Background
Gliomas are the most frequent (> 80%) primary malignant brain tumor in adults [
55]. They are classified into low-grade (I and II) and high-grade (III and IV) gliomas based on integrated classic histological/molecular features [
43]. Grade IV astrocytoma, the most prevalent glioma, known as glioblastoma multiforme (GBM), is one of the most devastating and malignant cancers [
55] and its incidence has increased relevantly in recent years, while in other gliomas remained stable [
57]. Despite significant advances in the knowledge of GBM pathophysiology, it remains an incurable disease with median survival after diagnosis of ~ 15 months [
54,
61]. Still, effective therapeutic targets are severely lacking, and, therefore, innovative therapeutic approaches are urgently needed [
62].
Growing evidence indicates that defects in the alternative splicing process are frequent in cancer, which has gained important attention in the past 10 years [
27,
53]. Moreover, our group and others have demonstrated that the spliceosome, the cellular machinery controlling the splicing process, is drastically altered in GBM and different cancer types [
7,
23,
31,
34,
67], leading to the appearance of aberrant/oncogenic splicing variants (SVs) from different genes [e.g.,
GFAP [
47]
/VEGF [
26]
/TP53 [
2]
/BCL2L1 [
72]
/TP73 [
23]]. Specifically, we have demonstrated that the dysregulation of the spliceosome is associated with GBM development/progression/aggressiveness, which could potentially be considered as a source of novel diagnostic/prognostic-biomarkers and therapeutic targets to combat this devastating pathology [
23].
The splicing-factor-3B-subunit-1 (SF3B1) is a core spliceosome component essential for splicing function [
66]. SF3B1 gained importance due to many functionally deleterious mutations found in various cancer types [
41] [i.e., myelodysplastic syndrome [
29]/breast cancer [
20]/prolactinomas [
38]/uveal melanoma [
32]/pancreatic ductal adenocarcinoma [
4]], which are associated with patient poor-prognosis/survival. Additionally, we have recently found that
SF3B1 is overexpressed and associated with malignant features in prostate cancer [
30] and hepatocellular carcinoma [
42], supporting that SF3B1 could represent a valuable therapeutic target in cancer. Accordingly, various drugs have now been designed to specifically target SF3B1, including pladienolide B, a selective inhibitor that disrupts the spliceosome assembly [
16,
35,
48]. However, to the best of our knowledge, the oncogenic implication of SF3B1, its somatic mutations, and expression profile or its association with molecular features and clinical parameters have not been characterized in GBM, nor its putative therapeutic potential. Therefore, different human cohorts and a dataset from different glioma mouse models were analyzed to determine the mutation frequency as well as the gene and protein expression levels between tumor and control samples of the SF3B1, an essential and druggable spliceosome component.
SF3B1 expression was also explored at the single-cell level across all cell subpopulations and transcriptomic programs. The association of
SF3B1 expression with relevant clinical data (e.g., overall survival) in different human cohorts was also analyzed. Moreover, several functional and molecular endpoints were measured in different GBM cell models (human primary cultures and two cell lines) after SF3B1 blockade (using pladienolide B treatment). In addition, tumor progression and initiation in response to SF3B1 blockade were examined in two GBM xenograft mouse models. These analyses unveil SF3B1 as a potential biomarker being a novel pharmacological target in this devastating tumor.
Discussion
Targeting the spliceosome machinery could become an innovative and successful therapeutic approach to treat incurable cancers like GBM. Indeed, the transcriptomic landscape of cancer cells makes them particularly vulnerable to pharmacological inhibition of splicing, which might have important therapeutic relevance in the near future as suggested by multiple ongoing clinical trials aimed to answer this question [
7]. Specifically, various drugs have been designed to target SF3B1 (a central/essential core-component of the spliceosome) [
16,
35,
48], making this spliceosome element the best candidate to study its translational oncogenic implication and therapeutic capacity in cancers wherein there are no successful treatments or cure. However, the data published so far focused on the potential oncogenic role and therapeutic effectiveness of the modulation of critical spliceosome components through pharmacological approaches is quite limited, fragmentary, and unclear [
7,
48,
52]. To the best of our knowledge, the oncogenic implication and therapeutic capacity of SF3B1, its somatic mutations, and expression profile have not been characterized in GBM, neither its association with molecular features nor clinical parameters.
Herein, we demonstrated that SF3B1 dysregulation clearly affects several cancer hallmarks including apoptosis/proliferation/migration/angiogenesis/splicing pattern and signaling among others, and, of particular clinical relevance, that it could be associated with the development of drug resistance. Since splicing perturbations are common in cancer, including brain tumors [
23], and are associated with mutations and/or altered expression of splicing machinery [
23,
34,
70], we determined the
SF3B1mut-frequency and whether these mutations were associated with glioma progression. Interestingly,
SF3B1mut-frequency was low in glioma patients (~ 1%) compared to other cancer pathologies [wherein
SF3B1mut range from 5% in breast cancer to 81% in myelodysplastic syndromes [
7,
36,
69]]. Moreover, no difference was observed in mean OS of glioma patients with
SF3B1mut compared to
SF3B1wt, an observation that is not similar to previous data in chronic lymphocytic leukemia indicating that
SF3B1mut are associated with rapid disease progression and unfavorable OS [
69]. The low
SF3B1mut-frequency found in glioma patients might suggest that the potential SF3B1role in glioma pathogenesis could be exerted through altered expression levels rather than somatic mutations. In fact, we demonstrate for the first time a drastic SF3B1 overexpression (at mRNA/protein levels) in different cohorts of GBMs vs. non-tumor tissues, which was also confirmed in EPed-glioma mouse models vs. control samples. Moreover, bioinformatic analyses revealed a potential diagnostic capacity of SF3B1 levels to discriminate between GBM/gliomas vs. control tissues from humans and mice, suggesting that GBM/glioma curse with a global dysregulation of SF3B1 in different species. Furthermore, our data revealed a potential utility of SF3B1 as aggressiveness biomarker in GBM which is supported by the direct and strong association found between SF3B1 expression levels and relevant development/progression tumor -markers (e.g., MKI67
/PDGFRA) [
59] and different oncogenic spliceosome components, including
SRSF3 (the most critical splicing machinery component in GBM recently identified by our group) [
23], in human GBM and tumor -samples from EPed-glioma mouse models.
Most importantly, this study revealed that high
SF3B1 expression is directly associated with a worse OS rate in GBM patients, certainly, the main clinical problem in this pathology. This finding was corroborated in two external patient cohorts with GBM (Rembrandt/CGGA-dataset) and further supported by similar observations found in other tumor pathologies [
3,
4,
32,
38,
49,
63]. Similarly, a higher
SF3B1 expression was observed in human and mouse classical and mesenchymal GBM (subtypes with poorer survival rate) compared to proneural GBM (subtype with better survival rate) or non-tumor samples, which reinforced the prognostic value and potential oncogenic role of
SF3B1 [
68]. To the best of our knowledge, this is the first report identifying the diagnostic and prognostic capacity of
SF3B1 in human GBM, and in glioma mouse models with different prognoses, wherein these observations suggest a causal link between
SF3B1 dysregulation and GBM aggressiveness. Notably, we also characterized
SF3B1 expression at the single-cell level demonstrating
SF3B1 was homogeneously expressed across all GBM cell populations/states, being higher in cells expressing a proliferative neural progenitors-like transcriptional program. These data are therapeutically important and have a potential translational/oncogenic implication since the current therapeutic strategies for GBM are not efficient at reducing tumor volume/growth or augmenting survival rate, which is likely due, in part, to the resistance acquired by tumors, particularly by neural progenitors-cells, to different current drugs [
51]. Therefore, our data showing that
SF3B1, a druggable spliceosome component, is homogeneously overexpressed in all GBM cell populations/states offer a novel opportunity and therapeutic approach to treat GBM. Remarkably, GDSC-dataset analysis also unveiled a potential implication of
SF3B1 dysregulation in different oncogenic pathways (e.g., mTOR-PI3K/cell cycle/DNA replication, etc.) to confer drug resistance in GBM, which further encourages the use of an SF3B1 specific-inhibitor in GBM.
Indeed, we demonstrate strong in vitro/in vivo antitumor actions of pladienolide B in GBM cells. Notably, SF3B1 blockade induced marked reductions in aggressiveness features of different GBM cell models [cell -lines and primary-GBM cell -cultures, i.e., inhibition of proliferation/migration/VEGF secretion, and increase of apoptosis]. Most notably, SF3B1 blockade strikingly decreased also GBM-stem/progenitor cells in terms of tumorspheres number and area, both relevant functional results that may help to explore the GBM onset and how to overcome the well-known GBM -resistance to different/current drugs [
21,
51]. It should be emphasized that our data also suggest that pladienolide B effects selectively impact on GBM cells and not non-tumor brain cells, which is clinically relevant and agrees with previous data in other cancer-types, where splicing inhibitors exert stronger, more selective actions on cancer cells than on non-transformed cells [
14]. Moreover, we demonstrate that SF3B1 is also an effective target in GBM in vivo since pladienolide B treatment effectively blocks GBM progression of already established GBM tumors and the GBM onset/formation in preclinical GBM mouse models. Indeed, pladienolide B clearly blunted tumor volume compared to control tumors (which drastically continued their progression), and markedly decreased tumor-weight/mitosis-number of GBM cells and vascularization and necrosis in vivo. Furthermore, pharmacological SF3B1 blockade also decreased the expression of key tumor progression markers and critical oncogenic spliceosome components in GBM cells in vitro and in vivo. Remarkably, pre-treatment with pladienolide B in vitro (24/48 h) was also capable to impair the in vivo onset/formation of GBM possibly through disruption of GBM stem-cell survival. Thus, all these robust in vitro/in vivo results, together with the extended OS observed in different human cohorts, unveiled an important pathophysiological role of SF3B1 in GBM. Although some aspects should be considered when using pladienolide B (e.g., specific concentration used, possible side effects in patients, etc), our data suggest that SF3B1 blockade could be a novel therapeutic avenue with relevant pathophysiological/clinical-potential to combat this devastating disease.
We also interrogated the signaling mechanisms underlying the antitumor actions of SF3B1 blockadge in GBM in vitro and in vivo. Our data revealed, for the first time in GBM, a striking alteration in relevant routes closely associated with GBM progression and initiation, especially the AKT-mTOR and ß-catenin signaling pathways [
12,
37,
46,
56], in response to SF3B1 blockade. In support of the link between SF3B1 activity and AKT pathway, it has been recently reported that
SF3B1K700E mutation can modulate the expression of key components of the AKT pathway with resulting increases in the migration/invasion of breast cancer cells [
39]. Specifically, we observed an overall downregulation in several critical points belonging to these pathways [i.e., total-protein -levels of AKT/MTOR/S6K1/CTNNB1/TP53/HIF1A; phosphorylated-protein levels of AKT/MTOR/S6K1/PDK1 and expression levels of
CCND1 and
MYC) and an upregulation of phosphorylated-TP53 levels, in different GBM models (in vitro and/or in vivo) in response to SF3B1 blockade. Moreover, our data indicate that pladienolide B inhibitory actions observed in AKT-mTOR/ß-catenin signaling pathways may likely be exerted through a significant down-regulation in
SRSF1-levels, a relevant pro-oncogene overexpressed in GBM [
1,
13,
23,
71,
74] which acts activating both signaling pathways simultaneously [
22,
64]. This idea is further supported by the fact that SRSF1 and SF3B1 are functionally connected since SRSF1 directly interacts with the U2-snRNP complex where SF3B1 takes part [
15], and by our data indicating that
SRSF1 expression is strongly correlated with
SF3B1,
MTOR, and
CCTNB1 expression in GBM. Therefore, these data provide original, compelling evidence that SF3B1 is functionally linked, likely via SRSF1 modulation, to these well-known relevant pro-oncogenic pathways (AKT-mTOR/ß-catenin) in GBM, which further supports the pathophysiological relevance of SF3B1 and the antitumor actions of SF3B1-blockade in GBM. Interestingly, SF3B1 blockade suppressed
SF3B1 -expression suggesting positive feedback that could enhance its antitumor effects.
SF3B1 blockade also exerted important molecular actions involving the splicing modulation of two clinically relevant SVs of
BCL2L1 (
Bcl-xL and
Bcl-xS) associated with cancer -development and known to play an oncogenic role and a tumor suppressor actions, respectively [
6,
10]. Specifically, SF3B1-blockade downregulated anti-apoptotic
Bcl-xL, while upregulated the pro-apoptotic
Bcl-xS, in GBM, both in vitro and in vivo, but not in non-tumor brain cell cultures. This idea was further corroborated by PSI analysis demonstrating a pladienolide B-induced reduction of
BCL2L1 A5SS splicing -event in GBM in vitro and in vivo. In line with these data, previous reports found that several apoptosis-regulatory genes, including
BCL2-related genes, generate alternatively SVs with opposite activities, which is a biological program often employed by cancer -cells to escape from intrinsically programmed cell death and radiotherapy/chemotherapy-induced cytotoxicity [
72]. In fact, our observations in
Bcl-xL/xS, together with the data demonstrating the implication of
SF3B1 dysregulation in different oncogenic -pathways that confers drug resistance in GBM (e.g., mTOR-PI3K/cell cycle/DNA -replication, etc.), might be clinically relevant because it has been demonstrated that
Bcl-xL is transcriptionally upregulated and associated with poor prognosis and chemoresistance in many cancers [
6,
10]. In this sense, it should be indicated that the use of two different ASOs that inhibited the pro-apoptotic
Bcl-xS variant and promoted the anti-apoptotic
Bcl-xL variant in response to pladienolide B treatment was able to significantly reduce the antitumor effect of pladienolide B on GBM cells. All these data demonstrate that changes in the splicing of
BCL2L1 seem to be one of the main molecular mechanisms underlying the link between SF3B1 blockade and the significant decrease in GBM onset, GBM progression, and aggressiveness features observed in response to pladienolide B treatment.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.