We obtained the mRNA expression data (FPKM) value and corresponding clinicopathologic information of bladder cancer in The Cancer Genome Atlas (TCGA) (https://portal.gdc.cancer.gov/). The cohort comprised 410 BLCA samples and 19 normal urothelium tissues. And then, the FPKM value of TCGA cohort was translated into transcripts per kilobase million (TPM) value before analysis.

The validation cohort was derived from the bladder cancer patients who underwent surgery in Xiangya Hospital. It comprised a total of 57 BLCA cancer samples and 13 normal urothelium tissues. Fresh tissues were collected and stored with liquid nitrogen immediately. First, total RNA was extracted from fresh tissues using TRIzol (Invitrogen, Carlsbad, California, United States). The total RNA was then quantified using NanoDrop and Agilent 2100 biological analyzers (Thermo Fisher Scientific, MA, United States). After constructing the mRNA library, we further purified and fragmented the total RNA into small pieces. After that, we synthesized the first-strand cDNA and the second-strand cDNA, and further amplified by PCR to construct the final library (single-stranded circular DNA). Finally, 57 BLCA samples and 13 normal tissues were qualified and sequenced on the BGISEQ-500 platform (BGI-Shenzhen, China).TPM value was also translated in Xiangya cohort. Besides, the data has been uploaded to the GEO database (GSE188715).

IMvigor210 was an immunotherapy cohort in which BLCA patients received anti-PD-1 therapy. We obtained the mRNA expression data and corresponding clinicopathologic information from http://research-pub.Gene.com/imvigor210corebiologies/based on the Creative Commons 3.0 License.

The effect of anticancer immunity was highly associated with the expression of immunomodulators, activity of the cancer immunity cycle, infiltration level of tumor infiltrating lymphocytes (TILs), and the expression of inhibitory immune checkpoints in an inflamed tumor microenvironment (TME). We collected the information of 122 immunomodulators in the previous study (Charoentong et al., 2017), and compared differential expression immunomodulators including MHC, receptors, chemokines, and immune stimulators between low and high ACER2 groups. We further explored the effect of ACER2 in impacting cancer immunity cycle in BLCA. The cancer immunity cycle was stepwise events proceed and expand iteratively comprising in 7 critical steps, which determined the fate of tumor cells (Chen and Mellman, 2013). Thereafter, we calculated the correlation between the infiltration level of TILs in TME and ACER2 expression using seven independent algorithms, including Cibersort-ABS, MCP-counter, quanTIseq, TIMER, xCell, TIP, and TISIDB (Newman et al., 2015; Becht et al., 2016; Li et al., 2016; Xu et al., 2018b; Finotello et al., 2019; Ru et al., 2019; Li et al., 2020). The relationships between ACER2 and the corresponding effector genes of these TILs were also analyzed. Moreover, we correlated the ACER2 expression with 22 common immune checkpoint inhibitors (ICIs), such as PD-1, PD-L1, CTLA-4, and LAG-3. Finally, we analyzed the T-cell inflamed score (TIS) and corresponding TIS-related effector genes in the TME, which represented pre-existing cancer immunity and predicted the clinical response of ICB (Ayers et al., 2017).

Consider that BLCA varied with high heterogeneity and differed distinctly in the treatment response and prognosis, several molecular subtype systems of BLCA had been constructed (Sjödahl et al., 2022), including UNC, Baylor, TCGA, MDA, Lund, CIT, and Consensus subtype systems. We used ConsensusMIBC and BLCA subtyping R packages to determine the molecular subtype systems, and depicted the relationship between the specific signatures of molecular subtypes and ACER2 expression (Hu et al., 2021). Receiver operating characteristic (ROC) curves were used for evaluating the accuracy of ACER2 in predicting BLCA molecular subtypes. The difference of neoadjuvant chemotherapy related mutation was evaluated in the high and low ACER2 groups. Subsequently, we explored several therapeutic responses to immunotherapy, targeted therapies and radiotherapy. Finally, drug-target genes were collected and analyzed in the Drug Bank database.

Bladder cancer and normal bladder cell lines were used to extracted total RNA using cell total RNA isolation kit (Foregene, China) according to manufacturer’s protocol. cDNA was synthesized using UeIris II RT-PCR System for First-Strand cDNA Synthesis (US Everbright, China). qRT-PCR was performed using SYBR Green qPCR Master Mix (US Everbright, China) on CFX Connect System (Bio-Rad, United States). Gene expression levels were normalized to the “housekeeping” gene GAPDH. The primers were designed and synthesized by Sangon Biotech (Shanghai, China) and detailed primer sequences were listed below: ACER2: (forward primer: 5′- CCT​TTG​GGT​TCT​GAT​GTG​TGC​TTT​G-3′; reverse primer: 5′-GGA​CAC​TGA​CCA​CCA​CCT​TGA​AC-3′); GAPDH: (forward primer: 5′- CAA​GGC​TGT​GGG​CAA​GGT​CAT​C-3′; reverse primer: 5′- GTG​TCG​CTG​TTG​AAG​TCA​GAG​GAG-3′).

Pearson or Spearman coefficients was calculated to explore correlations between variables. For variables fitting a normal distribution between binary groups, t-test was used to compare the differences. For categorical variables, chi-squared test or Fisher’s exact test was performed. Analyses with two-sided p = 0.05 were considered as the threshold of statistical significance. Receiver operating characteristic (ROC) curves were depicted to evaluate the predictive accuracy for molecular subtypes. All the statistical analyses and visualizations were performed in R software, Version: 4.2.2.

We performed pan-cancer analysis to illustrate the immunological role of ACER2, and screened cancer types which were impacted most by ACER2. Figure 1A revealed the relationship between ACER2 expression and immunomodulators across 37 cancer types. We found that ACER2 was positively associated with multiple immunomodulators in several cancer types, such as THYM, CHOL, ACC, and SARC. Remarkably, we noticed negative correlations in BLCA between ACER2and a majority of immunomodulators including immunomostimulators, MHC, receptors, and chemokines. Likewise, a negative relation was also observed in breast cancer. We then explored the relationship between ACER2 expression and several critical immune checkpoints. Of note, ACER2 expression in BLCA was negatively correlated with four immune checkpoints, namely, PD-L1, PD-1, CTLA-4, and LAG-3 (Figures 1B–E). Besides, we uncovered ACER2 expression in BLCA was negatively associated with the ESTIMATE score, Immune score and Stromal score in TME (Figures 1F–H). In brief, ACER2 was considered potential to be a biomarker to predict the TME status especially in BLCA. High ACER2 expression in BLCA may contribute to the formation of non-inflamed TME, as the result of the reduced level of immunomodulators, immune checkpoints, immune cells, and stromal cells in TME.

We assessed the expression characteristics of ACER2 in BLCA. The qPCR analysis of Xiangya cohort revealed that ACER2 was significantly elevated in tumor tissues (p = 0.0086) compared with normal urothelium (Figure 2A). Then we analyzed the differential expression of ACER2 in diverse cell lines. ACER2 was significantly higher expressed in bladder cancer cell lines in comparison to bladder epithelial cell lines (Figure 2B). Besides, single-cell analysis also indicated a higher expression in the malignant epithelial cells (Figure 2C). However, we did not observe the correlations between ACER2 expression and gender, T stage (Figures 2D,E). No significant difference was observed in the disease-specific survival between high- and low-ACER2 (p = 0.088) (Figure 2F).

ACER2 exhibited negative correlations with multiple immunomodulators (Figure 3A). The high ACER2 group had an obvious downregulation of the majority of chemokines including CCL4, CCL3, CCL24, and CCL26. Immunostimulators including CD80, CD86, ICOS, and TNFSF13B were also found negatively correlated with ACER2. Subsequently, four major steps in the anti-cancer immunity cycle were downregulated in the high ACER2 group (Figure 3B), which included the release of cancer cell antigens, priming and activation, immune cells recruiting and killing of cancer cells. Strikingly, we noticed that recruiting of multiple immune cells such as T-cell, CD8⁺ T-cell, macrophage cell, NK cell, Th1 cell, dendritic cell, neutrophil cell, eosinophil cell, basophil cell, Th17 cell, CD4 T-cell were significantly restrained in the high ACER2 group. It implied the reduction of infiltration levels of effector TILs in the TME.

For further validating the correlation between ACER2 and TILs in the TME, we calculated the infiltration level of TILs using seven independent algorithms. Similarly, ACER2 was negatively correlated with the infiltration level of CD8⁺ T-cell, macrophage, NK cell, Th1 cell, and dendritic cell (Figure 3C). Likewise, ACER2 was proved to be negatively correlated with the effector genes of TILs (Figure 3D). Moreover, we explored the relation between ACER2 and immune checkpoint inhibitors in the TME. ACER2 was found to be negatively correlated with a majority of usual immune checkpoint inhibitors including PD-L1, PD-1, CTLA-4, LAG-3, TIM-3, and TIGIT (Figure 3E). Finally, we further discovered ACER2 was negatively correlated with the TIS and corresponding TIS-related effector genes (Figures 4A,B). In summary, the results revealed that ACER2 contributed to the formation of non-inflamed TME in BLCA.

BLCA was a highly heterogeneous malignancy with various molecular subtypes which mainly differed in sensitivity to diverse therapeutic regimens. Therefore, we distinguished BLCA molecular subtypes in the TCGA within ACER2 expression. Figure 4C illustrated patients with low ACER2 expression were more likely to be basal subtype and characterized by basal differentiation, EMT differentiation, immune differentiation, and keratinization. As shown in previous studies, basal subtype was more sensitive to ICB with higher immune infiltration level and pathological response rates, which was consistent to our prior finding (Kamoun et al., 2020; Necchi et al., 2020). Thus, patients with lower ACER2 expression may benefit from ICB. Conversely, individuals with high ACER2 expression were inclined to luminal subtype with urothelial differentiation, Ta pathway, and luminal differentiation. Thereafter, ROC analysis was adopted to evaluate the predictive accuracy of ACER2 for molecular subtypes, and the area under the ROC curves was calculated ranging from 0.77 to 0.93 (Figure 4D).

In order to further explore the correlation between ACER2 with neoadjuvant chemotherapy (NAC), we found that low ACER2 group was more likely to carry neoadjuvant chemotherapy related mutation such as RB1 (23%), ATM (13%), and ERBB2 (10%). We also found that high mutation rates of ATM (12%), RB1 (12%), and ERBB2 (11%) in the high ACER2 group (Figure 4E). Notably, chemotherapy related mutation of RB1 was significantly higher in the low ACER2 group, which implied that tumors with low ACER2 may be more sensitive to NAC. Ulteriorly, we revealed enrichment scores for radiotherapy-predicted pathways and EGFR ligands was higher in the low ACER2 group (Figure 4F). And several enrichment scores for various immunosuppressive oncogenic pathways such as PPARG coexpressed genes, WNT-β-catenin pathway, and IDH1 were significantly higher in high ACER2 group, which also indicated the formation of non-inflamed TME in BLCA. Besides, we used the Drugbank database to determine that the low ACER2 group was sensitive to immunotherapy and ERBB therapy, whereas antiangiogenic therapy may be more suitable for high ACER2 group (Figure 4G). Collectively, neoadjuvant chemotherapy, adjuvant chemotherapy, immunotherapy, ERBB therapy could be used for patients with low ACER2.

Thereafter, we further validated the role of ACER2 in Xiangya cohort. ACER2 negatively correlated with the several critical steps of the anti-cancer immune cycles especially including the release of cancer cell antigens, and immune cells recruiting (Figure 5A). Consistently, ACER2 was also negatively associated with multiple ssGSEA immune cells including activated dendritic cell, macrophage cell, natural killer cell, regulatory T-cell, T follicular helper cell (Figure 5B). Subsequently, a negative relationship between ACER2 and immune checkpoints was observed in Xiangya cohort (Figure 5C). Besides, we confirmed that lower ACER2 was associated with high expression in T-cell inflamed scores effector genes including CD274, LAG-3, and TIGHT (Figure 5D). Consistent with the prediction for subtype in TCGA, high ACER2 group was more likely to be luminal subtype in Xiangya cohort (Figure 5E), and the accuracy of ACER2 in prediction for molecular subtypes was over 0.87 (Figure 5F). Likewise, high ACER2 was more likely to respond to immune inhibited oncogenic therapy (Figure 5G).

We further explored the effect of ACER2 in predicting clinical response to ICB in an immunotherapy cohort, the IMvigor210 cohort, in which patients received anti-PD-1 therapy. In the IMvigor210 cohort, ACER2 was also observed negative correlations with several critical steps of cancer immune cycles (Figure 6A), thus the infiltration level of TILs in TME was downregulated when ACER2 high expressed (Figure 6B). Whereafter, we found that ACER2 was negatively correlated with the expression of several ICI genes and the TIS genes (Figures 6C,D). In addition, high expressions of ACER2 were inclined to the luminal subtype which was consistent with the results captured from the TCCA cohort (Figure 6E). And the area under ROC curves of predictive accuracy for molecular subtypes was calculated ranging from 0.81 to 0.96 (Figure 6F).

In the IMvigor210 cohort, immunohistochemistry (IHC) was performed to detect the PD-L1 expression on immune cells or cancer cells. Based on the PD-L1 expression level, immune cells and cancer cells were classified into three groups. We discovered ACER2 was significantly expressed in the TC0 and IC0 groups (Figures 6G,H), which existed the lowest PD-L1 expression in the immune cells and cancer cells, respectively. And ACER2 was higher expressed in the desert TME compared with inflamed TME (Figure 6I). Moreover, we compared the ACER2 expressions in different clinical responses to ICB. ACER2 expression was lower in patients with desert TME with complete response (CR) compared with partial response (PR), progressive disease (PD), and stable disease (SD). Even so, we did not observe significant difference across these groups (Figure 6J).