Abstract
Background
Malignant pleural mesothelioma (MPM) is a rare and highly malignant thoracic tumor. Although alternative splicing (AS) is associated with tumor prognosis, the prognostic significance of AS in MPM is unknown.
Methods
Transcriptomic data, clinical information, and splicing percentage values for MPM were obtained from The Cancer Genome Atlas (TCGA) and TCGA SpliceSeq databases. Least absolute shrinkage and selection operator (LASSO) regression and multivariate Cox analyses were performed to establish a model affecting the prognosis of MPM. Survival and ROC analyses were used to test the effects of the prognostic model. LASSO/multivariate Cox analysis was used to construct the MPM prognostic splicing factor (SF) model. The SF–AS interaction network was analyzed using Spearman correlation and visualized using Cytoscape. The association between the MPM prognostic SF model and drug sensitivity to chemotherapeutic agents such as cisplatin was analyzed using pRRophetic.R.
Results
The LASSO/multivariate Cox analysis identified 41 AS events and 2 SFs that were mostly associated with survival. Nine prognostic prediction models (i.e., seven types of AS model, total AS model, and SF model) were developed. An MPM prognostic SF–AS regulatory network was subsequently constructed with decreased drug sensitivity in the SF model high‐risk group (p = 0.025).
Conclusion
This study provides the first comprehensive analysis of the prognostic value of AS events and SFs in MPM. The SF–AS regulatory network established in this study and our drug sensitivity analysis using the SF model may provide novel targets for pharmacological studies of MPM.
Prognostic signatures for alternative splicing events and splicing factors in malignant pleural mesothelioma.
1. INTRODUCTION
Malignant pleural mesothelioma (MPM) is a rare and extremely malignant tumor 1 that is mainly caused by asbestos. Despite the reduced use of asbestos, the mortality rate of patients with MPM is increasing every year. 2 , 3 Currently, there are limited treatment alternatives for advanced MPM, and the use of cisplatin/pemetrexed combination improves patient survival by 1–2 months; therefore, there is an urgent need for the development of new pharmacological targets. 3 , 4 Alternative splicing (AS) is a common regulatory process in gene expression that produces various mRNA isoforms using different splicing methods. 5 The occurrence of AS may affect tumor oncogene expression, leading to the development of cancer. 6 , 7 , 8 For example, AS of TrkA regulates the development of human neuroblastoma, 9 MTR4 drives liver tumorigenesis by promoting cancer metabolic conversion through AS, 10 and AKAP8 inhibits tumor metastasis by antagonizing EMT‐related AS. 11 Thus, targeted AS has been newly developed for treating tumors. 12 , 13 The proteins that regulate AS are known as splicing factors (SFs). Based on their functional roles, they can be categorized as small nuclear ribonucleoprotein particle (snRNP) and non‐snRNP protein factors. 14 Furthermore, a single SF may regulate multiple splicing events and play an essential role in tumor development, which is associated with several tumor prognoses. 15 , 16 , 17 To improve MPM prognosis and suggest novel ideas for MPM‐targeted pharmacological experiments, we analyzed MPM data in The Cancer Genome Atlas (TCGA). 18 TCGA SpliceSeq 2.0 19 was used via a bioinformatics approach to determine the AS events and SFs with considerable prognostic relevance during MPM development.
2. MATERIALS AND METHODS
2.1. Data download and preprocessing
RNA‐seq data (FPKM) and clinical information regarding MPM were downloaded from the TCGA database (https://portal.gdc.cancer.gov/) 18 and TCGA SpliceSeq database 2.0 (https://bioinformatics.mdanderson.org/public‐datasets/). 19 Samples with PSI values of >75% were selected to download the percentage of AS (PSI) values for MPM. Patients without follow‐up information and those with a survival of <30 days were excluded. The impute.knn function was used to add missing data and remove the samples with PSI values of <0.05 and AS events with standard deviations of <0.01, thus preventing interference from unchanged AS samples. AS events involved the alternative acceptor (AA) sites, alternative donor (AD) sites, alternative promoters (AP), alternative terminators (AT), exon skipping (ES), mutually exclusive exons (ME), and retained introns (RI).
2.2. Univariate Cox analysis to screen for prognosis‐related AS events
Univariate Cox analysis was performed for the preliminary screening of prognostic tumor correlates. Clinical data and PSI values obtained from the TCGA database were combined to initially screen AS events and determine overall survival (OS) using a one‐way Cox analysis. The AS events associated with OS (p < 0.05) were selected, and the results were visualized using the upset.R package. 20 We used the ggplot2.R package to clearly demonstrate the association between each AS event and MPM prognosis. 21 In this package, prognosis‐related AS events were presented as bubble plots and volcano plots. Univariate Cox analysis was used to screen for SFs that might be associated with MPM prognosis.
2.3. LASSO regression and multivariate Cox analyses to construct a prognostic model
To investigate each type of AS and the association between SFs and MPM prognosis, we individually modeled the prognosis for each type of AS event (including a total AS model). Overfitting AS events and SFs were excluded using LASSO regression. Multivariate Cox analysis was used to determine the final survival‐related AS events, estimate the OS risk values for MPM, and visualize these events using R. Moreover, it was used to screen for SFs that were strongly associated with MPM prognosis. Further, the MPM samples were stratified according to the AS type and categorized into high‐ and low‐risk groups based on the median risk values. Kaplan–Meier survival analysis 22 and ROC analysis were performed for both groups to test the validity of the model and reliability of model predictions, respectively.
2.4. Analysis of independent prognostic factors
To estimate the prognostic value of the AS model, we analyzed several clinical prognostic factors, including risk scores for sex, pathological stage, T category (to assess the aggressiveness of the tumor), N category (to assess the lymph node metastasis of the tumor), M category (to assess the distant metastasis of the tumor), and markers. These factors were considered as independent prognostic variables when the p value was <0.05 in univariate and multivariate analyses.
2.5. SF–AS interaction network analysis
We downloaded the currently known 404 SFs 23 to determine the SF expression values (FPKM) using RNA‐seq expression profiles of patients with MPM. Univariate Cox analysis was performed to screen for prognostic associations with SFs, and Spearman test analysis was used to identify the AS events and SFs that were strongly associated with OS. The SF–AS interaction network was mapped using the Cytoscape software. 24 When the SF is more connected and centrally located in the network, it is more likely to play a central role in the development of MPM.
2.6. Gene set enrichment analysis (GSEA) of SF
To investigate the molecular mechanism of SF affecting MPM prognosis and suggest ideas for subsequent basic experiments, we selected the core SFs in the SF–AS network, used the SF expression values in TCGA data (log2FPKM), categorized the samples into survival and death groups, and used an unpaired t‐test to screen for differences in the survival and death of SFs. After LASSO/multivariate Cox analysis, the SF prognostic model of MPM was constructed and tested using survival, ROC, and independent prognostic analyses. The association between the prognostic model and drug sensitivity to cisplatin was tested using the pRRophetic.R package. 25 Samples were classified into two groups based on their median core SF expression level. The pathways in which the core SF decreased the sensitivity of sensitivity of mesothelioma to drugs were analyzed using the GSEA software. 26 Major pathways were selected for visualization.
3. RESULTS
3.1. Primary screening of MPM prognosis‐associated AS events
RNA‐seq expression profiles and clinical information of 87 samples were obtained from the TCGA official website, and data on 43,433 cross‐splicing events were obtained from the TCGA SpliceSeq 2.0 database. A general flowchart is shown in Figure 1. The association of genes with cross‐splicing types is represented using an upset plot (Figure 2A). In total, 84 samples and 28,664 AS events remained after filtering samples with a total survival time of <30 days using R. A total of 3974 prognosis‐associated crossover splicing events were initially screened using the univariate Cox analysis. The upset plot showed that multiple crossover splicing events could occur for each gene. The most common type of crossover splicing was ES, whereas the least common type was ME (Figure 2B). The top 10–20 AS events were selected using the p‐values of one‐way Cox analysis and were presented using a volcano plot to clearly demonstrate the prognostic risk of various AS events in patients with MPM (Figure 3).
FIGURE 1.
Flowchart. The general flow of the article is shown in the figure. After obtaining MPM data from TCGA and TCGA SpliceSeq databases, AS event modeling was performed to suggest new ideas for MPM research.
FIGURE 2.
Upset plot for AS events and genes for MPM. Upset plots for all seven types of AS events for MPM (A) and MPM prognostic associations (B).
FIGURE 3.
Bubble and volcano plots for MPM prognosis‐associated AS events. The top 10–20 AS events ranked by the p‐values of the one‐way Cox analysis for types AA (A), AD (B), AP (C), AT (D), ES €, ME (F), and RI (G). Figure 3H shows a volcano map for prognosis‐related alternate splicing events. AA, alternate acceptor site; AD, alternate donor site; AP, alternate promoter; AT, alternate terminator; ES, exon skipping; ME, mutually exclusive exon; RI, retained intron.
3.2. Eight MPM prognosis AS models
Different AS types were combined to construct prognostic models for each of the seven AS types and an overall AS prognostic model. The AS events were again filtered using multivariate Cox/LASSO regression (Figures S1 and S2), leaving a total of 41 prognosis‐associated AS events (Table 1) to prevent event overfitting. The samples were categorized into high‐ and low‐risk groups based on the mean risk values. The risk curves of different prognostic models were plotted using R. The higher the risk value, the lower was the patients' survival rate (Figures S1 and S2). The survival curves of different prognostic models were constructed using the survivor.R package 22 to investigate the prognostic values of these models. All eight models had good prognostic values, and AA was most strongly associated with prognosis (p = 3.631e−11) (Figure 4). According to the results of the ROC analysis, the larger the area under the curve (AUC), the more accurate would be the prognostic model; the AA model (AUC = 0.925) had the highest prognostic value for predicting MPM (Figure 5).
TABLE 1.
Overall survival by LASSO/multivariate Cox analysis of prognosis‐related variable shear events
| AS event | AS Types | Coef | HR | HR.95 L | HR.95H | p value |
|---|---|---|---|---|---|---|
| PTPN7|9405|AA | AA | −16.4231 | 7.37E‐08 | 1.41E‐10 | 3.86E‐05 | 2.72E‐07 |
| ARL5A|55,593|AT | AT | −32.3272 | 9.13E‐15 | 1.01E‐20 | 8.28E‐09 | 3.86E‐06 |
| KANSL3|54,543|ES | ES | −8.55493 | 0.000193 | 2.87E‐06 | 0.012915293 | 6.69E‐05 |
| INO80E|36,010|ES | ES | −2.72802 | 0.065349 | 0.01674 | 0.255110399 | 8.64E‐05 |
| AGTRAP|668|ES | ES | −26.5163 | 3.05E‐12 | 4.06E‐18 | 2.29E‐06 | 0.000122 |
| C5orf38|71,503|AT | AT | −3.73193 | 0.023947 | 0.003358 | 0.170773829 | 0.000197 |
| LTC4S|74,924|AA | AA | −50.9154 | 7.72E‐23 | 1.60E‐34 | 3.72E‐11 | 0.000208 |
| GEMIN6|53,288|AD | AD | −13.3838 | 1.54E‐06 | 7.13E‐10 | 0.003326874 | 0.000634 |
| ACOXL|54,941|AT | AT | −2.74687 | 0.064128 | 0.012799 | 0.321300648 | 0.000835 |
| ARL5A|55,593|AT | ALL | −22.9286 | 1.10E‐10 | 1.40E‐16 | 8.66E‐05 | 0.000931 |
| IL32|33,418|AA | AA | −4.27077 | 0.013971 | 0.001066 | 0.183134759 | 0.001142 |
| PKMYT1|33,330|AA | AA | −6.30277 | 0.001831 | 3.63E‐05 | 0.092452858 | 0.001633 |
| MORC4|89,814|AD | AD | −7.19528 | 0.00075 | 8.20E‐06 | 0.068658607 | 0.001794 |
| MSS51|12,150|AA | AA | −29.2668 | 1.95E‐13 | 1.10E‐21 | 3.44E‐05 | 0.00252 |
| POLR2H|67,943|AA | AA | 13.73347 | 921236.6 | 113.109 | 7,503,174,698 | 0.002798 |
| TMEM33|69,137|RI | RI | −10.032 | 4.40E‐05 | 5.91E‐08 | 0.032739853 | 0.002946 |
| PKMYT1|33,330|AA | ALL | −4.94584 | 0.007113 | 0.000271 | 0.186437805 | 0.002999 |
| DUT|30,485|AP | AP | 3.904597 | 49.63006 | 3.392975 | 725.9536336 | 0.004338 |
| TBC1D7|75,381|ES | ES | −6.01096 | 0.002452 | 3.73E‐05 | 0.161100953 | 0.004878 |
| TMC7|34,279|AT | ALL | −3.58995 | 0.0276 | 0.002184 | 0.348767234 | 0.00554 |
| ACOXL|54,942|AT | ALL | 2.780201 | 16.12226 | 2.230998 | 116.5071299 | 0.005866 |
| MBD3|46,525|RI | RI | 12.82166 | 370148.9 | 39.66551 | 3,454,139,960 | 0.005976 |
| COPZ2|120,285|ES | ALL | 3.867227 | 47.80962 | 3.027093 | 755.1004487 | 0.006021 |
| EPN1|52,138|AA | AA | −12.3437 | 4.36E‐06 | 6.35E‐10 | 0.029901776 | 0.006169 |
| EPS15L1|48,158|AT | AT | −5.14956 | 0.005802 | 0.000126 | 0.267389956 | 0.008417 |
| PLSCR1|67,169|ES | ES | −7.04999 | 0.000867 | 3.85E‐06 | 0.195185382 | 0.010736 |
| TTC13|10,258|ME | ME | 2.245038 | 9.440775 | 1.630633 | 54.65867847 | 0.012221 |
| MEIS3|50,645|AA | AA | 1.965686 | 7.139811 | 1.479656 | 34.45185452 | 0.01437 |
| RPL13|392,312|AD | AD | 5.833494 | 341.5498 | 2.723692 | 42830.20911 | 0.01796 |
| MLST8|33,226|AA | AA | −14.6293 | 4.43E‐07 | 1.65E‐12 | 0.118941789 | 0.021802 |
| ZNF140|204,379|ME | ME | −5.46178 | 0.004246 | 3.94E‐05 | 0.458065023 | 0.022204 |
| SEC16A|88,181|AA | AA | 6.920263 | 1012.587 | 2.664695 | 384783.7242 | 0.02241 |
| ROMO1|59,225|RI | RI | 8.891291 | 7268.394 | 2.885455 | 18308912.13 | 0.02607 |
| SDR39U1|27,012|ME | ME | −1.36389 | 0.255663 | 0.073959 | 0.883787348 | 0.031148 |
| SRPK2|81,282|AT | AT | 12.33208 | 226858.6 | 2.816276 | 18,274,066,785 | 0.032386 |
| SMPD4|55,291|ME | ME | 2.817861 | 16.741 | 1.262212 | 222.0397519 | 0.032637 |
| GCFC2|54,161|AD | AD | −9.06928 | 0.000115 | 2.58E‐08 | 0.514822037 | 0.034449 |
| ZFP64|59,808|AP | AP | 4.577718 | 97.29208 | 1.151712 | 8218.849812 | 0.043139 |
| TRAPPC2|88,516|RI | RI | 3.184589 | 24.15735 | 1.092294 | 534.2676796 | 0.043816 |
| TIGD6|74,053|AD | AD | 1.824516 | 6.199792 | 1.024474 | 37.51918468 | 0.047001 |
| XRCC4|72,698|AA | AA | −1.18247 | 0.30652 | 0.095302 | 0.985860944 | 0.047272 |
FIGURE 4.
Kaplan–Meier curves for OS between patients in the high‐ and low‐risk groups.
FIGURE 5.
ROC curves of patients in the high‐ and low‐risk groups.
3.3. Independent prognosis AS models
The prognostic models were analyzed for multiple clinical factors using clinical data from the TCGA database. The AS prognostic model could be regarded as an independent prognostic factor for MPM when both univariate and multivariate analyses demonstrated a p value of <0.05. All eight AS models were independent prognostic factors for MPM (Figure 6).
FIGURE 6.
Independent prognostic analysis of AS prognostic models. The figure shows the results of the multivariate independent prognostic analysis of different AS prognostic models.
3.4. SF–AS interaction network
The SF expression values were obtained from TCGA RNA‐seq expression profile data. The R software was used to perform the Spearman test and analyze the p‐values of the AS of SF expression values (log2FPKM values) with the prognostic correlation obtained using the one‐way Cox analysis. When r was >0.6 and Spearman p value was <0.001, SF was considered to be associated with AS events. In the final results, 19 SFs (e.g., FAM50A and RBM47) were visualized using Cytoscape v.3.7.1 (Figure 7). According to the correlation and Cox analyses, the closer the SF proximity to the center of the network, the higher is the probability that it is prognostically meaningful for MPM. This helps in identifying relevant SFs and provides further directions for improving MPM prognosis.
FIGURE 7.
AS–SF interaction network diagram.
3.5. Core MPM prognosis‐associated SF and SF‐enriched pathways
Of the 19 SFs obtained from the SF–AS regulatory network, 11 SFs remained after screening using analysis of variance (unpaired t‐test) (p < 0.05). Multifactorial LASSO/Cox regression analyses finally selected FMP1 and PPM1G as the two SFs that were most strongly associated with MPM prognosis (Table S1). Survival analysis and ROC analysis demonstrated that the two SFs FMP1 and PPM1G had a better prognostic value among other SFs (p = 1.371e − 07, AUC = 0.768, Figure 8). In multivariate‐independent prognostic analysis, the SF prognostic model was an independent prognostic factor for MPM (Figure 8). Furthermore, drug sensitivity analysis of cisplatin revealed that patients with MPM and higher risk values had lower drug sensitivity (p = 0.025, Figure 8).
FIGURE 8.
Independent prognostic models for the SFs FMP1 and PPM1G. (A and B) show the results of the LASSO/multivariate Cox analysis. (C and D) show the Kaplan–Meier and ROC curves for OS between patients in the high‐ and low‐risk groups according to the SF prognostic model. (E and F) show the results of univariate Cox and multivariate Cox analyses, where the risk scores can be an independent prognostic factor for MPM. (G and H) represent a boxplot of the difference in drug sensitivity to cisplatin and docetaxel in the high‐ and low‐risk groups, respectively.
GSEA analysis can be used to investigate individual gene enrichment pathways and establish pathways and functions in which genes may be involved using large amounts of data. In our study, the gene expression profiles obtained from TCGA were used as GSEA background analysis data, which were classified into high‐ and low‐expression groups, to predict the specific pathways in which the two SFs FMP1 and PPM1G affected MPM prognosis. FMP1 and PPM1G were primarily enriched in several signaling pathways, such as TGF‐β, JAK–STAT, and RIG‐I‐like‐receptor (Table 2).
TABLE 2.
GSEA enrichment pathway of splicing factors FMP1 and PPM1G
| NAME | ES | p‐value | Group |
|---|---|---|---|
| KEGG_PROTEIN_EXPORT | 0.753822 | <0.001 | FMR1‐HIGH |
| KEGG_VALINE_LEUCINE_AND_ISOLEUCINE_DEGRADATION | 0.594927 | 0.001332 | FMR1‐HIGH |
| KEGG_RNA_DEGRADATION | 0.582613 | <0.001 | FMR1‐HIGH |
| KEGG_UBIQUITIN_MEDIATED_PROTEOLYSIS | 0.512201 | <0.001 | FMR1‐HIGH |
| KEGG_RIG_I_LIKE_RECEPTOR_SIGNALING_PATHWAY | 0.509803 | <0.001 | FMR1‐HIGH |
| KEGG_JAK_STAT_SIGNALING_PATHWAY | 0.434675 | 0.001101 | FMR1‐HIGH |
| KEGG_SPLICEOSOME | 0.410689 | 0.004566 | FMR1‐HIGH |
| KEGG_NEUROACTIVE_LIGAND_RECEPTOR_INTERACTION | −0.28096 | <0.001 | FMR1‐LOW |
| KEGG_FOCAL_ADHESION | −0.37563 | <0.001 | FMR1‐LOW |
| KEGG_TGF_BETA_SIGNALING_PATHWAY | −0.38041 | <0.001 | FMR1‐LOW |
| KEGG_VASCULAR_SMOOTH_MUSCLE_CONTRACTION | −0.38339 | <0.001 | FMR1‐LOW |
| KEGG_LEUKOCYTE_TRANSENDOTHELIAL_MIGRATION | −0.3853 | <0.001 | FMR1‐LOW |
| KEGG_CELL_ADHESION_MOLECULES_CAMS | −0.39206 | <0.001 | FMR1‐LOW |
| KEGG_LEISHMANIA_INFECTION | −0.39671 | <0.001 | FMR1‐LOW |
| KEGG_HYPERTROPHIC_CARDIOMYOPATHY_HCM | −0.40259 | <0.001 | FMR1‐LOW |
| KEGG_DILATED_CARDIOMYOPATHY | −0.41436 | <0.001 | FMR1‐LOW |
| KEGG_HEMATOPOIETIC_CELL_LINEAGE | −0.47273 | <0.001 | FMR1‐LOW |
| KEGG_ECM_RECEPTOR_INTERACTION | −0.62211 | <0.001 | FMR1‐LOW |
| KEGG_RIBOSOME | −0.6248 | <0.001 | FMR1‐LOW |
| KEGG_GLYCOSAMINOGLYCAN_BIOSYNTHESIS_CHONDROITIN_SULFATE | −0.63757 | 0.003344 | FMR1‐LOW |
| KEGG_STEROID_BIOSYNTHESIS | 0.654061 | 0.004301 | PPM1G‐HIGH |
| KEGG_SPLICEOSOME | 0.705826 | <0.001 | PPM1G‐HIGH |
| KEGG_RNA_POLYMERASE | 0.692128 | <0.001 | PPM1G‐HIGH |
| KEGG_RIBOSOME | 0.729201 | <0.001 | PPM1G‐HIGH |
| KEGG_PYRIMIDINE_METABOLISM | 0.628346 | <0.001 | PPM1G‐HIGH |
| KEGG_PURINE_METABOLISM | 0.489276 | <0.001 | PPM1G‐HIGH |
| KEGG_PROTEASOME | 0.814298 | <0.001 | PPM1G‐HIGH |
| KEGG_PROGESTERONE_MEDIATED_OOCYTE_MATURATION | 0.528318 | <0.001 | PPM1G‐HIGH |
| KEGG_PENTOSE_PHOSPHATE_PATHWAY | 0.583868 | 0.00655 | PPM1G‐HIGH |
| KEGG_OOCYTE_MEIOSIS | 0.56482 | <0.001 | PPM1G‐HIGH |
| KEGG_ONE_CARBON_POOL_BY_FOLATE | 0.651877 | 0.006565 | PPM1G‐HIGH |
| KEGG_NUCLEOTIDE_EXCISION_REPAIR | 0.565179 | <0.001 | PPM1G‐HIGH |
| KEGG_MISMATCH_REPAIR | 0.71348 | <0.001 | PPM1G‐HIGH |
| KEGG_HUNTINGTONS_DISEASE | 0.421105 | <0.001 | PPM1G‐HIGH |
| KEGG_HOMOLOGOUS_RECOMBINATION | 0.635229 | <0.001 | PPM1G‐HIGH |
| KEGG_GLUTATHIONE_METABOLISM | 0.532697 | <0.001 | PPM1G‐HIGH |
| KEGG_GLIOMA | 0.470497 | <0.001 | PPM1G‐HIGH |
| KEGG_DNA_REPLICATION | 0.837645 | <0.001 | PPM1G‐HIGH |
| KEGG_CELL_CYCLE | 0.665223 | <0.001 | PPM1G‐HIGH |
| KEGG_BASE_EXCISION_REPAIR | 0.698721 | <0.001 | PPM1G‐HIGH |
4. DISCUSSION
The development of high‐throughput sequencing has led to recent advances in the study of AS. Studies on MPM and AS events are scarce, and all of them suggest that the occurrence of different AS events affect the formation of mRNA heterodimers, thereby affecting the development of MPM. 27 , 28 , 29 , 30 Langerak et al. (1992) revealed that platelet‐derived growth factor (PDGF), a chain in MPM cell lines, can undergo various AS events and that MPM cells primarily produce mRNA without exon six‐derived sequences in the PDGF A‐chain. 31 Subsequently, Morrison et al. found that the BAP1 mutation C.2054A > T (p.Glu685Val) creates new splicing in MPM cell lines and thus affects mRNA expression. 32 Dong et al. demonstrated for the first time that different CDK4 splice variants are expressed at different levels in the lungs of MPM‐affected patients and in those of normal individuals by analyzing MPM bulk expression microarrays. 33 With the development of high‐throughput sequencing, Bueno et al. performed a comprehensive array analysis of 216 MPM samples in 2016 and identified 177 splicing events. They demonstrated that these AS events are differentially expressed in SF3B1 mutants and wild types. 34 However, the association of AS events with MPM prognosis has not been elucidated.
The present study used the TCGA data, RNA‐seq profiles of 84 MPM cases from the TCGA SpliceSeq database, clinical data, and AS percentage PSI data to investigate the effect of AS events and SF on the prognosis of MPM and construct eight distinct AS prognostic models of MPM and SF–AS regulatory network using the LASSO/Cox regression analysis. Furthermore, we performed a different analysis of SF in the network using the RNA‐seq data of SF in the TCGA database to explore the molecular mechanisms of SFs in the development of MPM. The LASSO/Cox regression analysis of the high‐ and low‐risk groups helped in the identification of the SF that was most strongly associated with MPM prognosis. The analysis revealed that the SFs FMR1 and PPM1G were most strongly associated with MPM prognosis. Drug sensitivity analysis suggested that FMR1 and PPM1G are associated with resistance to multiple MPM chemotherapeutic agents, including cisplatin. To further investigate the mechanisms by which FMR1 and PPM1G acquire MPM resistance, we predicted the pathways of FMR1 and PPM1G convergence in MPM using GSEA analysis. The results showed that FMP1 and PPM1G might acquire MPM resistance through TGF‐β, JAK–STAT, RIG‐I‐like‐receptor, pentose‐phosphate, and other signaling pathways. This enables the advancement of drug resistance in patients with MPM.
The fragile X mental retardation protein (FMR1) is an RNA‐binding protein that is generally believed to regulate the expression of genes. This protein is essential for neuronal growth and synaptic plasticity through transcription control. 35 , 36 Existing studies indicate that its G‐quartet motif can act as a control element (i.e., SFs) of AS, regulating the occurrence of AS events. 37 , 38 This is consistent with the findings of our Spearman test analysis. However, no study has yet described the role of FMR1 in MPM. Based on the results of our analysis, we can speculate that FMR1 affects patients with MPM prognosis through several signaling pathways, such as TGF‐β, JAK–STAT, and RIG‐I‐like‐receptor, or by regulating the occurrence of AS events. Protein phosphatase magnesium‐dependent 1 (PPM1G), a Ser/Thr protein phosphatase in the PP2C family, can affect AS by regulating the dephosphorylation of pre‐mRNA SFs. 39 , 40 When PPM1G is bound to molecules, including 7SK RNA, it prevents the assembly of SFs7SK snRNP and maintains transcriptional elongation. 41 This is another indication that PPM1G can affect the presence of AS events. Additionally, PPM1G may affect the prognosis of MPM primarily by affecting the pentose‐phosphate pathway or by affecting the AS events. However, these are speculations based on bioinformatics analysis, and experimental evidence is required.
5. CONCLUSION
Our study analyzed AS and prognostic values of SF for MPM as well as the potential pathways by which SF affects MPM. However, there are several limitations in our study: (1) the analytical results were obtained through extensive data analysis, which required the experimental evidence for specificity. (2) as MPM is a rare malignancy, the existing sample size of the MPM cohort was small, indicating bias in the results. (3) the novel therapeutic agent pemetrexed for MPM is yet to be analyzed for drug sensitivity.
AUTHORS' CONTRIBUTIONS
YJ analyzed the data and wrote the manuscript; CDZ collected and analyzed the data; YC and SYZ analyzed the data; YPH laid out of figures; JH designed the study and revised the manuscript. All authors read and approved the final manuscript.
FUNDING INFORMATION
This study was funded by the Health Commission of Mianyang City (grant number 202001).
CONFLICT OF INTEREST
The author reports no conflicts of interest in this work.
ETHICS APPROVAL AND CONSENT TO PARTICIPATE
Not applicable.
CONSENT FOR PUBLICATION
Not applicable.
Supporting information
Figure S1
Figure S2
Table S1
ACKNOWLEDGMENT
Not applicable.
Jiang Y, Zhang C, Chen Y, Zhao S, He Y, He J. Prognostic risk assessment model for alternative splicing events and splicing factors in malignant pleural mesothelioma. Cancer Med. 2023;12:4895‐4906. doi: 10.1002/cam4.5174
Yue Jiang and Chengda Zhang contributed equally to this work.
DATA AVAILABILITY STATEMENT
The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.
REFERENCES
- 1. Töreyin ZN, Ghosh M, Göksel Ö, Göksel T, Godderis L. Exhaled breath analysis in diagnosis of malignant pleural mesothelioma: systematic review. Int J Environ Res Public Health. 2020;17:1110. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Sung H, Ferlay J, Siegel RL, et al. Global cancer statistics 2020: Globocan estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin. 2021;71:209‐249. [DOI] [PubMed] [Google Scholar]
- 3. Zauderer MG. A new standard for malignant pleural mesothelioma. The Lancet. 2016;387:1352‐1354. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Disselhorst MJ, Baas P. Chemotherapy options versus “novel” therapies: how should we treat patients with malignant pleural mesothelioma. Transl Lung Cancer Res. 2020;9:S77‐S85. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Wang J, Tian GG, Zheng Z, Li B, Xing Q, Wu J. Comprehensive transcriptomic analysis of mouse gonadal development involving sexual differentiation, meiosis and gametogenesis. Biol Proced Online. 2019;21:1‐13. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Lin J‐C. Therapeutic applications of targeted alternative splicing to cancer treatment. Int J Mol Sci. 2018;19:75. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Peng W‐Z, Liu J‐X, Li C‐F, Ma R, Jie J‐Z. Hnrnpk promotes gastric tumorigenesis through regulating cd44e alternative splicing. Cancer Cell Int. 2019;19:1‐11. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Huang R, Li Z, Li C, et al. Germ cell‐specific gene 1‐like protein regulated by splicing factor cugbp elav‐like family member 5 and primary bile acid biosynthesis are prognostic in glioblastoma multiforme. Front Genet. 2020;10:1380. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Brodeur GM, Iyer R, Croucher JL, Zhuang T, Higashi M, Kolla V. Therapeutic targets for neuroblastomas. Expert Opin Ther Targets. 2014;18:277‐292. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Yu L, Kim J, Jiang L, et al. Mtr4 drives liver tumorigenesis by promoting cancer metabolic switch through alternative splicing. Nat Commun. 2020;11:1‐12. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Hu X, Harvey SE, Zheng R, et al. The rna‐binding protein akap8 suppresses tumor metastasis by antagonizing emt‐associated alternative splicing. Nat Commun. 2020;11:1‐15. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Oltean S. Modulators of alternative splicing as novel therapeutics in cancer. World J Clin Oncol. 2015;6:92‐95. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Martinez‐Montiel N, Rosas‐Murrieta NH, Anaya Ruiz M, Monjaraz‐Guzman E, Martinez‐Contreras R. Alternative splicing as a target for cancer treatment. Int J Mol Sci. 2018;19:545. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Anczuków O, Krainer AR. Splicing‐factor alterations in cancers. RNA. 2016;22:1285‐1301. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Mao S, Li Y, Lu Z, et al. Survival‐associated alternative splicing signatures in esophageal carcinoma. Carcinogenesis. 2019;40:121‐130. [DOI] [PubMed] [Google Scholar]
- 16. Jia K, Wu Y, Huang J, Wu H. Survival‐associated alternative splicing events in pan‐renal cell carcinoma. Front Oncol. 2019;9:1317. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Li X, Wang Y, Li B, Ma W. Prognostic value and potential regulatory mechanism of alternative splicing in geriatric breast cancer. Genes. 2020;11:200. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Hutter C, Zenklusen JC. The cancer genome atlas: creating lasting value beyond its data. Cell. 2018;173:283‐285. [DOI] [PubMed] [Google Scholar]
- 19. Ryan MC, Cleland J, Kim R, Wong WC, Weinstein JN. Spliceseq 2.0: A tool for analysis and visualization of splicing variation in rnaseq data. [DOI] [PMC free article] [PubMed]
- 20. Lex A, Gehlenborg N, Strobelt H, Vuillemot R, Pfister H. Upset: visualization of intersecting sets. IEEE Trans Vis Comput Graph. 2014;20:1983‐1992. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Wickham H. Ggplot2. Wiley Interdiscip Rev Comput Stat. 2011;3:180‐185. [Google Scholar]
- 22. Therneau TM, Lumley T. Package ‘survival’. R Top Doc. 2015;128:28‐33. [Google Scholar]
- 23. Seiler M, Peng S, Agrawal AA, et al. Somatic mutational landscape of splicing factor genes and their functional consequences across 33 cancer types. Cell Rep. 2018;23:282‐296.e4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Shannon P, Markiel A, Ozier O, et al. Cytoscape: a software environment for integrated models of biomolecular interaction networks. Genome Res. 2003;13:2498‐2504. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Geeleher P, Cox N, Huang RS. Prrophetic: an r package for prediction of clinical chemotherapeutic response from tumor gene expression levels. PloS One. 2014;9:e107468. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Subramanian A, Kuehn H, Gould J, Tamayo P, Mesirov JP. Gsea‐p: a desktop application for gene set enrichment analysis. Bioinformatics. 2007;23:3251‐3253. [DOI] [PubMed] [Google Scholar]
- 27. Chew SH, Okazaki Y, Akatsuka S, et al. Rheostatic cd44 isoform expression and its association with oxidative stress in human malignant mesothelioma. Free Radic Biol Med. 2017;106:91‐99. [DOI] [PubMed] [Google Scholar]
- 28. Rundlöf A‐K, Fernandes AP, Selenius M, et al. Quantification of alternative mrna species and identification of thioredoxin reductase 1 isoforms in human tumor cells. Differentiation. 2007;75:123‐132. [DOI] [PubMed] [Google Scholar]
- 29. Ivanov SV, Ivanova AV, Goparaju CM, Chen Y, Beck A, Pass HI. Tumorigenic properties of alternative osteopontin isoforms in mesothelioma. Biochem Biophys Res Commun. 2009;382:514‐518. [DOI] [PubMed] [Google Scholar]
- 30. Sebban S, Davidson B, Reich R. Lysyl oxidase‐like 4 is alternatively spliced in an anatomic site‐specific manner in tumors involving the serosal cavities. Virchows Arch. 2009;454:71‐79. [DOI] [PubMed] [Google Scholar]
- 31. Langerak AW, Dirks RP, Versnel MA. Splicing of the platelet‐derived‐growth‐factor a‐chain mrna in human malignant mesothelioma cell lines and regulation of its expression. Eur J Biochem. 1992;208:589‐596. [DOI] [PubMed] [Google Scholar]
- 32. Morrison A, Chekaluk Y, Bacares R, Ladanyi M, Zhang L. Bap1 missense mutation c. 2054 a> t (p. E685v) completely disrupts normal splicing through creation of a novel 5'splice site in a human mesothelioma cell line. PloS One. 2015;10:e0119224. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33. Dong L, Jensen RV, De Rienzo A, et al. Differentially expressed alternatively spliced genes in malignant pleural mesothelioma identified using massively parallel transcriptome sequencing. BMC Med Genet. 2009;10:1‐9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34. Bueno R, Stawiski EW, Goldstein LD, et al. Comprehensive genomic analysis of malignant pleural mesothelioma identifies recurrent mutations, gene fusions and splicing alterations. Nat Genet. 2016;48:407‐416. [DOI] [PubMed] [Google Scholar]
- 35. Zalfa F, Achsel T, Bagni C. Mrnps, polysomes or granules: Fmrp in neuronal protein synthesis. Curr Opin Neurobiol. 2006;16:265‐269. [DOI] [PubMed] [Google Scholar]
- 36. Garber K, Smith KT, Reines D, Warren ST. Transcription, translation and fragile x syndrome. Curr Opin Genet Dev. 2006;16:270‐275. [DOI] [PubMed] [Google Scholar]
- 37. Didiot M‐C, Tian Z, Schaeffer C, Subramanian M, Mandel J‐L, Moine H. The g‐quartet containing fmrp binding site in fmr1 mrna is a potent exonic splicing enhancer. Nucleic Acids Res. 2008;36:4902‐4912. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38. Pal S, Medatwal N, Kumar S, et al. A localized chimeric hydrogel therapy combats tumor progression through alteration of sphingolipid metabolism. ACS Cent Sci. 2019;5:1648‐1662. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39. Murray MV, Kobayashi R, Krainer AR. The type 2c ser/thr phosphatase pp2cγ is a pre‐mrna splicing factor. Genes Dev. 1999;13:87‐97. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40. Allemand E, Hastings ML, Murray MV, Myers MP, Krainer AR. Alternative splicing regulation by interaction of phosphatase pp2c γ with nucleic acid–binding protein yb‐1. Nat Struct Mol Biol. 2007;14:630‐638. [DOI] [PubMed] [Google Scholar]
- 41. Gudipaty SA, McNamara RP, Morton EL, D'Orso I. Ppm1g binds 7sk rna and hexim1 to block p‐tefb assembly into the 7sk snrnp and sustain transcription elongation. Mol Cell Biol. 2015;35:3810‐3828. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Figure S1
Figure S2
Table S1
Data Availability Statement
The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.