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

Figure 1. Flowchart of this study

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).

Figure 2. Identification of genes related to unfavorable prognosis via WGCNA (A) Dendrogram of samples and trait heatmap for TCGA-ACC. (B) Cluster dendrogram derived from genes within the coexpression network. (C) Association between gene modules and unfavorable prognosis in ACC. (D) Scatter plots illustrating the distribution of genes linked to unfavorable prognosis. (E,F) Results of GO (E) and KEGG (F) enrichment analyses corresponding to genes connected in the brown module.

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.

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))

DUnivariate COX
VariablepvalueHRL95CIH95CI
TCGA-ACC.
Age3.79E-011.010.991.040
Gender9.99E-011.000.472.14 I--I
Stage9.39E-062.631.714.03I- 1
clinical_M5.14E-055.302.3611.88- - 1
pathologic_N2.48E-011.740.684.47 F1
pathologic_T9.28E-073.041.954.74I- -I
FSCN12.26E-051.761.362.29I
GSE10927
Age5.19E-011.010.981.05.
Gender4.75E-011.470.514.24 1-I
Stage2.09E-021.821.093.02-l. -1
FSCN14.67E-024.451.0219.38I 1
GSE19750
Age9.73E-021.040.991.08
Gender6.50E-011.260.463.44 I--l
Stage8.73E-011.030.741.43 F1
FSCN11.08E-021.581.112.25'+ 1
Meta Adult
Age1.72E-011.020.991.04.
Gender7.54E-011.120.552.26 I-1
Stage3.73E-011.110.881.41
FSCN11.83E-031.671.212.311
VariablepvalueHRL95CIH95CI
TCGA-ACC
Stage2.82E-010.510.151.75+
clinical_M3.67E-012.070.4310.10I- -I
pathologic T3.21E-033.341.507.431- -1
FSCN16.80E-041.681.242.26
GSE10927.
Stage1.01E-022.051.193.55'+
FSCN12.37E-027.871.3247.04· I
GSE19750.
Age4.15E-021.041.001.09
FSCN16.23E-031.791.182.73.I
Meta Adult
Age6.44E-021.021.001.05
FSCN11.09E-031.781.262.52101
-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_idRangeLogFCadj.P.Val
ACC_338075593335~55938361.0825.55E-02
ACC_338085594070~55945712.1051.30E-03
ACC_338095594601~55951021.7981.43E-03
ACC_338105595480~55959811.9271.51E-03
ACC_338115596006~55965072.2293.68E-05
ACC_338135598752~55992531.0628.27E-02
ACC_338145603706~56042071.5902.75E-02
ACC_338165607141~56076421.5403.40E-02
ACC_338175608168~56086691.5553.04E-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).

Figure 5. Experimental validation of FSCN1 (A) Representative images from IHC staining and a combined box-and-whisker and scatter plot illustrating the quantitative analysis between noncancerous tissues and the ACC. (B) Western blot analysis was used to detect the expression level of FSCN1 in SW-13 and H295R cell lines. (C) Transwell assays were used to detect the invasion ability of SW-13 and H295R cells. (D) CCK-8 assays were used to detect the proliferation of SW-13 and H295R cells. * P< 0.05; ** P< 0.01; *** P < 0.001; **** P < 0.0001. (E) Gross observation of xenograft tumor size in BALB/c-nude mice. (F,G) NP-G2-044 inhibited tumor growth, including tumor volume (P< 0.001) and weight (P= 0.045, n = 5).

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

Figure 6. Functional enrichment analysis of FSCN1 (A) Representative outcomes of GO. (B) The enriched GO pathways identified via GSEA are highlighted. (C) Representative outcomes of the KEGG analysis. (D) KEGG pathway enrichment analysis via GSEA.

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.

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.

Figure 7. Immunological attributes associated with FSCN1 in ACC patients (A) Correlation heatmap illustrating the relationship between FSCN1 expression and immune cell infiltration. (B) Scatter plots depicting the associations between FSCN1 expression and CD8+ T cell infiltration score across the four algorithms. (C) Violin plots demonstrating CD8+ T cell infiltration scores across the groups with low and high FSCN1 expression. (D) Immunohistochemical analysis of CD8+ T cell in the human ACC. (E) Scatter plots depicting the association between the FSCN1 IRS and the CD8+ IRS (n = 5).

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

Figure 8. Characterization of mutations and copy number variations associated with FSCN1 (A) Mutation profiles of the top 20 most frequently mutated genes and copy number variation patterns of the top 6 amplified and homozygously deleted chromosomal fragments, as differentiated between the high- and low-FSCN1 expression groups.

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