Emerging roles of the spliceosomal machinery in myelodysplastic syndromes and other hematologic disorders

Abstract

In humans, the majority of all protein-coding transcripts contain introns that are removed by mRNA splicing carried out by spliceosomes. Mutations in the spliceosome machinery have recently been identified using whole exome/genome technologies in myelodysplastic syndromes (MDS) and in other hematologic disorders. Alterations in Splicing Factor 3 Subunit b1 (SF3b1) were the first spliceosomal mutations described, immediately followed by identification of other splicing factor mutations, including U2 Small Nuclear RNA Auxillary Factor 1 (U2AF1) and Serine Arginine Rich Splicing Factor 2 (SRSF2). SF3b1/U2AF1/SRSF2 mutations occur at varying frequencies in different disease subtypes, each contributing to differences in survival outcomes. However, the exact functional consequences of these spliceosomal mutations in the pathogenesis of MDS and other hematologic malignancies remain largely unknown and subject to intense investigation. For SF3b1, a gain of function mutation may offer the promise of new targeted therapies for diseases that carry this molecular abnormality that can potentially lead to cure. This review aims to provide a comprehensive overview of the emerging role of the spliceosome machinery in the biology of MDS/hematologic disorders with an emphasis on the functional consequences of mutations, their clinical significance, and perspectives on how they may influence our understanding and management of diseases affected by these mutations.

Introduction

Protein synthesis is a carefully regulated mechanism that begins with replication, followed by transcription, and culminating with translation of the protein. Post-transcriptional modifications, a myriad of intermediary steps that occur mainly between the transcription and translation process, ensure the integrity, diversity, and fidelity of the final protein product. One of the crucial post-transcriptional processes is RNA splicing, whereby non-coding sequences called introns are removed from the primary transcript (pre-mRNA). This process is carried out by a series of proteins consisting of small nuclear ribonucleoproteins (snRNPs) and small nuclear ribonucleic acids (snRNA) which form spliceosomes.¹ Alternative splicing is a process that increases genomic diversity through alterations in the composition of the exons by using alternative 5′ and 3′ splice sites, retained introns, and unconventional exons.² Since RNA splicing is a ubiquitous process in eukaryotic cells, it is not surprising that dysfunction of this pathway can lead to disease. Polymorphisms or mutations altering regulatory sequences or producing different splice variants ultimately lead to hereditary diseases and cancer. This phenomenon has been observed in neuromuscular, psychiatric, X-linked and neoplastic disorders. ³

Recently, whole genome sequencing as a genetic tool has been instrumental in the discovery of novel mutations in genes such as DNMT3A in hematologic malignancies, including acute myeloid leukemia (AML). ⁴ A similar approach was utilized to further elucidate the pathogenesis of other hematologic malignancies which led to the identification of mutations in spliceosomal genes, in particular SF3b1, first in myelodysplastic syndromes (MDS), myeloproliferative neoplasms (MPN), MDS/MPN overlap, and finally in chronic lymphocytic leukemia (CLL) and other hematologic disorders. Subsequently, mutations in other spliceosomal genes like U2AF1 and SRSF2 were identified (Table 1). Within individual disease subtypes, mutations of specific genes confer varying functional and clinical consequences. The discovery of this new class of molecular lesions represents a leap in our understanding of the biology of hematologic diseases and their pathogenesis.

Table 1.

Spliceosomal Genes

Gene Name	Reference
	Visconte et al.
	Yoshida et al.
	Papaemmanuil et al.
	Malcovati et al.
SF3b1	Lasho et al.
	Damm et al.
	Patnaik et al.
	Rossi et al.
	Wang et al.
	Quesada et al.
	Yoshida et al.
U2AF1	Graubert et al.
	Makishima et al.
U2AF2	Yoshida et al.
	Yoshida et al.
SRSF2	Makishima e al.
ZRSR2	Yoshida et al.
	Makishima et al.
SF3a1	Yoshida et al.
PRPF40B	Yoshida et al.
SF1	Yoshida et al.
PRPF8	Makishima et al.
LUC7L2	Makishima et al.
SAP130	TCGA#
HCFC1	TCGA
SFRS6	TCGA
SON	TCGA
U2AF26	TCGA

SF3b1, splicing factor 3b subunit 1; U2AF1/2, U2 Small Nuclear RNA Auxillary Factor 1/2; SRSF2, serine argine rich splicing factor 2; ZRSRS2, zinc finger (CCCH type), RNA-binding motif and serine/arginine rich 2; SF3a1, splicing factor 3a, subunit 1, PRPF40B, PRP40 pre-mRNA processing factor 40 homolog B, SF1, splicing factor 1; PRPF8, PRP8 pre-mRNA processing factor 8 homolog, LUC7L2, putative RNA-binding protein Luc7-like 2 SAP130, sin3A-associated protein, 130kDa; HCFC1, host cell factor C1; SFRS6, splicing factor, arginine/serine-rich 6; SON, DNA binding protein; U2AF26, U2 small nuclear RNA auxiliary factor 1-like 4;

^#TCGA, The Cancer Genome Atlas

Overview of the biology of spliceosomes

The dynamic process of intron splicing is made possible through the active participation of two types of spliceosomal complexes, major and minor spliceosomes. Major spliceosomes, which consist of U1, U2, U4/U6 and U5snRNPs, catalyze most of the splicing processes. Minor spliceosomes, consisting of U11 and U12snRNPs and U4atac and U6atac snRNAs, carry out the splicing of minor class introns.⁵ this process is responsible for the excision of ~1 in 300 introns from human pre-mRNA. Although generally perceived as a post-transcriptional process, splicing can also occur as a co-transcriptional process in the nucleus.⁶ Most of the individual components of the major spliceosomes are partly synthesized in the cytoplasm but mature forms are mainly localized in the nucleus.

Spliceosomes, which catalyze the essential process of RNA splicing and ligation of flanking exons, rely on specific recognition sites in the target pre-mRNA transcript for appropriate binding and assembly, namely the 5′ end and the 3′ end splice sites. The 5′ end splice site marks the exon-intron junction at the 5′ end of the intron. The 3′ end splice site marks the exon-intron junction site at the 3′ of the intron and consists of the branch point sequence (BPS), a polypyrimidine tract. Dinucleotides that are located at 5′ and 3′ sites are donor and acceptor splice sites for intron removal. Major spliceosomes remove introns with the canonical dinucleotide, GU-AG, and rarely AT-AC and GC-AG dinucleotides. Minor spliceosomes carry out the removal of AT-AC and GT-AC introns. ^7,8,5 The process of spliceosome assembly is a systematic process that begins with U1snRNP binding at the 5′ end splice site of the pre-mRNA to form an early or E complex. This E complex requires ATP and commits the pre-mRNA to splicing. U2snRNP is subsequently recruited to the branch region and becomes tightly associated with the BPS through an ATP-dependent process to form complex A. The duplex formed between U2snRNP and the pre-mRNA branch region forces out the branch adenosine and commits it for the first transesterification process. A pseudouridine residue in U2 snRNA changes the duplex conformation, making it more accessible for the first step of splicing. The tri-snRNP U4/U5/U6 are recruited to the spliceosome to form complex B, which ultimately becomes rearranged and forms complex C, characterized by the loss of the U1 and U4 snRNPs from the complex and replacement of U1snRNP at the 5-splice site with U6 snRNP (Figure 1).

Figure 1. Alternative splicing pattern and disease causation.

Left panel: Spliceosomes catalyze RNA splicing which leads to the ligation of two flanking exonic regions. The GU dinucleotide at the 5′ end, BPS and terminal AG dinucleotide at the 3′ end serve as specific recognition sites. Three different complexes (E, A, and B) are subsequently formed by the spliceosome assembly. In order, U1, U2, U4/U5/U6snRNP all cooperate in the formation of each complex. SF3b1 encodes a protein responsible for the binding of the U2snRNP to the branch point at the 3′ splicing site. Right panel: alternative splicing is the process which generates variability in transcripts, ultimately leading to biological differences in protein function and structure through different mechanisms (A–G). The first type is called A) exon skipping or sometimes called exon cassette which involves either retention or splicing out of the involved exon from the primary transcript. It is the most common mode of alternative splicing in mammalian pre-mRNA. In B) mutually exclusive exons, one of two exons are retained in the primary transcript after splicing. In C) competitive 5′ splice site, a new 5′ splice junction is used. In D) competive 3′ splice site, a new 3′ splice junction is used. In E) retained intron, a sequence is spliced out or kept. In F) multiple promoters, a transcriptional process where different starting points in the transcription can lead to transcripts with different 5′ sites. In G) multiple Poly-A sites, is the process that generates different 3′ ends. Several hypotheses are illustrated to explain the disease causation. (Illustrated by Ramon V. Tiu, MD)

U2snRNP is formed by a complex of proteins including SF3a, SF3b1, and the 12s RNA unit. SF3b is a 450 kDa multiprotein complex composed of SF3b1, SF3b2, SF3b3, SF3b4, SF3b5, SF3b14, and PHF5A.⁹ SF3b1 allows for the binding of the spliceosomal U2snRNP to the branch point near the 3′ splicing site. The main function of this complex is to prevent an inappropriate nucleophilic attack by additional components of the spliceosome prior to the initial transesterification reaction that must occur in order to achieve RNA splicing. Splicing can be affected by mutations in the components important for pre-mRNA processing (cis-acting elements) or in components implicated in the regulation (trans-acting elements) of splicing. Splicing requires accuracy in the recognition of the exon-intron boundaries and insertion or removal of nucleotides by mutations leading to a shift in the open reading frame of the future protein. Mutations affecting alternative splicing, intronic regions, recognition sites, exonic regions, and changes in the abundance and ratio of different splicing isoforms have been previously described and implicated in the biogenesis of several human diseases. ¹⁰ Recently, SF3b1 has been found to be highly mutated in MDS and related disorders. ^11–13 Yoshida et al. also described mutations in SF3a1, albeit at lower frequency. ¹²

Mechanisms of spliceosome regulation

The process of spliceosome assembly has been associated with kinases and phosphatases. Thus, mechanisms of sequential phosphorylation and dephoshorylation regulate components of the spliceosomes and dictate the steps in spliceosome assembly. SR proteins are factors essential in both the constitutive and alternative RNA splicing. The phosphorylation usually occurs at the serine resideues in the RS domain promoting protein interactions that lead to spliceosome assembly.¹⁴ Subsequently, SR proteins must be dephosphorylated to allow the step of transesterification to proceed. Several kinases have been described to phosphorylate SR proteins, such as SRPK, PRP4, Clk1-4, Cdc2 kinase and topoisomerase 1. All these kinases target serine/threonine residues and are localized in the cytosol and nuclear speckles. Several kinase inhibitors have been developed including TG003, a derivate of benzothiazol which is a selective pharmacologic inhibitor of Clk1 activity suppressing the serine/arginine rich domains and SR phosphorylation. Different phosphatases, such as PP2Cg, PP1, and PP2A have also been implicated in the early step and in the second catalytic step of spliceosome assembly. ¹⁵

Current evidence and perspectives on spliceosome component mutations: biological and clinical significance

Splicing Factor 3 subunit b1 (SF3b1)

Splicing mutations have been found to lead to hereditary diseases, such as X-linked disorders of copper metabolism and retinitis pigmentosa.¹⁶ Somatic mutations of vital components of the spliceosome machinery were recently discovered by several investigators using whole-exome and genome sequencing in hematologic malignancies, first in MDS and MDS/MPN overlap syndromes, particularly refractory anemia with ring sideroblasts (RARS) and refractory anemia with ring sideroblasts with thrombocytosis (RARS-T) and then in other hematological diseases. Mutations in SF3b1 were first reported by three groups. ^11–13 Beside SF3b1, mutations were found in other genes involved in RNA splicing machinery, such as U2AF1, SRSF2, ZRSR2, SF3a1, PRPF40B, U2AF65, and SF1. ^12,17

Mutational frequencies for SF3b1 ranging from 68–75% were reported for RARS and 81% in RARS-T. In our laboratory, SF3b1 mutations were always associated with acquired rather than with hereditary cases of refractory anemia, since no mutation was observed in a small group of patients with congenital sideroblastic anemia. Lower mutational frequencies of SF3b1 (0–7%) were noted in MDS and MDS/MPN with <15% RS. ^11,12 The notable difference in the prevalence of SF3b1 mutations in MDS and MDS/MPN with >15% RS led some investigators to believe that this mutation is important in the pathogenesis of RARS/ RARS-T.

The clinical entity RARS was first defined in 1982, a time when a set of new diagnostic criteria was created for the classification of MDS.¹⁸ The traditional distinction of ≥15% RS in erythroid precursors was an arbitrary distinction without a biologic basis; however, it now appears that this cut-off does have some biological significance, as evidenced by the higher frequency of SF3b1 mutations. RS are an abnormal localization of ferritin iron in the mitochondria of erythroid precursors and are usually an index of altered erythropoiesis.¹⁹ Although clearly important in the subsequent morphologic classification of RARS and RARS-T, the presence of RS is not in itself pathognomonic for MDS. As the World Health Organization (WHO) has clearly defined, definitive evidence of dysplasia, cytopenias, and certain types of cytogenetic abnormalities remain crucial in the diagnosis of MDS and related diseases like MDS/MPN overlap neoplasms.

As mentioned, these iron deposits are not unique to RARS and RARS-T but sometimes are observed in other disease entities, especially congenital sideroblastic anemias. The molecular pathogenesis of certain types of congenital sideroblastic anemias are more clearly defined. For example, germ-line mutations in aminolevulinate deltha-synthase (ALAS2) ultimately result in X-linked sideroblastic anemia.²⁰ It is imperative therefore to assume that certain types of mitochondrial genes may be at fault for RARS and RARS-T. Several genes, included ATP-binding cassette, subfamily B, member 7 (ABCB7), Pseudouridine synthase-1 (PUS-1), and Ferrochelatase (FECH) are potential candidates, but to date no mutations in these genes have been associated with acquired cases of RARS and are confined solely to hereditary cases of sideroblastic anemia.²¹. RS can appear in other clinical conditions such as copper deficiency, chronic neoplastic diseases, and exposures to ethanol, acetaldehyde and other toxins. ^22–24 Few reports have described the presence of rare RS in the BM of patients with copper deficiency and in patients with chronic alcohol abuse. Very little is known regarding the mechanism/s leading to the RS formation in this context. Although, in general, trancriptional and post-transcriptional regulation has been characterized in the iron utilization through iron-regulatory proteins.^22,25

The high frequency of SF3b1 mutations in RARS/ RARS-T makes this gene a very strong candidate for the pathogenesis of these diseases. Indeed, we are currently investigating this particular association using specific pharmacologic inhibitors of the spliceosome machinery, such as meayamycin, which specifically targets the SF3b complex. We also hypothesized that structural differences in iron distribution may be observed between SF3b1 mutant and wild type cases. This is being investigated using transmission electron microscopy. We are also performing experiments in Sf3b1 heterozygous mice ²⁶ to further clarify the association between SF3b1 mutations and ring sideroblasts formation. Part of the data were presented in the recent 53^rd ASH meeting. ²⁷ Furthermore, we found two cases of rare diseases with SF3b1 mutations. The two informative cases were: a patient post-polycythemia vera myelofibrosis in a cohort of 30 patients with MPN¹¹ and a patient with paroxysmal nocturnal hemoglobinuira (PNH) in a group of aplastic anemia (AA, N=22), T cell large granular lymphocyte leukemia (T-LGL, N=17), and paroxysmal nocturnal hemoglobinuria (PNH, N=24).²⁷ Interestingly, both cases presented RS with no other morphologic or cytogenetic features suggesting concomitant MDS, further supporting the role of SF3b1 in RS biogenesis.

In terms of clinical significance, the two studies that initially explored the effects of SF3b1 mutation on overall survival had slightly different results. The study by Visconte et al. did not initially find any difference in overall survival. In contrast, a larger study by Papaemmanuil and colleagues showed better overall survival (p=.01), leukemia-free survival (p=.05), and event free survival (p=.008) in patients with SF3b1 mutations compared to wild type patients. Papaemmanuil and Malcovati also noted that SF3b1 mutations are associated with lower risk of evolution to AML.^13,28 Several other studies subsequently followed exploring the clinical correlations of SF3b1. Damm et al. studied 317 patients with MDS, although clinical data was only analyzed in 253 patients, and found no difference in overall survival and rate of AML transformation.²⁹ The multivariate analysis performed by Patnaik et al. determined that the prognostic value of SF3b1 mutations was completely accounted for the WHO morphologic grouping. ³⁰ In opposition to these negative findings and our own original negative findings in a small cohort of patients, we recently presented data on 511 cases of MDS, MDS/MPN, and other hematological disorders (PNH, AA, T-LGL, Mast cell diseases, and MPN), finding better overall survival in MDS and MDS/MPN cases, ¹⁴ further supporting the results of Papaemmanuil et al. We also observed that the rate of overall survival is better in SF3b1 mutants versus wild type patients within the subset of RARS/RARS-T.

It is possible that the good prognosis seen in SF3b1 mutants is related to its close association with a clinic-pathologic entity that is known to have good outcomes, but recent data from our group also demonstrates that it may be independent of disease classification. Biologically, it is possible that SF3b1 mutations result in aberrations in alternative splicing, leading to a loss of activity of protein transcripts from pathway signals that typically portend poor outcomes in AML. It is also possible that alternative splicing results in activation of tumor suppressor gene pathways, making these cells less likely to acquire poor prognostic chromosomal and genetic mutations, further supported by the absence of SF3b1 mutations at the time of AML transformation in serial studies. In-vivo studies by Isono et al. have demonstrated that Sf3b1 is important in Polycomb (PcG)-mediated repression of Hox genes in mice. Sf3b1 heterozygous mice develop abnormal skeletal phenotypes similar to what is observed in PcG complex mutant cases. Several inactivating mutations in genes of the PcG complex have already been associated with myeloid malignancies, such as ASXL1 in chronic myelomonocytic leukemia (CMML) and EZH2 in a variety of diseases mostly associated with increased platelets.^31,32 Other components of the polycomb-repressor complex (PRC2) have also been found to be infrequently mutated in myeloid malignancies. ³³ Since the function of the PcG complex seems to be dependent on SF3B as demonstrated in the mouse studies, it is possible that this is what leads to the subsequent development of MDS in SF3b1 mutated cases. However, ASXL1 and EZH2 mutations are almost always associated with more aggressive MDS phenotypes which tend to progress to AML, which is contrary to the phenotype where SF3b1 mutations is frequently found, such as low risk MDS (RARS/ RARS-T). This is further supported but the very low frequency of SF3b1 mutations in AML (de-novo=2.6%; secondary=4.8%).

Initial studies from our group also suggest its potential link to increased risk of thrombosis within these patients. Further analysis has clarified that approximately 40% of mutant patients developed thrombosis, mainly arterial. ³¹ How thrombosis occurs in the setting of SF3b1 mutations is unclear and will be the subject of further investigation.

Another study screened for the presence of SF3b1 mutations in 155 patients with primary myelofibrosis (PMF) and also found a low frequency of mutations in this group of patients (10/155; 6.5%), a majority of which (60%) carried a concomitant JAK2V617F mutation. In 6 patients with bone marrow available for Prussian blue staining, they also found that all patients with SF3b1 mutations had RS. Neither clinicopathologic nor overall survival differences were observed between patients with SF3b1 mutations and their wild type counterparts.³⁴ Furthermore, no mutations have been found so far in our hands in patients with essential thrombocytopenia (ET, N=15), polycythemia vera (PV, N=24), and in very few cases of chronic myelogenous leukemia (CML, N=7). Mutations in SF3b1 were also found to be rare in much larger cohort of MPN as reported by Yoshida et al. (0%), and Papaemmanuil et al. (PMF, 4.4%, PV, 0%, and ET, 3.1%, CML, 4.7%). Interestingly, we found 2/32 (6.2%) patients with mastocytosis with SF3b1 mutations (Visconte et al., unpublished data). The low frequency of SF3b1 mutations in MPN further supports the selective specificity of the presence of SF3b1 mutations for RARS and RARS-T. The majority of the studies utilized whole exome sequencing as their initial mutation screening tool. Few studies performed whole genome amplification. ^{13, 17,35} In the vast majority of the studies, SF3B1 was sequenced for all the exons in a small target cohort preliminarily then a much more focusing sequencing of the most frequently mutated exons were performed. More recently, it has been reported that aside from RARS and RARS-T, the disease with the most frequent mutations in SF3b1 is CLL. ^35–37 Papaemmanuil et al. reported SF3b1 mutations in 2/40 CLL patients, although only in one patient was the mutation somatically confirmed. ¹³ A second independent study initially looked at a limited series of 11 fludarabine –refractory CLL by whole-exome sequencing and found mutations in 3/11 cases. ³⁶ This lead to the screening of a larger cohort of 393 CLL patients (fludarabine refractory [N=59]; newly diagnosed and previously untreated [N=301]; and clonally related Richter’s syndrome [N=33]), which found a statistically significant higher frequency of SF3b1 mutations in patients with fludarabine refractory CLL (10/59; 17%) compared to just 5% (17/301) in newly diagnosed, previously untreated CLL (p=.002). A majority of the mutations were missense (9/10; 90%). Patients with clonally related Richter’s syndrome have a mutational frequency of 6% (2/33). In this same study, multivariate analysis revealed that the presence of SF3b1 mutations independently predicted for increased risk of death (HR: 3.02 [CI: 1.24–7.35], p=.015). The group also studied 163 cases of mature B cell neoplasms, including mantle cell lymphoma, follicular cell lymphoma, diffuse large B cell lymphoma, splenic marginal zone lymphoma, extranodal marginal zone lymphoma, hairy cell lymphoma and multiple myeloma and found no SF3b1 mutations in these cases, supporting its specificity in CLL within lymphoid neoplasms. ³⁶ Another group, using whole exome and targeted sequencing of a smaller cohort of CLL patients (N=91) found SF3b1 mutations in 15% (14/91) of cases. ³⁵ Several clinical associations were observed in this study, including a higher frequency of mutated patients having a concomitant deletion in chromosome 11q, and K700E mutants were associated with unmutated immunoglobulin heavy chain gene (IGHV) status (p=.048). Moreover, Cox multivariate regression model found that the presence of SF3b1 mutations in these patients was predictive of an earlier need for treatment (HR: 2.2; p=.03) independent of other factors including IGHV mutation status, del17p or ataxia telangiectasia mutated (ATM) gene mutation. Using intron retention of two endogenous genes (BRD2 and RIOK3), they demonstrated that the spliceosome inhibitor E7107 leads to disruption of splicing of BRD2 and RIOK3 in both normal and CLL cells. SF3b1 mutated patients have aberrant endogenous splicing activity in tumor samples compared to the wild type cases and the ratio of unspliced to spliced mRNA forms of BRD2 and RIOK3 was significantly higher in SF3b1 mutation carriers versus wild type cases (median ratio: 2:1 vs. 0.5:1, p=<.001 and 4.5:1 vs. 2.1:1, p=.006).³⁵ Finally, a recent paper was published examining 105 CLL patients in whom whole exome sequencing was performed, and found SF3b1 to be mutated in 9.7%. Similar to earlier findings, the presence of SF3b1 mutations were associated with worse 5 year time to progression in Binet stage A patients (34% vs 73%, p=.002) and 10 year overall survival (30% vs 77%, p=.0002). ³⁷ Overall, all three studies reported SF3b1 mutations in codons which have been described to be mutated also in MDS and other disorders.

U2 Small Nuclear RNA Auxillary Factor 1 (U2AF1)

Mammalian U2 small nuclear ribonucleoprotein auxiliary factor (U2AF) is a heterodimer composed of a 65-kDa subunit (U2AF⁶⁵; U2AF2) and a 35-kDa subunit (U2AF³⁵; U2AF1). U2AF⁶⁵ contacts the pyrimidine tract while U2AF³⁵ interacts with the AG splice acceptor dinucleotide of the target intron at the 3′ splice site. Recently, whole exome/genome sequencing has helped in the discovery of somatic mutations of U2AF in MDS. Mutations were found mostly in U2AF1 with frequencies of 8.7–11.6-% in de-novo MDS. Yoshida et al. also reported 4 patients with mutations in U2AF2. ^{12, 17} Mutations of this gene are distributed among a myriad of myeloid malignancies and mainly involve two conserved amino acid positions (S34 and Q157) located within zinc finger domains.^12,17 Mutational frequencies were found to be 12% in MDS without RS, 8% in CMML and 10% in cases of MDS-derived secondary AML (10%). A lower frequency was observed in primary AML and MPNs. Recent evidence shows that a p.Ser34Phe mutation occurs in 8.7% of de novo MDS which progress to sAML. ¹⁷ Subsequently, Makishima et al. extended the study to juvenile myelomonocytic leukemia (JMML) and pediatric AML, finding no mutations in these disease entities. ³⁸ Yoshida et al. first reported that U2AF1 mutant transduced TF-1 and HeLa cells present with a decrease in cell proliferation rather than a growth advantage, suggesting a loss of function mutation. U2AF1 mutants also displayed lower reconstitution capacity by competitive reconstitution assay in mice. ¹² Graubert et al. described a significant increase in exon skipping when the mutant p.Ser34Phe cDNA was transiently expressed in vitro, suggesting a gain of function.¹⁷ Clinically, no correlation with response to hypomethylating agents was found in a group of 92 patients.³⁹ Makishima et al. reported a significant association with worse overall survival in patients with MDS/MPN.⁴⁰ In the context of spliceosome regulation and interaction, U2AF⁶⁵ usually binds to SF1 factor, increasing the affinity of SF1 factor for the pre-mRNA branch point sequence. Later during splicing, SF1 is usually displaced from U2AF⁶⁵ and replaced by the U2snRNP protein, SF3b1. It is of great interest to understand how genes that share almost the same mechanism are pathognomonic of different diseases.

Serine Argine Rich Splicing Factor 2 (SRSF2.)

SRSF2 encodes a member of the serine/arginine (SR)-rich family of pre-mRNA splicing factors. A majority of these factors contains an RNA recognition motif (RRM) and a RS domain. The presence of the RS domain helps the interaction between different SR splicing factors. SRSF2 has been reported to be critical for constitutive and alternative mRNA splicing. Interestingly, SRSF2 is believed to be a key element in the acetylation/ phosphorylation network and an important regulator of the DNA stability.⁴¹ Yoshida was first to report a mutation involving this gene in hematologic malignancies, particularly in MDS and related diseases. It is most frequent in CMML, with frequencies ranging between 28.4–30%.¹² Nagata et al. further confirmed these results (Table 2). ⁴² The most common mutation occurs at amino acid position P95 between the RRM motif and the RS domain. Lower mutations frequencies were reported in other malignancies (Table 2). A presentation at the 2012 annual ASH meeting reported a frequency of 47.2% in their cohort of patients with CMML.⁴³ More recently it has been emphasized that DNA hypermutability might explain the co-existence of SRSF2 with other mutations. ⁴² Depletion of SRSF2 leads to genomic instability and might possibly explain the worse outcome in patients with SRSF2 mutations. ^44,45 There are some clinical associations linking this mutation to a clinical phenotype associated with increased age, higher hemoglobin levels, and coincidental TET2 and RUNX1 mutations. It was also noted that it seems to be mutually exclusive from mutations in EZH2, a poor prognostic marker. SRSF2 mutations showed better outcomes in CMML patients with concomitant RUNX1 mutations. ⁴³

Table 2.

Mutations of splicing factor genes in myelodysplastic syndromes and other hematologic disorders

Spliceosome Genes	Studies	Discovery Tool	% Mutations within disease groups	Effects on Overall survival	Effects on AML evolution/ any progression	Other clinical Associations
SF3b1	Visconte et al. (N=56)	Whole Exome	MDS: RARS (68%), RAEB1/2: (2.9%) RCMD/RCUD (4.3%) MDS/MPN: RARS-T (81%), CMML (6.8%), MDS/MPN-U (0) AML: 1o AML (4.7%), 2o AML (5.9%) BMF (1.6%) MPN (1.4%)	No difference	No difference	Higher risk of thrombosis Very uncommon in <15% RS
	Yoshida et al. (N=582)	Whole Exome	MDS: RARS (82.6%), RCMD-RS (76%), MDS without RS (6.5%) MDS/MPN: CMML (4.5%) AML: 1o AML (2.6%) 2o AML/MDS (4.8%) MPN (0%)	NA	NA	NA
	Papaemmanuil et al. (N=354)	Whole Genome		Better	Lower risk of progression	NA
	Malcovati et al. (N=564)	Whole Genome		Better	Lower risk of progression	NA
				Better	Lower risk of progression
	Damm et al. (N=317)	Direct Sequencing		No difference	No difference	NA
	Lasho et al. (N=155)	Direct Sequencing		No difference	NA	NA
	Rossi et al. (N=59/301/33)*	Whole Exome	Fludarabine-refractoryCLL (15%) CLL at diagnosis (5%)	Poor	NA	NA
	Wang et al. (N=91)	Whole Exome	CLL (15%)	NA	NA	Association with del(11q) and unmutated IGHV status. Predictor factor for an earlier treatment
	Quesada et al. (N=279)	Whole Exome	CLL (9.7%): Unmutated IGHV (7.9%) Mutated (20.5%)	Poor	Short time to progression	High β2-microglobuli n, unmutated IGHV
U2AF1	Yoshida et al. (N=582)	Whole Exome	MDS no RS (11.6%) MDS/MPN: CMML (8%) AML: 1o AML (1.3%), 2o AML (9.7%) MPN: (1.9%)	NA	NA	NA
	Makishima et al. (N=310)	Whole Exome	Low Risk MDS (6%) High Risk MDS (11%) MDS/MPN: CMML 17% AML: 9% JMML/pediatric AML: (0%)	Worse	NA	NA
	Graubert et al. (N=150)	Whole Genome	MDS de novo (8.7%) MDS/sAML (15.2%)	No difference	Increased probability to progress to sAML	More frequent in del 20q and Monosomy 20 patients
SRSF2	Yoshida et al. (N=582)	Whole Exome	MDS no RS (11.6%) RCMD/RCMD-RS (5.5%) MDS/MPN: CMML (28.4%) AML: 1o AML (0.7%), 2o AML (6.5%) MPN (1.9%)	NA	NA	NA

NA, not available; RS, ring sideroblasts

^*The numbers represent three different cohorts utilized by Rossi et, Blood 2011

Other spliceosome genes

Alterations of components of the spliceosome machinery might be a key biological process in the pathogenesis of many hematological diseases, where each gene might be determinant for different clinical phenotypes. Yoshida et al. published the largest comprehensive map of splicing factors genes found to be mutated in 29 patients with myelodysplasia by whole-exome sequencing. Other spliceosomal-related genes were found mutated in cases of MDS in addition to the well known SF3b1, U2AF1, and SRSF2 and include U2AF65, ZRSR2, SF1, SRSF1, SF3a1, and PRF40B. More recently, Makishima et al. added PRPF8 and LUC7L2 to the list of somatic mutations in spliceosomal genes (Table 1). ⁴⁰

Spliceosomal mutations and potential mechanisms of disease pathogenesis

Besides the strong correlation between spliceosomal mutations and certain hematological diseases, the mechanism through which these mutations perturb the process of RNA splicing and subsequently lead to disease is still unknown. Physiologically, multiple protein isoforms are generated by the process of differential splicing of RNA transcripts, also known as alternative splicing. Alternative splicing determines the final set of coding sequences that will ultimately be translated and contributes to the differential biological and chemical function of the final protein product. Different patterns of alternative splicing can occur and include: 1) exons that are not always consistently included in the final mRNA transcript, called cassette exons, may be involved, 2) certain cassette exons may be mutually exclusive, thus if two or more cassette exons are present, only one will be incorporated in the mRNA, 3) alternative 5′ splice sites, 4) alternative 3′ splice sites, 5) intron retention, 6) presence of multiple promoters and 7) multiple polyadenylation sites. ⁴⁶ It has been reported that almost 50% of mutations in spliceosomal genes affect splicing and regulation of alternative splicing leading to causative defects and contributing to increased susceptibility to diseases (Figure 1). ^3,47 Diseases that result from aberrant alternative splicing patterns include neuropsychiatric diseases like frontotemporal dementia with Parkinsonism-17 (FTDP-17) characterized by misregulation of tau exon 10 alternative splicing, ⁴⁸ amyotrophic lateral sclerosis where abnormal patterns of intron retention and exon skipping in astrocytes were observed, ⁴⁹ and schizophrenia which has an abnormally higher frequency of exon selection and skipping.^50,51 This mechanism has also been observed in solid cancer like Wilm’s tumor, where misregulation of alternative splicing of WT1 occurs and glioblastoma where reduced splicing of FGF-R1 alpha exon is seen. ^52,53 It is therefore conceivable that dysfunction or changes in alternative splicing patterns can be caused by mutations in the spliceosome machinery (SF3b1, U2AF1, SRSF2, and others) and can lead to disease manifestations observed in hematologic neoplasms like MDS, CLL, and others.

In support of this hypothesis, experimental studies in the nervous system demonstrated that alternative splicing can lead to exacerbation or inhibition of certain apoptotic pathways. ⁵⁴ Increased apoptosis has been shown to be important in the pathogenesis of MDS, supported by data showing overexpression of proapoptotic proteins like Minichromosome Maintenance Protein 2 (MCM2). ⁵⁵ It is possible that spliceosomal mutations leading to aberrant alternative splicing may be responsible for the dysregulation in apoptotic pathways seen in MDS. Similarly, it is possible that mutations of the spliceosomal pathway can result in variable effects on splicing enhancers and splicing repressors, leading to differential expression of protein isoforms that may have oncogenic potential and have the capacity of evading the normal process of nonsense-mediated decay. It is also plausible that these abnormal protein isoforms can lead to upregulation of certain pathways already known to be abnormal in MDS like the high aberrant pSTAT5 activation seen in some JAK2/MPL negative RARS-T. Conversely, these mutations can possibly lead to repression of certain metabolic pathways (decreased conversion of 5-MC to 5-HMC in some cases of TET2 wild type MDS) relevant in MDS as was recently reported by Jankowska et al. ⁵⁶ A similar mechanism may be responsible for the altered iron deposition seen in patients with RS. Mitochondria with their supporting proteins play a crucial role in iron passage, leading to the final incorporation of iron in hemoglobin during normal heme biosynthesis. However, mutations of mitochondrial genes are not frequent in RARS and RARS-T and therefore it is intuitive to assume that an alternative mechanism to the altered iron deposition may be at play. ¹³ Furthermore, Yoshida et al. studied the impact of the spliceosomal mutations expressing mutant and wild-type U2AF1 in Hela cells and found that mutants cells display an over-representation of non-sense mediated mRNA decay genes by gene expression analysis. ¹² Graubert et al. proposed a mechanism of exon-skipping in mutant U2AF1 cell lines.¹⁷ Exon skipping is a process which results in the loss of an exon or in a lengthened/ shortened (alternative 5′ or 3′ splicing) alternatively spliced mRNA. Mutant U2AF1 cell lines exhibited a high proportion of transcripts with exon skipping compared to wild-types. More comprehensive analyses will help clarify these possibilities. In addition, introns might not be spliced out from the RNA transcript and retained in the mRNA as part of an exon. Up to 15% of human diseases are due to mis-splicing. Mutations in certain genes can change the DNA sequences and results in inefficient recognition of canonical sites by the spliceosome generating a mutated pre-mRNA. Genetic therapies have actually been used to targeted mis-spliced transcripts in certain diseases. ⁵⁷

Drugs that target the spliceosome machinery

Identification of a novel gene typically altered in a specific disease group offers the promise of potential targeted therapy. We now know that SF3b1 mutations are mainly heterozygous in MDS, CLL, and other hematologic neoplasms. Disruption of the remaining allele using a pharmacologic inhibitor may induce cell killing specific for carriers of the mutations and can hypothetically spare normal cells because of the presence of two normal SF3b1 alleles. Alternatively, if the disease seen in SF3b1 mutants is related to downstream abnormalities in splicing and restoration of normal splicing is necessary to improve disease, then the fundamental knowledge of whether this mutation is a gain of function or a loss of function mutation is paramount. If SF3b1 mutations are gain of function mutations and result in the aberrant alternative splicing seen, one can hypothesize that pharmacologic therapy can restore the normal splicing patterns of these individuals by inhibiting the gain of function effects of the mutation. Conversely, if this is a loss of function mutation, then the addition of a pharmacologic inhibitor that disrupts the function of the remaining normal allele can worsen disease manifestations. However, the process and degree of alternative splicing as a measure of function of SF3b1 is complex. There are multiple regulators of alternative splicing aside from SF3b1, ranging from other snRNPs but more importantly the pre-mRNP itself, and the type of alternative splicing observed are tissue specific as can be seen in neural and muscle tissues. Knowing the crucial role played by the spliceosomes in the essential cellular process of protein production, several groups have developed therapeutic agents that induce cell killing by inhibition of the spliceosomal pathway. Spliceosome-targeted pharmacologic compounds, including Meayamycin, Spliceostatin, FR901464, E7107, pladienolide and Sudemycins, are currently being investigated.⁵⁸ Two of these, FR901464 and a pladienolide derivative called E7107, target the SF3b component of the spliceosome. FR901464 is a natural product derived from the bacterium Pseudomonas and it has recently found to have antiproliferative activity against human cancer cell lines. Three chemical syntheses of FR901464 have been reported including a compound called Meayamycin.⁵⁹ Meayamycin seems to be more potent and has enhanced stability in cell culture media as compared to the analog FR901464. Meayamycin specifically targets the SF3b complex. Various tumor cell lines (MCF-7, HCT-116, MDA-MB231, A549, and DU-145) were found to be differentially sensitive to picomolar concentrations of Meayamycin. Elegant studies by Dr. Koide’s group illustrated the role of Meayamycin in pre-mRNA splicing as well as alternative splicing.⁵⁹ The low toxic effect of Meayamycin against IMR-90 human lung fibroblasts and the selectivity against transformed cells might be due to the splicing inhibition. Pharmacologic analogues called Sudemycins were also developed that target SF3b and are capable of modulating alternative splicing in human tumor xenografts. ⁶⁰ Drugs such as Pladienolide and FR901464 represent novel ways of targeting the spliceosome with potent cytotoxic effects at lower IC50. FR901464 is also capable of arresting the cell cycle at G1/ G2/M phases.

Future directions/ summary

Aberrant splicing has been described in many cancers. The discovery of mutations in the spliceosome pathway and their specificity in RARS, RARS-T, and CLL offers a new area of investigation for explaining disease biology, diagnostics, and possibly therapeutic interventions for these conditions. New information regarding the crucial targets may be helpful to clarify the exact mechanism that leads to the pathophysiology of many disease subtypes.

Our group and others are actively searching for viable targets of the SF3b1 gene. In vitro SF3b1 knock-down in cell lines have addressed some important issues related to U2-dependent introns. In fact, downregulation of SF3b1 mRNA leads to a lower splicing activity of U2 introns as compared to U12-introns. ³¹ Global analysis of the transcriptome by RNA sequencing is being performed in our laboratory to map the targets of SF3b1 mutants. Functionally, spliceosome assembly and splicing catalysis mechanisms may be key biological processes in SF3b1 regulation. SF3b1 contains threonine-proline dipeptides at the N-terminal chain which are “hotspots” of phosphorylation. In splicing catalysis, phosphorylation of SF3b1 is functionally important. An orchestra of substrates, proline-kinases, serine-threonine-kinases, cyclin E, and phosphatase may work in the regulation of SF3b1. In fact, we also sequenced DYRK1A, a protein discovered to phosphorylate SF3b1 in vitro without finding any mutations, leading us to theorize that regulation of SF3b1 occurs at the protein rather than at the genetic level. Therefore, our laboratory is currently investigating whether specific kinases inhibitors, including TG003, and phosphatase inhibitors, such as calyculin, might reproduce/restore the RARS phenotype.