Systematic pan‐cancer analysis identifies RBM39 as an immunological and prognostic biomarker

Abstract RNA‐binding Motif Protein39 (RBM39) is identified as a splicing factor and transcription coactivator. Despite mounting evidence that RBM39 plays a critical role in the development of specific malignancies, no systematic pan‐cancer investigation of RBM39 has been conducted. As a result, we set out to investigate RBM39’s prognostic significance and putative immunological activities in 33 different cancers. Based on TCGA and CCLE, GTEx, cBioportal and HPA, we used a series of bioinformatics approaches to explore the potential oncogenic role of RBM39, including analysis of the expression of the pan‐cancer species RBM39, the prognostic relationship between RBM39 expression and overall survival (OS), disease‐specific survival (DSS) and progression‐free interval (PFI), the relationship between RBM39 expression and clinical phenotype, analysis of the relationship between RBM39 expression and tumour mutational burden (TMB), microsatellite instability (MSI), DNA methylation and immune cell infiltration. Our results showed that RBM39 is overexpressed in most cancers. RBM39 was positively or negatively correlated with the prognosis of different tumours. RBM39 expression was associated with TMB and MSI in 9 and 12 cancer types. In addition, RBM39 expression was associated with DNA methylation in almost all tumours. There are eight tumours were screened for further study, including BRCA, COAD, HNSC, LIHC, LUSC, SKCM, STAD, UCEC. In the screed tumours, RBM39 was found to be negatively correlated with the infiltration of most immune cells. In addition, the correlation with RBM39 expression varied by immune cell subtype. Based on RBM39’s role in tumorigenesis and tumour immunity, we suggest it can serve as a surrogate prognostic marker.


| INTRODUC TI ON
Cancer, as a highly complex disease, is the leading cause of death in every country in the world and an important obstacle to prolonging life expectancy. By 2020, it is estimated that there will be nearly 10 million cancer deaths and 19.3 million new cancer cases worldwide. 1 Recently, tumour immunotherapy has become a prominent cancer treatment approach. As public databases, like TCGA and CCLE, con-  3,4 This protein, also known as RBM39, HCC1, Caperα, FSAP59 and RNPC2, was found to be distributed in a very speckled network within the nucleus and to colocalize with SC35 and uridine-rich small nuclear RNA. 5 RBM39 is involved in transcriptional regulation and RNA splicing. Selective splicing has been shown to be a widespread phenomenon that promotes the multiplication of gene products in a tissue-specific manner. 6,7 Recent findings reveal diverse ways in which splicing is pathologically changed to facilitate the initiation and/or retention of cancer. 8 It contains a trans-activating structural domain, a serine-arginine-rich pair region and an RNA recognition motif, which consists of two highly conserved peptide motifs with RNA and single-stranded cDNA binding activity. 9 It has been shown that RBM39, a transcriptional coactivator of activator protein-1 (AP-1/Jun), oestrogen receptor and NF-kB, [10][11][12] is involved in energy and redox homeostasis and cancer cell proliferation, in addition to its role in vascular endothelial growth factor splicing. 13 Some studies have demonstrated that RBM39 is a proto-oncogene that plays an important role in the development of several malignancies. In many preclinical models, loss of RBM39 leads to aberrant splicing events and differential gene expression, which inhibit cell cycle progression and lead to tumour regression.
So far, most studies on the role of RBM39 in tumours have been limited to one specific type of cancer. There are no pan-cancer studies on the relationship between RBM39 and various cancers. Therefore, we analysed the expression levels of RBM39 in different types of malignancies and its relationship with prognosis using multiple databases, including TCGA, Cancer Cell Line Encyclopedia (CCLE), Genotype-Tissue Expression (GTEx), cBioPortal and Human Protein Atlas (HPA). In addition, we explored the relationship of RBM39 expression with DNA methylation and immune infiltration.
We introduced co-expression analysis and enrichment analysis of genes to investigate the biological functions of RBM39 in tumours. Our results suggest that RBM39 may play an important role in tumours and serve as a prognostic factor. Our study provides a theoretical basis to gain insight into the role of RBM39 in tumour immunotherapy.

| Analysis of the connection between RBM39 and prognosis and clinical phenotype
TCGA data on each sample's survival and clinical features were obtained. Three indicators, OS, DSS and PFI, were considered when analysing the correlation between RBM39 and survival. Kaplan-Meier (p < 0.05) survival analysis was used. Age, tumour stage and clinical phenotypes were analysed to see whether they were associated with RBM39 expression (p < 0.05).

| Correlation analysis of RBM39 with the tumour mutational burden (TMB) and microsatellite instability (MSI)
For the purpose of calculating TMB for each tumour specimen independently, Spearman rank correlation coefficient was utilized. TMB is a biomarker that reflects mutations of tumour cells. A microsatellite instability (MSI) occurs when an insertion or deletion of repetitive units results in a change in microsatellite length as compared with normal tissues. Spearman rank correlation coefficient was used to analyse the correlation between RBM39 expression and MSI.

| Correlation analysis of RBM39 with DNA Mismatch Repair Genes and Methyltransferases
Mismatch repair is an intracellular process that allows mismatches to be repaired. DNA replication errors are not repaired when this mechanism fails, resulting in an increase in somatic mutations. The relationship between five of the MMRS genes (MLH1, MSH2, MSH6, PMS2, EPCAM) and RBM39 expression was studied using the TCGA data. The chemical modification of DNA known as DNA methylation can affect epigenetics and can influence gene expression without changing the DNA sequence. We used this study to examine whether RBM39 expression correlated with that of the four methyltransferases.

| Correlation analysis of RBM39 with the immune microenvironment
Tumour-infiltrating lymphocytes are independently correlated with the status of primary lymph nodes and survival, and tumour immune cell scores are determined using immune scores and interstitial scores. We used software estimation to analyse the correlation between gene expression and immune cell score. We used CIBERSORT, a metagene tool that predicts immune cell phenotypes, to assess the correlations between RBM39 levels and immune cell infiltration in cancer in 33 tumours. (p < 0.05 was considered significant). Furthermore, we looked at the co-expression of RBM39 and immune-related genes, such as those encoding major histocompatibility complex (MHC), immune activation, immunosuppression, chemokine and chemokine receptor proteins. Gene Set Enrichment Analysis of RBM39 across cancers. The biological activities of RBM39 in malignancies were investigated using Gene Set Enrichment Analysis (GSEA).

| Expression of RBM39 in pan-cancer
We analysed the expression levels of the physiological RBM39 gene across tissues through the GTEx database ( Figure 1A). RBM39 was expressed at the highest and lowest levels in the oviductal and heart, respectively. Figure 1B depicts the relative RBM39 expression levels across various cell lines based on CCLE data. Then, we used the TCGA database to obtain the differential expression patterns of RBM39 in cancer and paraneoplastic tissue samples ( Figure 1C).
It can be noted that most tumour tissues have greater RBM39 expression than surrounding normal tissues. We investigated changes in RBM39 expression in 27 tumour tissues by merging data from GTEx normal tissues and TCGA tumour tissues ( Figure 1D). Furthermore, to assess RBM39 protein expression, we matched IHC results from the HPA database to RBM39 gene expression data from the TCGA database. The results showed that normal breast, lung, liver and colon tissues showed low levels of RBM39 staining, whereas the corresponding breast, lung squamous, liver and colon malignancies had moderate to high expression, which was consistent with the expression pattern of RBM39 in the TCGA database for the five tumours (Figure S1A-E).

| Prognostic value of RBM39 across cancers
To assess the link between RBM39 expression levels and prognosis, we ran a survival association study for each cancer, including OS, DSS and PFI. As shown in Figure
We also looked at the association between RBM39 expression and tumour stage. RBM39 expression was shown to be substantially linked with tumour stage in four malignancies studied ( Figure S4E-H), including LIHC, KICH, THCA and KIRC. Significant differences in RBM39 expression occurred between stage I and II tumours in KICH and THCA were observed. Interestingly, in contrast to LIHC and KIRC, whose RBM39 expression increased significantly with increasing tumour stage between stage I and III, most other types of cancer did not show a statistically significant difference in terms of tumour stage.

| Expression of RBM39 in pan-cancerous tissues correlates with TMB and MSI
Tumour mutational burden is typically defined as the total amount of nonsynonymous mutations occurring on an average of 1 M bases in the tumour cell genome's coding region, and the mutation types primarily comprise single nucleotide variants (SNVs) and minor insertions/deletions (INDELs). TMB is a quantifiable biomarker that indicates the number of mutations discovered in tumour cells. As shown in Figure 3A, the association between TMB and RBM39 expression was statistically examined using Spearman rank correlation coefficients for each tumour type independently. Notably, RBM39 expression was positively correlated with TMB in ESCA, LAML, LGG, SKCM and STAD, and negatively correlated with UCS, UVM, UCEC, BRCA, THCA and COAD. MSI is a term that describes any change in microsatellite length caused by the insertion or deletion of repeat units in tumours compared with normal tissue, as well as the introduction of novel microsatellite alleles. As shown in Figure 3B, the correlation between RBM39 and MSI was analysed using the Spearman rank correlation coefficient. The results showed that  Figure S5B, implying that RBM39 may play a role in carcinogenesis and progression by modulating the epigenetic status of human pan-cancer.

| Relationship between RBM39 and tumour microenvironment
We calculated the immunological score and interstitial score of in-

| Expression of RBM39 in pan-cancer is correlated with immune neoantigens and immune checkpoint genes
We next examined the correlation between RBM39 with more than

| Relationship between the RBM39 expression and tumour immune cell infiltration
Next, we investigated the association of RBM39 expression with the level of infiltration of 24 immune-related cells. Our data showed that the level of immune cell infiltration was substantially linked with RBM39 expression in the majority of malignancies (Table S1). We screened eight tumours for further analysis, including BRCA, COAD, HNSC, LIHC, LUSC, SKCM, STAD and UCEC, for which the correlation between RBM39 expression and the level of immune cell infiltration was high ( Table 1)

| GSEA enrichment analysis
We used GSEA enrichment analysis to investigate the biological im-

| DISCUSS ION
An increasing amount of evidence reveals that RBM39 is substantially linked to cancer growth in a range of malignancies. As an RNAbinding protein, RBM39 is extensively involved in selective splicing of RNA. 14 It has been demonstrated that RBM39 is essential for RNA splicing in neuroblastoma cells. 15 Of the 27 tumours we studied in the GTEx and TCGA databases, the RBM39 gene was highly expressed in six cancers and lowly expressed in 19 tumours. Until now, there have been some studies on the relationship between RBM39 and tumours. For example, Mercier I et al. 16 showed that RBM39 was not detected or expressed at low levels in normal breast tissue with a cytoplasmic localization. In contrast, it was expressed at high levels in breast ductal carcinoma in situ specimens and was mainly distributed in the nucleus. RBM39 is a coactivator of ER, ER, and Activator Protein-1 (AP-1) component c-Jun, which binds transcription and mRNA precursor processing together and effectively boosts their transcriptional activity, hence encouraging breast cancer growth. 17 Another study reported that RBM39 acts as a major transcriptional regulator and interacts with the MLL1 complex to promote breast cancer cell proliferation. 18 In patients with hepatocellular carcinoma, RBM39 was first identified as an autoantigen whose overexpression reduced tumour angiogenesis and growth and inhibited v-Relmediated lymphocyte transformation. [19][20][21] Previous studies have shown that RBM39 is involved in biological processes in colorectal adenocarcinoma development, such as cell survival and anchoring non-dependent growth. 22 Using TCGA data, we performed a Kaplan-Meier survival analysis and discovered that greater RBM39 expression was associated F I G U R E 4 Correlation analysis between RBM39 expression in Pan-cancer and immune score (A), stromal score (B) and estimate immune score (C) with a poor prognosis for OS, DSS, and PFI in LIHC. RBM39 was discovered as an autoantigen in a cirrhotic patient who later acquired hepatocellular cancer. 23 These findings suggest that RBM39 may be an oncogene during liver carcinogenesis. We also found that RBM39 expression was associated with age in certain types of cancer. Our study showed that RBM39 expression was lower in older patients in HNSC, KIRP and SARC, while in UCEC patients, reduced expression of RBM39 was associated with younger age. These results may have important implications for guiding the selection of immunotherapy regimens for patients of different ages. Our study also found that the expression of RBM39 in most cancers correlated with tumour stage, and the difference in expression was particularly significant in stage I and II tumours. For example, RBM39 expression was significantly lower in stages I and II of KICH. These findings imply that RBM39 might be employed as a biomarker to predict the prognosis of different malignancies.
Tumour mutational burden is an intriguing predictive biomarker with the potential to usher immuno-oncology into the era of precision medicine. 24 Previous studies have shown that in melanoma, whole-exome sequencing (WES) of TMB is a predictor of increased survival in patients treated with ipilimumab or tremelimumab. 25 Another study used TMB as a biomarker in non-small cell lung cancer patients receiving pembrolizumab and WES and found an overall response rate (ORR), progression-free survival (PFS) and long-term therapeutic benefit in patients with a high somatic nonsynonymous mutation burden. 26 Thus, tumours with higher TMB are assumed to have more neoantigens that can be recognized by the immune system in response to immune checkpoint inhibition. MSI is also a useful biomarker for immune checkpoint inhibitors (ICI). MSI in tumour DNA is defined as the presence of alternate-sized repetitive DNA sequences that are not seen in the corresponding germline DNA.
Depending on the type and the number of microsatellites analysed, widely variable results on the frequency of MSI in different tumour types have been published. 27,28 It has been shown that MSI is the first DNA marker that can be used to identify hereditary colorectal cancer. 29 Our study showed that RBM39 expression correlated with TMB in 9 cancer types and with MSI in 12 cancer types. This might imply that RBM39 expression levels influence tumour TMB and MSI, and consequently the patient's response to immune checkpoint inhibition medication. This would give a new basis for immunotherapy prognosis. In addition, we also investigated the correlation of RBM39 expression with five MMRs and four DNA methyltransferases.
Our results suggest that RBM39 plays an important role in cancer immunity. The tumour microenvironment (TME) not only influences tumorigenesis, development and metastasis, but also has a profound impact on therapeutic results. 30 According to the assessment scores, RBM39 is considered as a novel human tumour-associated antigen, and its specific immunity has been reported in a variety of tumours. 32 We analysed the relationship between RBM39 expression and the number of neoantigens in 19 tumours. We found that RBM39 was positively correlated with the number of neoantigens in LGG, and negatively correlated with the number of neoantigens in BRCA. A major advance in cancer treatment is the development of immune checkpoint inhibitors (ICIs) 33 . Our results also showed that RBM39 expression was positively correlated with the expression levels of immune checkpoint genes in four types of cancer, suggesting that RBM39 may regulate tumour immune patterns through the number of tumours neoantigens and by regulating the expression levels of immune checkpoint genes. It has been demonstrated that when the interaction between checkpoint ligands and their cognate receptors on effector cells is blocked, a robust and long-lasting antitumor response can be observed. 34,35 Tumour-infiltrating immune cells have an important impact on tumorigenesis and progression and can antagonize or promote tumorigenesis and progression. 36 The presence of certain immune subgroups usually exhibits a favourable prognostic effect in a malignancy. 37 Recent studies have found that T lymphocyte subsets predict response to existing and emerging immunotherapies, highlighting the importance of studying tumour-associated immune cells as potential predictive biomarkers. 38,39 Previous studies have reported that immunomodulatory drugs, such as sulfonamides, can recruit the splicing factor RBM39 to the E3 ligase substrate receptor DCAF15, leading to ubiquitination and degradation of RBM39, which resulted in altered RNA splicing and death in some cancer cell lines. [40][41][42][43] Moreover, RBM39 expression plays a role in intracellular immunosuppression. It has been shown that RBM39 orchestrates a splicing program that is essential for the survival of AML cells according to CRISPR/Cas9 screens. 44, 45 Thomas R et al. 46 have demonstrated that RBM39 is a critical RBP required for AML survival and F I G U R E 6 Co-expression of RBM39 and immune-related genes that RBM39 deficiency causes leukaemia progression to be slowed and overall survival to be improved. Jia Tong et al. 47 have found that RBM39 was overexpressed in myeloma cells and associated with poor prognosis of patients, and they demonstrated that the α-DARS-AS1-RBM39 axis of HIF-1 is a potential target for the treatment of multiple myeloma.
Our study further elucidates the broader tumorigenic applicability of RBM39 and confirms that RBM39 expression is closely associated with the biological processes of immune cells and immune-related molecules in most tumours. In addition, our study revealed that RBM39 is co-expressed with genes encoding MHC, immune activation, immune suppression, chemokine and chemokine receptor proteins. These results suggest that RBM39 expression is closely associated with immune infiltration of tumour cells, affecting patient prognosis and providing a new target for immunosuppressant development. Furthermore, our enrichment analysis suggests that RBM39 may influence the aetiology or pathogenesis of cancer by acting in immune cell infiltration, immune chemokine transduction and immune anti-infection pathways.
In conclusion, our first pan-cancer RBM39 analysis revealed that it is differentially expressed in tumour and normal tissues and that there is a link between RBM39 expression and clinical prognosis. Our findings imply that RBM39 can function as an independent prognostic biomarker in a wide range of malignancies. RBM39 expression leads to different prognostic outcomes for different tumours, which requires further investigation of the specific role of RBM39 in each tumour. In addition, RBM39 expression is associated with the tumour microenvironment and immune cell infiltration in different cancer types. Its effect on tumour immunity also varies by tumour type. These findings may help to elucidate the role of RBM39 in tumour development and provide a reference for achieving more precise and personalized immunotherapy in the future.

ACK N OWLED G EM ENT
This work was supported by grants from the National Natural Science Foundation of China (No. 81972165, No. 81972845).

CO N FLI C T O F I NTE R E S T
The authors report no conflicts of interest in this work.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that supports the findings of this study are available in the supplementary material of this article.