulation of cytokine expression is a key element in understanding lupus pathogenesis (King & Chen, 2014;Lin et al., 2011).
Increased expression of X-linked genes along with abnormal X chromosome inactivation (XCI) was observed when analyzing immune cells purified from SLE patients. In this regard, Yu et al. described the implication of tripartite motif containing 28 (TRIM28), a transcription factor with RNA-binding functions (Fernandez-Marrero et al., 2018), in B-cell XCI maintenance. TRIM28 is a B cell-specific protein that binds XIST (X-inactive specific transcript), required for XCI in a subset of X-linked immune genes such as TLR7, an X-linked immune gene whose biallelic expression is associated with female-biased autoimmunity. Interestingly, duplication of the TLR7 gene is enough to drive lupus-like symptoms, whereas deletion of TLR7 in mouse SLE models ameliorates the disease (Christensen et al., 2006;Subramanian et al., 2006).
Another interestingly RBP implicated in SLE is Ro60, an evolutionarily conserved protein present in cells as both free protein and as a component of a ribonucleoprotein complex (RNP) that include a noncoding RNAs called Y RNAs. Ro60 is involved in the cellular response to stress (Chen et al., 2003), and has been proposed to function in ncRNAs quality control (Sim & Wolin, 2011;Wurtmann & Wolin, 2010). Interestingly, Greiling and colleagues identified Ro60 orthologs in commensal bacterial species colonizing human skin, oral cavity, and gut. After identifying these orthologs in SLE patients, and considering that anti-Ro60 autoantibodies are present in asymptomatic individuals years before SLE onset (Arbuckle et al., 2004), the authors proposed that individual predisposed to autoimmunity, who are chronically colonized by Ro60 commensals, may develop antibodies against a bacterial Ro60 ortholog that leads to autoimmunity via cross-reactivity and epitope spreading (Greiling et al., 2018). Furthermore, given that specific autoimmune disorders such as neonatal lupus (Weston et al., 1982) and subacute cutaneous lupus erythematosus (Sontheimer et al., 1982) are strongly associated with anti-Ro antibodies, immune complexes containing Ro60 have been proposed to contribute to SLE pathogenesis (Clark et al., 1969;Hung et al., 2015). Until now, only few studies have analyzed the role of RBPs in SLE pathogenesis and additional analysis are required to better understand the functions of these proteins in SLE onset and progression.

| Rheumatoid arthritis
RA is a long-term autoimmune condition that primarily affects joints. It is characterized by synovial inflammation, local infiltration of immune, and nonimmune cells producing pro-inflammatory cytokines leading to chronic joint inflammation and damage, bone destruction, and eventually joint deformity and progressive disability (Solus et al., 2015). RA fibroblast-like synoviocytes (FLS) play a central role in the initiation and persistence of RA-activating NF-κB pathway, promoting local proliferation, production of proinflammatory cytokines, and cartilage invasion (Bugatti et al., 2019). The expression of several RBPs was found to be dysregulated in RA, and their downregulation using silencing protocols affected key regulatory mechanisms in arthritis pathogenesis, both in vitro and in vivo (Chen Cascio et al., 2013;Nieminen et al., 2008). Here, we analyze the function of key RBPs in RA (Table 1).

| RBPs and rheumatoid arthritis
One of the best characterized RPB in RA is TTP (tristetraprolin), a protein involved in the regulation of AU-rich mRNAs (Ross et al., 2017). TPP is a suppressor of the production of the intercellular adhesion molecule-1 (ICAM-1)  through different mechanisms: reduction of ICAM1 mRNA stability and translation through the AUrich ICAM1 3 0 UTR and by recruiting deadenylases that shorten the poly(A) tail . ICAM-1 (along with VCAM-1) is involved in multiple mechanisms that promote inflammation, predominantly by promoting local infiltration of inflammatory cells (Klimiuk et al., 2002). Supporting the relevance of TTP in RA, Yang et al. reported a reduction of TTP mRNA in human PBMCs purified from RA patients and an association between a SNP located in the gene encoding for TTP (rs3746083) and RA predisposition in Chinese RA patients (Yang et al., 2021).
HuR binds to and regulates (typically promotes) the posttranscriptional expression of mRNAs encoding cytokines and inflammatory factors implicated in RA. One critical mRNA target of HuR is the nucleotide-binding oligomerization domain, leucine-rich repeat, and pyrin domain-containing 3 (NLRP3) protein component of the inflammasome. The inflammasome plays a key role in inflammation and autoimmunity due to its ability to upregulate pro-inflammatory cytokines such as IL1β and IL18. In FLS, HuR knockdown significantly upregulated in RA synovial tissue, and led to reduced caspase-1 p20 protein expression resulting in decreased secretion of IL1β (Liu et al., 2019).
The cold-inducible RNA-binding protein (CIRP) belongs to the family of heat shock proteins that respond to cold stress. Under cellular stress, CIRP migrates from the nucleus to the cytoplasm, promoting the translation of target transcripts including CIRP mRNA itself. Extracellular CIRP, identified as a damage-associated molecular pattern, triggers inflammatory responses during RA (Yoo et al., 2018). Furthermore, Yoo et al. found that serum levels and expression of CIRP in CD14+ monocytes are elevated in patients with RA as compared to patients with osteoarthritis, and that CIRP level correlated with disease activity (Yoo et al., 2018).
Sam68 (Src-associated substrate during mitosis of 68 kDa) is an RBP ubiquitously expressed that plays highly specialized roles in different cells. In many cells, Sam68 localizes in the nucleus and regulates mRNA processing, including transcription, alternative splicing and nuclear export. After translocation to the cytoplasm, Sam68 regulates mRNA translation efficiency through association with the mRNA translation machinery (Frisone et al., 2015). In RA, Sam68 is overexpressed in FLS cells and promotes RA-associated synovial inflammation via the NF-κB p65 pathway. In addition, the inhibition of Sam68 expression reduced TNF-α-induced inflammation in FLS suggesting its possible use as new therapeutic target for RA .
Methyltransferase-like 3 (METTL3), is a crucial component of the N 6-methyladenosine methyltransferase complex implicated in the most common internal mRNA modification identified in eukaryotes, m 6 A (Bokar, 2005); m6A modifications are linked to RNA stability, splicing, and translation (Huang et al., 2018;Xiao et al., 2016). Using human RA synovial tissues and the adjuvant-induced arthritis (AIA) animal model of RA, Shi et al. investigated the role of METTL3 in the inflammatory response and proliferation, invasion, and migration of FLSs during RA. METTL3 expression was significantly upregulated in both human RA synovial tissues and the rat AIA model. Additionally, they found that METTL3 knockdown reduced the production of IL6, matrix metalloproteinase (MMP) 3, and MMP9 levels in human RA-FLSs and rat AIA-FLSs (Shi et al., 2021).

| Type 1 diabetes
T1D is a complex metabolic disorder characterized by abnormal high blood glucose due to the destruction of pancreatic β islet cells via autoimmunity that reduce or eliminate insulin production (Atkinson & Eisenbarth, 2001). T1D is associated with several complications including neuropathy, cardiovascular disease, and peripheral vascular disease inducing to a considerable morbidity and mortality worldwide (Melendez-Ramirez et al., 2010). Genetic, epigenetic, as well as environmental factors have been implicated in the autoimmune pathogenetic mechanism leading to T1D; from the genetic perspective, more than 50 candidate genes act on both the immune system and pancreatic β cells to influence T1D onset (Gale, 2002). The first pathological hallmark of T1D is the inflammation of pancreatic islets caused by infiltration of immune cells such as CD4+ and CD8+ T cells and B cells (Coppieters et al., 2012) and the presence of autoantibodies directed against β cells in peripheral blood and lymph nodes (Willcox et al., 2009). As for other ADs, RNA regulatory networks controlled by RBPs can be altered in T1D and contribute to disease. Here, we discuss how changes in RBPs function or binding influence T1D onset or progression (Table 1).

| RBPs and type 1 diabetes
The functions of β cells, which are required to maintain blood glucose homeostasis, are characterized by a specific set of RBPs performing a variety of functions. Interestingly, analysis of RNA-seq data from human tissues revealed that β cells share RBPs pattern with the brain (Alvelos et al., 2018;Juan-Mateu et al., 2017), in keeping with the fact that neurons and β cells share extensive transcriptional networks (Van Arensbergen et al., 2010). Among the prominent RBPs that regulate mRNAs in pancreatic β cells there is HuD (human antigen D/ELAVL4) . In normal β cells, HuD expression is dependent on glucose and regulated through the insulin receptor (INSR) signaling pathway. Additionally, HuD regulates mRNAs essential for β cell functions including the Preproinsulin2 (Ins2) mRNA translation, the Autophagy-related Gene 5 (ATG5) mRNA and the mitochondrial gene Mitofusin2 (Mfn2) mRNA (Baltrusch, 2016;Fujimoto et al., 2009;Lee et al., 2012). Another RBP family with prominent function in β cells functions are the polypyrimidine-tract-binding proteins (PTBs). In pancreas, PTB proteins were shown to regulate insulin mRNA in both human (INS mRNA) and rodent (Ins1 and Ins2 mRNAs) (Fred et al., 2010;Tillmar et al., 2002).
There is growing interest in uncovering the role of alternative splicing (AS) in T1D and its regulation by RBPs. Indeed, dysregulation of AS in T1D pathogenesis was identified by different research groups; they found that RBPs with splicing functions, including RBPs NOVA 1 and 2 (NOVA Alternative Splicing Regulator 1 and 2), RBFOX1 (RNA Binding Fox-1 Homolog 1), RBFOX2 (RNA Binding Fox-1 Homolog2), and HuD, were preferentially enriched in β cells (Juan-Mateu et al., 2017). Changes in splicing may represent a mechanism of stress response affecting β cells at the onset of diabetes; thus, the study of AS regulation in β cells may represent a critical step to understand the pathogenesis of T1D and discover new drug treatments. Another RBP implicated in AS and T1D is CUGBP Elav-Like Family Member 1 (CELF1), a multi-functional protein with roles in AS, mRNA stability and translation. Verma and colleagues showed that the functions of CELF1 and RBFOX2 increase in T1D mouse skeletal and heart muscle, and are regulated via phosphorylation by the protein kinase C (PKC) (Verma et al., 2013). Importantly, increased CELF1 levels in T1D may impair glucose metabolism and insulin receptor signaling through regulation of AS to produce proteins with a roles in glucose metabolism (Nutter et al., 2016;Verma et al., 2013). The splicing factor SRSF6 (Serine and Arginine Rich Splicing Factor 6) is regulated by GLIS3 (GLIS Family Zinc Finger 3), a transcription factor encoded by a T1D susceptibility gene. Recently, using a cell line (EndoC-βH1), that recapitulates glucose uptake and insulin secretion human islet β cells, Alvelos and colleagues performed iCLIP analysis to identify SRSF6 target mRNAs in basal conditions. Several SRSF6-binding sites in T1D susceptibility genes were found, suggesting that SRSF6 may modulate the splicing of multiple T1D susceptibility genes. This scenario strongly support the presence of an AS-regulated network with a key role in T1D risk (Alvelos et al., 2020).
Chronic hyperglycemia due to destruction of β cells in T1D is associated with several systemic complications which are influenced by RBPs (Nutter et al., 2019). HuR, for example, is a major player in diabetic retinopathy and nephropathy. In the kidney of diabetic patients, HuR is upregulated in the cytoplasm, associated with stabilization of target mRNAs including those that encode CTGF (Cellular Communication Network Factor 2), TGFB1 (Transforming Growth Factor Beta 1), FOS (Fos Proto-Oncogene, AP-1 Transcription Factor Subunit), and SNAIL (Snail Family Transcriptional Repressor 1), and linked to epithelial-to-mesenchymal transition (EMT) and diabetic nephropathy (Yu et al., 2015). Thus, increased HuR levels in the cytoplasm facilitated EMT in renal epithelial cells, while suppression of HuR partially inhibited EMT of cells stimulated with high glucose. Furthermore, HuR bound to 3 0 UTRs of mRNAs encoding key transcription factors and cytokines involved in EMT (Yu et al., 2015). Along these lines, it was found that depletion of HuR improves retinal damage in diabetic mice and ameliorated kidney proteinuria, inflammation, and hypertrophy (Shang et al., 2015). HuR was also implicated in cardiovascular disease, as upregulation of HuR in diabetic human hearts correlated with increased inflammatory markers that can lead to cardiomyocyte death, while silencing HuR in a mouse model reduced infarct size after myocardial infarction. Altogether, these results indicate that HuR may be a general factor exacerbating diabetic complications in multiple tissues (Jeyabal et al., 2016;Krishnamurthy et al., 2010).

| AUTOIMMUNE DISEASE-ASSOCIATED VARIANTS IN RBPS BINDING SITES
We have discussed critical roles for RBPs in gene expression programs driving ADs and other diseases (Gebauer et al., 2021). From a genetic perspective, the ability of RBPs to affect disease onset can be influenced by genetic variability in three ways: (i) rare high-penetrance variants can alter RBPs function with a causal role in disease pathogenesis; (ii) common variants may modulate disease risk through alterations of RBPs function or expression; and (iii) diseaseassociated variant could affect a binding site for RBPs.
Regarding (i), the best-known high-penetrance variants altering RBPs function with a causal role in disease pathogenesis is the FMR1 (Fragile X Messenger Ribonucleoprotein 1) gene, which encodes the RBP FMRP involved in mRNA transport and translation in neurons. The presence of CGG triplet expansions in the 5 0 UTR of the FMR1 gene leads to Fragile X syndrome (FXS) and correlates with disease severity (Peprah, 2012). In this regard, no similar examples in ADs have been reported until now. Concerning (ii), common variants putatively modulate disease risk through alterations of RBP function or expression. For examples, SNPs located in the in CELF1 locus were associated with risk of Alzheimer's disease (Karch et al., 2016), and genetic variants located in the PUS10 (Pseudouridine synthases) locus were associated with MS risk (International Multiple Sclerosis Genetics Consortium, 2019). Unfortunately, until now the causal variant and the specific biological role for most identified SNPs are unknown.
Regarding (iii), the presence of disease-associated variants altering RBP binding sites and the extensive availability of GWAS identifying hundreds of causative loci, have allowed the identification of many risk variants for human diseases. Furthermore, the simultaneous large-scale discovery of binding sites of many RBPs through CLIP experiments permit the identification of genetic variants associated with human disease, including ADs, potentially altering RBP binding sites. To further investigate this possibility, we searched for risk variants associated with the four ADs analyzed here in order to find those that fall into RBP-binding sites. First, binding sites for 150 RBPs were downloaded from the ENCORE project website, which has reported a total of 226 eCLIP assay developed in three different cell types: HepG2 (n = 103), K562 (n = 121) and SM-9MVZL (n = 2) (Table S1). Filtering the data in the GWAS catalog by the disease name (including child traits) we found 2.014 unique variants with p-value <5 e-08 associated to the four considered ADs. Among these, 395 were associated with MS, 562 with RA, 694 with SLE and 363 with T1D. Next, we selected autoimmune associated variants located within the RBP binding sites, and found a total of 14 distinct variants, of which three different variant are associated with MS, two with RA, seven with SLE, and two with T1D (Table 2). Interestingly, the RBP with more disease-associated variants within the binding sites of RBP is the Pre-mRNA Processing Factor (PRPF) family; specifically, we identified the four different variants for PRPF8 and one for PRPF4. PRPF4 and PRPF8 associated with the mRNA to regulate spliceosome activation (Li et al., 2013). According to our data, five variants within five different PRPF protein binding sites are associated with RA, MS and SLE. Next, the Splicing Factor 3b Subunit 4 (SF3B4) is characterized by the presence of three variants within three different binding sites associated with MS and SLE. All the other RBPs present one or two variants that localize in binding site of target RNA (Table 2).
We extended the list by including all the AD-associated variants in linkage disequilibrium, LD, (1000G EUR superpopulation, r2 ≥ 0.8) with variants present within the RBP binding sites, and identified a total of 142 associations, of which 46 correspond to MS, 22 to RA, 50 to SLE, and 24 to T1D (Table S2). The identification of AD-associated variants in LD with variants present that fall within RBP binding sites further support the relevance of genetic variation in target gene beyond that explained by GWAS associations. Altogether, our analysis support future studies aimed at assessing the relevance of genetic variants which follow within the RBP binding sites and to investigate them role in RBP regulatory pathways (Figure 2).

| CONCLUSIONS AND PERSPECTIVES
In this review, we have discussed the roles of RBPs in four major ADs and seek to identify genetic variants relevant in this context (Figure 3).
Over the past decades, several RBPs have been found to have clear roles in the regulation of several immune pathways and ADs. These RBPs can regulate target mRNA fate by influencing post-transcriptional events like mRNA splicing, transport, stability, and translation. The precise regulation of these processes prevents aberrant activation of the immune, endocrine, musculoskeletal, and nervous systems, and consequently the onset and progression of ADs. Indeed, several RBPs have been shown to be involved in the development, homeostasis, and regulation of immune responses as well as autoimmune pathogenesis (Yoshinaga & Takeuchi, 2019). Proper RBP functions prevent the abnormal activation of the immune responses, thereby preventing the development and/or progression of ADs (Yoshinaga & Takeuchi, 2019). Although there are common features and molecular mechanisms driving disease onset in the different ADs analyzed here, and among ADs in general, it is important to highlight that there is a great level of variability among them, even within the same disease. Common elements can also be linked to similar regulations driven by RBPs. For example, HuR has been implicated in the binding and regulation of mRNAs encoding for several cytokines including IL-1β, IL-6 IL-17, TNF (Chen Cascio et al., 2013). IL-6, a multifunctional pro-inflammatory cytokine, is secreted by T cells and macrophages to stimulate immune response during inflammation and infection; it is involved in the pathogenesis of various autoimmune and chronic inflammatory diseases including MS, T1D, SLE, and RA analyzed in this manuscript (Ishihara & Hirano, 2002). Thus, HuR-mediated de-regulation of IL-6 may influence several ADs and affect many aspects of immune regulation. Also, deregulation of the cytokines BAFF, mediated by loss of NF90 DDX3X Target mRNA: CDKN1B

5ʹUTR 3ʹUTR
rs34330 F I G U R E 2 Schematic representation of the RBP DDX3X and its target CDKN1B mRNA, an example of our analysis. The red star indicates a SNP identified by our analysis associated with SLE risk and a binding site for DDX3X.
binding site has been associated with increased MS and SLE risk (Idda et al., 2018). These examples emphasize the overlap in the immune-deregulation pathway implicated in ADs and highlight the need of accurate analysis to understand deseases pathogenesis and, thus identify proper therapies. Furthermore, while most RBPs were shown to be critical in AD pathogenesis in several ways, only few examples have been shown to be associated with increased genetic susceptibility to ADs in human (e.g., PUS10 in MS and EIF5A in T1D) ( F I G U R E 3 Schematic representation of the RBPs analyzed in this manuscript and involved in the regulatory mechanism of MS, SLE, RA and T1D. Abbreviations: RBP: RNA binding protein; MS, Multiple sclerosis; RA, Rheumatoid arthritis; SLE, Systemic lupus erythematosus; T1D, Type 1 diabetes histocompatibility complex (MHC) locus provides the greatest genetic risk factor for AD development and is a key link between different ADs. Non-HLA proteins such as CTLA4 (Cytotoxic T-Lymphocyte Associated Protein 4), PTPN22 (Protein Tyrosine Phosphatase Non-Receptor Type 22), and TNF have also been associated with different ADs (Tavares et al., 2015) but clear mechanisms leading to ADs still need to be identified. Further studies are necessary to investigate and properly understand the pathogenic roles of RBPs in human diseases. Yet, the increasing interest on the role of RBPs in AD pathogenesis have prompted scientists to characterize disease-associated variants affecting RBP levels and function. Therefore, we searched for genetic variants associated with AD risk that can alter the binding site of RBPs, and thereby affect its regulatory function on target RNAs. Our preliminary findings suggest that the impact of genetic variability on the binding of RBPs to target transcripts warrant further investigation.
Until few years ago, RBPs were considered "undruggable," due to a lack of enzymatic pockets, the presence of unstructured regions in RBPs, and the high structural similarity among members of the same RBP family. With the development of new approaches based on the modulation of RNA-RBP interactions by chemically modified antisense oligonucleotides (ASOs) that block the access of the RBPs to the target binding sites, or by small molecules such as target site blockers (TSBs) that alter RNA-protein interactions, the landscape of RBPs as therapeutic targets has changed drastically, and many emerging studies have focused on RBPs as drug targets for human diseases. RBPs modulate different mRNAs involved in several pathways implicated in the pathogenesis of ADs, supporting their use as targets to control disease progression and response to therapy (Yoshinaga & Takeuchi, 2019). In the future, additional studies in animal models and in humans are necessary to better understand the mechanism underlying RBP dysfunctions in AD pathogenesis, and the possible role of RBPs as a therapeutic target.
To close, our report underlines the roles of RBPs and post-transcriptional events in four AD pathologies-MS, RA, T1D and SLE. Recent studies reported in this manuscript clearly support the role of RBPs in rodent and human models of ADs and the possibility of modulating them to ameliorate AD symptoms using modified oligonucleotides or ASOs or TSBs that can exploit the interactions of RBPs with target mRNAs. Importantly, as for miRNA, attention is necessary for drug design targeting RBPs as the same RBP can have multiple targets and different roles (in some case, even opposite roles) in different tissues. We anticipate that deeper knowledge of the molecular biology of RBPs with a more detailed understanding of the genetics of RBPs, will pave the way to new therapeutic approches with increased efficiency and precision.

RELATED WIREs ARTICLES
Advances and challenges in the detection of transcriptome-wide protein-RNA interactions