Alternative splicing is an essential process in post-transcriptional mRNA processing, and produces various mature mRNAs with different structures and functions. In this process, exons are taken together in different combinations and introns are removed. Recent data indicate that each transcript of protein-coding genes contain 11 exons and produce 5.4 mRNAs on average [1]. So far, seven basic types of alternative splicing have been identified, including exon skipping, alternative 5′-splice site, alternative 3′-splice site, mutually exclusive exons, intron retention, alternative promoter, and alternative polyadenylation [2] (Figure 1). A notable example of alternative splicing is the human gene TTN which encodes muscle protein titin and contains 364 coding-exons and 4039 different splicing events which have been identified by RNA-sequencing [3]. Most genes generate at least two transcript variants. The alternative spliced mRNAs are further translated into many protein variants which differ in function and structure. The precision and diversity of alternative splicing events are aided by many significant factors, such as the strength or weakness of splice sites, the concentration and combination of enhancing and silencing splicing factors, chromatin modifications, and RNA secondary architectures [3]. Since the activity of splicing factors and the spliced variants change the developing process of diseases, they can also serve as experimental indicators or biomarkers for diagnosis.

Cancer is a complex disease which can be caused by various factors. It has been determined that gene expression profiles of tumor cells are different from normal samples. Instable cellular homeostasis is an important cause of cancer, and has been reported to be closely related to aberrant alternative splicing. Since alternative splicing plays a key role in post-transcriptional regulation and controls the formation of spliced variants, the mutations and changed level of splice factors may contribute to tumorigenesis.

Abnormal expressions of specific splicing isoforms can cause specific cancers but also can serve as a biomarkers and therapeutic targets. The TCGA database allows one to analyze the expression of alternative splicing patterns during cancer progression [11]. Increasing evidence demonstrates the dysregulation of alternative splicing leads to the production of tumor-associated isoforms that further impact cellular activity [4], such as sustaining proliferation, preventing cell death, rewiring cell metabolism, promoting angiogenesis, enabling cell invasion and metastatic dissemination, and enabling drug resistance [12]. For example, the mutation of SF3B1 is highly related with several cancers including chronic lymphocytic leukemia (CLL), cutaneous melanomas, and uveal melanomas. SF3B1 contains an essential region which interacts with SF3B14a and further forms a complex with U2AF2, playing a key role in BPS recognition. However, the mutation of SF3B1 at exons 12–15 disturbs the interaction between SF3B1 and SF3B14a, leading to the prevention of BPS recognition and 3′-ss mis-selection. Moreover, the mutated SF3B1 functions as an anti-apoptotic factor which is regularly detected in various cancers in order to sustain the cell proliferation [13]. SR proteins are responsible for multiple cellular physiological process and involved in the alteration of gene expression in tumors. SRSF1 is often upregulated in breast tumors through binding with MYC, leading to increasing of cell proliferation and decreasing of cell apoptosis. Additionally, SRSF1 overexpression in lung cancer leads to resistance to the chemotherapy drugs cisplatin and topotecan. Previous studies have demonstrated that the SRSF2 mutant is associated with myelodysplastic syndromes (MDS), since the mutated SRSF2 expression alters the binding specificity, inducing the inclusion of a premature termination codon (PTC). This PTC locates in EZH2 region which encoding a histone methyltransferase related to the pathogenesis of MDS. SRSF6 overexpression is detected in skin cancer, promoting the splicing of cassette exons and inducing hyperplasia [14]. The hnRNPs play an important role in regulation of pre-mRNA splicing. Abnormal expression of hnRNPs affects RNA splicing, results in alteration of RNA expression levels, and further causes the occurrence of cancer. As the most abundant hnRNPs, hnRNP A1/A2 has been reported to be upregulated in lung cancer. Overexpression of hnRNP A1/A2 may function as carcinogenic factor to promote cell proliferation. Moreover, hnRNP A1 and hnRNP A2 participate in recognizing and protecting telomeric sequences. Thus, they are related to cancer regulation. In contrary, silencing of hnRNPA1/A2 expression causes tumor cell apoptosis [15]. In addition, miRNAs can modulate splicing-factor expression and function as oncogenes. miR-30a-5p and miR-181a-5p regulate SRSF7 in renal tumors, thereby altering the splicing pattern ^[16][17][16,17]. Overexpression of splicing associated miRNAs has been detected in a variety of cancers. For example, repression of SRSF1 results in upregulation of miR-10a and miR-10b, thereby promoting terminal differentiation of neuroblastoma cells.

Alternative splicing events generate protein isoforms related to cancer hallmarks, promoting tumorigenesis. Gene RPSkKB1 alternatively encode two protein isoforms, full-length RPS6KB1-1 and RPS6KB1-2 lacking kinase domain. The production of RPS6KB1-2 is regulated by SRSF1 mediated AS, and contributes to tumor growth in lung and breast cancer, while RPS6KB1-1 suppresses proliferation of cancer cells ^[15][18][15,18]. Similarly, oncogene CCND1 encoding cyclin D1 has two isoforms, namely conventional cyclin D1a and cyclin D1b lacking the C-terminal protein domains. Cyclin D1b C-terminal domain encoded by the exon 5 is a GSK-3β phosphorylation site, allowing the cyclin D1b to be transported from the nucleus to the cytoplasm. Altered selection of 5′-ss induce exclusion of exon 5, thereby causing cyclin D1b to be trapped in the nucleus. As opposed to cyclin D1a, the overexpression of cyclin D1b variant was observed in breast cancer tissues compared to normal breast tissues, which characterizes metastasis and invasive migration of cancer cells mediated by αvβ3 and toll-like receptor 4 (TLR4) ^{[19][20][21][22][23]}[19,20,21,22,23] (Figure 2).

Neurodegenerative disease constitutes a variety of mental and neuromuscular disorders, including Alzheimer’s disease (AD), Parkinson’s disease (PD), spinal muscular atrophy (SMA), and familial dysautonomia (FD). It has been found that alternative splicing plays an important role in neurological diseases.

Neuro-oncological ventral antigens 1 and 2 (NOVA1 and NOVA2) are two members of the NOVA family, regulating neuron-specific alternative splicing. They belong to splicing factors and mainly enriched in brain tissue and are responsible for neuronal viability and maturity. NOVA1 is distributed in the cerebellum and spine, whereas NOVA2 is located in the cortex. NOVAs regulates various receptors and voltage-gated channels which are critical for the signal transduction of neuromuscular junctions. As an alternative splicing method for a key regulator of glycine α2 exon 3A (GlyRα2E3A) pre-mRNA splicing, NOVA1 recognizes an adjacent intronic splice site, alternatively inducing exon 3A inclusion. NOVA1, together with TDP-43, participates in RNA recognition and induces various neurodegenerative diseases such as schizophrenia and amyotrophic lateral sclerosis [24]. NOVA1 participates in the splicing of RNA-binding protein RBM8A and regulates RNA metabolism, causing neurological damage-like symptoms of AD.

Alternative splicing events frequently occur in the progression of AD. FynT is an isoform of tyrosine kinase protein Fyn, which is upregulated in AD patients and correlated with chronic neuro-inflammation [25]. Alternative splicing of exon 15 of Amyloid precursor protein (APP) transcript changes the ratio of APP proteolytic fragments betaA4 and p3, which contributes to the formation of amyloid plaque and aggravate AD [26]. Another major toxic in AD is microtubule-stabilizing protein tau. Tau has two isoforms, four microtubule repeats (4R-tau), and three microtubule repeats (3R-tau). The generation of both isoforms are determined by the inclusion or skipping of exon 10, which are further regulated by splicing factor SRSF6 [27]. The two isoforms expression level of 3R-tau and 4R-tau have to maintain a balance in healthy individuals. However, dysregulation of tau splicing disturbs the ratios, thereby causing neurofibrillary degeneration in AD patients [28]. AMPA serves as a critical mediator of synaptic transmission. AMPA receptor subunits of PD patients have been regulated alternatively, which are potential biomarkers for PD diagnosis and therapies [29]. SMA is a common autosomal recessive disease which is partially caused by the decrease of SMN protein. Human beings encoded two SMN genes, namely SMN1 and SMN2. They are almost identical and encode SMN protein, which is crucial for snRNP assembly. However, SMN2 is vulnerable to mutate in exon 7 due to the replacement of C by T at position 6. Hence, during the post transcriptional processing, exon 7 is skipped in SMN2 pre-mRNA splicing, generating truncated isoforms which cannot compensate the function of SMN ^[30][31][30,31]. Sodium butyrate has been detected to modulate exon 7 inclusion of SMN2 transcript, thereby producing the full-length SMN protein [32]. Familial dysautonomia (FD) is a recessive disease mainly caused by the mutation of the i-kappa-B kinase complex associated protein (IKBKAP). The mutated IKBKAP loses its function, and thereby FD patients show a demyelination phenotype. Single-base changes in less well-conserved nucleotides affecting the selection of splice site are responsible for numerous genetic diseases. In this case, the mutation of nucleotide T to C occurring at position 6 of intron 20 disrupts the interaction between IKBKAP pre-mRNA and U1 snRNA, and weakens the splicing from the intronic 5′-ss. Additionally, computational analyses identified that upstream 3′-ss also contains weak splicing signals, further causing the defining of exon 20 almost impossible. [33] (Figure 3).