RNA-Seq data (FPKM/UQ-FPKM/TPM formats) and clinical data from TCGA and GTEx were obtained via the UCSC Xena browser to ensure data consistency and comparability. Prior to differential expression analysis, we filtered the gene expression matrix, retaining only genes with expression levels $>$0.1 in over 50% of samples to remove low-expression and unreliable measurements. Due to distinct batch effects arising from different research platforms, we applied batch correction to the merged TCGA and GTEx expression matrices using the “ComBat” algorithm to eliminate non-biological variation introduced by platforms and research centers.

Data obtained from Xena had already been uniformly processed into log2(TPM + 1) or log2(FPKM-UQ + 1) formats, and we directly utilized this standardized data for analysis. For differential analysis, we employed the R package ‘DESeq2’, which internally utilizes its own normalization method (Median of ratios). For gene set variation analysis (GSVA), we used the ‘GSVA’ R package, whose default method converts expression matrices to rank data and calculates enrichment scores.

Differentially expressed genes (DEGs) were filtered using the criteria: $|$log2(Fold Change)$|$ $>$1 and a Benjamini-Hochberg (BH)-corrected false discovery rate (FDR) 0.05. In survival analysis, following log-rank testing, we similarly applied BH correction to p-values for all candidate genes or gene sets to control false positive rates during multiple testing across the genome or transcriptome. All reported significance p-values represent values corrected for multiple testing.

In correlation analysis, we calculated Spearman correlation coefficients and performed BH correction on significance p-values.

We looked at overall survival (OS), disease-specific survival (DSS), disease-free interval (DFI), and progression-free interval (PFI). We used univariate Cox regression and Kaplan-Meier modeling to analyze ADCY5’s prognostic influence on specific prognostic types for each malignancy, and the results were presented as a heatmap. Kaplan-Meier curves were generated using the survminer program.

Spearman correlation analysis was used to detect expression correlations among ADCY family genes, and the results were presented as heatmaps. Meanwhile, the co-expression of ADCY5 and other ADCY family genes in ten tumors was investigated, and the findings were shown using the ggplot tool. The predictive usefulness of ADCY family members was confirmed by assessing their overall survival using one-way Cox regression.

Pathologic images of ADCY5 in five cancers were obtained through the Human Protein Atlas database (HPA, https://www.proteinatlas.org/).

In the gene set enrichment analysis (GSEA), we utilized Hallmark gene sets, Kyoto Encyclopedia of Genes and Genomes (KEGG) metabolic pathway gene sets, immune-related gene sets, and aging-related gene sets from the Molecular Signatures Database (MSigDB). For various tumor types, samples were stratified into high and low expression groups based on the upper and lower 30th percentiles of gene expression levels, respectively. To conduct a variance analysis, the limma package was utilized to derive the log2 fold change (log2FC) for each gene. Subsequently, all genes were ranked based on their log2FC values. TheGSEA was executed using the GSEA function from the clusterProfiler package, focusing on hallmark gene sets. Enrichment analysis for hallmark and KEGG metabolic gene sets was performed with the GSEA function in clusterProfiler, which involved calculating the enrichment score (ES) for each gene set and conducting significance testing with multiple hypothesis correction. The results were visually represented through bubble plots. To refine the gene set scoring, we applied the scale function for normalization, and subsequently, we calculated the Pearson correlation coefficients between ADCY5 genes and each respective gene set score.

In the GSVA analysis, we selected 14 functional state gene sets from the CancerSEA database (http://biocc.hrbmu.edu.cn/CancerSEA), covering diverse tumor cell behaviors such as apoptosis, invasion, and stemness. Using the z-score method from the GSVA R package, we calculated the composite z-score for each gene set across all samples. Subsequently, the scale function was applied to standardize the data and generate gene set scores. Subsequently, Pearson correlation coefficients between ADCY5 expression and each gene set score were calculated to assess functional associations.

To identify potential compounds targeting ADCY5, we performed CMap analysis. For each cancer type, the top 150 differentially expressed genes between high and low ADCY5 expression groups were selected to construct an ADCY5-related gene signature. This signature was compared against compound-induced gene expression profiles from the CMap database (https://www.broadinstitute.org/connectivity-map-cmap) using the eXtreme Sum (XSum) algorithm. Similarity scores were generated for 1288 compounds. A lower score suggests a higher potential for the compound to reverse ADCY5-driven oncogenic signatures.

We employed the Sparkle database (https://grswsci.top) to conduct correlation analyses utilizing data obtained from GSE167297 within the TISCH2 database. The TISCH2 database (http://tisch.comp-genomics.org/) serves as a repository for single-cell RNA sequencing (scRNA-seq) datasets derived from both human and mouse tumors. To assess the variations in ADCY5 expression across distinct cell types, we applied the Kruskal-Wallis rank sum test (commonly referred to as the Kruskal test). Based on the expression status of ADCY5, all cells were classified into two groups: those that were expression-positive and those that were expression-negative. Subsequently, the limma package was utilized to evaluate the differences in scoring between these two groups. We further examined various biological pathways related to immunity, metabolism, signaling, proliferation, cell death, and mitochondrial functions, employing the AUCell package for scoring these pathways.

We performed the analysis using SangerBox 3.0 (http://sangerbox.com) [21]. The TCGA Pan-Cancer dataset, retrieved from the UCSC (https://xenabrowser.net/) database and uniformly normalized, was used. Differences in gene expression were calculated for each tumor in samples at various clinical stages. The relationship between ADCY5 and immune cells was evaluated using the Pearson correlation coefficient. In addition, Tumor Mutation Burden (TMB), Microsatellite Instability (MSI) scores, and Immune Neoantigen data for each tumor were calculated using the tmb function from the R package maftools. Additionally, expression levels for the ADCY5 gene alongside 44 marker genes associated with three distinct categories of RNA modification genes were extracted from each sample. Subsequently, the Pearson correlation between ADCY5 and these RNA modification genes was analyzed. Role of ADCY5 in immunotherapy cohorts

We analyzed the role of ADCY5 in the immunization cohort using the Biomarker Exploration for Solid Tumors (BEST) database (https://rookieutopia.com/app_direct/BEST/) [31].

Space the transcriptome analysis is to use Sparkle database (https://www.grswsci.top/) and SpatialTME (https://www.spatialtme.yelab.site/) [32]. We visualized gene expression landscapes in each microregion from spatial transcriptome data. Liver hepatocellular carcinoma (LIHC) data were used from GSE203612-GSM6177612, and CRC samples were derived from publicly available data from Wu et al. [33]. The predominant cell type within each microregion was determined, and the SpatialDimPlot function from the Seurat package was employed to illustrate the maximal cellular composition of these microregions. The SpatialFeaturePlot function was used to visualize gene expression patterns across microregions, while Spearman correlation analysis assessed the relationships between cellular composition and ADCY5 expression across all spots. The visual representation of these findings was accomplished using the LINKET package.

This study has been approved by the Ethics Committee of the First Hospital of China Medical University (Shenyang, China), and all participating subjects have signed informed consent forms. Gastric cancer tissues were collected from patients undergoing subtotal gastrectomy. The selection criteria for these patients were: no other primary tumors and no prior radiotherapy or chemotherapy before surgery. Gastric cancer and adjacent tissue specimens from all patients were obtained surgically and subsequently subjected to independent histopathological diagnosis by two senior gastrointestinal pathology specialists. Concurrently, multiple clinical and pathological parameters were collected during the study. Subsequent follow-up was conducted via telephone to collect prognostic information, including survival status, overall survival duration, and date of death. The clinicopathological features of the 61 gastric cancer (GC) patients in the validation cohort are summarized in Supplementary Table 1. Follow-up visits were scheduled every six months. Immunohistochemical (IHC) staining techniques were employed to evaluate the protein expression levels of ADCY5 within gastric cancer tissue samples. These sections underwent a deparaffinization process, followed by rehydration through a series of ethanol gradients, and were subsequently incubated in an Ethylenediaminetetraacetic acid (EDTA) solution. The activity of endogenous peroxidase was inhibited by treatment with a 3% hydrogen peroxide solution. Tissue collagen was disrupted using 10% normal goat serum to reduce nonspecific binding. Using ADCY5 antibody (1:300, Cat No. 30153-1-AP, Proteintech, China) as the primary antibody, samples were incubated at room temperature for 1 h. After washing with PBS, biotin-labeled secondary antibody and streptavidin-horseradish peroxidase were added to the samples, respectively, and the samples were incubated for 10 min at room temperature each time. The samples were then stained with 3,3’-Diaminobenzidine (DAB), dehydrated, and fixed with resin. Finally, the stained tissue sections were observed under a microscope by two experienced pathologists. Five distinct areas were randomly chosen from the 400$\times$ magnification field of view of the microscope. The immunoreactivity score was determined by integrating the percentage of positive cells and the staining intensity. The categorization of positive cell percentages was defined as follows: 10% corresponds to a score of 0; 10–25% indicates a score of 1; 26–50% equates to a score of 2; 51–75% results in a score of 3; and 76–100% receives a score of 4. Staining intensity was classified such that 0 represents no staining, 1 signifies light yellow, 2 denotes yellow, and 3 indicates brown. Ultimately, the final score is derived from the multiplication of the two individual scores.

MKN-45 cells were purchased from the Cell Resource Center, Institute of Basic Medical Sciences, Chinese Academy of Medical Sciences (Beijing, China) and came with an STR identification certificate. Cells were cultured in RPMI1640 (11875093, Vivacell, USA) containing 10% FBS at 37 °C with 5% CO2. Mycoplasma contamination was ruled out using the Mycoplasma PCR Detection Kit (C0301S, Beyotime, China).

The antisense sequence of siRNA is (5^′ to 3^′): GCAAGAUGAUGGACACUAUTT. The primer sequences used are: Forward: 5^′-TGCTTCTGGTCGTGGCTGTC-3^′, Reverse: 5^′-ACGACAGCATCGAGGACAAC-3^′. Prepare a 6-well plate with a cell density of 4.0 $\times$ 10⁵ cells/well for siRNA transfection. siRNA was mixed with jetPRIME® buffer, and transfection reagent (101000046/114-15, jetPRIME, MA, USA), incubated for 10 minutes, and added to the cells. RNA and protein were extracted 48–72 hours post-transfection to assess transfection efficiency.

Cellular lysis was achieved using TRIZOL reagent (R0016, Beyotime, Wuhan, China), which was allowed to react for a duration of 5 minutes at ambient temperature. Following this, chloroform was incorporated, the mixture was agitated and subsequently incubated at room temperature for another 5 minutes before being subjected to centrifugation at 12,000 rpm for 20 minutes at 4 °C. The resultant supernatant was carefully removed, thoroughly mixed, and combined with an equal volume of pre-chilled isopropanol. This mixture was left undisturbed at room temperature for 10 minutes, followed by centrifugation at 12,000 rpm for 20 minutes at 4 °C. The supernatant was then discarded, and 1 mL of 75% ethanol was introduced to the precipitate at the bottom of the tube, which was then washed several times. The precipitate was subsequently resuspended and centrifuged at 12,000 rpm for 5 minutes at 4 °C. The ethanol was removed, and the precipitate was allowed to dry completely at room temperature before the addition of 50 µL of enzyme-free sterile water to facilitate resuspension and full dissolution. Total cellular RNA was extracted utilizing TRIzol reagent (R0016, Beyotime, Wuhan, China) and reverse transcribed into single-stranded complementary DNA (cDNA) employing a PrimeScript reverse transcription reagent kit (DP117, TIANGEN) in accordance with the manufacturer’s guidelines. For PCR amplification, the AceQ qPCR SYBR Green Master Mix (RR047A, TaKaRa, Japan) was employed on an ABI QuantStudio 3 PCR system (Thermo Fisher Scientific Inc., China) under specified conditions: an initial denaturation at 95 °C for 3 minutes, followed by 40 amplification cycles consisting of 95 °C for 10 seconds, 58 °C for 20 seconds, and 72 °C for 30 seconds. The expression data were normalized against $\beta$-actin levels and quantified utilizing the 2^{-Δ⁢Δ⁢Ct} method.

The assessment of cell proliferation was conducted utilizing the Cell Counting Kit (CCK-8 Solution; A311-01/02, Vazyme, China). In summary, cells were seeded into a 96-well plate at a concentration of 2 $\times$ 10³ cells per well and subsequently incubated in a hydrogen-rich environment for a duration of 24 hours within a humidified chamber maintained at 37 °C with 5% CO2. After incubation, 10 µL of CCK-8 reagent and 90 µL of fresh medium were added to each well, followed by a 2-hour incubation and subsequent measurement of the optical density at 450 nm using a microplate reader (Thermo Fisher Scientific Inc., China).

Kaplan-Meier analysis combined with log-rank test was used for survival analysis. Statistical analysis was performed by R, and p 0.05 was considered statistically significant (*p 0.05, **p 0.01, ***p 0.001).

This study examined the expression and mutation of ten genes belonging to the ADCY family in cancer. As shown in Fig. 1A, the ADCY3, ADCY7, and ADCY10 expression levels in most cancer tissues were upregulated when compared with those in non-cancerous tissues. The expression of most members of the ADCY family was upregulated in kidney renal clear cell carcinoma (KIRC) and cholangiocarcinoma (CHOL). Analysis of The Cancer Genome Atlas-Gene Tissue Expression (TCGA_GTEx) data confirmed that ADCY family genes were differentially expressed between 18 tumor tissues and their matched samples (Fig. 1B).

Fig. 1.

Expression of adenylate cyclase (ADCY) family genes in pan-cancer species. (A) Colors represent the difference between the mean value of the gene in the tumor group compared to the mean value in the normal group for each cancer type. A positive difference is colored in red, while a negative difference is colored in blue. The larger the absolute value of the difference, the darker the color. (B) Differential expression of ADCY family genes in paired samples. The left side is the adjacent tissue of the cancer, and the right side is the tumor tissue from the same patient. The paired samples are connected by line segments. ns: p $>$ 0.05, * p 0.05, ** p 0.01, *** p 0.001.

The most commonly mutated ADCY family member in the pan-cancer dataset was ADCY8, followed by ADCY2 (Fig. 2A). The ten members of the ADCY family were frequently mutated in uterine corpus endometrial carcinoma (UCEC) but less frequently mutated in thyroid carcinoma (THCA) and kidney chromophobe (KICH). Analysis of the copy number revealed deletions in ADCY1, ADCY5, and ADCY8 (Fig. 2B). The copy number variations of these genes were significantly and positively correlated with their expression levels (Fig. 2C). Additionally, the methylation levels of the ADCY1, ADCY5, and ADCY8 promoter regions were downregulated in the pan-cancer dataset. However, the methylation levels in the ADCY8 promoter were significantly upregulated in UCEC and cervical squamous cell carcinoma and endocervical adenocarcinoma (CESC) (Fig. 2D). The DNA methylation levels of the ADCY3, ADCY4, and ADCY5 promoter regions in the pan-cancer dataset were significantly and negatively correlated with their mRNA expression levels (Fig. 2E).

Fig. 2.

Mutation and co-expression profiles of the ADCY gene family. (A) Mutation frequency map of ADCY family genes. (B) CNV frequency plot of ADCY family genes in different cancer types. (C) Correlation between CNV and mRNA expression of genes in pan-cancer. (D) Differential DNA methylation in the promoter region of ADCY family genes in pan-cancer. (E) Association between promoter methylation and mRNA expression of ADCY5. (F) Correlation between ADCY family genes in ten types of carcinomas. (G) Heatmap of co-expression between ADCY family genes in ten types of carcinomas.

Previous studies have demonstrated that ADCY members coordinate with each other to regulate critical cellular functions. The correlation of ADCY family expression was examined in ten different cancers using TCGA data. The expression of ADCY family members was strongly correlated in all these cancers (Fig. 2F). The co-expression heatmap is shown in Fig. 2G.

Next, the correlation of ADCY family members with the prognosis of different tumors was examined using univariate Cox proportional risk regression models. The dysregulation of ADCY family expression was associated with OS in patients with cancer (Fig. 3). ADCY6 and ADCY7 were associated with prognosis in up to seven cancers. The upregulated expression of ADCY10 could predict poor prognosis in adrenocortical carcinoma (ACC), KIRC, brain lower-grade glioma (LGG), mesothelioma (MESO), and UCEC. The specific predictive values of the ADCY family members for various types of cancer are shown in the Forest plots.

Fig. 3.

One-way Cox regression analysis of prognosis of ADCY family members in pan-cancer. Red color indicates HR greater than 1 and p 0.05, green color indicates HR less than 1 and p 0.05.

The ability of the ten ADCY family members to predict prognosis was good in the pan-cancer dataset (Fig. 4A–O), especially in colon adenocarcinoma (COAD), lung adenocarcinoma (LUAD), bladder urothelial carcinoma (BLCA), and LIHC (Fig. 4F,I,M). The specific genes with good predictive ability for different cancers were as follows: ADCY1: glioma (GBMLGG) and CHOL; ADCY3: stomach adenocarcinoma (STAD), head and neck squamous cell carcinoma (HNSC), rectum adenocarcinoma (READ), GBMLGG, CHOL, and COAD; ADCY8: LUAD, THCA, and breast invasive carcinoma (BRCA); ADCY5: BRCA, prostate adenocarcinoma (PRAD), pancreatic adenocarcinoma (PAAD), BLCA, READ, GBMLGG, CHOL, COAD, and CESC. These results suggest that the ADCY family members are excellent predictive markers for different cancers.

Fig. 4.

Diagnostic potential of ADCY family genes across cancers. ADCY family gene prediction in (A) Breast Invasive Carcinoma. (B) Prostate adenocarcinoma. (C) Stomach adenocarcinoma. (D) Thyroid carcinoma. (E) Kidney renal clear cell. (F) Bladder urothelial carcinoma. (G) Pancreatic adenocarcinoma. (H) Lung adenocarcinoma. (I) Liver hepatocellular carcinoma. (J) Head and neck squamous cell carcinoma. (K) Prostate adenocarcinoma. (L) Glioblastoma and low-grade glioma. (M) Colon adenocarcinoma. (N) Cholangiocarcinoma. (O) Cervical squamous cell carcinoma.

To comprehensively profile ADCY5 expression, we analyzed its levels across normal tissues (from GTEx) and tumors (from TCGA). This integrated analysis revealed generally low and dysregulated ADCY5 expression in cancers of the stomach, lung, and intestine (Fig. 5A). Immunohistochemical staining further confirmed moderate to high ADCY5 protein expression in various tumor cells (Fig. 5B), consistent with transcript levels observed in five representative cancers (Fig. 5C). We next investigated the clinical relevance of ADCY5 by assessing its relationship with cancer stage, gender, and age. ADCY5 expression varied significantly with tumor stage in several cancers, including COAD, BRCA, Stomach and Esophageal carcinoma (STES), STAD, KIRC, Ovarian serous cystadenocarcinoma (OV), and Pan-kidney cohort (KIPAN) (Fig. 5D). A notable sex-based difference was observed, with male patients exhibiting significantly higher ADCY5 expression than females in COADREAD, BRCA, Sarcoma (SARC), and kidney renal papillary cell carcinoma (KIRP) (Fig. 5E). Furthermore, ADCY5 expression exhibited distinct age-associated patterns, showing a negative correlation with age in GBMLGG but a positive correlation in Thymoma (THYM), OV, and BRCA (Fig. 5F).

Fig. 5.

Pan-cancer expression and clinical correlations of ADCY5. (A) For comparative visualization, ADCY5 expression across tumor and normal tissues from various organs was normalized to unitary Z-scores, and the median values for each group were plotted. (B) ADCY5 shows moderate to high intensity staining in different parts of the tumor cells. (C) Immunohistochemical images of ADCY5 in the Human Protein Atlas (HPA) database. (D,E) Correlation between ADCY5 expression stage and gender. (F) Correlation between ADCY5 expression and age. * p 0.05, ** p 0.01, **** p 0.0001.

The effect of ADCY5 on the OS, DSS, DFI, and PFI in the pan-cancer dataset was examined. ADCY5 upregulation predicted poor prognosis in patients with COAD, STAD, and MESO (Fig. 6A). In contrast, ADCY5 upregulation exerted protective effects in ACC, KIRC, and LGG. The specific effect of ADCY5 expression on OS in each tumor type is shown in the Forest plot (Fig. 6B). Kaplan-Meier survival analysis confirmed that ADCY5 is a prognostic biomarker for ACC, BLCA, KIRC, LGG, MESO, and STAD (Fig. 6C).

Fig. 6.

Prognostic value of ADCY5 in pan-cancer. (A) Correlation of ADCY5 expression level with overall survival (OS), disease-specific survival (DSS), disease-free interval (DFI), and progression-free interval (PFI). Red indicates risk factors (high expression is associated with poorer survival), green indicates protective factors (high expression is associated with better survival), gray indicates no calculations were performed or missing data, and white indicates no significant association. Calculations were performed using both univariate Cox and Log-rank methods. (B) Prognostic value of ADCY5 in patients with various malignancies. The impact of individual factors on overall survival was evaluated using univariate Cox proportional hazards models from the R “survival” package. (C) Overall survival stratified by ADCY5 expression in six cancer types. The blue dashed line represents the high-expression group, while the orange dashed line denotes the low-expression group. Survival curves between the two groups were compared using the Log-rank test, with p 0.05 indicating statistically significant differences.

To explore the potential function of ADCY5 in different types of cancers, we performed gene set enrichment analysis. GSEA analysis demonstrated the important pathways and biological functions that ADCY5 may be involved in pan-cancer (Fig. 7). Epithelial-mesenchymal transition and myogenesis were significantly enriched in the ADCY5 high expression group. We also calculated the correlation between ADCY5 and the different functional status scores of 14 tumor cells in pan-cancer, and the results showed that ADCY5 had a strong positive correlation with angiogenesis, differentiation and stemness (Supplementary Fig. 1). We performed pan-cancer GSEA enrichment analysis on the Hallmark gene set, metabolic gene set, immune-related gene set, and senescence-related gene set (Supplementary Fig. 2). In most tumors, ADCY5 was associated with prostaglandin and leukotriene metabolism in senescence, epithelial-mesenchymal transition, hedgehog signaling, B-cell differentiation, B-cell receptor signaling pathway, drug metabolism, and biocarta eicosanoid pathway. Thus, ADCY5 may exert its effects in cancers through these pathways.

Fig. 7.

Differences in the enrichment of ADCY5 were analyzed across 50 Hallmark pathways and 83 metabolic genome datasets. Based on gene expression levels, samples were dichotomized into high- and low-expression groups (top and bottom 30%, respectively). Differential expression was analyzed with the limma package, followed by Gene Set Enrichment Analysis (GSEA) using clusterProfiler. The enrichment score (ES) for each gene set was computed, followed by significance testing and adjustment for multiple hypotheses regarding the ES values. Visualization of the results was facilitated through bubble plots. A negative normalized enrichment score (NES) denotes significant pathway enrichment in the ADCY5 low-expression group, whereas a positive NES suggests significant enrichment in the ADCY5 high-expression group. The intensity of the color represents the magnitude of the absolute value of the ES, while the degree of scatter among the bubbles correlates with the significance of the adjusted p-value.

We further analyzed the relationship between ADCY5 and response to immunotherapy in a clinical trial. In Kim cohort 2019 (anti-PD1/PD-L1), patients with a high response rate to immunotherapy had lower ADCY5 expression (Fig. 8A). Further, we performed ROC curve analysis based on ADCY5 expression to measure the predictability of ADCY5 response to immunotherapy (Fig. 8B). The area under the curve in Kim cohort 2019 (anti-PD1/PD-L1) was 0.783. In addition, survival analysis showed that anti-PD-L1-treated patients with high ADCY5 expression had shorter OS (Fig. 8C).

Fig. 8.

Association of ADCY5 with Immunotherapy and the Tumor Immune Microenvironment. (A) High ADCY5 expression was associated with a reduced response rate to immunotherapy. (B) ROC curves based on ADCY5 expression in patients in the immunotherapy cohort. (C) Patients exhibiting elevated levels of ADCY5 expression experienced a reduced survival rate in response to this immunotherapy treatment. (D–F) The relationship between ADCY5 expression and tumor mutational burden, neoantigen expression, and microsatellite instability was assessed across various cancer types. (G) The association between ADCY5 expression and the prevalence of immune-related cell populations within tumors was analyzed. (H,I) In the spatial transcriptome map, each point corresponds to a sequenced microregion, and colors denote different cell populations; a deeper red hue signifies an increased ADCY5 expression within that particular spot. (J) Spearman correlation analysis was employed to determine the relationships between cell content and cell content, as well as between cell content and gene expression across all identified spots. * p 0.05, ** p 0.01, *** p 0.001, **** p 0.0001.

TMB, MSI, and Neoantigen (NEO) are considered to have important roles in predicting tumor immunotherapy response. We observed a significant negative correlation of ADCY5 expression with TMB and MSI in more than half types of tumors (Fig. 8D,E), especially STAD, Lymphoid Neoplasm Diffuse Large B-cell Lymphoma (DLBC), and STES. ADCY5 showed a significant positive correlation with NEO in testicular germ cell tumors (TGCT) and THCA, and a significant negative correlation with NEO in THYM and READ (Fig. 8F). Spatial transcriptomics is a powerful technique that enables the simultaneous acquisition of cellular spatial localization and gene expression profiles, serving as a pivotal research tool for elucidating cellular functions within tissues and their interactions with the microenvironment. In this study, we aimed to visualize the gene expression landscapes across various microregions derived from spatial transcriptomic datasets. To achieve this, we utilized the SpatialFeaturePlot function available in the Seurat package, which allowed us to illustrate the enrichment scores corresponding to each cell type. A darker hue in the visualization indicates a higher enrichment score, reflecting an increased abundance of that specific cell type within the designated spot. At the same time, we analyzed the correlation between the cell content ADCY5 expression in all the spots was calculated (Fig. 8H–J). Highly consistent with the results we obtained using transcriptomic data: in both CRC and LIHC (Fig. 8G), The positive correlation of ADCY5 with CD4+ T cells, CD8+ T cells, and dendritic cells indicates its association with tumor immune infiltration.

To further explore potential drugs targeting ADCY5, CMap analysis was used. In a comprehensive analysis encompassing over 15 different cancer types, MK.886 emerged as the most promising compound capable of correcting the dysregulation of ADCY5 and alleviating its associated oncogenic impacts (Fig. 9A). Additionally, we identified compounds that specifically target ADCY5 across six distinct cancers, highlighting its prognostic significance in these malignancies (Fig. 9B). These results lend considerable credence to our predicted efficacy; however, additional research is warranted to clarify the mechanisms.

Fig. 9.

Identification of potential therapeutic compounds targeting ADCY5. (A) Identification of ADCY5 target compounds by Connectivity Map (CMap) analysis. (B) Target compounds in six cancers.

Uniform Manifold Approximation and Projection (UMAP) was utilized to visualize the distribution of single-cell data following the dimensionality reduction of the GSE167297 dataset, leading to the segregation of distinct cell types based on their specific expression profiles (Fig. 10A,B). Cells were classified into two categories: expression-positive and expression-negative, contingent upon the presence or absence of ADCY5 expression. The relative proportions of each cell type were quantified between the two groups; notably, the proportion of Fibroblasts in the ADCY5-expression-positive cohort was significantly elevated compared to that in the ADCY5-expression-negative cohort (Fig. 10C). Furthermore, we analyzed the variations in ADCY5 expression across different cell types (Fig. 10D–F). We also evaluated scores associated with immune responses, metabolic processes, signaling pathways, proliferation, cell death, and mitochondria-related biological pathways, comparing the scoring discrepancies between the ADCY5 expression-negative and ADCY5 expression-positive groups (Fig. 10G). Elevated mitochondrial pathway scores in ADCY5-positive endothelial cells implicate a potential mechanistic role for ADCY5 in gastric cancer.

Fig. 10.

Analysis of ADCY5 in single-cell data of gastric cancer. (A,B) Through the application of Uniform Manifold Approximation and Projection (UMAP) for dimensionality reduction, distinct cell populations were delineated based on their characteristic expression profiles. (C) In this analysis, the red hue denotes the group exhibiting positive expression of ADCY5, while the blue hue signifies the group with negative ADCY5 expression. (D–F) The variations in the expression levels of select genes across different cell types were evaluated utilizing the Kruskal-Wallis rank sum test (commonly referred to as the Kruskal test). (G) The scoring of biological pathways associated with immune responses, metabolism, signaling, proliferation, cellular apoptosis, and mitochondrial functions was conducted using the AUCell package. The red color represents pathway scoring increased in the ADCY5 expression-positive group (activation) and the blue color represents pathway scoring decreased in the ADCY5-positive group (inhibition).

ADCY5 upregulation predicts poor OS in patients with gastric cancer. To examine if ADCY5 can serve as a biomarker for gastric cancer, the effects of ADCY5 expression on DSS and PFI in patients with gastric cancer were examined (Fig. 11A,B). ADCY5 upregulation predicted poor prognosis in patients with gastric cancer. Additionally, ADCY5 expression could predict the prognosis of patient subgroups stratified according to clinical characteristics, including male patients, patients without Helicobacter pylori infection, and patients with completely resected tumors (Fig. 11C–E).

Fig. 11.

ADCY5 as an independent prognostic biomarker in gastric cancer. (A,B) DSS and PFI analysis of ADCY5 in the TCGA-STAD cohort. (C–E) The survival value of ADCY5 in different clinical subgroups, including male patients, patients without Hp infection, and patients with complete tumor resection. (F–H) Association of ADCY5 expression with clinicopathologic parameters, including T-stage and age. (I,J) To identify independent prognostic factors, we conducted univariate and multivariate Cox regression analyses on the TCGA-STAD cohort. (K) Development of a column-line graphical model for ADCY5. (L) Presentation of calibration curves for the 1-year, 3-year, and 5-year intervals corresponding to this column-line graphical model. ** p 0.01.

Analysis of the correlation of ADCY5 with clinical characteristics in the TCGA-STAD cohort revealed that ADCY5 expression in patients with stage T3 and T4 tumors was significantly higher relative to T1–T2 stage tumors (Fig. 11G). ADCY5 expression was inversely associated with patient age, showing higher levels in the $\le$65-year cohort (Fig. 11H). Heatmaps of clinically relevant information in the TCGA-STAD cohort demonstrated that patients with stage T3 and T4 tumors were mainly distributed in the ADCY5 high-expression group (Fig. 11F). Univariate and multivariate Cox regression analyses indicated that ADCY5 was an independent prognostic factor for OS in patients with gastric cancer (Fig. 11I–J). Next, a nomogram model was developed (Fig. 11K), which exhibited good predictive ability. The model demonstrated well-calibrated prediction accuracy (Fig. 11L).

To determine the clinical significance of ADCY5 in gastric cancer, we utilized pathologic tissue specimens collected from 61 gastric cancer patients for the analysis of ADCY5 expression and clinical significance. Based on the immunohistochemical staining results (Fig. 12A), the pathological tissue samples obtained from patients with gastric cancer were categorized into two groups: one exhibiting high expression levels and the other demonstrating low expression levels. Elevated ADCY5 expression predicted poorer overall survival in gastric cancer (Fig. 12B), which further demonstrated the value of ADCY5 as a biological marker for gastric cancer. Forest plots demonstrating the results of univariate and multivariate regression analyses indicated that ADCY5 may be a potential and promising marker for predicting poor prognosis in gastric cancer (Fig. 12C,D). Despite these bioinformatics findings suggesting a link between ADCY5 and gastric cancer progression, its specific biological function in this cancer remains unclear. To address this, we performed in vitro functional experiments. ADCY5 expression was silenced using siRNA-mediated gene knockdown, and the knockdown efficiency was confirmed by quantitative real-time PCR (Fig. 12E). Cell proliferation was assessed using the CCK-8 assay, which revealed that ADCY5 knockdown significantly suppressed the proliferation of MKN-45 cells (Fig. 12F) (p 0.01).

Fig. 12.

Experimental validation of ADCY5 as a prognostic marker and functional oncogene in gastric cancer. (A) Representative immunohistochemical (IHC) of ADCY5 in gastric cancer tissues. ADCY5 Low (20$\times$ ): scale bar = 100 µm; ADCY5 High (40$\times$ ): scale bar = 50 µm. (B) Patients were grouped according to their ADCY5 expression status: low ADCY5 and high ADCY5. A comparative analysis of survival duration was conducted between these two groups. (C,D) Forest plots demonstrate the results of unifactorial and multifactorial analysis. (E) qPCR analysis confirming the knockdown efficiency of ADCY5 in MKN-45 cells using siRNA. (F) Cell Counting Kit (CCK-8) assay showing that ADCY5 knockdown reduces cell viability. ** p 0.01, *** p 0.001, ns: p $>$ 0.05.