Dysregulation of splicing machinery components
Alterations in the expression of components of the molecular machinery that operate and control the splicing process have already been described in an extensive list of diseases, including many tumor pathologies [
35,
38‐
40]. In PDAC, pioneering studies by Carrigan et al. evaluated expression levels of selected genes in human pancreatic cancer cell lines, discovering a downregulation of 30% of spliceosomal genes, revealing a clear repression of splicing machinery components [
41]. More recently, Wang et al. validated some of these changes in PDAC human samples, establishing an expression signature of the spliceosome and splicing regulatory genes that discriminated with high accuracy between tumor and healthy samples [
42]. Interactions and implications of alternative splicing in PDAC pathogenesis have been subsequently reviewed [
43], emphasizing the promising prospect of dissecting its role in this pathology.
Further work has provided a deeper understanding of the dysregulations in the expression of the splicing machinery in PDAC, revealing that they frequently consist in an altered expression of spliceosome components and/or splicing factors (Table
1), which usually leads to an imbalanced profile of splice variants and/or the appearance of aberrant variants. Considering that the correct functioning of the splicing process regulates the overall balance of RNA variants in the cell, it is not to be unexpected that changes in the expression of splicing-related proteins could dramatically modify cell homeostasis, including key processes in PDAC evolution. This is the case of the overexpression of splicing machinery components that are associated with proliferation and apoptosis, such as
SRPK1 [
44,
45],
CLK1 [
46],
HNRNPK [
47],
PTBP3 [
48],
HNRNPL [
49],
HNRNPA2B1 [
50] and
ESRP1 [
51]; with metastasis and invasion, such as
SF3B1 [
52],
CLK1 [
46],
PRPF40A [
53],
ESRP1 [
51],
SRSF6 [
54] and
RBFOX2 [
55]; with the acquisition of chemotherapy resistance, such as
SRPK1 [
44,
45],
SRSF1 [
56],
PTBP1 [
57] and
SRSF3 [
58]; with autophagy, such as
SFPQ [
59] and
PTBP3 [
48]; with ubiquitination and degradation of some proteins, such as
HRNPU [
36]; or with the activation of relevant pathways in PDAC, such as the signaling mediated by
KRAS, driven by
HNRNPK [
47], or the MAP kinases pathway by
METTL3 [
60], or the pancreatitis and KRASG12D-mediated cancer promoted by
SRSF1 [
61].
Table 1
Dysregulations in the expression of the splicing machinery in PDAC
CLK1 | Upregulated | METTL14exon10 Cyclin L2exon6.3 | Growth and metastasis, and regulation of m6A methylation | |
(CDC Like Kinase 1) |
ESRP1 | Upregulated | FGFR-2 IIIb FGFR-2 IIIc | Cell growth, migration, invasion, and metastasis | |
(Epithelial Splicing Regulatory Protein 1) |
HNRNPA2B1 | Upregulated | Bcl-x(s) Bcl-x(L) | Apoptosis, proliferation, and metastasis | |
(Heterogeneous Nuclear Ribonucleoprotein A2/B1) |
HNRNPK | Upregulated | GTPase Activating Proteins | Tumor growth and sensibility to spliceosome inhibitors | |
(Heterogeneous Nuclear Ribonucleoprotein K) |
HNRNPL | Upregulated | | Migration and epithelial mesenchymal transition | |
(Heterogeneous Nuclear Ribonucleoprotein L) |
HNRNPM | Downregulated | | Adaptation to a hypo-vascular environment | |
(Heterogeneous Nuclear Ribonucleoprotein M) |
HNRNPU | Upregulated | | Putative biomarker | |
(Heterogeneous Nuclear Ribonucleoprotein U) |
METTL3 | Upregulated | | Mitogen-activated protein kinase cascades, ubiquitin-dependent process, and RNA splicing | |
(Methyltransferase Like 3) |
PTBP1 | Upregulated | PKM | Drug resistance | |
(Polypyrimidine Tract Binding Protein 1) |
PTBP3 | Upregulated | | Drug resistance | |
(Polypyrimidine Tract Binding Protein 3) |
PRPF40A | Upregulated | | Putative biomarker | |
(Pre-MRNA Processing Factor 40 Homolog A) |
RBM5 | Downregulated | | KRAS expression, lymph node and distant metastases, stage, and nerve and venous invasion | |
(RNA Binding Motif Protein 5) |
RBM10 | Upregulated | TERT | Telomere shortening | |
(RNA Binding Motif Protein 10) |
SF3B1 | Upregulated | BCL-XS/BCL-XL KRASa/KRAS Δ133TP53/TP53 | Tumor grade, lymph node involvement | |
(Splicing Factor 3b Subunit 1) |
SF3B4 | Downregulated | | Cell growth, proliferation, and migration | |
(Splicing Factor 3b Subunit 4) |
SFPQ | Upregulated | | Autophagy and cell growth | |
(Splicing Factor Proline And Glutamine Rich) |
SRPK1 | Upregulated | | MAPK and AKT signaling modulation | |
(SRSF Protein Kinase 1) |
SRSF1 | Upregulated | MNK2b | Drug resistance | |
(Serine And Arginine Rich Splicing Factor 1) |
SRSF3 | Upregulated | CDKN2B-AS1 | Drug resistance | |
(Serine And Arginine Rich Splicing Factor 6) |
SRSF6 | Upregulated | | Epithelial-to-mesenchymal transition, metastasis | |
(Serine And Arginine Rich Splicing Factor 6) |
RBFOX2 | Downregulated | RHO GTPase pathways | Metastasis, cytoskeletal organization and focal adhesion formation | |
(RNA Binding Fox-1 Homolog 2) |
Although current evidence supports that overexpression of splicing machinery components is more frequent in PDAC than initially observed, there are numerous examples where such components are repressed, indicating a putative tumor suppressor role. For example,
ESRP1 regulates the expression pattern of
FGFR-2 isoforms, attenuating cell growth, migration, invasion, and metastasis in PDAC cell lines [
51]. Likewise,
SRSF6 hinders pancreatic cancer cells migration and invasion by regulating
ECM1 alternative splicing isoforms [
54]. Further,
RBM10 promotes the appearance of the
TERT splicing isoform
TERT-S, which in contrast with
TERT-FL, is not able to maintain telomeres [
62].
SF3B4 inhibits the growth and migration of cancer cells preventing STAT3 phosphorylation [
63]. In the same line, downregulation of
HNRNPM and
RBM5 is associated with an increase in tumor aggressiveness. Specifically,
HNRNPM is implicated in the adaptation to a hypovascular environment [
64]; while
RBM5 expression inversely correlates with
KRAS levels and is associated with clinicopathological features and appears to promote tumor progression [
65]. Splicing factor
CELF2 is downregulated in PDAC and associated to PDAC progression, where its downregulation affects the splicing pattern of
CD44, thereby regulating endoplasmic reticulum-associated degradation [
66].
Mutations in splicing factors in PDAC
Mutations in genes specifically involved in the splicing process are increasingly recognized as a source of pathological effects in a range of diseases, including cancer [
17,
35]. These types of mutations are particularly frequent in pathologies such as acute myeloid leukemia or myelodysplastic syndromes, where they are tightly linked to etiology and offer therapeutic opportunities [
67‐
69]. Whole-exome sequencing of PDAC has revealed a number of mutations in key oncogenes and tumor suppressor genes, as
KRAS or
TP53 [
70], which are known to be among the most commonly mutated genes in PDAC [
4]. Nonetheless, those studies also identified mutations, although with lower frequency, in genes involved in other essential processes including splicing (Table
2), which conferred a higher tumor heterogeneity in PDAC. Genomic studies have also identified recurrent mutations affecting the early components of the RNA splicing machinery, such as
SF3B1 [
6]. Mutations in
SF3B1 are the most common across multiple tumor types, mainly found in myelodysplastic syndrome [
71,
72] and other hematologic malignancies, uveal melanoma and breast cancer [
73,
74], where its high mutational frequency altered the capacity of the spliceosome to recognize the pre-RNA pattern [
25,
75].
SF3B1 mutations are generally heterozygous, major hotspots identified matched with codon positions K700 and R625, which correspond with deleterious mutations [
76‐
78]. However,
SF3B1 mutations differ between each pathology, suggesting context-depending functional differences. For example, it has recently been described that
SF3B1 K700E mutation increased glycolysis and the Warburg effect in PDAC, by promoting aberrant splicing of
PPP2R5A and therefore activating c-Myc signaling, a positive regulator of glycolysis [
79]. Thus, while the knowledge on the effects of
SF3B1 mutations in PDAC is still limited, it is worth pointing out that several such mutations have been identified in a small but appreciable percentage of patients, such as G740E, N763S, K843R [
80], P342T [
7], K700E, L773R [
6], K700E, Q699_K700delinsHE, N763S, K741K [
81].
Table 2
Splicing machinery mutations found in PDAC
SF3B1 (Splicing Factor 3b Subunit 1) | P342T L415P R625C H662Q K666R Q699_K700delinsHE K700E N763S K843R L773R K741K K946T | |
SF3A1 (Splicing Factor 3a Subunit 1) | S58I | |
U2AF1 (U2 Small Nuclear RNA Auxiliary Factor 1) | A47V S34F | |
U2AF2 (U2 Small Nuclear RNA Auxiliary Factor 2) | | |
SRSF1 (Serine And Arginine Rich Splicing Factor 1) | | |
PRPF40B (Pre-MRNA Processing Factor 40 Homolog B) | L265M | |
SF1 (Splicing Factor 1) | R380Q R662* Q269_P273del | |
RBM6 (RNA Binding Motif Protein 6) | | |
PABPC1 (PolyA Binding Protein Cytoplasmic 1) | M158Nfs*8 D165G E345* E345K L562S T319I P402L F335Lfs*19I R475Q R481C | |
RBMX (RNA Binding Motif Protein X-Linked) | D312N P106Ffs*32 A78T R341W | |
In line with the findings on
SF3B1, mutations have also been described in other splicing factors like
U2AF2, which is involved in pre-RNA
branch site (BS) binding;
SRSF1, a SR protein that generally promotes exon inclusion [
42,
82];
PABPC1, required for poly(A) shortening, the first step in RNA decay [
4,
83]; or
RBMX, a RNA-binding protein that plays a crucial role in alternative splicing of several pre-RNAs [
84].
Collectively, these studies provide compelling evidence that mutations and particularly altered expression in specific components of the splicing machinery are a common feature in PDAC, and that such dysregulations often result in pathological consequences. These observations also point to the aforementioned alterations as potential targets for therapeutic intervention, which is already being exploited via diverse strategies, as discussed below. Nevertheless, the splicing machinery, which integrates all spliceosome components and their associated splicing factors, is extraordinarily complex and there is still much to be learned on its precise expression and regulation in PDAC. A comprehensive understanding of its abnormal functioning will likely help to better comprehend the development and progression of PDAC and will facilitate the discovery of new molecular targets and tools.
Alterations in splicing variants in PDAC
Defects in alternative splicing often result in the appearance of abnormal splicing variants that can play an oncogenic function by conferring advantages to cancer cells. Observations from early studies on alternative splicing in PDAC, employing expression microarray techniques, prompted further analysis that applied more sophisticated bioinformatic approaches to explore the pattern of splicing events and signatures in PDAC cell lines [
41] and human tissue [
42,
85]. The landscape of alternative splicing in PDAC shows that the most common alterations in the protein-coding genes are skipped exon and alternative first exon, followed by intron retention [
42].
In the last two decades, numerous studies have aimed to achieve a deeper understanding of the precise regulation of alternative splicing of individual genes and its mechanistic basis and pathological implications in PDAC (Table
3). A case as paradigmatic as intricate is the study of
CD44, a multifunctional cell surface glycoprotein involved in structural and functional roles in cell-cell and cell-matrix interactions. The standard isoform of
CD44 (
CD44s or
CD44h) only contains the five first exons [
1‐
5] and last five exons [
16‐
21], while the alternative variants CD44v have variable exons (v1-v10) that are alternatively spliced and incorporated between the exons 5–16, conditioning its final structure and thus its biological role [
86]. The
CD44 variants
CD44v2 and
CD44v6, can be detected in human PDAC tissue by immunohistochemistry, where their expression is connected to an increase in mortality rate [
87‐
89]. Recently, Zhao et al. delved into the potential role of
CD44 isoforms in PDAC cell lines, linking these to an EMT phenotype and higher invasiveness and chemoresistance features [
90]. Another study by Zhu et al. described that
CD44v3 is associated with poor prognosis in PDAC, with its generation being regulated by splicing factor
U2AF1 [
91]. A study more focused on metastasis by Xie et al. showed that isoform
CD44v6 is essential for liver fibrosis and metastasis from PDAC and could be used as metastasis and prognosis biomarker [
92]. Actually, it was demonstrated that peptide inhibitors of this isoform block tumor growth and metastasis in rodent models of PDAC [
93]. Thus, although it is still unknown whether and how expression of
CD44 variants specifically affect the cellular function of PDAC cells in vivo in patients, alternative splicing of this gene and their variant products comprise likely actionable targets and tools in PDAC.
Table 3
Altered or aberrant isoforms in PDAC
CD44 | CD44v2 CD44v6 | EMT to MET transition and tumor invasiveness | |
CCK2 | CCK2i4svR | Cholecystokinin and gastrin mediated pathways | |
PRLR | PRLR-SF | Proliferation, tumor growth | |
FGFR1 | FGFR1-IIIb FGFR1-IIIc | Cell proliferation, adhesion, and movement | |
MUC4 | MUC4/Y MUC4/YX | Cell adhesion, immune response, and cell signaling | |
BCL2L1 | BCL-xL BCL-xS | Apoptosis | |
TF | flTF asTF | Blood coagulation cascade and vascularization | |
VEGFA | VEGF-111 VEGF-145 | Transition from other lesions to PDAC | |
KRAS | KRAS4A KRAS4B | Cell proliferation and apoptosis | |
Splicing variants of relevant receptors have also been described in PDAC. An aberrant variant of secretin receptor (encoded by the
SCTR gene) was found, where the third exon is spliced out and therefore residues 44–79 from the NH [2]-terminal tail are eliminated, blocking secretin binding, and thus prompting tumor growth and progression [
94,
95]. Moreover, the potential of the secretin receptor variant as an early diagnostic serum biomarker has been proposed [
96]. Likewise, an aberrant variant of the cholecystokinin and gastrin receptor,
CCKR2, which retains the fourth intron, was identified in PDAC, but was absent in normal pancreas. This variant is constitutively active, may contribute to pathological features in vitro, and it might be associated to a polymorphism of
U2AF35, affecting the splicing regulation of this receptor [
95]. Furthermore, Ryberg and collaborators reported three novel
CCKR2 splice-forms in PDAC, different from the better known
CCK2i4svR variant, which might have similar functions [
97]. Another example is the prolactin receptor,
PRLR, that has been previously linked to carcinogenesis [
98]. This gene undergoes alternative splicing in PDAC, allowing for the formation of several splicing isoforms that differ from each other in the intracellular domain and thus they promote the activation of different downstream signaling pathways [
99]. The most abundant and best-known isoform is
PRLR-LF, by which prolactin mainly transmit its signals. In contrast, the short isoform,
PRLR-SF, is not as well understood. A recent study demonstrated that
PRLR-SF reduces nucleotide synthesis by inhibiting the pentose phosphate pathway (PPP) through the NEK9-Hippo pathway in PDAC cells and in xenografted tumors in mice, hindering proliferation and tumor growth.
PRLR-SF regulates the PPP pathway by reducing the expression of two rate-limiting enzymes G6PD (Glucose-6-phosphate dehydrogenase) and TKT (transketolase). PPP generates both pentose phosphate for nucleic acid synthesis and NADPH for fatty acid synthesis, being a key pathway for cell proliferation which is also upregulated in PDAC cancer stem cells (CSCs) [
100]. Therefore,
PRLR-SF might play an important role in metabolic reprograming, thus preventing PDAC tumor progression [
101].
Fibroblast Growth Factor Receptors comprise a family of tyrosine kinase receptors (
FGFR1-4), whose presence, signaling, and therapeutic potential in PDAC has been recently reviewed [
102]. Early work demonstrated the existence of different splicing isoforms for
FGFR1,
2 and
3 in PDAC [
103], where their differential expression was linked to tumor biology. Thus, whereas
FGFR1-IIIb and
FGFR1-IIIc isoforms are mainly expressed in epithelial and mesenchymal cells, respectively, they are co-expressed in PDAC cells, promoting tumorigenicity by modulating cell proliferation, adhesion, and movement, possibly via activation by FGF5 [
104‐
108].
Also related with epithelial cells are mucins, which protect and lubricate the ducts and are involved in the differentiation and renewal of the epithelium and the modulation of cell adhesion, immune response, and cell signaling. The expression of mucin subtypes and their splice variants are used to classify PDAC in four different subtypes, which were differentially associated to patient survival [
109]. Specifically,
MUC4 presents several splicing variants, differing in the lack of exons 0–21, being
MUC4/Y and
MUC4/X [
110] the best known isoforms. Interestingly,
MUC4 isoforms are mainly expressed in PDAC and correlated with tumor malignancy, while the canonical isoforms are not detectable in normal pancreas [
111‐
113].
The BCL2 family is composed by a number of proteins that play critical roles in apoptosis. Alternative splicing of one of its members,
BCL2L1, results in the production of two variants with opposite functions,
BCL-xL (anti-apoptotic) and
BCL-xS (pro-apoptotic), due to the retention/lack of exon 2, respectively. In PDAC,
BCL2 dysregulation has been associated with apoptosis resistance, due to
BCL2L1 anti-apoptotic isoform overexpression in human tumor tissue [
114]. Furthermore, its expression has been related to the progression to high-grade PanINs in mice [
115]. The presence of specific
BCL2 isoforms is differentially regulated by several splicing factors, where
HNRNPF,
HNRNPH,
KHDRBS1,
RBM11, or
RBM25 promote the short variant, whereas
SRSF1,
SRSF9, or
SF3B1 promote the longer variant [
116]. In PDAC, this regulation has been examined in model cell lines, where the role of
SF3B1 in the 5’splice site activation of
BCL-xS has been confirmed [
52,
117].
Tissue Factor (
TF) is a glycoprotein primarily involved in the blood coagulation cascade. Its alternative splicing isoform, known as
asTF, excludes the fifth exon and exhibits low prothrombogenic potential [
118]. Regulation of
asTF has been linked with several splicing factors, specifically with the SR family:
SRSF6. In PDAC,
asTF has been identified in tumor tissue, correlating with tumor infiltration. Moreover, in PDAC cell lines,
asTF promotes tumor vascularization and tumorigenesis by different pathways, such as EGFR and EMT [
119]. Likewise, asTF plasmatic levels are found at higher levels in PDAC patients than in healthy subjects, suggesting a potential use as a biomarker [
120].
The vascular endothelial growth factor A (
VEGFA) gene has 8 exons that can be alternatively spliced in multiple ways. Isoforms generated differ from each other in their affinity for binding sites, tissue localization, and their capacity to be diffusible [
121].
VEGFA was previously described as a potential biomarker for benign pancreatic serous cystic neoplasm, which could help differentiate them from other types of lesions that can evolve into PDAC, like intraductal papillary mucinous neoplasms and mucinous cystic neoplasms [
121]. Recently, it was observed that
VEGFA spliced isoforms show different expression levels in normal pancreas, benign pancreatic serous cystic neoplasms, mucinous cystic neoplasms and intraductal papillary mucinous neoplasms [
122].
KRAS mutations are prevalent in more than 95% of pancreatic cancers [
123]. The
KRAS gene encodes two splicing isoforms,
KRAS4A and
KRAS4B, products of alternative splicing of the fourth coding exons 4 A and 4B, being mutually exclusive. While
KRAS4B is one of the most studied oncogenes, the role of
KRAS4A is much less known. In PDAC, both isoforms are detectable, but their specific role has not yet been elucidated [
52], although it may parallel that found in colorectal carcinoma, where
KRAS4A has been associated to a suppressive and pro-apoptotic activity, while
KRAS4B would play an anti-apoptotic effect [
124].
Study of EMT-related alternative splice events unveiled that specific splice events from
TMC7 and
CHECK1 were associated with metastatic PDAC. Additionally, the inclusion of exon 17 of
TMC7 was associated to poor prognosis in PDAC, meanwhile its knockdown reduced tumor-related properties [
125]. In addition, aberrantly spliced variants have also been described and their role examined in PDAC. For example, while searching genomic variants that could be linked to splicing alterations, an allele was found to promote the generation of a truncated splice variant of the Elongator Acetyltransferase Complex Subunit 2 (
ELP2). This aberrant variant acts as tumor suppressor for PDAC by blocking STAT3 oncogenic pathway [
126].
As the precision and breadth of RNA sequencing approaches improve and the analysis of splicing receives more attention, the discovery of novel variants and the study of their potential pathogenic role advances further in PDAC will certainly heighten.