ABBS
Original Article
Integrated multi-omics and experimental approaches identify fascin actin-bundling protein 1 as an unfavorable prognostic biomarker in adrenocortical carcinoma
Pingkaiqi He1,2,+, Yihao Chen3,+, Ming Xi4+, Shanshan Mo2, Jiahong Chen3, Chuanfan Zhong5, Fengping Liu6, Weide Zhong6,7, Le Zhang8, Junhong Deng2, Jianming Lu2,7,*, and Chao Cai1 .*
1Department of Urology, Minimally Invasive Surgery Center, The First Affiliated Hospital of Guangzhou Medical University, Guangdong Key Laboratory of Urology, Guangzhou Institute of Urology, Guangzhou 510120, China, 2Department of Andrology, Guangzhou First People’s Hospital, Guangzhou Medical University, Guangzhou 510180, China, 3Department of Urology, Huizhou Municipal Central Hospital, Huizhou 516001, China, 4Department of Urology, Huadu District People’s Hospital, Southern Medical University, Guangzhou 510800, China, 5Department of Urology, Zhujiang Hospital, Southern Medical University, Guangzhou 510282, China, 6State Key Laboratory of Quality Research in Chinese Medicine, Macau University of Science and Technology, Macau 999078, China, 7Department of Urology, Guangdong Key Laboratory of Clinical Molecular Medicine and Diagnostics, Guangzhou First People’s Hospital, Guangzhou Medical University, Guangzhou 510180, China, and 8Institute for Integrative Genome Biology, University of California, Riverside 92507, USA These authors contributed equally to this work.
*Correspondence address. Tel: +86-13512780911; E-mail: chaocai85@hotmail.com (C.C.) / Tel: +86-18029160464; E-mail: louiscfc8@gmail.com (J.L.)
Received 23 June 2024 Accepted 18 November 2024 Published 19 May 2025
Abstract
Adrenocortical carcinoma (ACC) is a rare epithelial tumor originating from adrenal cortical cells, notable for its high degree of malignancy and poor prognosis. Owing to heterogeneity, patient outcomes vary significantly. Current biomarkers for ACC risk stratification have notable limitations. However, with the advancement of multi-omics sequencing technology, we can utilize multi-omics data to explore the heterogeneity of ACC, thereby identifying novel biomarkers. In this study, we establish multicenter transcriptomics and ATAC-seq data from the TCGA and GEO databases to perform weighted gene coexpression network analysis (WGCNA) clustering and conduct com- prehensive analyses of various ACC samples. These findings are integrated with univariate Cox regression, receiver operating characteristic (ROC) curve analysis, and survival analysis to identify potential biomarkers. We establish FSCN1 as an independent risk factor associated with poor ACC prognosis. ATAC-seq data demonstrate higher chromatin accessibility of FSCN1 in ACC patients with progressive disease. Immunohistochemical analysis con- firms the expression of FSCN1 at the protein level, while functional cell assays reveal its role in promoting tumor invasion and proliferation. Functional enrichment analyses highlight the biological characteristics of FSCN1, and estimation of TME-infiltrating cells suggests that FSCN1 expression contributes to poor prognosis by inhibiting CD8+ T-cell infiltration within the ACC microenvironment. Finally, multi-omics analyses elucidate the role of FSCN1 at the mutation level. Taken together, our findings highlight FSCN1 as a promising novel biomarker and potential therapeutic target, underscoring its value in guiding the strategic management of ACC.
Key words adrenocortical carcinoma, ATAC-seq, FSCN1, multi-omics, prognosis
Introduction
Adrenocortical carcinoma (ACC) is an uncommon yet markedly malignant epithelial tumor that arises from the cells of the adrenal
cortex. It accounts for a minor 0.2% of total cancer-related mortality, yet its malignancy and aggressive invasion ability are noteworthy. It is frequently discovered at advanced stages, often
@ The Author(s) 2025. This is an open access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs License (https:// creativecommons.org/licenses/by-nc-nd/4.0/).
with distant metastasis, which leads to a discouraging late-stage survival rate of less than 30% [1,2]. Surgical intervention currently offers the only potential curative approach for ACC. However, this option is feasible only for stage I, II and some stage III patients who are candidates for complete resection. Despite surgical intervention, more than 50% of patients experience either recurrence or metastasis postoperatively. Unfortunately, nonsurgical treatment modalities are limited in their scope and often yield severe side effects. For example, mitotane, a pharmaceutical therapy, has limited efficacy [2]. Conventional therapeutic strategies, such as chemotherapy and radiotherapy, still lack robust evidence support- ing their effectiveness in primary ACC [2,3].
Presently, biomarkers for ACC demonstrate inadequate specificity and sensitivity. A minority of these biomarkers are supported by robust evidence, leading to limited clinical utility. Consequently, a deeper understanding of the mechanisms underlying ACC progres- sion, alongside the identification of novel prognostic indicators and therapeutic targets, is crucial for improving patient outcomes.
In recent years, the application of high-throughput sequencing technologies has resulted in exponential growth in biological data. In 2000, the National Center for Biotechnology Information (NCBI) in the United States launched the Gene Expression Omnibus (GEO) database, a comprehensive repository of high-throughput gene expression data provided by international research organizations [4]. Additionally, in 2006, the United States initiated the TCGA project, which offered an extensive range of publicly accessible multi-omics data from cancer patients, complemented by compre- hensive prognostic follow-up information [5]. These expansive datasets have unequivocally enhanced analysis in the realm of cancer bioinformatics.
Fascin Actin-bundling Protein 1 (FSCN1) is an actin-binding protein that plays crucial roles in cellular migration, motility,
adhesion, and cytoskeletal organization [6]. It is a key determinant of the adverse prognosis of various cancers. The involvement of FSCN1 in the ACC has been reported in few studies. A study by Liang et al. [7] confirmed the overexpression of FSCN1 in ACC tissue compared with non-cancerous adrenal cortex tissue. Patients with higher FSCN1 expression had significantly poorer prognoses. FSCN1 may lead to adverse outcomes by modulating the tumor immune microenvironment in the ACC, particularly affecting the infiltration levels of CD8+ T cells. Studies by Poli et al. [8] and Cantini et al. [9] suggested that FSCN1 is a novel, independent prognostic marker for ACC. FSCN1 potentially promotes the invasiveness of ACC cells by supporting the overexpression of SF-1. However, the role of FSCN1 in ACC remains inadequately explored because of the lack of multicenter datasets and experi- mental validation. In our investigation, we confirmed FSCN1 as an autonomous risk factor and a prospective biomarker indicating an unfavorable prognosis in ACC, spanning various datasets and multi- omics levels.
Materials and Methods
The study workflow
As delineated in the workflow shown in Figure 1, our study adopts a multi-strategy to identify potential biomarkers for ACC. First, we analyzed TCGA-ACC transcriptomic data. Using weighted gene coexpression network analysis (WGCNA) based on prognostic outcomes, we identified gene modules associated with poor prognosis in ACC patients. We integrated these findings with differentially accessible peak-associated genes from ATAC-seq data, as well as candidate genes implicated in ACC progression identified through receiver operating characteristic (ROC) curve and Cox regression analyses. This approach led to the identification of FSCN1 as a potential biomarker for ACC. FSCN1, a novel biomarker
The Cancer Genome Atlas
&
TCGA-ACC GSE10927 GSE19750
5
Gene Expression Omnibus
Dwumatin Hotone Structure
ROC’s AUC
8
WGCNA
ATAC-seq
COX Regression
Kaplan-Meier Survival Analysis
Fascin Actin-Bundling Protein 1 (FSCN1)
Functional Enrichment
Experiments
Immune cell infiltration
Multiomics
for ACC, was subsequently investigated via functional enrichment, immune infiltration correlation, mutational genomics, immunohis- tochemistry and functional in vivo and in vitro experiments.
Access and processing of publicly available data
Spectra of somatic mutation, copy number variation (CNV), RNA- seq, and ATAC sequencing data from the TCGA-ACC dataset and their corresponding clinical annotations were obtained from the UCSC Xena platform (https://xena.ucsc.edu/). Expression micro- array datasets, including two adult adrenal cortical carcinoma cohorts (GSE10927 and GSE19750), were retrieved from the GEO database. The Meta-GEO cohort, consisting of two cohorts of adult ACC patients (GSE10927 and GSE19750), was established. The initial data underwent additional processing via the robust multi- array average (RMA) algorithm integrated into the “affy” (version 1.72.0) R package [10]. Batch effects were adjusted through the application of the “ComBat” function within the “sva” (version 3.42.0) R package. The baseline clinical information of this study is detailed in Supplementary Table S1.
WGCNA analysis
To identify clusters of co-expressed genes, we used the “WGCNA” (version 1.72-1) R package to construct a gene co-expression network from the mRNA expression matrix of the TCGA-ACC cohort [11,12]. The outliers were identified and excluded through hierarchical clustering of the samples via the “hclust” function. We selected an appropriate soft-thresholding power to ensure strong correlation among genes within modules, penalize low correlation, and achieve a scale-free network. Pearson correlation coefficients were calculated for the gene pairs to construct an adjacency matrix. This matrix was subsequently converted into a topological overlap matrix (TOM) and its corresponding dissimilarity matrix (1-TOM). Network construction and module detection were performed via the “DynamicTreeCut” algorithm. Module eigengenes (MEs) were computed as representative expression profiles for each module. Candidate modules, characterized by a substantial correlation coefficient between the ME profile and clinical feature information, were pinpointed and selected for subsequent analysis.
Analysis of functional enrichment
Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analyses can integrate genomic information with high-level functional insights to elucidate gene functions. GO enrichment categorizes gene functions into biological process (BP), cellular component (CC), and molecular function (MF) categories. The Sanger Box bioinformatics analysis online tool [13], which uses the “clusterProfiler” (version 3.14.3) R package [14], was employed for enrichment analysis and result visualization. Gene sets containing between 5 and 5000 genes, with a P value < 0.05 and a false discovery rate (FDR) < 0.25, were deemed statistically significant.
Chromatin accessibility variance analysis
On the basis of ATAC-seq data from the TCGA database, we identified 14 samples with matched RNA-seq and clinical data. The 14 samples were categorized into control and progression groups using the median PFI time (derived from 79 TCGA-ACC samples) as a threshold. Differentially accessible peaks (DAPs) between the control and progression groups were identified via the “DESeq2” (1.34.0) R package [15]. The DAP results were then visualized via
the “ggplot2” R package. Statistical significance in the DAP analysis was defined as an adjusted P value < 0.05 and |log2FC| > 2. The DAPs and all the peaks were subsequently annotated via the “TxDb. Hsapiens.UCSC.hg38.knownGene” [16], “org.Hs.eg.db” (3.14.0), and “ChIPseeker” (1.30.3) R packages [17]. The DAP ID was matched to the association data provided by Corces et al. [18] on “peak-gene” links to obtain differentially accessible peak-related genes (DPGs). IGV software was used to visualize peak region accessibility corresponding to DPGs [19].
Identification and validation of pivotal prognostic genes
To identify pivotal genes associated with prognosis, we utilized the “timeROC” (0.4) and “survival” (3.5-5) R packages to compute the area under the receiver operating characteristic (ROC) curve (AUC). We subsequently conducted a univariate Cox regression analysis, employing the progression-free interval (PFI) and overall survival (OS) as criteria for discerning prognostic genes. The predetermined threshold for the AUC was set at ≥ 0.7 for the median survival time, along with an HR > 1 and a P value < 0.05 in Cox regression. To assess the prognostic predictive capacity of the candidate module genes, we concurrently calculated the AUC in three GEO adult ACC cohorts. Genes with an AUC ≥ 0.7 in all three cohorts were filtered, identifying FSCN1 as a prognostic gene for the PFI and OS. The prognostic ability of FSCN1 was validated via Kaplan-Meier (KM) survival curves (P < 0.05) [20].
Immunohistochemistry (IHC) analysis
The ACC and adrenal adenoma tissue samples used in this study were procured from Huizhou Municipal Central Hospital, with ethical approval granted by the Ethics Committee of Huizhou Municipal Central Hospital. The samples were fixed in 4% polyformaldehyde before being embedded in paraffin. Each tissue block was sliced into 4-um-thick sections, treated with a 1% H2O2 solution, and then blocked with non-immune goat serum. The sections were incubated with a primary antibody overnight at 4℃ and then incubated with a biotin secondary antibody at room temperature for 30 min to bind with the primary antibody. The primary antibodies used were an anti-FSCN1 antibody (YT5212; Immunoway, Shanghai, China) and an anti-CD8a antibody (YM8067; Immunoway). The biotin secondary antibody used were UltraSensitive™M SP (mouse/rabbit) IHC Kit (KIT-9720; MXB Biotechnologies, Fuzhou, China). The specific IHC staining procedure aligns with that used in our previous study [21]. The final immunoreactive score (IRS) was obtained by adding the scores for the percentage of positively stained cells and the staining intensity. The scoring definitions for the percentage of immunor- eactive cells were as follows: 0 (0%), 1 (1%-10%), 2 (11%-50%), and 3 (> 50%). Visual scoring and stratification for staining intensity were performed as follows: 0 (negative), 1 (weak), 2 (moderate), and 3 (strong) [22].
Cell lines and cell transfection
The ACC cell lines SW-13 and H295R were acquired from Procell Biotechnology (Wuhan, China). SW-13 cells were cultured in DMEM (BC-M-005; Bio-Channel, Nanjing, China) supplemented with 10% fetal bovine serum (A5256701; Gibco, Carlsbad, USA) and 1% penicillin-streptomycin (15140122; Gibco). H295R cells were maintained in DMEM/F12 (BC-M-002; Bio-Channel) enriched with 10% fetal bovine serum (Thermo Fisher Scientific, Waltham,
USA), 1% ITS-G (41400045; Gibco), and 1% penicillin-streptomy- cin (15140122; Gibco). The cells were incubated at 37°℃ in a 5% CO2 atmosphere. For the transfection experiments, a negative control (NC) and FSCN1-siRNA#1/2/3 (Genepharma, Suzhou, China) were used according to the manufacturer’s instructions with siRNA-mate (Genepharma). The sequences of the siRNAs used were as follows: FSCN1-siRNA#1: 5’-GAGCAAAGAGCUCGUCCUU dTdT-3’; FSCN1-siRNA#2: 5’-GGCGUCCAAUGGCAAGUUUdTdT-3’; FSCN1-siRNA#3: 5’-CGACUAUAACAAGGUGGCCAUdTdT-3’; siNC: 5’-UUCUCCGAACGUGUCACGUdTdT-3’.
Western blot analysis
The cells were collected and lysed in RIPA buffer containing protease inhibitors. The extracted protein samples were separated by SDS-PAGE and then transferred onto polyvinylidene fluoride (PVDF) membranes (IPVH00010; Millipore, Billerica, USA). The membranes were subsequently blocked with a 5% non-fat milk solution for 1 h. The membranes were subsequently incubated overnight at 4℃ with primary antibodies against FSCN1 (YT5212; Immunoway) and GAPDH (YN5585; Immunoway). The mem- branes were then incubated with horseradish peroxidase-conju- gated secondary antibodies against rabbit (BA1055; Boster, Wuhan, China) or mouse IgG (BA1051; Boster). Each experiment was conducted in triplicate. The bands were visualized via a chemilu- minescence imaging system (CLINX ChemiScope Touch, Shanghai, China) and quantified via ImageJ software.
Transwell assay
Approximately 4 × 104 transiently transfected cells were cultured in 200 µL of serum-free medium in the upper chamber of a transwell apparatus (353097; Corning, Steuben County, USA), while 500 uL of complete medium was placed in the lower chamber. The cells were then incubated at 37°℃ in a 5% CO2 environment for 48 h. After incubation, the cells that invaded to the lower surface of the membrane were washed with PBS, fixed with paraformaldehyde, and stained with 0.1% crystal violet solution. Finally, images of the cells were captured under an inverted microscope, and the stained cells were counted.
Cell counting kit-8 (CCK8) assay
Transfected cells were seeded at a density of 1000 cells per well in a 96-well plate. At 24, 48, and 72 h post-seeding, complete medium containing 10% CCK-8 solution (C0042; Beyotime Biotechnology, Shanghai, China) was added to each well. The absorbance at 450 nm was measured via an iMark microplate reader (Bio-Rad, Hercules, USA) after an additional 2-h incubation. Each transfection was performed in quintuplicate.
Gene set enrichment analysis
To confirm the biological characteristics of FSCN1, we calculated its correlation with other mRNAs via the TGGA-ACC mRNA expression matrix. The sorted mRNA list was input into a gene set enrichment analysis (GSEA) to investigate whether genes strongly correlate with FSCN1 aggregated in meaningful functional pathways.
Estimation of tumor microenvironment-infiltrating cells
The IOBR R package integrates multiple open-source tumor microenvironment (TME) deconvolution algorithms [23]. We used four TME deconvolutions (TIMER, MCPcounter, xCell, CIBER-
SORT) of the IOBR R package (0.99.9) (https://sourceforge.net/ projects/estimateproject/) to estimate the immune cell infiltration score in TCGA-ACC. We correlated the expression level of FSCN1 in each sample with its immune infiltration score to explore the relationship between FSCN1 and the infiltration level of immune cells in the TME.
Copy number alteration analysis
For the analysis of copy number alterations, we employed the GISTIC 2.0 platform available on the GISTIC online analysis website (https://www.genepattern.org/) to pinpoint significantly amplified and deleted genomic regions. To visualize the mutational landscape of FSCN1 expression, we highlighted the top 20 genes on the basis of copy number amplifications and deletions. We examined the expression variations of each gene across different FSCN1 expres- sion groups to elucidate the involvement of FSCN1 in copy number alterations (CNAs).
Tumor xenografts in nude mice
Female BALB/c-nude mice (5 weeks old) were obtained from GemPharmatech (Nanjing, China). SW-13 cells (5.5 x 106) were injected subcutaneously into the right axilla. On day 15 post- injection, NP-G2-044 dissolved in DMSO was given once a day by gavage at 100 mg/kg per mouse. The control solvent was given to the control group of mice once a day by gavage. On day 30 post- injection, the mice were euthanized, and the tumor tissues were dissected. The primary tumor volume was calculated as length x width2 x 7/6. The animal experiment was approved by the ethics committee of Huizhou Central Hospital (No. KYLL2023105).
Statistical analysis
Data are presented as the mean ± standard deviation (SD) and were analyzed via GraphPad Prism version 9.0 (GraphPad Software, La Jolla, USA). To evaluate normalized data, Student’s t test was employed. The correlation between FSCN1 and the expression of other genes, as well as its relationship with immune cell infiltration levels in the tumor microenvironment, was assessed via Spear- man’s correlation test. Differences between two groups were determined via the Wilcoxon rank-sum test. The Kaplan-Meier (KM) method was used to estimate the survival curves. All the statistical tests were two-sided, and a P value of less than 0.05 was considered statistically significant.
Results
Identification of the brown module as a key component associated with prognosis via WGCNA
Initially, in an attempt to identify prognostic biomarkers related to ACC, we chose the TCGA-ACC cohort. All the samples were divided into control and progression groups on the basis of the median PFI (Figure 2A). We chose a soft-thresholding power of b = 16 to achieve a scale-free network and further transformed the adjacency matrix into a topological overlap matrix (Supplementary Figure S1). After the gene characteristics of the module were calculated, highly co-expressed genes were clustered into the same module via the dynamic tree cutting method (Figure 2B). The results revealed a total of seven co-expression modules. A heatmap of module feature correlations revealed that the brown module is highly correlated with the progression group of the ACC (Figure 2C). Similar results were also shown in the scatter plot of gene significance and module
membership (Figure 2D). Therefore, the 4323 genes in the brown module were defined as ACC progression-related genes.
To explore the biological functions of ACC progression-related genes, we performed GO and KEGG enrichment analyses. GO analysis revealed that ACC progression-related genes are enriched primarily in the regulation of DNA replication, chromosome separation, RNA splicing, and nuclear division. Moreover, the
KEGG analysis results indicated that these mRNAs are predomi- nantly involved in the cell cycle, spliceosome, ubiquitin-mediated protein hydrolysis, and ribosome pathways. Overall, the results of the GO and KEGG analysis suggested that ACC progression-related genes might play roles in tumorigenesis, development, and the proliferation and division of cancer cells (Figure 2E,F and Supplementary Tables S3 and S4).
A
Sample dendrogram and trait heatmap
D
Module membership vs. gene significance
140
0.6
120
Height
TCGA-OR-A5JO-01A
TOGA-OR-ASKO-01A
100
TCGA-OR-A5J8-01A
TCGA-P6-A5OG-01A
TCGA-OR-A5J5-01A
TCGA-OR-AD
TCGA-OR-A5L3-01A
TCGA-OR-A5LD-01A
TEGA-OR-A5KV-01A
TCGA-OR-ASLL-01A
TCGA-OR-A5KX-01A
PGA-OR-A5IR-01A
TCGA-OR-AS.IV-01A
TCGA-PK-A5H8-01%
CEL-OR-ASK3-014
TCGA-OR-A5K3-01A
0.4
TOGA-OR-AS IK-01A
TCGA-OR-ABANT TCGA-OR-AMOJA
TOGATOTTVIA
TCGA-OR-ASJD-01/
TCGA-OR-A5JZ-01A
TCGA-ORTASLP-01A
TCGA-PK-ASHA-
TCGA-OR ADLS-01A
TOGA PIEDI
TCGA-OR-A5LK-01A
TCGA-OR-ASJI-01A
TCGA-OR-A5LN-01A
TCGA-OR-A5JY-01A
TOGA-OR-A5KZ-01A
TCGA-PK-A5HB-01A
TCGA-OR-A5J1-01A
TCGA-OR-A5K9-01A
TCGA-OR-A5LC-01A
TCGA-OR-A5JS-01A
TCGA-OR-ASIM-DI
TČGA-OR-A5K6-01A
TČGA-OR-A5LE-01A
YOGA-OR-ASK5-01A
TCGA-OR-A5J7-01A.
TCGA-OR-AGIE-OTA
TCGA-OR-A5K2-01A
TCGA-OR-A5KU-01/
TCGA-ORSPEKT-
TCGA-OR-A5LJ-01A
TCGA-OR-ADJE-01A
TCGA-OR-A5JL-01A
TCGA-ORTASLO-01
TCGA-OR-A5KO-01A
TCGA-OU-A5PI-01A
TCGA-OR-A5JF-01A
TCGA-P6-A5OF-01A
TCGA-OR-A5JW-01A
TCGA-OR-A5LB-01A
TCGA-OR-A5JG-01A
TOGA-OR-A5KY-01A
TCGA-OR-A5J3-01A
TCGA-OR-A5JA-01A
TCGA-OR-A5KW-01A
TOGA-OR-A5J9-01/
TCGA-OR-A5JJ-01A
Gene significance
60
0.2
0.0
cor=0.61,
p<1e-200
Progression
0.3
0.5
0.7
0.9
Module Membership in brown module
B
Cluster Dendrogram
E
GO
1.0
p.adjust
organelle fission
4.9e-42
ribonucleoprotein complex
biogenesis
0.8
3.4e-25
nuclear division
ncRNA metabolic process
Height
6.9e-25
0.6
1.0e-24
RNA splicing
chromosome segregation
1.4e-24
ncRNA processing
0.4
Count
mitotic nuclear division
DNA replication
100
0.2
nuclear chromosome segregation
125
ribosome biogenesis
Module colors
150
double-strand break repair
175
sister chromatid segregation
Merged colors
mitotic sister chromatid
200
segregation
DNA-dependent DNA replication
0.05
0.04
0.03
GeneRatio
C
Module-PFI relationships
F
KEGG
Cell cycle
MEgrey
MEbrown
MEblack
MEyellow
MEgreen
MEred
MEpink
p.adjust
Spliceosome
Ubiquitin mediated proteolysis
2.5e-05
Ribosome
Oocyte meiosis
5.0e-05
Progression
Nucleocytoplasmic transport
7.5e-05
mRNA surveillance pathway
-0.05(0.7)
0.69(2e-09)
0.29(0.02)
-0.25(0.05)
0.18(0.2)
0.18(0.2)
-0.21(0.1)
RNA degradation
Fanconi anemia pathway
Count
DNA replication
20
Nucleotide excision repair
40
Homologous recombination
60
Base excision repair
Protein export
-1
-0.5
0
0.5
1
Mismatch repair
0.04
0.03
0.02
0.01
GeneRatio
ATAC-seq analysis screens out different peak genes
ATAC-seq is a technique that explores the openness of genes upstream and downstream of binding sites by studying chromoso- mal accessibility. Using ATAC-seq data from TCGA-ACC, we analyzed the differences in chromatin accessibility by dividing the ATAC-seq samples into progression and control groups. The ACC progression group contained ten samples, whereas the ACC control group contained four samples (Supplementary Figure S2A). After differential analysis, a total of 23,303 DAPs were obtained. We subsequently used the R package “ChIPseeker” to annotate these DAPs and all peaks. The results revealed that the percentage of promoter elements < 1 kb in length in the DAPs was lower than that in all the peaks (Supplementary Figure S2B). This result indicates that promoter elements < 1 kb in length are more sensitive to ACC progression. Additionally, using the ‘peak-gene’ link in this study, we obtained a total of 86 DPGs corresponding to the DAPs (Supplementary Figure S2C,D).
Cox regression and ROC curve validation of ACC prognosis-associated biomarkers
The mRNA expression data from the TCGA-ACC cohort were selected on the basis of the recommendations for ACC with OS and PFI as clinical endpoints for univariate Cox regression and calculation of the AUC. We used the following criteria: Cox regression HR>1, P value <0.05, and AUC≥0.7. Finally, by intersecting the genes identified by Cox regression and ROC curve analysis, with a focus on OS and PFI prognosis, we identified 22 overlapping genes associated with poor prognosis in ACC patients from the WGCNA hub genes and DPGs (Figure 3A). To refine the selection of ACC progression-related genes, AUC value heatmaps were generated from data from three independent adrenocortical carcinoma cohorts: GSE10927, GSE19750, and a meta dataset from two GEO cohorts, with the latter compiled after batch effect removal (Figure 3B and Supplementary Figure S3). With the screening criterion of an AUC > 0.7, the gene FSCN1 was highlighted. These results underscore the potential of FSCN1 as a robust and reliable prognostic biomarker for ACC progression. In addition, univariate Cox regression analyses of the TCGA and GEO datasets revealed that FSCN1 could significantly predict patient prognosis (Figure 3C, D). After the effects of age, sex, clinical stage, and pathological stage were excluded from multivariate Cox regression, FSCN1 was identified as an independent factor affecting the prognosis of ACC patients (Figure 3E).
FSCN1 with outlier chromatin accessibility is associated with an unfavorable prognosis in ACC patients
To confirm the association between FSCN1 expression and unfavorable prognosis in ACC patients, we performed Kaplan- Meier (KM) survival analysis. KM survival analyses across the TCGA-ACC, GSE10927, GSE19750, and meta cohorts suggested that elevated FSCN1 expression predicts worse OS and PFI outcomes (Figure 4A-E). Notably, our prior differential peak analysis revealed that FSCN1 upregulation in the ACC progression group was accompanied by increased chromatin accessibility and mutation rates. We subsequently illustrated the chromatin accessibility within the peak regions linked to FSCN1. Each of the regions corresponding to FSCN1, namely, ACC_33807 to ACC_33817, displayed increased chromatin accessibility in the ACC progression group compared with the control group (Figure 4F,G). These
findings suggest an association between increased chromatin accessibility of FSCN1 and adverse prognostic outcomes in ACC.
Experimental validation of FSCN1
We conducted IHC assays on ACC samples and noncancerous adrenal gland samples, which produced representative images of the IHC results. FSCN1, an actin-binding protein, is expressed in the adrenal cell cytoplasm and membrane. The expression of FSCN1 in ACC samples was significantly greater than that in non-cancerous adrenocortical samples from the control group (Figure 5A).
To investigate the function of FSCN1 in vitro, specific siRNAs targeting FSCN1 (si-FSCN1#1/2/3) and a negative control siRNA were transfected into the ACC cell lines SW-13 and H295R. The knockdown efficiency of si-FSCN1#1/2/3 was confirmed by western blot analysis. si-FSCN1#2/3 was selected as the most effective siRNA for subsequent experiments (Figure 5B). CCK-8 and Transwell assays were then performed, and the results demonstrated that FSCN1 knockdown significantly inhibited the invasiveness and viability of both the SW-13 and H295R cell lines (Figure 5C,D).
To assess the role of FSCN1 in ACC in vivo, we established an ACC xenograft tumor model by subcutaneously injecting SW-13 cells into BLAB/c nude mice. The FSCN1 inhibitors NP-G2-044 and placebo were administered to the experimental group and the control group, respectively, to observe the differences in tumor growth. Consistent with expectations, xenograft growth was slower in the mice treated with NP-G2-044 (Figure 5E,F). The tumor weight in this group was also lower than that in the control group (Figure 5G).
Biological characteristics of FSCN1
On the basis of the correlation of FSCN1 with other genes, functional enrichment analysis was used to explore pathways involving FSCN1 (Supplementary Tables S5 and S6). As shown in Figure 6A,B, the significantly enriched genes are highly correlated with FSCN1, including the activation of pathways such as cytoplasmic translation, mitotic sister chromatid segregation, mRNA processing, mitotic nuclear division, and chromosome segregation, and the inhibition of pathways such as leukocyte cell-mediated cytotoxicity, the MHC class II protein complex, B-cell- mediated immunity, positive regulation of leukocyte proliferation and the humoral immune response. The KEGG pathways shown in Figure 6C,D include the activation of pathways such as steroid biosynthesis, proteasome, ribosome, spliceosome, and DNA repli- cation; the inhibition of pathways such as complement and coagulation cascades, the metabolism of xenobiotics by cytochrome P450, antigen processing and presentation, and drug metabolism by cytochrome P450.
FSCN1 expression correlates with immune-related characteristics
On the basis of the pathways enriched with FSCN1-related genes suggested by GSEA, we hypothesized that the expression of FSCN1 promotes ACC progression by regulating immune cells in the tumor immune microenvironment. We carried out a correlation analysis of FSCN1 expression and tumor immune cell infiltration (Figure 7A and Supplementary Table S7). After that, we summarized the associations of FSCN1 expression with immune cell infiltration scores via CIBERSORT, TIMER, MCPcounter, and xCell (Supple- mentary Figure S4). With respect to the highly enriched MHC class II protein complexes and B-cell-mediated immunity identified via
A
B
C
DPGs
FSCN1
0.757
0.809
0.733
WGCNA
ZIC2
0.715
0.650
0.608
AUC_OS(Median)>0.7
VAV2
0.611
0.704
0.634
0
98
5
URB2
0.632
0.719
0.642
111
1
258
SMNDC1
0.736
0.710
0.697
10
126
SLIT2
0.764
0.586
0.544
22
11
SIM2
0.403
0.585
21
AUC_PFI(Median)>0.7
PLPP2
523
22
0.566
0.765
0.568
74
0
PLCL2
0.729
0.737
0.641
5
30
10
NSG1
0.500
0.709
0.529
928
NFATC4
0.785
0.616
0.646
757
527
155
LEF1
0.698
0.639
0.603
PFI COX HR>1
KHNYN
0.639
0.570
0.546
KDM5A
0.681
0.559
0.460
OS COX HR>1
IGDCC4
0.535
0.674
0.580
HPCAL1
0.701
0.717
0.648
GATA3
0.740
0.652
0.685
FAM171A1
0.639
0.604
0.516
DZIP1
0.767
0.580
0.532
CTNNB1
0.625
0.514
0.516
BCL11A
0.615
0.844
0.674
ABCB4
0.729
0.592
0.676
Cohort
AUC value
Condition
GSE10927
1.0
Risky
GSE19750
0.9
Not significance
Meta Adult
None
0.8
0.7
E
Multivariate COX
TT
n
-1
0
1
2
3
4
log2(Hazard Ratio(95%CI))
| D | Univariate COX | ||||
|---|---|---|---|---|---|
| Variable | pvalue | HR | L95CI | H95CI | |
| TCGA-ACC | . | ||||
| Age | 3.79E-01 | 1.01 | 0.99 | 1.04 | 0 |
| Gender | 9.99E-01 | 1.00 | 0.47 | 2.14 I- | -I |
| Stage | 9.39E-06 | 2.63 | 1.71 | 4.03 | I- 1 |
| clinical_M | 5.14E-05 | 5.30 | 2.36 | 11.88 | - - 1 |
| pathologic_N | 2.48E-01 | 1.74 | 0.68 | 4.47 F | 1 |
| pathologic_T | 9.28E-07 | 3.04 | 1.95 | 4.74 | I- -I |
| FSCN1 | 2.26E-05 | 1.76 | 1.36 | 2.29 | I |
| GSE10927 | |||||
| Age | 5.19E-01 | 1.01 | 0.98 | 1.05 | . |
| Gender | 4.75E-01 | 1.47 | 0.51 | 4.24 1- | I |
| Stage | 2.09E-02 | 1.82 | 1.09 | 3.02 | -l. -1 |
| FSCN1 | 4.67E-02 | 4.45 | 1.02 | 19.38 | I 1 |
| GSE19750 | |||||
| Age | 9.73E-02 | 1.04 | 0.99 | 1.08 | |
| Gender | 6.50E-01 | 1.26 | 0.46 | 3.44 I- | -l |
| Stage | 8.73E-01 | 1.03 | 0.74 | 1.43 F | 1 |
| FSCN1 | 1.08E-02 | 1.58 | 1.11 | 2.25 | '+ 1 |
| Meta Adult | |||||
| Age | 1.72E-01 | 1.02 | 0.99 | 1.04 | . |
| Gender | 7.54E-01 | 1.12 | 0.55 | 2.26 I- | 1 |
| Stage | 3.73E-01 | 1.11 | 0.88 | 1.41 | |
| FSCN1 | 1.83E-03 | 1.67 | 1.21 | 2.31 | 1 |
| Variable | pvalue | HR | L95CI | H95CI | |
|---|---|---|---|---|---|
| TCGA-ACC | |||||
| Stage | 2.82E-01 | 0.51 | 0.15 | 1.75 | + |
| clinical_M | 3.67E-01 | 2.07 | 0.43 | 10.10 | I- -I |
| pathologic T | 3.21E-03 | 3.34 | 1.50 | 7.43 | 1- -1 |
| FSCN1 | 6.80E-04 | 1.68 | 1.24 | 2.26 | |
| GSE10927 | . | ||||
| Stage | 1.01E-02 | 2.05 | 1.19 | 3.55 | '+ |
| FSCN1 | 2.37E-02 | 7.87 | 1.32 | 47.04 | · I |
| GSE19750 | . | ||||
| Age | 4.15E-02 | 1.04 | 1.00 | 1.09 | |
| FSCN1 | 6.23E-03 | 1.79 | 1.18 | 2.73 | .I |
| Meta Adult | |||||
| Age | 6.44E-02 | 1.02 | 1.00 | 1.05 | |
| FSCN1 | 1.09E-03 | 1.78 | 1.26 | 2.52 | 101 |
| -2 -1 0 1 2 3 4 5 log2(Hazard Ratio(95%CI)) | |||||
Figure 3. Discovery and appraisal of the prognostic value of FSCN1
(A) Venn diagram outlining genes linked to unfavorable prognosis. (B) AUC heatmaps for ROC curves across the GSE10927, GSE19750, and meta-datasets. (C) Cox regression analysis across the GSE10927, GSE19750, and meta-datasets. (D,E) Outcomes of both univariate (D) and multivariate (E) Cox regression analyses of OS in the TCGA-ACC, GSE10927, GSE19750, and combined meta-cohorts.
A
B
C
GSE10927
GSE19750
Meta
1.0
FSCN1
1.0
FSCN1
1.0
FSCN1
Low
Low
Low
Survival probability
0.8
High
Survival probability
0.8
High
Survival probability
0.8
High
0.5
0.5
0.5
0.3
0.3
0.3
0.0
logrank test p=0.03
0.0
logrank test p=0.01
0.0
logrank test p=3.3e-4
Number at risk
Number at risk
Number at risk
Low
8
3
1
1
1
Low
9
6
4
3
1
Low
18
11
5
3
1
High
16
2
1
1
1
High
12
3
1
1
1
High
27
2
1
1
1
0
37
74
111
148
0
54
108
162
216
0
54
108
162
216
OS.time(Months)
OS.time(Months)
OS.time(Months)
D
E
F
TCGA-ACC
TCGA-ACC
1.0
FSCN1
Low
1.0
FSCN1
Low
Survival probability
0.8
High
Survival probability
0.8
High
0.5
0.5
0.3
0.3
0.0
logrank test p=2.0e-9
0.0
logrank test p=7.9e-13
Number at risk
Number at risk
Low
50
29
12
4
1
Low
46
23
10
4
1
High
29
12
3
2
1
High
33
1
1
1
1
0
38
76
114
152
OS.time(Months)
0
38
76
114
152
PFI.time(Months)
G
chr7
5,600 kb
TCGA-PK-A5H8-01A
TCGA-OR-A5J6-01A
TCGA-OR-A5J2-01A
TCGA-OR-A5J3-01A
TCGA-OR-A5J9-01A
TCGA-OR-A5JZ-01A
TCGA-OR-A5KX-01A
Control
Progression
Refseq Genes
FSCN1
XM_017012908.1
| Peak_id | Range | LogFC | adj.P.Val | |
|---|---|---|---|---|
| ACC_33807 | 5593335~5593836 | 1.082 | 5.55E-02 | |
| ACC_33808 | 5594070~5594571 | 2.105 | 1.30E-03 | |
| ACC_33809 | 5594601~5595102 | 1.798 | 1.43E-03 | |
| ACC_33810 | 5595480~5595981 | 1.927 | 1.51E-03 | |
| ACC_33811 | 5596006~5596507 | 2.229 | 3.68E-05 | |
| ACC_33813 | 5598752~5599253 | 1.062 | 8.27E-02 | |
| ACC_33814 | 5603706~5604207 | 1.590 | 2.75E-02 | |
| ACC_33816 | 5607141~5607642 | 1.540 | 3.40E-02 | |
| ACC_33817 | 5608168~5608669 | 1.555 | 3.04E-02 |
Figure 4. Survival analysis via KM estimation and ATAC-seq profiling of FSCN1 (A-E) KM survival estimates for groups with high and low FSCN1 expression across the TCGA-ACC, GSE10927, GSE19750, and combined meta-cohorts. (F) ATAC-seq tracks focused on the FSCN1 locus in the ACC. (G) Differentially accessible peaks related to FSCN1 are emphasized in green, accompanied by peak IDs (ranging from ACC_33807 to ACC_33817).
A
10X
20X
12.5*
non- Tumor
10.0·
Immunoreactive Score.of.FSCN1
7.5.
5.0·
10X
2.5.
ACC
0.0-
non-Tumor
ACC
B
Relative FSCN1 expression
1.0
SW-13
Relative FSCN1 expression
1.5-
H295R
SW-13
H295R
0.8
ns
*
NC
si-1
si-2
si-3
NC
si-1
si-2
si-3
1.0-
0.6
*
FSCN1
**
54kDa
GAPDH
0.4-
37kDa
0.5-
0.2
C
NC
si-2
si-3
0.0
0.0
NC
Si-1
Si-2
Si-3
NC
Si-1
Si-2
Si-3
500
SW-13
200
H295R
SW-13
Cell numbers
400-
Cell numbers
150
*
**
300-
*
*
100-
200-
H295R
100-
50-
D
SW-13
E
0
0
NC
Si-2
Si-3
NC
Si-2
Si-3
2.5
si-NC
2.0-
si-2
OD(450nm)
1.5-
si-3
Control
1.0-
NP-G2-044
0.5-
0.0
0
1 cm
2
3
4
5
6
7
8
9
10
11
12
13
14
15
0
1
2
3
days
H295R
1.5
F
G
si-NC
*
2500-
2.0
si-2
Control
·
OD(450nm)
*
Tumor volume(mm3)
1.0-
si-3
2000-
NP-G2-044
Tumor weight(g)
1.5-
1500-
0.5
1000-
1.0-
500-
0.5-
0.0
0
1
2
3
0
0.0
15
18
21
24
27
30
Control
NP-G2-044
days
days
A
C
activated
suppressed
activated
suppressed
GOCC MHC CLASS II PROTEIN COMPLEX
KEGG ALLOGRAFT REJECTION
GOBP CYTOPLASMIC TRANSLATION -
KEGG STEROID BIOSYNTHESIS
GOBP MITOTIC SISTER CHROMATID SEGREGATION -
KEGG GRAFT VERSUS HOST DISEASE
GOBP ANTIGEN PROCESSING AND PRESENTATION OF
EXOGENOUS PEPTIDE ANTIGEN
Gene Set
KEGG PROTEASOME
GOBP MRNA PROCESSING-
50
KEGG RIBOSOME
Gene Set
GOBP NUCLEAR CHROMOSOME SEGREGATION -
100
KEGG TERPENOID BACKBONE BIOSYNTHESIS
20
GOBP MITOTIC NUCLEAR DIVISION
150
KEGG SPLICEOSOME
40
GOBP CHROMOSOME SEGREGATION
GOBP ANTIGEN PROCESSING AND PRESENTATION OF
200
KEGG DNA REPLICATION -
60
EXOGENOUS ANTIGEN
KEGG TYPE I DIABETES MELLITUS
250
80
GOBP CHROMATIN REMODELING
KEGG COMPLEMENT AND COAGULATION CASCADES
GOBP MITOTIC CELL CYCLE PHASE TRANSITION
KEGG INTESTINAL IMMUNE NETWORK FOR IGA PRODUCTION -
p.adjust
GOBP DNA REPLICATION
p.adjust
KEGG CELL CYCLE
GOBP NCRNA PROCESSING
KEGG DRUG METABOLISM CYTOCHROME P450-
0.0020
GOBP ANTIGEN PROCESSING AND PRESENTATION -
KEGG AUTOIMMUNE THYROID DISEASE-
0.0015
GOBP LEUKOCYTE MEDIATED CYTOTOXICITY
1.292768e-08
KEGG RNA DEGRADATION -
0.0010
GOBP B CELL MEDIATED IMMUNITY
KEGG HEMATOPOIETIC CELL LINEAGE
0.0005
GOBP ANTIGEN PROCESSING AND PRESENTATION OF
PEPTIDE ANTIGEN
KEGG METABOLISM OF XENOBIOTICS BY CYTOCHROME P450-
GOBP POSITIVE REGULATION OF LEUKOCYTE
PROLIFERATION
KEGG OOCYTE MEIOSIS
GOBP REGULATION OF LEUKOCYTE PROLIFERATION -
KEGG NUCLEOTIDE EXCISION REPAIR
B
GOBP HUMORAL IMMUNE RESPONSE
D
KEGG ANTIGEN PROCESSING AND PRESENTATION -
0.4 0.5 0.6 0.7 0.8
0.4 0.5 0.6 0.7 0.8
0.3 0.4 0.5 0.6 0.7
0.3 0.4 0.5 0.6 0.7
GeneRatio
GeneRatio
GOBP_CHROMATIN_REMODELING
GOBP_CHROMOSOME_SEGREGATION
KEGG_CELL_CYCLE
GOBP_CYTOPLASMIC_TRANSLATION
KEGG_RIBOSOME
0.4
GOBP_DNA_REPLICATION
KEGG_SPLICEOSOME
0.4
Running Enrichment Score
Running Enrichment Score
0.0
0.0
KEGG_ALLOGRAFT REJECTION
GOBP_ANTIGEN_PROCESSING_AND_PRESENTATION
GOBP_ANTIGEN_PROCESSING_AND_PRESENTATION OF_EXOGENOUS ANTIGEN
KEGG_ANTIGEN_PROCESSING AND PRESENTATION
GOBP_ANTIGEN_PROCESSING_AND_PRESENTATION_OF_EXOGENOUS_PEPTIDE_ANTIGEN
KEGG_AUTOIMMUNE_THYROID DISEASE
-0.4
-0.4
GOBP_ANTIGEN_PROCESSING_AND_PRESENTATION_OF_PEPTIDE_ANTIGEN
KEGG_DRUG_METABOLISM_CYTOCHROME_P450
KEGG_GRAFT_VERSUS_HOST_DISEASE
GOBP_B_CELL_MEDIATED_IMMUNITY
KEGG_HEMATOPOIETIC_CELL_LINEAGE
GOBP_HUMORAL_IMMUNE_RESPONSE
-0.8
KEGG_TYPE_I_DIABETES_MELLITUS
V
IL
11
1
Ranked List Metric
I
I
1.0
Ranked List Metric
1.0
0.5
0.5
0.0
0.0
0.5
0.5
5000
Rank in Ordered Dataset
10000
15000
5000
10000
Rank in Ordered Dataset
15000
GSEA, we performed an analysis of the correlation between CD4+ T- cell and B-cell immune infiltration scores and FSCN1 expression (including scatter plots and violin plots). However, the results indicated that the immune infiltration scores of CD4+ T cells and B cells are only correlated with FSCN1 expression according to one or two algorithms (Supplementary Figure S5). In contrast, the immune infiltration scores of CD8+ T cells are negatively correlated with high FSCN1 expression across all four algorithms (Figure 7B,C). There- fore, we explored the relationship between FSCN1 expression and CD8+ T-cell infiltration. We found that CD8+ T cells are negatively correlated with FSCN1 expression in all four algorithms, and both the Spearman correlation analysis and the Mann-Whitney U test, which is based on the median FSCN1 expression, are statistically significant (Figure 7B,C). The infiltration of CD8+ T cells in tumors is generally considered indicative of a good prognosis [24]. To verify the role of FSCN1 in immune cells, we conducted CD8+ IHC staining on five ACC tissues from the Central Hospital of Huizhou and compared the results with those from the corresponding FSCN1 IHC sections. We found that the CD8+ IRS is significantly negatively correlated with the FSCN1 IRS (P= 0.0012) (Figure 7D,E). These findings imply that high FSCN1 expression may lead to poor prognosis through the inhibition of CD8+ T cell infiltration in the
ACC microenvironment.
FSCN1-related mutation and CNV
We subsequently used multi-omics data to explore the role of FSCN1 in ACC. At the mutation level, we plotted the mutation conditions of the top 20 frequently mutated genes (FMG) in ACC patients and generated CNV landscape maps. We detected 20 commonly mutated genes, namely, TP53 (17%), CTNNB1 (16%), MUC16 (16%), TTN (11%), CNTNAP5 (8%), HMCN1 (8%), PKHD1 (8%), APOB (7%), KMT2B (7%), NF1 (7%), PRKAR1A (7%), SVEP1 (7%), TUT7(7%), ASXL3 (5%), MEN1 (5%), CMYA5 (5%), FRAS1 (5%), LRP1 (5%), STAB1 (5%), and ZNRF3 (4%) (Figure 8). We found that TP53 has the highest mutation frequency in ACC (17%), and the mutation frequency in the high FSCN1 expression group is greater than that in the low FSCN1 expression group. Genes such as TP53, CTNNB1, TTN, PKHD1, APOB, KMT2B, NF1, and SVEP1 also present higher mutation frequencies in the high-expression group than in the low-expression group. We also depicted the CNV status of the top 6 AMP and Homdel chromosomal segments between the two FSCN1 expression groups, and deletions at 4q35.1 and 4q34.3 differ between the high FSCN1 and low FSCN1 groups.
A
N
T
Infiltration Score
FSCN1:
N:
T:
Stage:
Stage
Low
NO
T1
stage i
Gender Age
-1.0 -0.5 0.0 0.5 1.0
High
N1
T2
stage ii
T3
stage iii
FSCN1
T4
stage iv
T_cells_CD8 ***
T_cells_CD4_memory_restin
Gender:
Age:
female
CIBERSOFT
T_cells_follicular_helper **
male
T_cells_gamma_delta **
203040506070
TIMER
NK_cells_resting **
CIBERSORT
B
0.01
R = - 0.44, p = 5.5e-05
10.0
NK_cells_activated ****
R = - 0.36, p=0.0012
Monocytes_CIBERSORT
7.51
7.51
Macrophages_MO ****
Macrophages_M1 ***
FSCN1
FSCN1
5.01
5.01
Macrophages_M2
Dendritic_cells_activated ***
2.51
2.51
Eosinophils **
0.0
0.01
MCPcounter
T_cells
0.17
0.20
0.23
0.0
CD8+ T Cell
0.1
0.2
CD8+ T
0.3
CD8_T_cells ***
Cytotoxic_lymphocytes *** NK_cells*
MCPCounter
XCell
10.0
R =- 0.39, p=0.00046
10.0
R= - 0.4, p= 0.00024
Monocytic_lineage **
Myeloid_dendritic_cells’
7.51
7.51
Fibroblasts **
FSCN1
FSCN1
5.01
5.01
B_cell ****
TIMER
T_cell_CD4
2.51
2.51
T_cell_CD8 ****
Macrophage *** DC*
0.0
0.0
0.0
0.5
1.0
1.5
2.0
2.5
3.0
0.00
0.05
0.10
0.15
CD8+ T
CD8+ T
CD4 ._ memory_T.cells ***
C
0.23
Wilcoxon, p = 1e-05
Wilcoxon, p = 0.0035
CD4 ._ naive_T.cells ***
T_cell_CD8_TIMER
T_cells_CD8_CIBERSORT
0.22
0.3
CD4 ._ Tem*
CD8 ._ naive_T.cells **
0.21
0.2
CD8 ._ T.cells ***
0.20
CD8 ._ Tcm **
0.19
0.1
XCELL
cDC **** DC ***
0.18
0.0
Endothelial_cells*
Low FSCN1 High FSCN1
Low FSCN1 High FSCN1
Fibroblasts ***
CD8_T_cells_MCPcounter
Wilcoxon, p = 0.00015
0.15
Wilcoxon, p = 0.0014
Macrophages **
Macrophages_M1 **
CD8 ._ T.cells_xCell
Macrophages_M2’
2
0.10
NK_cells
NKT ****
0.05
Th1_cells ****
1
Th2_cells ***
Tregs
0
Low FSCN1 High FSCN1
0.00
Low FSCN1 High FSCN1
D
FSCN1 IRS = 12
FSCN1 IRS = 6
E
CD8+ Immunoreactive Score
10
R =- 0.98, p=0.0012
665035
8
CD8+
553680
6
628360
662203
4
2-
568594
FSCN1
0
0
5
10
15
FSCN1 Immunoreactive Score
Low
High
FSCN1
17%
TP53
16%
CTNNB1
16%
MUC16
11%
TTN
8%
CNTNAP5
8%
HMCN1
8%
PKHD1
7%
APOB
7%
KMT2B
7%
NF1
7%
PRKAR1A
7%
SVEP1
7%
TUT7
5%
ASXL3
5%
MEN1
5%
CMYA5
5%
FRAS1
5%
LRP1
5%
STAB1
4%
ZNRF3
76%
12q14.1-Amp
76%
12q14.3-Amp
76%
12q15-Amp
75%
5p15.33-Amp
69%
5q35.3-Amp
63%
5p15.31-Amp
56%
22q12.1-Del
43%
1p36.23-Del
29%
4q35.1-Del
29%
4q34.3-Del
28%
9p21.3-Del
25%
3q13.31-Del
80%
CDK4
O
5 O
1
80%
PCT
OS9
80%
AGAP2
56%
ZNRF3-AS1
FSCN1
Alterations
CNA (arm-level)
CNA (gene-level)
☐ Low
☐ Gain
☐ Gain
☐ High_balanced_gain Loss
☐ High
☐ Mutated
☐ Loss
☐ Loss
☐ High_balanced_loss
Discussion
The prognosis of ACC is generally poor, and owing to its heterogeneity, there are substantial individual variations in the progression, recurrence, and survival of ACC [25]. Such inter- individual variations may originate from differences in the indivi- dual’s epigenetic composition [26]. Our multi-omics approach can facilitate the understanding of this heterogeneity. Advancements in genetic analysis technologies, exemplified by next-generation sequencing, and the development of bioinformatics tools can be used to identify different diagnostic and prognostic factors as well as therapeutic targets. By integrating multi-omics data with clinical prognosis data, we can identify common markers associated with poor prognosis amid tumor heterogeneity, with the hope of personalizing medicine [27]. After conducting a clinical prognosis analysis by combining multiple transcriptomes and ATAC-seq data, we identified FSCN1 as a potential ACC biomarker. A comprehensive analysis of independent cohorts for ACC prognosis indicated that the overall prognosis for the high FSCN1 expression group is worse than that for the low FSCN1 expression group.
FSCN1 is a highly conserved actin-binding protein. Studies have shown that FSCN1 is overexpressed in various tumors and that its overexpression in tumor cells is associated with tumor growth, invasion, and metastasis [6]. Currently, FSCN1 is thought to promote the progression of esophageal squamous cell carcinoma (ESCC) progression by downregulating PKT6 expression level and to increase the invasive capability of pituitary adenomas by modulating the expression of the Notch/DLL pathway [28,29]. Research suggests that FSCN1 acts as a transcriptional binding site for the ß-catenin protein and promotes ACC invasion, but its role in ACC prognosis has not been studied from a multi-omics perspective [8,30]. Our present research fills this gap.
In our study, a multicenter cohort demonstrated that high FSCN1 expression in ACC is associated with poor OS and PFI. Additionally, in our clinical samples from the Chinese population, FSCN1 expression was significantly elevated in ACC tissues compared with noncancerous adrenal tissues, which is consistent with reports from European cohorts [8]. These consistent findings across different countries and ethnicities underscore the potential of FSCN1 as a prognostic marker for ACC. In vitro and in vivo experiments by Huang et al. [31] revealed that FSCN1 knockdown inhibited the invasion and proliferation of lung adenocarcinoma cells in mice, whereas Chen et al. [32] reported that FSCN1 knockdown suppressed the invasion of breast cancer cells. Similarly, our in vitro assays demonstrated that FSCN1 knockdown significantly inhibited the invasion and proliferation of SW-13 and H295R cells. These findings in ACC cell lines align with those reported in lung adenocarcinoma and breast cancer, highlighting the crucial role of FSCN1 in tumor progression. In our in vivo experiments, the FSCN1 inhibitor NP-G2-044 inhibited the growth of ACC-transplanted tumors in mice. These findings suggest a potential role for FSCN1-targeting drugs in ACC therapy.
The GSEA results revealed that FSCN1-related genes are significantly enriched in the ACC, particularly in the activation of biological processes and pathways such as cytosolic ribosomes, spliceosomes, DNA replication, chromosome separation regulation, cytoplasmic transport, and the cell cycle. Owing to the rapid proliferation of tumor cells, more ribosome biogenesis is required to enable the cells to pass the G1-S phase checkpoint, thus activating the cell cycle process, and the spliceosome promotes the conversion
of heterogeneous nuclear RNA (hnRNA) into mature mRNA, DNA replication, mitosis, the transport of more intracellular signaling molecules and metabolites, and a series of biological processes [33,34]. However, cancer cells may result in a loss of negative regulation of ribosome production through TP53 mutations [35]. In the CNV data from TCGA, we found that the frequency of TP53 mutation in the high FSCN1 expression group is greater than that in the low FSCN1 expression group. These findings suggest that FSCN1 may promote the development of tumors mainly by regulating the occurrence of ribosomes.
CD8+ T cells are key immune cells involved in controlling tumor growth. During the development of a tumor, cancer immune editing involves three phases: elimination, equilibrium, and escape [36]. The TME is composed of tumor cells and various types of cells around tumor cells that promote or inhibit tumor growth [36]. Several immune cells, including CD8+ T cells and their subgroups, can be detected in the TME [36,37]. FSCN1 has been reported to be correlated with CD8+ T-cell infiltration scores across the TIMER algorithm [7]. In our study, FSCN1 expression was found to be negatively correlated with the infiltration score of CD8+ T cells across various algorithms. IHC analysis of sections from ACC patients also revealed a negative correlation between FSCN1 expression and CD8+ T-cell infiltration. CD8+ T-cell infiltration in the tumor microenvironment is regulated by various factors, including chemical substances, cytokines, chemokines, and im- mune checkpoint molecules expressed in the tumor microenviron- ment [38]. Our GSEA results also revealed that FSCN1-related genes are suppressed in the chemokine signaling pathway and antigen presentation pathway. These findings indicate that in the ACC, FSCN1 may inhibit CD8+ T-cell infiltration through mechanisms such as suppressing antigen presentation and inhibiting chemokine secretion. Research has shown that tumors infiltrated by CD8+ T cells respond better to treatments aimed at suppressing tumor growth [16]. In the immunotherapy of ACC, immune checkpoint inhibitors (ICIs) have shown efficacy in only a few cases, and FSCN1 may hold potential value in predicting the efficacy of immunotherapy for ACC. Public ACC immunotherapy cohorts are currently limited, but we aim to further explore this area once more public data become available.
Nevertheless, there are several limitations in our study. The number of ACC samples available in public databases is limited, with only 79 samples from TCGA-ACC. More ACC cohorts are needed for further validation of the role of FSCN1 in subsequent studies. Finally, the lack of ACC immunotherapy cohorts to study the impact of FSCN1 expression inhibition on CD8+ T cells and the effectiveness of ICIs will be the focus of future research. Although a comprehensive analysis with multi-omics from independent data- sets was conducted, the molecular mechanism of FSCN1 in ACC progression should be validated in other experiments.
Supplementary Data
Supplementary data is available at Acta Biochimica et Biophysica Sinica online.
Funding
This work was supported by the grants from the National Natural Science Foundation of China (Nos. 82073294 and 82003271), Youth Medical Innovation and Practice Research Program of Guangzhou (No. 2023QNYXZD001), Guangzhou Planned Project of Science and
Technology (Nos. 2023A04J1269 and 202102010152), Guangzhou Municipal Science and the Technology Key Project (No. 202102080624), and Guangzhou Medical Key Subject Construction Project [2021-2023].
Conflict of Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
References
1. Favia G, Lumachi F, Carraro P, D’Amico DF. Adrenocortical carcinoma. our experience. Minerva Endocrinol 1995, 20: 95-99
2. Almeida MQ, Bezerra-Neto JE, Mendonça BB, Latronico AC, Fragoso MCBV. Primary malignant tumors of the adrenal glands. Clinics 2018, 73: e756s
3. Catalano R, Giardino E, Treppiedi D, Mangili F, Morelli V, Elli FM, Serban AL, et al. The cytoskeleton actin binding protein filamin A impairs both IGF2 mitogenic effects and the efficacy of IGF1R inhibitors in adrenocor- tical cancer cells. Cancer Lett 2021, 497: 77-88
4. Barrett T, Wilhite SE, Ledoux P, Evangelista C, Kim IF, Tomashevsky M, Marshall KA, et al. NCBI GEO: archive for functional genomics data sets- update. Nucleic Acids Res 2013, 41: D991-D995
5. Tomczak K, Czerwińska P, Wiznerowicz M. Review the cancer genome atlas (TCGA): an immeasurable source of knowledge. Contemp Oncol (Pozn) 2015, 1A: 68-77
6. Li Z, Shi J, Zhang N, Zheng X, Jin Y, Wen S, Hu W, et al. FSCN1 acts as a promising therapeutic target in the blockade of tumor cell motility: a review of its function, mechanism, and clinical significance. J Cancer 2022, 13: 2528-2539
7. Liang J, Liu Z, Wei X, Zhou L, Tang Y, Zhou C, Wu K, et al. Expression of FSCN1 and FOXM1 are associated with poor prognosis of adrenocortical carcinoma patients. BMC Cancer 2019, 19: 1165
8. Poli G, Ruggiero C, Cantini G, Canu L, Baroni G, Armignacco R, Jouinot A, et al. Fascin-1 is a novel prognostic biomarker associated with tumor invasiveness in adrenocortical carcinoma. J Clin Endocrinol Metab 2019, 104: 1712-1724
9. Cantini G, Fei L, Canu L, De Filpo G, Ercolino T, Nesi G, Mannelli M, et al. Circulating fascin 1 as a promising prognostic marker in adrenocortical cancer. Front Endocrinol 2021, 12: 698862
10. Liu Z, Guo CG, Dang Q, Wang L, Liu L, Weng S, Xu H, et al. Integrative analysis from multi-center studies identities a consensus machine learning-derived lncRNA signature for stage II/III colorectal cancer. EBioMedicine 2022, 75: 103750
11. Langfelder P, Horvath S. WGCNA: an R package for weighted correlation network analysis. BMC BioInf 2008, 9: 559
12. Langfelder P, Horvath S. Fast R Functions for robust correlations and hierarchical clustering. J Stat Soft 2012, 46: i11
13. Shen W, Song Z, Zhong X, Huang M, Shen D, Gao P, Qian X, et al. Sangerbox: a comprehensive, interaction-friendly clinical bioinformatics analysis platform. iMeta 2022, 1: e36
14. Wu T, Hu E, Xu S, Chen M, Guo P, Dai Z, Feng T, et al. clusterProfiler 4.0: a universal enrichment tool for interpreting omics data. Innovation 2021, 2: 100141
15. Love MI, Huber W, Anders S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol 2014, 15: 550
16. Gajewski TF, Schreiber H, Fu YX. Innate and adaptive immune cells in the tumor microenvironment. Nat Immunol 2013, 14: 1014-1022
17. Yu G, Wang LG, He QY. ChIPseeker: an R/Bioconductor package for ChIP peak annotation, comparison and visualization. Bioinformatics 2015, 31: 2382-2383
18. Corces MR, Granja JM, Shams S, Louie BH, Seoane JA, Zhou W, Silva TC,
et al. The chromatin accessibility landscape of primary human cancers. Science 2018, 362: eaav1898
19. Robinson JT, Thorvaldsdóttir H, Wenger AM, Zehir A, Mesirov JP. Variant review with the integrative genomics viewer. Cancer Res 2017, 77: e31- e34
20. Liu J, Lichtenberg T, Hoadley KA, Poisson LM, Lazar AJ, Cherniack AD, Kovatich AJ, et al. An integrated TCGA pan-cancer clinical data resource to drive high-quality survival outcome analytics. Cell 2018, 173: 400-416. e11
21. Zhong C, Long Z, Yang T, Wang S, Zhong W, Hu F, Teoh JYC, et al. M6A- modified circRBM33 promotes prostate cancer progression via PDHA1- mediated mitochondrial respiration regulation and presents a potential target for ARSI therapy. Int J Biol Sci 2023, 19: 1543-1563
22. Lu J, Dong W, He H, Han Z, Zhuo YJ, Mo RJ, Liang Y, et al. Autophagy induced by overexpression of DCTPP1 promotes tumor progression and predicts poor clinical outcome in prostate cancer. Int J Biol Macromol 2018, 118: 599-609
23. Zeng D, Ye Z, Shen R, Yu G, Wu J, Xiong Y, Zhou R, et al. IOBR: multi- omics immuno-oncology biological research to decode tumor microenvir- onment and signatures. Front Immunol 2021, 12: 687975
24. van der Leun AM, Thommen DS, Schumacher TN. CD8+ T cell states in human cancer: insights from single-cell analysis. Nat Rev Cancer 2020, 20: 218-232
25. Ismail A, Bateman A. Expression of TBX2 promotes anchorage-indepen- dent growth and survival in the p53-negative SW13 adrenocortical carcinoma. Cancer Lett 2009, 278: 230-240
26. Baudin E. Adrenocortical carcinoma. Endocrinol Metab Clin N Am 2015, 44: 411-434
27. Shankaran V, Ikeda H, Bruce AT, White JM, Swanson PE, Old LJ, Schreiber RD. IFNy and lymphocytes prevent primary tumour develop- ment and shape tumour immunogenicity. Nature 2001, 410: 1107-1111
28. Cai H, Wang R, Tang Z, Lu T, Cui Y. FSCN1 promotes esophageal carcinoma progression through downregulating PTK6 via its RNA-Binding protein effect. Front Pharmacol 2022, 13: 868296
29. Liu C, Gao H, Cao L, Gui S, Liu Q, Li C, Li D, et al. The role of FSCN1 in migration and invasion of pituitary adenomas. Mol Cell Endocrinol 2016, 419: 217-224
30. Ruggiero C, Tamburello M, Rossini E, Zini S, Durand N, Cantini G, Cioppi F, et al. FSCN1 as a new druggable target in adrenocortical carcinoma. Intl J Cancer 2023, 153: 210-223
31. Huang Y, Shan G, Yi Y, Liang J, Hu Z, Bi G, Chen Z, et al. FSCN1 induced PTPRF-dependent tumor microenvironment inflammatory reprogram- ming promotes lung adenocarcinoma progression via regulating macro- phagic glycolysis. Cell Oncol 2022, 45: 1383-1399
32. Chen C, Xie B, Li Z, Chen L, Chen Y, Zhou J, Ju S, et al. Fascin enhances the vulnerability of breast cancer to erastin-induced ferroptosis. Cell Death Dis 2022, 13: 150
33. Derenzini M, Montanaro L, Chillà A, Tosti E, Vici M, Barbieri S, Govoni M, et al. Key role of the achievement of an appropriate ribosomal RNA complement for G1-S phase transition in H4-II-E-C3 rat hepatoma cells. J Cell Physiol 2005, 202: 483-491
34. Derenzini M, Montanaro L, Trerè D. Ribosome biogenesis and cancer. Acta Histochemica 2017, 119: 190-197
35. Brighenti E, Calabrese C, Liguori G, Giannone FA, Trerè D, Montanaro L, Derenzini M. Interleukin 6 downregulates p53 expression and activity by stimulating ribosome biogenesis: a new pathway connecting inflamma- tion to cancer. Oncogene 2014, 33: 4396-4406
36. St. Paul M, Ohashi PS. The roles of CD8+ T cell subsets in antitumor immunity. Trends Cell Biol 2020, 30: 695-704
37. Frankiw L, Li G. The tumor microenvironment’s role in malignant progression and treatment response. Cancer Lett 2022, 548: 215888
38. Chen DS, Mellman I. Elements of cancer immunity and the cancer- immune set point. Nature 2017, 541: 321-330