Precursor (pre) mRNA splicing is an intricate biological process involving intron excision and exon ligation to generate mature mRNA products. Splicing recognizes exon–intron boundaries and is catalyzed and controlled by complex ribonucleoprotein complexes known as spliceosomes (7). The spliceosome consists of at least 300 associated proteins and five uracil-rich small nuclear ribonucleoproteins (U1, U2, U4, U5, and U6 snRNPs) (8). Proper spliceosome assembly at the splice sites (ss) is the core reaction mechanism. The 5’ss is located at the 5’ end of each intron, the 3’ss is located at the 3’ end of each intron, and the branch point (BP) site is located 18–40 nucleotides upstream of each 3’ss (9). Spliceosome assembly begins with adenosine triphosphate (ATP)-independent recognition of the 5’ss by U1 snRNP. U2 auxiliary factor (U2AF) then binds the 3’ss and the polypyrimidine region. Splicing factor 1 (SF1) then binds the BP ( Figure 1 ) (8, 10). These steps form the stable early complex E, which then triggers ATP-dependent U2 snRNP recruitment, SF1 replacement at the BP, and conversion into the pre-spliceosome (complex A) (10). The pre-assembled U4/U6–U5 tri-snRNP complex is recruited and forms a pre-catalytic spliceosome (complex B), which is then converted into a catalytically active complex B* by removing U1 and U4. Two subsequent transesterification steps and conformational and compositional rearrangements form complex C. Exon ligation follows, and mature mRNA is formed (11).

Splicing may be constitutive or alternative. Constitutive splicing events usually generate single transcripts from the pre-mRNAs. By contrast, alternative splicing events occur when at least two 5’ss or 3’ ss compete (12). Constitutive splicing usually occurs in the presence of strong splice sites containing consensus sequences that are easily recognized by the spliceosome. However, alternatively spliced exons usually contain weaker splice sites that can only be recognized via the mediation of additional splicing regulatory elements (SREs).

Alternative splicing arises from specific exons/introns that may or may not be included in a mature mRNA transcript. The five basic alternative splicing patterns are exon skipping, mutually exclusive exons, alternative 5’ss, alternative 3’ss, and intron retention (12). Alternative splicing is controlled by cis-regulatory elements and trans-acting factors. Certain epigenetic factors, such as transcription, chromatin structure, DNA methylation, and histone modifications, may also affect SRE function. Depending on their relative positions and activities, cis-regulatory elements are classified as exonic splicing enhancers (ESE), intronic splicing enhancers (ISE), exonic splicing silencers (ESS), or intronic splicing silencers (ISS) (13). Cis-regulatory elements can recruit trans-acting splice factors (RNA-binding proteins or RBPs) either to promote or suppress the recognition of nearby splice sites. Therefore, various RBP categories, expression levels, and activities may influence the regulation of alternative splicing. Thus far, many RBPs have been found to modulate the splicing mechanism. These include serine/arginine-rich (SR) proteins, heterogeneous nuclear ribonucleoproteins (hnRNPs), neuro-oncological ventral antigens (NOVA) proteins, polypyrimidine track binding proteins (PTB), forkhead boxes (FOX) proteins, and others. Of these, the most extensively studied are the SR proteins and the hnRNPs (14).

SR proteins or SR-rich splicing factors (SRSFs) structurally resemble each other and are characterized by the RS domain [rich in arginine (R) and serine (S) residues] at the C-terminal and one or two RNA recognition motifs (RRMs) at the N-terminal (13). The SR protein is usually a positive splicing factor and it facilitates splice site recognition. When it binds the ESE, the SR protein recruits U1 and U2 snRNPs and auxiliary factors, such as U2AF. In this way, it induces spliceosome assembly followed by exon inclusion (15). SR protein mutation or aberrant post-translational modification may lead to various diseases. Correct SR protein phosphorylation is vital in the regulation of the splicing event and alterations in subcellular localization, activity, and stability that may affect protein–protein and protein–RNA interactions (16). By contrast, hnRNPs are usually considered negative splicing factors that inhibit splice site recognition. For example, hnRNP blocks spliceosome access to the polypyrimidine tract by binding the ESS and promoting exon exclusion.

Alternative splicing is a pivotal mechanism widely acknowledged to regulate gene expression and promote protein diversity. The maintenance of functional immune responses requires protein diversity and flexibility, which are essential in alternative splicing regulation. Many genes involved in innate or adaptive immune signaling undergo varying degrees of alternative splicing. Ergun et al. determined by RNA sequencing and microarray that alternatively spliced isoforms occur in 60% of all T- or B-lymphocyte genes (17). Based on spliced gene functions, alternative splicing may affect the physiological function of the immune system in different ways.

Basic alternative splicing patterns and its regulatory mechanisms are widely acknowledged. In autoimmunity, the disease-related pathogenesis varies from one to another. Here, we conclude the common features of alternative splicing in the pathological changes of autoimmune diseases.

IRF5 is a transcription factor belonging to the IRF family. IRF5 mainly regulates interferon expression by activating the innate immune response to viruses or inflammatory stimuli. Moreover, IRF5 modulates cellular maturation, differentiation, and apoptosis as well as proinflammatory cytokine generation (41). Both innate and adaptive immune signaling activate IRF5 via the TLR and MyD88 pathways (42). IRF5 overexpression is associated with clinical SLE progress. The human IRF5 gene has nine coding exons and at least four non-coding alternative exons designated 1A, 1B, 1C, and 1D (41, 43). By differential selection of the coding or non-coding alternative exons, human IRF5 could theoretically generate multiple distinct functional isoforms. Some of these have been identified and were designated V1–V11. The spliced variants generated via the inclusion of alternative exon 1B (v2, v9, and v10) are strongly linked to IRF5 upregulation and SLE susceptibility. However, genetic polymorphisms in the human IRF5 locus disrupt normal splicing and increase the risk of aberrant spliced variant formation. The distinct haplotypes of four functional SNPs increase susceptibility to SLE and other autoimmune diseases. For example, the T allele of rs2004640 creates a novel 5’ss at the intron-exon boundary of alternative exon 1B. Thence, exon 1B is included in transcript production. The G allele of rs10954213 destroys the polyadenylation (polyA) site in the 3’ UTR region of exon 9. This mechanism arrests transcription and results in a relatively long, unstable transcript (44). A 5-bp (CGGGG) indel (insertion/deletion) located in the proximal promoter creates a novel binding site for the Sp1 transcription factor, which leads to alternative splicing (45). Furthermore, 30-bp indel polymorphisms in exon 6 select two alternative 3’ss. Insertion of these 3’ss produces V5 and V6 isoforms, whereas their deletion generates V1 and V4 isoforms (41). These isoforms are structurally and functionally distinct and have different expression levels. Hence, they influence IRF5 signaling and its downstream immune responses.

Murine IRF5 (MuIRF5) shares approximately 87% amino acid sequence homology with human IRF5. Different from the heavily spliced human IRF5, MuIRF5 primarily generates a full-length transcript. Nevertheless, Paun et al. have detected a splicing variant of IRF5 from the bone marrow of C57BL/6J mice, termed as BMv, which contains a deletion of the entire exon 5 and part of exon 4, 6 (46). In the MyD88 activated innate immune signaling, BMv serves as a relatively weaker activator to the downstream IFNA4 promoter, compared to the full-length transcript.

Leukocyte immunoglobulin-like receptors (LILRs) are also known as CD85. This family of leukocyte receptors has 13 members and includes the two pseudogenes LILRP1 and LILRP2. LILRA2 is a stimulatory receptor that delivers positive signals and is widely expressed on B cells, dendritic cells, monocytes, macrophages, and a subset of natural killer (NK) cells. LILRA2 cross-links with the immunoreceptor tyrosine-based activation motif (ITAM) of Fc-γ receptor (FcRγ) and induces calcium mobilization and subsequent signaling activation (47). Genetic polymorphisms of human LILRA2 were associated with increased SLE susceptibility in Japanese populations (48). The SNP rs2241524 G>A at the intron 6–exon 7 junction disrupts the consensus splicing acceptor motif and interferes with normal splicing. This mechanism results in a novel transcript with an in-frame deletion of three amino acids (Ala-Ser-Leu) at position 419–421 in the linker region (48). The impact of this spliced variant (Δ419–421 isoform) on ligand binding is unknown.

The B-cell scaffold protein with ankyrin repeats 1 (BANK1) gene encodes a scaffold adaptor protein expressed primarily in B cells. Stimulation of B-cell receptors (BCRs) induces tyrosine phosphorylation in BANK1, which then amplifies B-cell signaling by facilitating the activation of downstream signal molecules, such as Src-family kinases, the inositol 1,4,5-trisphosphate receptor (IP3R) calcium channel, phospholipase Cγ2 (PLCγ2), protein kinase C β (PRKCB), and Ras (49, 50). BANK1 variants have been linked with susceptibility to multiple autoimmune diseases, such as SLE, RA, and SS (51–53). Kozyrev et al. analyzed BANK1 cDNA and identified a novel spliced transcript (50) designated Δ2 isoform because it lacks the entire exon 2. It encodes a protein without putative IP3R-binding or PH domains. This defect attenuates BANK1-mediated signaling. The Δ2 isoform is also detected from chimpanzee and mouse spleen. Three SNPs (rs10516487 G>A in exon 2, rs17266594 T>C in intron 1, and rs3733197 G>A in exon 7) associated with SLE might also influence Δ2 isoform splicing. The rs17266594 is localized to a putative splice BP of exon 2. This mutation affects relative splicing efficiency and upregulates the Δ2 isoform. The rs10516487 is associated with a decrease in exon 2 splicing (54). The precise role of BANK1-mediated signaling downregulation in SLE is unclear, but it may participate in SLE pathogenesis as B cells have vital functions in autoimmunity.

Ras guanyl releasing protein 1 (RasGRP1) is a guanine-nucleotide-exchange factor that is first cloned in the brain and is then highly expressed on T cells (55). RasGRP1 is linked to autoimmunity. RasGRP1 deletion may lead to a deficiency of thymocyte differentiation or T-cell development in mice (56). In humans, RasGRP1 deficiency may interfere with T-cell and B-cell proliferation and activation, thereby causing various autoimmune disorders (57). T-cell receptor (TCR) stimulation drives protein tyrosine kinase activity and diacylglycerol (DAG) binding, which, in turn, recruit son of sevenless (SOS) to the plasma membrane and activate Ras (58, 59). RasGRP1 participates in TCR-induced signaling and activates Ras. RasGRP1 consists of a DAG-binding domain and two calcium-binding EF hands. Yasuda et al. identified 13 new spliced human RasGRP1 transcripts generated by alternatively including or deleting exons 5–17. These novel transcripts constitute an isoform family (splice variants A–M) (59). Nine of these were translated into truncated nonfunctional RasGRP1 isoforms because of a premature translation termination codon. The remaining four, however, did not cause any frameshift abnormalities or induce early translation termination. Compared with normal individuals, SLE patients lack the functional RasGRP1 isoform and have elevated levels of defective variant isoforms.

Splicing factors, such as SRSF1, regulate alternative splicing. The RasGRP1 transcript lacking exon 11 (splice variant A) is the type most observed in the T cells of SLE patients. SRSF1 directly binds the exon 11 of RasGRP1 mRNA, thereby promoting exon 11 inclusion. Kono et al. found that compared with the T cells of healthy individuals, those of SLE patients presented with significant SRSF1 downregulation. SRSF1 knockdown in healthy human T cells via small interfering RNAs (siRNAs) increased the splice variant A:normal RasGRP1 isoform ratio and downregulated RasGRP1 (60). Moreover, hnRNP-K protein was correlated with RASGRP1 expression and extracellular signal-regulated kinase (ERK) phosphorylation (61).

The TCRζ chain is involved in TCR/CD3 complex assembly and surface expression (62). The cytoplasmic domain of TCRζ includes three ITAMs that are phosphorylated when TCR is activated. This mechanism leads to the recruitment of other signaling molecules and, in turn, the phosphorylation or activation of downstream signaling transduction molecules (63–65). T cells are central to SLE pathogenesis. In SLE, T cells display multiple abnormalities in their antigen‐mediated signaling. Recent studies showed that SLE patients had lower TCRζ mRNA expression levels than healthy subjects (66–68), possibly because of low transcription activity, splice variant generation, and other factors (69, 70). Alternative splicing generated a series of transcript variants of TCRζ, including TCRη (exon 8 replaced by exon 9), TCRι (exon 8 replaced by exon 10), an isoform lacking exon 7, and another with a short 3’UTR (69, 71–74). Similar variation of TCRζ transcript also exists in animal models. For example, TCRη contains completely uniform alternative exon 8 selection to human, and is observed in both mice and swine, while TCRι is observed in mice (75, 76). TCRη and TCRι are natural isoforms with normal function, whereas the other two are relatively unstable, more readily degraded than their full-length transcript counterparts, and detected exclusively in SLE patients. Variant isoforms affect the assembly of the TCR/CD3 complex and its expression on the cell surface. In addition, it changes the Treg/Th17 balance. Consequently, they attenuate normal signaling and result in T-cell dysfunction and insufficient IL-2 production (77–79).

A previous study demonstrated that epigenetic modification or changes in the levels of splicing factors may affect the mechanism regulating alternative splicing. In human T cells, SRSF1 can bind the 3’UTR region of TCRζ mRNA, thereby favoring the expression of TCRζ isoforms with full-length 3’UTR over unstable short ones (80). SLE patients usually present with relatively low SRSF1 levels and enhanced SRSF1 ubiquitination (81, 82). SRSF1 may increase total TCRζ mRNA expression, whereas SRSF1 knockdown may decrease it. The specific mechanism involved in SRSF1 regulation remains to be determined. However, SRSF1 might be a suitable therapeutic target for the restoration of normal T-cell function.

Many studies have identified numerous loci of disease susceptibility-related genes associated with certain specific splicing patterns. In addition to the foregoing gene loci, there are several less common spliced variant isoforms related to SLE susceptibility.

CD72 is an inhibitory receptor expressed mainly on B cells. The CD72 SNP affects normal splicing and generates a spliced isoform lacking exon 8 (CD72Δex8) that modulates antibody production. Polymorphisms of the Complement C3d Receptor 2 (CR2) gene also regulate alternative splicing. A haplotype formed by minor CR2 SNP alleles (rs17615 and rs1048971 in exon 10 and rs4308977 in exon 11) might affect splicing efficiency and significantly lowers the risk of SLE (83). The SNP of CR2 known as rs3813946 is located in the 5’UTR and may also influence CR2 transcription. CTLA4, also known as CD152, is a protein receptor expressed mainly on T cells. Variants of human CTLA4 are associated with a wide range of autoimmune diseases, including SLE, RA, T1DM, GD, and MS. Alternative splicing of human CTLA4 generates various transcripts, such as a membrane-bound form (mCTLA4), a soluble form (sCTLA4), and four other rare transcript types. Al Fadhli et al. reported that patients with SLE or RA express neither mCTLA4 nor sCTLA4 (84). sCTLA4 may interfere with the B7:CTLA4 interaction and block negative signals. However, the roles of various CTLA4 isoforms in autoimmune disease mechanisms remain to be clarified and additional studies are required to explore these signaling pathways.

CD44 is a cell-surface glycoprotein involved in cell survival, proliferation, migration, and traffic signal transmission, cell–cell interactions, and programmed cell death (88, 89). CD44 comprises 10 constant and 10 variant (v1–v10) exons inserted between the 5’ and 3’ ends (90). Standard CD44 (CD44s) is the most ubiquitous form. It contains only constant exons and is expressed mainly on hematopoietic cells. Alternative splicing of the variant exons and N-glycosylation, O-glycosylation, and glycosaminoglycanation lead to numerous variant isoforms (CD44v). However, only a few dozen of these have been identified thus far (91). CD44 is closely related to multiple autoimmune disorders, including RA, SLE, MS, and inflammatory bowel disease (IBD). Compared with osteoarthritis patients, RA patients have relatively higher synovial CD44s, CD44v4, CD44v6, and CD44v7–8 transcript levels (92, 93). CD44v3 and CD44v6 are associated with increased tumor cell migration and invasion because they add adhesion and MMP docking domains. A similar mechanism occurs in RA fibroblast-like synoviocytes (FLS), and it augments their ability to invade Matrigel (94). Elevated serum levels of soluble CD44 (sCD44) variant isoforms, such as sCD44v5 and sCD44v6, have been detected in RA patients. Serum sCD44v5 is correlated with inflammation and could serve as a predictive factor in responses to RA therapy (95, 96).

Tumor necrosis factor (TNF)-α is a pleiotropic proinflammatory cytokine implicated in inflammation, cell proliferation, differentiation, and apoptosis (97). TNF-α is one of the most potent proinflammatory cytokines in RA. TNF signaling is mediated by the receptors TNFR1 and TNFR2. Both of these belong to the TNFR superfamily and have high affinity for TNF-α (98). TNFR1 is widely expressed on most cells and is both proinflammatory and apoptotic. TNRF2 expression is limited mainly to endothelial cells and certain immunocytes and participates in anti-inflammatory processes and cell proliferation (99, 100). Proteolytic cleavage produces soluble forms of TNFR (sTNFR). Elevated serum sTNFR is observed in certain infections, cancers, and autoimmune disorders, such as RA, SLE, and SS (101, 102) Recent studies identified an alternatively spliced TNRF2 isoform lacking exons 7 and 8 that encode transmembrane and cytoplasmic domains (DS-TNFR2) (103, 104). TNRF2 is a soluble receptor that acts as an antagonist in TNF‐α-mediated signaling and blocks apoptosis (104). Cañete et al. demonstrated that alternatively spliced DS-TNFR2 constitutes the majority of the sTNFR2 in RA patients, and serum sTNFR2 is closely associated with RA activity and severity (101, 103).

IL-6 is a pleiotropic cytokine that is crucial in chronic inflammation and multifactorial autoimmune disorders, such as RA, IBD, MS, and Castleman’s disease. IL-6 overexpression may contribute to clinical RA symptoms. In classic signaling, IL-6 binds the cognate ligand-binding subunit (IL‐6R, CD126). This mechanism results in the recruitment of the signal-transducing components gp130 and CD130, which, in turn, promote anti-inflammatory responses (105). Gp130 is widely expressed on most cell types, whereas IL-6R is detected on only a few specific cells, such as monocytes, macrophages, neutrophils, lymphocytes, and hepatocytes (106). IL-6R can exist in both membrane-bound (mIL-6R) and soluble (sIL-6R) form. About 90% of all sIL-6Rs are transformed from mIL-6R via limited proteolysis of metalloproteinase domain-containing protein 17 (ADAM17) (106, 107). The SNP rs8192284 A > C localized to the proteolytic cleavage site of exon 9 was associated with an increase in mIL-6R shedding (108, 109). The exclusion of exon 10 encoding the transmembrane region and IL-6R mRNA splicing generate a minor proportion of the sIL-6R (110). The loss of mIL-6R disrupts classic IL-6 signaling whereas formatted sIL-6R is biologically active. By binding IL-6, the sIL‐6R/IL‐6 complex can interact with gp130, which leads to intracellular signaling in cells that do not otherwise express IL-6R (111–113). This process is known as IL-6 trans-signaling and it induces proinflammatory responses (114). Upregulated sIL-6R has been observed in both RA and osteoarthritis. Hence, sIL-6R is implicated in joint inflammation and destruction (115). IL-6R inhibitors, such as tocilizumab and sarilumab, target binding to mIL-6R and sIL-6R. Therefore, they block both classic IL-6 signaling and trans-signaling. Their efficacy in RA treatment has been demonstrated (116). Recently, soluble gp130 (sgp130) variants arising from alternative splicing or polyadenylation have been identified. Sgp130 variants are considered natural trans-signaling inhibitors. Their three known isoforms are full-length sgp130, sgp130-RAPS, and sgp130-E10 (117, 118). Fc-dimerized sgp130 protein (sgp130Fc) blocks IL-6 trans-signaling and has been evaluated in the treatment of various inflammatory disease models. It is currently undergoing clinical trials as a drug candidate for IBD (119) and is also a potential therapeutic agent for RA.

Along with the foregoing soluble isoforms, alternative splicing generates others associated with RA such as sIL-7R (excision of exon 6 containing the transmembrane domain) (120), sCD137 (lacking the 414–545 nucleotide region and characterized by a frameshift) (121), sCD1d (excision of exons 4 and 5 containing the transmembrane domain) (122), and FasΔTM (excision of exon 6) (123).

Vascular endothelial growth factor (VEGF) is a highly specific vascular endothelial cell mitogen. In vivo, VEGF induces microvascular permeability and plays central roles in regulating angiogenesis and tumor neovascularization (124). Prior studies demonstrated that VEGF-mediated angiogenesis also occurs in the synovial tissues of RA patients. In RA, VEGF polypeptide is highly expressed on synovial macrophages, fibroblasts, vascular smooth muscle cells, and synovial lining cells and in synovial fluids (125, 126). Human VEGF comprises eight exons and seven introns. Alternative selection of exons 6–8 and VEGF mRNA splicing generate the variant transcripts VEGF−A121, VEGF−A145, VEGF−A165, VEGF−A189, and VEGF−A206. These are generically designated VEGF−Axxx where xxx indicates the number of amino acids in the mature protein (127). VEGF121 and VEGF165 are the only VEGF-spliced isoforms expressed in RA. VEGF165 has 15 basic amino acids in 44 residues encoded by exon 7 and it binds heparan sulfate. By contrast, VEGF121 lacks this region and its diffusibility differs from that of VEGF165 (128).

Fibronectin (Fn) is a multifunctional glycoprotein that is widely expressed in the extracellular matrix, plasma, synovial fluid, and other fluids. From a single gene, alternative Fn mRNA splicing generates multiple distinct isoforms that play vital roles in embryonic development, inflammation, wound repair, malignant metastasis, and thrombosis. Recognized alternative splicing sites include extra domain A (EDA), extra domain B (EDB), and type III connecting segment (IIICS) (129, 130). EDA and EDB each encode a single homology type III repeat (131). RA patients present with elevated synovial EDA(+)FN. This marker is correlated with accelerated rheumatoid joint destruction. The EDA(+)FN synovial fluid level is proportional to the severity of RA signs and symptoms and is promising as a predictor of RA progression (132). In animal model, Gondokaryono et al. found that EDA(+)FN might induce joint swelling in a mast cell TLR4-dependent manner by injecting EDA(+)FN to mast cells that lacked W/Wv mice (133). Removal of EDA(+)FN from the plasma can efficiently and effectively mitigate RA symptoms. EDA(+)FN cryofiltration and selective EDA(+)FN adsorbents have been developed for this purpose (134). Furthermore, heparin has high affinity for FN and can selectively bind and remove EDA(+)FN from the plasma.