Construction of an immune‐related LncRNA signature with prognostic significance for bladder cancer

Abstract Bladder cancer (BLCA) is one of the most common urological cancer with increasing cases and deaths every year. In the present study, we aim to construct an immune‐related prognostic lncRNA signature (IRPLS) in bladder cancer (BLCA) patients and explore its immunogenomic implications in pan‐cancers. First, the immune‐related differentially expressed lncRNAs (IRDELs) were identified by ‘limma’ R package and the score of IRPLS in every patient were evaluated by Cox regression. The dysregulation of IRDELs expression between cancer and para‐cancer normal tissues was validated through RT‐qPCR. Then, we further explore the biological functions of a novel lncRNA from IRPLS, RP11‐89 in BLCA using CCK8 assay, Transwell assay and Apoptosis analysis, which indicated that RP11‐89 was able to promote cell proliferation and invasive capacity while inhibits cell apoptosis in BLCA. In addition, we performed bioinformatic methods and RIP to investigate and validate the RP11‐89/miR‐27a‐3p/PPARγ pathway in order to explore the mechanism. Next, CIBERSORT and ESTIMATE algorithm were used to evaluate abundance of tumour‐infiltrating immune cells and scores of tumour environment elements in BLCA with different level of IRPLS risk scores. Finally, multiple bioinformatic methods were performed to show us the immune landscape of these four lncRNAs for pan‐cancers. In conclusion, this study first constructed an immune‐related prognostic lncRNA signature, which consists of RP11‐89, PSORS1C3, LINC02672 and MIR100HG and might shed lights on novel targets for individualized immunotherapy for BLCA patients.

the gold standard, and the most common treatment offered for the management of primary MIBC. 4 Recently, studies have increasingly focused on novel therapeutic and diagnostic methods for BLCA. [5][6][7] However, effective potential targets and prediction models remain largely undetermined.
Long non-coding RNAs (lncRNAs) are non-coding RNAs ranging in length from 200 nucleotides to 100 kilobases that have diverse regulatory mechanisms in gene expression, such as chromatin modification and transcriptional and post-transcriptional processing. 8,9 Recently, lncRNAs have been found to be associated with the tumour progression and immune microenvironment. [10][11][12] For example, seminal work from Wang et al showed lnc-DC is required for normal dendritic cell differentiation and function. 13 Furthermore, lnc-THRIL regulates tumour necrosis factorα (TNFα) expression through interactions with hnRNPL during innate activation of THP1 macrophages. 14 Additionally, a great number of lncRNAs with prognostic significance have been identified, 10,15,16 which emphasizes the need to identify more accurate biomarkers and establish an effective prediction model of BLCA.
Increasing evidence indicates that complex interactions are involved between cancer cells and immunity in regards to immune checkpoint genes, 17 tumour-infiltrating lymphocytes (TILs) 18 and tumour-predicted neoantigen. 19,20 In recent years, immunotherapy has dramatically improved the treatment options for various cancers and has significantly prolonged the overall survival of treated patients.
Pharmacological manipulation of the physiological immune checkpoints is one of the most promising immunotherapeutic approaches.
Researchers are attempting to approve antibodies targeting checkpoint molecules such as cytotoxic T-lymphocyte antigen 4 (CTLA4), 21 programmed cell death 1 (PD1) 22 and programmed cell death ligand 1 (PD-L1), to block major antitumour activity. However, a limited number of patients with advanced/metastatic cancer respond to ICIs, 23 thus exposing the remaining patients to potentially ineffective, toxic and expensive treatments. 24,25 The identification of predictive factors determining the response efficiency to immunotherapy and immunogenomic landscape analysis for cancers are becoming increasingly critical. In the evolving era of immunotherapy, this work is devoted to exploring novel immune-related lncRNAs, which have potential prognostic value for BLCA patients and might facilitate evidence-based guidance for personalized immunotherapy. We hypothesized IRPLS as novel potential immune checkpoint targets, which may indicate therapeutic response and provide clinical strategies for the individualization of immunotherapy.

| Data collection and processing
We downloaded gene expression data sets of bladder urothelial carcinoma (BLCA cohort) from The Cancer Genome Atlas (TCGA) Research Network. We also obtained BLCA lncRNA expression data (GSE89006) from the Gene Expression Omnibus (GEO). 26 We collected 49 bladder cancer samples and their adjacent normal   tissue samples from patients with BLCA after radical resection in   Fudan University Shanghai Cancer Center (FUSCC) and obtained   human bladder cancer cell lines RT-4, UM-UC-3, 5637, SCaBER,   SW780, T24, MGH-U3 and human immortalized normal  . The SV-HUC-1 cell was maintained in F12K medium (Gibco, China). All medium was supplemented with 1% penicillin G sodium/ streptomycin sulphate and 10% foetal bovine serum (FBS) (Gibco, Australia). All cells were grown in a humidified atmosphere consisting of 5% CO2 and 95% air at 37 ℃.

| Quantitative reverse-transcription polymerase chain reaction (qRT-PCR) and analysis
Total RNA from bladder cancer cell lines and SV-HUC-1, BLCA tumour tissue and their adjacent normal tissue was extracted using TRIzol (Invitrogen), reverse-transcribed to cDNA. Human Student's t test (normally distributed) or the Mann-Whitney U test (non-normally distributed) was utilized to analyse differences.

| Identification of immune-related differentially expressed lncRNAs
The 'limma' R package 27 was used to generate the p-value and fold change (FC) for each lncRNA between BLCA tissue and normal tissue. And we defined those with p-value ≤ 0.05 and |log2 FC| ≥ 1 as differentially expressed lncRNAs. The single-sample gene set enrichment analysis (ssGSEA) algorithm 28 was carried out to evaluate the immune relevance of patients in TCGA-BLCA cohort by quantifying the enrichment levels of the 29 immune-associated gene sets 29 (Table S1). And the patients' score was provided in Table S2. The 'limma' R package was used again to obtain immune-related lncRNAs (p-value ≤ 0.05 and |log2 FC| ≥ 1) between high and low score groups.
The overlapping lncRNAs among different groups were determined via Venn diagrams.

| Construction of immune-related lncRNA signature
We performed univariable Cox proportional hazards regression model analyses for overall survival (OS) and recurrence-free survival (RFS) to obtain prognosis-related lncRNAs. Multivariable Cox proportional hazards regression model analyses were used 30 to build immune-related lncRNA signature (IRPLS) and determine the best cut-off value to distinguish BLCA patients into high-risk group and low-risk group. Kaplan-Meier (KM) method and log-rank tests were used to calculate differences between groups. The receiver operating characteristic curve (ROC) was constructed to validate the predictive ability of the IRPLS using 'survival ROC' R package.

| Cell counting kit (CCK)-8 assay
The cell proliferation ability was tested by Cell Counting Kit-8 (Dojindo, CK04). First, we seeded cells into 96-well plates (5000 cells/well) with complete growth medium. After incubation for 24, 48, 72 and 96 h, respectively, 10 mL CCK-8 was added into each well and then, the cells were cultured for an additional 2 h. Finally, the absorbance of the samples was measured at 450 nm using Microplate Spectrophotometer (BioTek, VT, USA).

| Transwell assay
Cell invasive capacity was determined using Transwell chambers (BD Biosciences). A total of 20 000 cells were plated in the top of a polycarbonate Transwell filter with 200 mL medium without foetal bovine serum. We fill the lower compartment with 500 mL culture medium with foetal bovine serum. After incubation for 24 hours (for 5637 cells) and 36 hours (for T24 cells), we stained the migrated cells using crystal violet and counted using Imagine J.

| Cell apoptosis assay
Cell apoptosis was evaluated by an Annexin V-APC/7-AAD kit. Flow cytometry was used to examine cells harvested after incubation with Annexin V-FITC/PI double staining.

| RNA immunoprecipitation (RIP) assay
We validated the relationship between lncRNA RP11-89 and miR-27A-3p via a Magna RIP RNA-Binding Protein Immunoprecipitation Kit (Millipore, USA). Anti-AGO2 and control IgG (Millipore, USA) were utilized for the RIP assay, and we evaluate the coprecipitated RNAs via cDNA synthesis and qRT-PCR.

| Immune-related analysis in bladder cancer
CIBERSORT algorithm in R software was utilized to estimate the abundance of tumour-infiltrating immune cells. 31 And we calculated immune scores and stromal scores for each sample applying the 'Estimation of STromal and Immune cells in MAlignant Tumours using Expression data' (ESTIMATE) algorithm. 18

| Gene set enrichment analyses
We performed the R category (version 2. were conducted to finish statistical analysis and graphical plotting.

| Immunogenomic landscape analysis in pancancers
The ssGSEA analysis was utilized to evaluate expression of checkpoint genes in pan-cancer of TCGA cohort according to 29 immune-associated gene sets 29 (Table S1). Abundance of 28 kinds of tumour-infiltrating lymphocytes (TILs) and predicted SNV (singlenucleotide variants)-derived Neoantigen of pan-cancer in TCGA cohorts were obtained from 'https://tcia.at/home' and 'https://gdc.
cancer. gov/about-data/publications/panimmune'. Pearson correlation tests were performed to evaluate association of expression between lncRNAs and checkpoint genes, tumour-infiltrating lymphocytes (TILs) and tumour-predicted SNV neoantigen. were identified by mapping the ENSEMBL ID with references to the Gencode database (https://www.genco degen es.org/human/). As Figure 2E shows, 1,347 differentially expressed lncRNAs (DELs) between BLCA tissue and normal tissue were identified, and 48 candidate lncRNAs overlapped between the two major cohorts.

| Identification of immune-related differentially expressed lncRNAs
Furthermore, 477 lncRNAs were recognized as immune-related lncRNAs by ssGSEA ( Figure 2F). Finally, 13 candidates were identified for further research.

| Construction of IRPLS and validation of lncRNAs dysregulation in BLCA
As shown in Table 1, we conducted multivariate Cox regression to identify four significant lncRNAs from 13 hub lncRNAs. The IRPLS was constructed by these four lncRNAs containing RP11-89, PSORS1C3, LINC02672 and MIR100HG. All patients in the TCGA cohort were stratified into low-risk or high-risk groups using a median IRPLS risk scores, which was calculated as follows: riskScore = expression level

| Confirmation of IRPLS as an independent predictor and establishment of a novel nomogram for predicting OS in BLCA
The entire TCGA-BLCA cohort was randomly divided into training and testing cohorts at a cut-off value of 3:2. In the training cohort

F I G U R E 2
Identification of immune-related differentially expressed lncRNAs (ICDELs). A and B, The genomic heat map and volcano plot was presented to show the 143 differentially expressed lncRNAs identified from the TCGA cohort. C and D, The genomic heat map and volcano plot were presented to show 1,252 differentially expressed lncRNAs identified from the GSE89006 cohort. E and F, A total of 48 candidate lncRNAs were overlapped between the two major cohorts, and 13 candidates were identified as the immune-related differentially expressed lncRNAs

| RP11-89 promotes BLCA progression via increasing cell proliferation and invasive capacity while suppressing cell apoptosis
We found one of IRPLS lncRNAs, RP11-89 was rarely investigated in BLCA or other tumours and evaluated its biological function in BLCA.

| RP11-89 targeted miR-27a-3p and up-regulate PPARγ expression in BLCA
Through the miRcode database (Table 3) The same trends were also observed in T24 cell transfected with overexpression control plasmid in contrast to T24 cell treated with overexpression plasmid (P <.05). In order to validate the relationship between miR-27a-3p and RP11-89, we performed the AGO2 immunoprecipitation assay, which showed that the AGO2 antibody was able to pull down both endogenous miR-27a-3p and RP11-89 ( Figure 6A).
We performed bioinformatic method to analysis targeted genes of miR-27a-3p in starBase database (http://starb ase.sysu.edu.cn/), which showed that miR-27a-3p was able to bind with PPARγ in 3′-UTR region and caused the negative regulation of PPARγ ( Figure 6D). It could be concluded from the findings mentioned above that RP11-89 might 'sponge' miR-27a-3p and up-regulate PPARγ expression in BLCA.

| Immune relationship between IRPLS and BLCA
As shown in Figure 7A, the distribution of immune cells was significantly different between the high-risk and low-risk groups. The fraction of T cells in low-risk group was remarkably higher than high-risk group (P <.001). According to the ESTIMATE algorithm ( Figure 7B-C), the high-risk group showed higher stroma scores and higher immune

| Immune landscape analysis of IRPLS in pancancers
RP11-89 was significantly correlated with the expression of CD44, CD70 and LAGLS9 in pan-cancers ( Figure 8A) and was significantly associated with the abundance of activated CD4 T cells, activated CD8 T cells and effector memory CD4 T cells in pan-cancers ( Figure 8B). RP11-89 was also significantly correlated with predicted SNV-derived neoantigens ( Figure 8C) of BLCA (P <.05) and rectal adenocarcinoma (P <.01). MIR100HG was significantly correlated with the expression of CD200, CD274 and CD276 in pan-cancers ( Figure 8D) and was significantly associated with the abundance of central memory CD8 T cells, effector memory CD4 T cells and effector memory CD8 T cells in pan-cancers ( Figure 8E). Furthermore, MIR100HG was significantly correlated with predicted SNV-derived neoantigens ( Figure 8F) of head and neck squamous cell carcinoma (P <.05), thyroid carcinoma (P <.05), stomach adenocarcinoma and breast cancer (P <.05). LINC02672 was significantly correlated

| D ISCUSS I ON
BLCA is one of the most common urological cancers and is a major An increasing number of studies have revealed that PPARγ, one of peroxisome proliferator-activated receptors, plays essential role in cancer metabolism. 35 Subsequently, more evidence has indicated PPARγ activation as a potential tumorigenic trigger in BLCA. 36,37 Otherwise, for its crucial role in cellular energy homeostasis, immune cells also require PPARγ activation to meet energy demands and regulate lipid metabolism and cell fate, which is involved in tumour immune microenvironment. 38,39 The heterogeneity of immune status in BLCA is particularly high. We evaluated the type and number of infiltrating immune cells between the high-risk and low-risk groups. Accumulating data suggest that neoantigens are relevant targets for personalized anti-cancer therapies. 40,41 Given the analysis of our findings, IRPLS, if validated, not only has roles in predicting the survival outcome and immune status of BLCA patients but is also tightly correlated with pan-cancer immunity. Interestingly, the high relevance of MIR100HG with immunotherapy, which is quite 'hot' in the presented heat map, has been confirmed in multiple cancers including osteosarcoma, laryngeal squamous cell carcinoma, triple-negative breast cancer and gastric cancer. [42][43][44][45] Furthermore, for the first time we reveal the commonality and heterogeneity of tumour immunity through immunogenomic landscape analysis of the lncRNA signature, which has the potential to contribute to the implementation of immunotherapy approaches for patients, especially those diagnosed with multiple primary cancers.
In conclusion, this study first constructed an immune-related prognostic lncRNA signature (IRPLS), which consists of RP11-89, PSORS1C3, LINC02672 and MIR100HG. As an oncogenic lncRNA, RP11-89 is able to promote cell proliferation and invasive capacity while inhibit cell apoptosis in BLCA. IRPLS significantly predicts poor clinical outcomes for bladder cancer patients and immune microenvironment disorders in pan-cancers, which might shed lights on novel targets for individualized immunotherapy.

ACK N OWLED G EM ENTS
We thank the TCGA databases and GEO databases for providing BLCA gene expression profiles.

CO N FLI C T O F I NTE R E S T S
The authors declare no competing interests.

E TH I C A L A PPROVA L
The Ethics approval and consent to participate of the current study were approved and consented by the ethics committee of Fudan University Shanghai Cancer Center.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data sets during and/or analysed during the current study available from the corresponding author on reasonable request.