A novel risk score model based on five angiogenesis-related long non-coding RNAs for bladder urothelial carcinoma

Tumour tissues and adjacent normal tissues of 138 BUC patients who underwent tumour excision or tissue biopsy in the First Affiliated Hospital of Chongqing Medical University between March 2019 and February 2021 were collected. (Table 1) Patients with severe underlying diseases or other primary cancers were excluded. The collected clinical tissues were utilized to analyze the expression of lncRNAs and percentage of endothelial cells (ECs). All patients included in this study provided written informed consent, and this study was approved by the Medical Ethics Committee of the First Affiliated Hospital of Chongqing Medical University (IRB:2021-085). All clinical data were reviewed according to medical records.

SV-HUC-1, BUC cell lines (T24, UM-UC-3, 5637, J82 and TCC-SUP) and HUVEC were purchased from the American Type Culture Collection (Manassas, Virginia, USA). Cells were cultured in DMEM (SV-HUC-1, UM-UC-3, T24 and HUVEC), McCoy’s 5A (J82) and RPMI-1640 (5637 and TCC-SUP) basal medium (Gibco, Gaithersburg, MD, USA), which were supplemented with 10% fetal bovine serum (FBS, MilliporeSigma, Burlington, MA, USA), 100 U/ml penicillin and 0.1 mg/ml streptomycin (Beyotime, Beijing, China). Cells were incubated at 37 °C in 5% CO₂ incubator. The medium was changed every 1–3 days.

Bladder specimens were obtained fresh from the operating field where grossly apparent tumour tissue or adjacent tissue not grossly affected by tumour (Par-cancer tissue). These tissues were transported at room temperature immersed in the RPMI-1640 medium (Gibco, Gaithersburg, MD, USA) with 10% FBS (MilliporeSigma, Burlington, MA, USA). Once received, tumour tissues were divided into two parts, one of which was cut into approximately 1 mm³ pieces and enzymatically digested to single cell suspensions using MACS tumor dissociation kit (Miltenyi Biotec) for 1 h on a rotor at 37 °C for further flow cytometry/FACS analysis, another part of tumour tissues and all par-cancer tissues were frozen in liquid nitrogen immediately and then stored at − 80 °C until lncRNA extraction.

The single cell suspensions were filtered with screen cloth and cell surface staining was performed in FACS buffer containing CD31 antibody (Cat# 303102, BioLegend, USA) on ice for 1 h. Following washing twice with FACS buffer, the percentages of EC subtype (CD31 positive) in these single cell suspensions were detected using a FACS flow cytometry system (Cytoflex, Beckman Coulter, USA).

The siRNAs of AC005625.1 and AC008760.1 were used to silence the expression of AC005625.1 and AC008760.1. The sequences used were: si-AC005625.1 (sense: 5′-GCUUCACAGCCACCAUCUATT-3′, antisense: 5′-UAGAUGGUGGCUGUGAAGCT T-3′); si-AC008760.1 (sense: 5′-GACAGGUAGUCACGACUAUTT-3′, antisense: 5′-AUAGUC GUGACUACCUGUCTT-3′). For transient transfection, T24 cells were added into 6-well plate (1 × 10⁵ cells per well). When cells grew to 50–60% confluence, cells were transfected with 10 μl siRNAs (20 nM) using 5 μl Lipofectamine 3000 (Invitrogen, USA) for 48 h according to the manufacturer. Finally, we used RT-qPCR to evaluate the expression levels of lncRNAs.

We used Trizol (Takara) to extract total lncRNAs from tissues and cell lines under various experimental conditions. cDNA Synthesis Kit (Takara) combined with lncRNAs (1 μg) was utilized to reverse transcribed cDNA. The quantitative polymerase chain reaction (qPCR) was performed on an ABI 7500 real-time PCR system (Applied Biosystems) by SYBR-Green method (Takara). The values of Ct were calculated with the 2^−ΔΔCt method and normalized to the expression levels of β-actin. The expression levels of lncRNAs were relative to the fold change of their controls which were defined as 1. The primer sequences were shown in Table 2. Three assays were conducted per cDNA sample.

100 μl ice‐cold Matrigel was added into a well of a 48‐wells plate for the HUVEC cells tube formation assay. 1 h later, 1 × 10⁵ HUVEC cells resuspended in 100 μl DMEM medium were added into the 48‐wells plate. After incubating at 37 °C in 5% CO₂ incubator for 4 h, the tube formation status was photographed by light microscopy. The numbers of branch points were counted and analyzed by image J. 5 fields (200×) per chamber were observed for counting invaded cell numbers.

For the migration assay, 1 × 10⁵ HUVEC cells were suspended in 100 μl medium without FBS and seeded into an upper chamber (Corning, USA). Then, 800 μl complete DMEM medium containing 10% FBS was added to the lower chamber. For the invasion assay, Diluted Matrigel (1:5 dilution with the DMEM medium) was added into the cell culture inserts. Four hours later, 1 × 10⁵ HUVEC cells in 100 μl non-FBS DMEM medium were added into the upper chamber, while the lower chamber was filled with 800 μl of DMEM medium with 10% FBS. After culturing 24 h for migration test and 48 h for invasion test, the cells culture inserts were washed by PBS and stained with 4% paraformaldehyde. Then inserts were dyed by 0.1% crystal violet solution for 20 min. The dyed cells were photographed by light microscopy. 5 fields (200×) per chamber were observed for counting invaded cell numbers.

Immunohistochemistry (IHC) analysis was performed as described previously [24]. IHC was performed using antibodies against CD31 (Cat# ab9498, Abcam, UK). Images were scanned using Pannoramic SCAN.

Transcriptome RNA-sequencing data of 414 BUC tumour tissues and 19 normal tissues were downloaded and extracted from The Cancer Genome Atlas (TCGA) data portal (https://​portal.​gdc.​cancer.​gov/​). We excluded patients whose OS ≤ 30 days from this study because they might die of unpredictable factors such as hemorrhage and infection. The data utilized in the study were updated in March 21, 2021. Raw data of BUC patients were collected for further analyses. Transcriptome RNA-sequencing results and clinical data of BUC patients were combined into a matrix file by a merge script in the Perl language (http://​www.​perl.​org/​).

Angiogenesis-related genes (ARGs) were extracted from The Molecular Signatures Database v4.0 (ANGIOGENESIS M14493 and WP_ANGIOGENESIS M39556, http://​www.​broadinstitute.​org/​gsea/​msigdb/​index.​jsp). Then the angiogenetic scores of these ARGs were calculated according to their expression levels in BUC tissues. To further identify the ARLNRs, we conducted the Pearson correlation analysis to clarify the correlation between angiogenetic score and the expression of lncRNA in BUC tissue. A standard of |r| > 0.6 and P < 0.05 was used to screen the ARLNRs. Besides, we selected the sARLNRs by univariate COX analysis and survival packages of R software (P < 0.01). sARLNRs were further divided into deleterious and protective portions by the Hazard ratio (HR).

Through multivariate COX regression analysis, we established the ARRSM based on the selected sARLNRs. The score in the ARRSM was calculated based on the expression of sARLNRs together with the Cox regression coefficients. The formula was as followed, [Expression levels of KIRREL1-IT1 * (0.266525)] + [Expression levels of AC005625.1 * (− 0.135165)] + [Expression levels of AC018809.1 * (− 0.170812)] + [Expression levels of AC008760.1 * (− 0.133273)] + [Expression levels of AC083862.2 * (− 0.221852)]. BUC patients were separated into the high-risk group and the low-risk group according to the median score.

Receiver operating characteristic (ROC) curves were used to assess the sensitivity and specificity of the ARRSM and drawn by survival ROC package of R software. Gene set enrichment analysis (GSEA) was used to detect the different pathways of ARRSM. We evaluated the survival probabilities of patients in different risk groups by Kaplan–Meier survival curves. We conducted the univariate and multivariate Cox regression analyses to verify the independent prognostic factor of BUC. Nomograms were drawn to predict the survival probabilities of BUC patients by the rms package of R software.

Transcriptome RNA-sequencing data and clinical data of BUC patients were downloaded from TCGA database. Following that, lncRNAs data were extracted from the transcriptome data and 1282 differentially expressed lncRNAs were identified by limma algorithm, of which 310 lncRNAs were down-regulated and 972 lncRNAs were up-regulated (Fig. 1A, B). Then, we screened 72 ARGs in the ANGIOGENESIS M14493 and WP_ANGIOGENESIS M39556 of Molecular Signatures Database. After analyzing the correlations of the 1282 differentially expressed lncRNAs and 72 ARGs by Pearson correlation analysis, we found that 98 lncRNAs were associated with angiogenesis (|r| > 0.6 and P < 0.05). To better investigate the clinical significances of these ARLNRs, we then conducted the univariate COX regression analysis. Finally, 10 ARLNRs were identified to be relevant to OS of BUC patients (sARLNRs: AL157392.3, AP000346.1, KIRREL1-IT1, AP002840.2, AC005625.1, AL356512.1, AC018809.1, IGBP1-AS1, AC008760.1 and AC083862.2) (P < 0.05). As shown in the forest map, all sARLNRs but aside from KIRREL1-IT1 were regarded as protective factors (Fig. 1C).

To further validate the survival and prognostic relevance, we conducted multivariate COX regression analysis. Finally, KIRREL1-IT1, AC005625.1, AC018809.1, AC008760.1 and AC083862.2 were employed to establish the ARRSM (Table 3). Besides, the correlations of the five sARLNRs and OS were further illustrated via the survival curves in which the higher expression levels of AC005625.1, AC018809.1, AC008760.1 and AC083862.2 were correlated with the longer survival time, while the expression levels of KIRREL1-IT1 showed the opposite conclusion (Additional file 1: Fig. S1). According to the median risk score, BUC patients were divided into the high-risk group and the low-risk group (Fig. 1D). With the increasing risk score, the mortality rate of BUC patients and the expression levels of KIRREL1-IT1 constantly increased, while AC005625.1, AC018809.1, AC008760.1 and AC083862.2 expressed decreasingly (Fig. 1E). Kaplan–Meier analysis illuminated that BUC patients in the low-risk group exhibited a longer survival time than those in the high-risk group (Fig. 1F).

In order to explore the relevance of sARLNRs and clinical features of BUC patients, we used ggpubr package of R language to analyze the correlations of ARRSM and the clinical characteristics, such as grade, stage, TNM stage. We found the expression levels of AC018809.1 and AC008760.1 decreased in patients with advanced grade (Fig. 2A). The expression levels of AC005625.1 and AC008760.1 increased in the more advanced stage while KIRREL1-IT1 dropped (Fig. 2B). Besides, the expression level of KIRREL1-IT1 was elevated in the advanced T-stage but AC008760.1 decreased (Fig. 2C). AC005625.1 expressed more in the advanced N-stage and M-stage (Fig. 2D, E).

Next, we performed univariate analysis to analyze the independent risk factor of BUC patients, and the results showed gender, stage, T-stage, N-stage, M-stage and risk score were significantly correlated to prognosis (P < 0.05). However, through the multivariate analysis, only risk score could be served as an independent risk factor of BUC patients (Table 4). Next, we calculated the areas under curves (AUCs) for receiver operating characteristic (ROC) curves of ARRSM and clinical characters, and found the AUCs of risk score, age, gender, grade, stage, T-stage, M-stage and N-stage were 0.741, 0.57, 0.486, 0.422, 0.659, 0.651, 0.521 and 0.629 respectively (Fig. 2F). The highest AUC was risk score, which meant the ARRSM was the most accuracy independent risk factor of BUC patients. We normalized the points of each BUC patient to a distribution ranging from 0 to 100, and then we calculated the 1-year, 3-year and 5-year survival probabilities of BUC patients by drawing a vertical line (clinical features and the expression levels of sARLNRs) between the total points axis and each prognosis axis (Fig. 3A, B). The nomograms will provide a novel diagnostic method for clinical workers to assess the prognosis of BUC patients. These results illustrated that the ARRSM could be served as a promising risk score model in predicting the prognosis of BUC patients.

Then, we used kyoto encyclopedia of genes and genomes (KEGG) pathway analysis of GSEA to further investigate the underlying mechanisms of ARRSM. We found that the high-risk group was closely related to the VEGF signaling pathway (Fig. 3C). These results motivated us to further discover the underlying mechanisms of angiogenesis and immune cell infiltration in the future studies.

In order to verify the expression levels of sARLNRs in various BUC cell lines and clinical samples, we detected the expression levels of AC005625.1 and AC008760.1 in SV-HUC-1, different BUC cell lines, tumour tissues and adjacent normal tissues. Compared with SV-HUC-1, the expression levels of AC005625.1 and AC008760.1 were significantly lower in T24, 5637, UMUC-3, J82 and TCC-SUP BUC cell lines (Fig. 4A). Besides, we found that AC005625.1 and AC008760.1 expressed lower in tumour tissues than those in adjacent normal tissues (Fig. 4B), with the lower expression in more advanced T stage (Fig. 4C).

To further investigate the clinical significance of the sARLNRs, we first divided the 138 BUC patients into sARLNRs-high-expression (n = 69) and sARLNRs-low-expression (n = 69) groups according to the average expression of AC005625.1 and AC008760.1 (the cut-off value: 0.4057). As illustrated in Table 1, the expression of AC005625.1 and AC008760.1 was significantly related to ECs percentage, tumour size and muscle invasion status, while there was no remarkable correlation with gender, age, distant metastasis, lymph node status, or nidus number. To further discover the correlation of the sARLNRs and tumour-related angiogenesis, we detected the percentage of EC subtype in tumour tissues by FACS. As illustrated in Fig. 4D, E, a prominently higher proportion of EC was observed in tumours with lower average expression of AC005625.1 and AC008760.1. The result of Pearson correlation analysis suggested the significant negative correlation between sARLNRs expression and ECs percentage (r = − 0.599, P < 0.001) (Fig. 4F). Furthermore, the higher percentage of EC in tumour tissues with lower expression of AC005625.1 and AC008760.1 was also validated by IHC (Fig. 4G).

To further determine the effects of AC005625.1 and AC008760.1 in tumour-related angiogenesis, we knocked down the AC005625.1 and AC008760.1 using siRNAs in T24 cells (Fig. 5A) followed by the co-culture of T24 cells and HUVEC cells to simulate tumour microenvironment of BUC (Fig. 5B). Next, we performed HUVEC cells tube formation and Transwell experiments to explore the angiogenesis-related roles of AC005625.1 and AC008760.1 in tumour microenvironment of BUC. As shown in Fig. 5C, D, the tube formation, migration and invasion abilities of HUVEC cells co-cultured with siRNAs-treated T24 cells were prominently enhanced. Then, we found that the expression of angiogenesis-related proteins (CD31 and VEGF-A) (Fig. 5E) and the key proteins in the epithelial–mesenchymal transition pathway (VIMENTIN and NCADHERIN) (Fig. 5F) increased in HUVECs which were co-cultured with siR-AC005625.1 and siR-AC008760.1-treated T24 cells, while E-CADHERIN decreased. These results demonstrated that knockdown of AC005625.1 and AC008760.1 in BUC cells promoted tumour-related angiogenesis by augmenting the tube formation, invasion and migration abilities of HUVECs in the TME.