ARNTL deregulation promotes ovarian cancer progression and metastasis by activating cancer-associated fibroblasts

Abstract

Background

Circadian disruption fuels tumor growth, but the role of the core clock gene aryl hydrocarbon receptor nuclear translocator-like (ARNTL) in ovarian cancer is still unclear. We investigated how ARNTL affects ovarian cancer cell proliferation, invasion, and migration, and revealed how it shapes these behaviors by acting on the fibroblast-rich tumor microenvironment.

Method

We engineered SKOV3 cells to either overexpress or silence ARNTL and then measured how fast they migrated, invaded, and proliferated. RNA sequencing (RNA-seq) and pathway analyses were performed, while Microenvironment Cell Populations-counter (MCP-counter) scored fibroblast abundance. In vivo, ovarian cancer cells were subcutaneously inoculated into immunodeficient mice to monitor tumor growth kinetics and quantify proliferative capacity at the organismal level. Single-cell RNA-seq of the tumors and in vitro co-cultures with cancer-associated fibroblasts (CAFs) uncovered how ARNTL re-programs the fibroblast-rich microenvironment to suppress cancer progression.

Results

Here, we found that ARNTL, a circadian core gene, is downregulated in ovarian cancer, and its abundance is negatively correlated with tumor proliferation, invasion, and metastasis. In vitro and in vivo models demonstrated that restoring ARNTL not only inhibits cancer cell proliferation and metastasis, but also reshapes the tumor microenvironment by reducing CAFs recruitment and activation. Mechanistically, ARNTL knockdown promotes PI3K-Akt/TGF-β signaling in ovarian cancer cells and induces CAF activation. Activated CAFs exhibit increased secretion of Stromal Cell-Derived Factor-1 (SDF-1) and Platelet-Derived Growth Factor (PDGF), which reciprocally enhance ovarian cancer cell proliferation and metastatic capacity. Our study indicates that ARNTL functions as a potent tumor suppressor by concurrently repressing cancer cell-intrinsic proliferation-associated genes and reprogramming the phenotype and subset composition of cancer-associated fibroblasts (CAFs) within the tumor microenvironment.

Conclusion

Collectively, our findings indicate the mechanisms of ARNTL in inhibiting ovarian cancer progression and metastasis via the ARNTL-CAFs axis, and provide a promising strategy for ovarian cancer treatment.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13048-026-02044-7.

Introduction

Ovarian cancer remains the deadliest gynecological malignancy; its insidious progression and high metastatic potential impose a disproportionate burden on women worldwide [1–3]. Owing to the absence of early symptoms, approximately 70% of patients present with advanced-stage disease, resulting in a five-year survival rate below 30% [4, 5]. Although cytoreductive surgery followed by platinum-taxane chemotherapy is the standard of care, chemoresistance and relapse remain formidable obstacles, underscoring the urgent need to identify novel molecular drivers and strategies to therapeutically target the tumor microenvironment [6, 7].

The circadian clock governs fundamental cellular processes, including cell-cycle progression, DNA repair, metabolism, and apoptosis, through a transcriptional-translational feedback loop [8–10]. Disruption of this circuitry has been causally linked to tumorigenesis across multiple organs. ARNTL, the core transcriptional activator of the circadian network [11–13], is indispensable for maintaining rhythmic homeostasis [14, 15].

The ARNTL locus produces three alternatively spliced transcripts encoding two protein isoforms, BMAL1a and BMAL1b, that differ exclusively in their C-terminal domains. Both isoforms form heterodimers with CLOCK to drive circadian gene transcription and maintain the core feedback loop, exhibiting comparable subcellular localization and transcriptional activity [16, 17].The functional role of ARNTL in cancer is highly context-dependent. In oral cancer, small cell lung cancer, endometrial carcinoma, Pancreatic cancer, ARNTL downregulation has been associated with reduced proliferation and enhanced chemosensitivity, mediated through the circRNA-miRNA-PTEN/PI3K axis [18–21]. In contrast, ARNTL overexpression in nasopharyngeal carcinoma and colorectal cancer promotes invasive and metastatic potential through the AMOTL2-LATS-YAP pathway, serving as an independent marker of poor prognosis [22, 23]. Despite this emerging understanding, the expression pattern, functional roles, and microenvironmental regulatory mechanisms of ARNTL in ovarian cancer remain incompletely characterized.

Over the past decade, cancer-associated fibroblasts (CAFs) have emerged as pivotal architects of the ovarian cancer microenvironment, driving metastasis and chemoresistance [24–29]. CAFs constitute a heterogeneous population that can be stratified into functionally distinct subtypes: pro-inflammatory CAFs (iCAFs), antigen-presenting CAFs (apCAFs), vascular CAFs (vCAFs), proliferative CAFs (dCAFs), and tumor-like CAFs (tCAFs). Each subset contributes to tumor progression via secretion of cytokines, extracellular-matrix remodeling, and metabolic reprogramming [30, 31]. Recent studies have shown that the expression level of ARNTL can play an important role in oral squamous cell carcinoma [32, 33] and pan-cancer by regulating CAFs [34]. Therefore, the abundance of ARNTL expression determines the pro-tumor and anti-tumor effects of CAFs. However, whether circadian genes systematically shape CAFs heterogeneity to precisely regulate the progression of ovarian cancer remains to be elucidated.

In this study, we integrated transcriptomic profiles from ovarian cancer cell lines and clinical specimens to identify ARNTL as one of the most significantly downregulated genes, with low expression correlating with poor patient prognosis across multiple independent cohorts. In vitro and in vivo experiments demonstrated that ARNTL overexpression suppresses ovarian cancer cell proliferation, invasion, and metastasis, while ARNTL silencing enhances these malignant phenotypes. Single-cell RNA-sequencing analyses further revealed that ARNTL reprograms the tumor microenvironment by attenuating activated fibroblast subsets, particularly inflammatory and tumor-like CAFs. Collectively, these fundings establish ARNTL as both a potential prognostic biomarker and therapeutic target, and suggest a novel strategy for tumor microenvironment modulation through circadian rhythm regulation.

Method

Cell lines and culture conditions

Human SKOV3 (ATCC HTB-77) and HEK293T (ATCC CRL-3216) cells were obtained from the American Type Culture Collection (ATCC) and authenticated by short tandem repeat profiling. Cells were maintained in high-glucose Dulbecco’s Modified Eagle Medium (DMEM; Thermo Fisher Scientific, Cat. 12800017) supplemented with 10% (v/v) fetal bovine serum (FBS; HyClone, Cat. SH30088.03) and 1% (v/v) penicillin–streptomycin (Invitrogen) in a humidified incubator at 37 °C with 5% CO₂. Primary normal ovarian fibroblasts (≤ passage 6) were cultured under identical conditions.

Cell construction

To establish stable gain- and loss-of-function ARNTL models in SKOV3 cells, the full-length CDS of human ARNTL (NCBI RefSeq: NM_001178.5) was PCR-amplified and inserted into pLVX-Puro via EcoRI/BamHI; the insert was verified by Sanger sequencing and restriction mapping. Lentivirus was produced by co-transfecting 293T cells with pLVX-ARNTL-Puro, psPAX2 and pMD2.G (4:3:1) using Lipofectamine 3000, and 48 h supernatants (0.45 μm-filtered) were used to transduce SKOV3 cells in 8 µg/mL polybrene, followed by 5-day selection with 2.5 µg/mL puromycin; overexpression was confirmed by qRT-PCR and Western blot, yielding ARNTL. For knockdown, three ARNTL-targeting shRNAs were cloned into pLKO.1-Puro; the most efficient was packaged into lentivirus as above and used to generate sh-ARNTL after identical transduction and selection. Both lines were maintained in 1.0 µg/mL puromycin.

RNA extraction and RT-PCR

Use quantitative polymerase chain reaction to detect mRNA levels such as ARNTL and CLOCK. The detailed information of the primers used is in Tables S5 and S6. Total RNA extraction was performed using TRIzol reagent (Invitgen) according to the manufacturer’s protocol. Total RNA purified by reverse transcription using random primers was synthesized using a cDNA synthesis kit (Thermo Fisher Science, Rockford, Illinois, USA). Using this as a template, SYBR Green (Thermo Fisher Science, Rockford, IL, USA) was used for real-time quantitative polymerase chain reaction (Real time PCR) to detect the mRNA levels of the genes shown. Actin serves as an internal control. All reactions were performed on an ABI7300 real-time polymerase chain reaction machine (Applied Biosystems, Foster, California, USA) with the following cycle parameters: 95℃ for 10 min, followed by 15s and 45s cycles at 95℃ and 60℃, respectively, for 45 times. Calculate the relative gene expression using the comparative CT method. All data represent the average of three repetitions.

Western blotting

Western Blot was used to detect the protein expression of Arntl and α-SMA. Extract cellular proteins from the corresponding treated cells using RIPA buffer (50 mM Tris HCl [pH 7.5], 150 mM sodium chloride, 1% Triton X-100, and 0.5% Na deoxycholate). The BCA protein detection kit (Thermo Fisher Science) is used for protein concentration determination. 10 micrograms of lysates were separated on 10% SDS-PAGE gel and transferred to PVDF membrane. Incubate the membrane with the primary antibody overnight at 4℃, then incubate with the primary antibody, and finally incubate with the coupled secondary antibody. Use chemiluminescent substrates (ECL, Bio Rad, Richmond, California, USA) to detect signals. Finally, measure the strip strength using Image J software. The primary antibodies ARNTL(sc-373955) and α-SMA (sc-53142) were purchased from Santa Cruz. GAPDH (5174T) and actin (5125 S) were used as internal controls, both purchased from CST Biotech.

Wound healing for detecting cell migration

Vehicle, ARNTL, sh-Vehicle and sh-ARNTL cells were seeded in 6-well plates (5 × 10⁵ cells per well) and cultured at 37℃, 5% CO₂ until they reached ~ 80% confluence. A straight scratch was generated in the monolayer using a sterile 200µL pipette tip. After two gentle PBS washes to remove detached cells, fresh complete medium was added. Plates were returned to the incubator and wound closure was monitored at 0 h (baseline), 24 h and 48 h under a phase-contrast microscope; images were captured at ×100 magnification using identical fields at each time point.

Cell proliferation (CCK-8) assay

Vehicle, ARNTL, sh-Vehicle and sh-ARNTL cells were harvested, counted, and seeded in 96-well plates at 1 × 10³cells per well (n = 5 per group). After overnight attachment, cells were incubated at 37℃/5% CO₂ for 24, 48, 72–96 h. At each time point, 10µL CCK-8 reagent was added, and the plates were returned to the incubator for an additional 3 h. Absorbance at 450 nm (A450) was measured on a microplate reader (BioTek). Percent proliferation was calculated relative to the 0 h baseline.

Transwell Matrigel invasion assay

Matrigel (Corning) was thawed at 4 °C overnight and diluted 1:8 in ice-cold serum-free medium. A 50µL aliquot was evenly layered on the upper surface of 8 μm-pore Transwell inserts (Corning) and polymerized at 37 °C/5% CO₂ for 3 h. After two PBS washes, inserts were rehydrated with 100µL serum-free medium for 30 min; absence of leakage was confirmed before cell seeding. Vehicle, ARNTL, sh-Vehicle and sh-ARNTL cells were trypsinized, washed, and resuspended in serum-free medium to 2.5 × 10⁵ cells/mL. A 200 µL volume of cell suspension was added to the upper chamber, and 500 µL complete medium (10% FBS) was placed in the lower chamber. After 24 h incubation, non-invading cells were removed with a cotton swab. Invading cells were fixed with 4% paraformaldehyde, stained with 0.1% crystal violet, and counted in five random ×200 fields per insert. Data are presented as mean invaded cells per field ± SD.

Animal model establishment and monitoring

Female BALB/c nude mice (4–5 weeks old, 18–20 g) were purchased from Beijing Huafukang Biotechnology Co., Ltd. and housed at the SPF level barrier facility (AAALAC certified) of Sichuan University Experimental Animal Center. Environmental setting: 22 ± 2℃, 12 h light/dark cycle, IVC cage (Tecniplast Green Line, 60 air changes/h), 3 per cage. Free consumption of irradiation sterilized AIN-93G feed (14% fat supply) and high-pressure sterilized tap water, changed twice a week and recorded daily consumption, with no difference between groups. Before the experiment, the random area was divided into 4 groups, with 6 animals in each group. Then, Vehicle, ARNTL, sh-Vehicle and sh-ARNTL cells (1 × 10⁵ per mouse, ≥ 95% viability) were suspended in 100µL PBS/Matrigel (1:1) and inoculated subcutaneously into the right dorsal flank (n = 6 per group). Tumor volume (V = 0.5×length×width²) was measured with digital calipers every 5 days, and body weight was recorded simultaneously. Mice were euthanized when any tumor dimension reached 15 mm or at the end of the 5-week observation period, whichever occurred first. Deep anesthesia was induced by intraperitoneal pentobarbital sodium (100 mg kg⁻¹; Nembutal, 60 mg mL⁻¹); after loss of the pedal withdrawal reflex, death was accomplished by cervical dislocation. Cessation of heartbeat and respiration was verified to confirm death. All procedures followed the AVMA Guidelines for the Euthanasia of Animals (2020) and were approved by the Institutional Animal Care and Use Committee (protocol No. IACUC20220301146).

RNA-seq sequencing

Total RNA from Vehicle, ARNTL, sh-Vehicle and sh-ARNTL cells and Vehicle, ARNTL, sh-Vehicle and sh-ARNTL inoculated tumor tissue (n = 3) was extracted using TRIzol reagent (Invitrogen). The library construction starts with 1 µg of total RNA (replaced with ≥ 200 ng SMART Seq ultra-low quantity scheme if insufficient), and only high-quality samples with Agilent 2100 RIN ≥ 7 and 28 S/18S ≥ 1.8 are selected for subjected to 150-bp paired-end sequencing on an Illumina NovaSeq 6000 platform. Raw reads were trimmed with Trimmomatic (v0.39) and mapped to the hg38 reference genome using STAR (v2.7.9a). Gene counts were quantified with featureCounts (Subread v2.0.1) and normalized via DESeq2 (v1.34.0). Differentially expressed genes (DEGs) were defined as those with |log₂FC|≥1 and FDR < 0.05. GO and KEGG enrichment analyses were performed using clusterProfiler (v4.2.2) with Benjamini-Hochberg correction. To quantify immune and stromal infiltration, the MCP-counter algorithm (v1.2.0) was applied to the log₂-transformed TPM matrix, yielding absolute abundance scores for 10 immune and 2 stromal populations. The association between ARNTL expression levels and each cell type was assessed by fitting a linear regression model adjusted for batch effects (limma v3.50.1). Subsequently, fibroblasts were re-classified into five functionally distinct subtypes according to their unique transcriptional signatures (Table S4): antigen-presenting CAFs (apCAFs), dividing CAFs (dCAFs), inflammatory CAFs (iCAFs), tumor-like CAFs (tCAFs) and vascular CAFs (vCAFs). Subtype-specific scores were calculated using gene set variation analysis (GSVA, v1.44.2).

Single cell RNA sequencing

Tumors were dissociated into single cells (Meitianli kit), filtered (40 μm), counted, and processed with the 10x Chromium v3.1 workflow; libraries were sequenced (NovaSeq, ≥ 50 k reads/cell). Reads were aligned and analyzed in Seurat v4.3.0: doublets removed (DoubletFinder), QC filters applied, data log-normalized and Harmony-integrated, PCA/UMAP on 2 000 variable genes, clusters called at resolution 0.8 and annotated with canonical markers/Single R. Fibroblasts were re-clustered (resolution 0.8) and subtyped with signature genes. DEGs were identified (FindMarkers, Wilcoxon, |log₂FC|≥0.25, FDR < 0.05) and CytoTRACE inferred stemness. Monocle 2 (DDRTree on top 1000 variable genes) reconstructed pseudotime trajectories and branch expression analysis modelling identified fate-transition genes.

Isolation of primary human ovarian fibroblast

Primary ovarian fibroblasts were isolated from non-neoplastic ovarian tissues obtained from three patients undergoing oophorectomy. Specimens were minced into 1–2 mm³ fragments and digested in serum-free DMEM containing 1 mg/mL collagenase IV and 0.5 mg/mL DNase I at 37 °C for 30 min with gentle agitation. The cell suspension was filtered through a 70 μm strainer, pelleted at 500×g for 10 min, and washed twice in DMEM. Cells were resuspended in complete DMEM (10%FBS, 1% penicillin-streptomycin) and cultured at 37 °C, 5%CO₂. Normal Fibroblasts (NFs) were verified by vimentin-positive, immunofluorescence and used at passage ≤ 6 for all experiments.

Co-culture systems

For separate co-culture, 2 × 10⁴ ovarian cancer cells were seeded in the top of the insert chamber while 4 × 10⁴ NFs were seeded in the bottom of 24-well culture plate. The insert chamber was put into 24-well plate after one day incubation and media of NFs were replaced with serum-free DMEM media. After 48 h co-culture, NFs were harvested for Western blot analysis of α-SMA (Sigma-Aldrich, #SAB5500002) and qRT-PCR quantification of PDGFR, PDGF, SDF and FAP mRNA.

The conditioned media (CM) form SKOV3 cells were used to treat NFs. 5 × 10⁵ ovarian cancer cells were seeded into the 6-well-plate and maintaining in DMEM media for 24 h then incubated in the serum-free DMEM for another 24 h. The supernatants were collected after incubation and centrifuged at 2,000 rpm for 10 min. NFs in the 6-well-plate were preincubated in the serum-free DMEM overnight and then treated with CM for 48 h. NFs were harvested and subjected for Western blot analysis of α-SMA (Sigma-Aldrich, #SAB5500002) and qRT-PCR quantification of PDGFR, PDGF, SDF and FAP mRNA. Meanwhile, in the co-culture system, recombinant SDF-1 (Thermo Fisher Scientific, #300–28 A-10UG) and SDF-1 inhibitor NOX-A12 (Merck Millipore, # B09050T12IM) were added for Transwell and clone formation experiments to investigate the regulatory role of SDF-1 in the ARNTL-CAF axis.

Immunofluorescence

1–2 × 10⁴ cells were seeded onto glass coverslips in 24-well plates. After treatment, cells were washed 3× with PBS and fixed with 4% paraformaldehyde at RT for 30 min. Following another PBS rinse, cells were permeabilized with 0.01% Triton X-100 (30 min) and blocked with 5% goat serum (EMD Millipore, S26) for 1 h at RT. They were then incubated overnight at 4 °C with primary antibody α-SMA (Sigma-Aldrich, #SAB5500002), washed 3×with PBS, and labeled with the Donkey Anti-Rabbit IgG, FITC-conjugate(Sigma-Aldrich, #AP182F) for 30 min at 37 °C. Nuclei were counterstained with DAPI (Solarbio, C0060) before imaging on a Zeiss confocal microscope.

Statistical analysis

Differences between two groups were assessed using a two-tailed Student’s t-test. ARNTL mRNA levels were analyzed in TCGA, GTEx and CCLE datasets. Statistical analyses and graphs were performed with R 3.6.2 and GraphPad Prism 8.0. Sample sizes were based on previous studies; no randomisation or blinding was applied, and no additional inclusion/exclusion criteria were used. All experiments were replicated at least three times; data are presented as mean ± SD. Only datasets with similar variances that met test assumptions were compared. Significance thresholds: *P < 0.05, **P < 0.01, ***P < 0.001.

Result

ARNTL expression in ovarian cancer

To explore the expression pattern of ARNTL in ovarian cancer, we analyzed the expression of ARNTL in ovarian cancer based on high-throughput RNA-sequencing data from The Cancer Genome Atlas project (TCGA, https://tcga-data.nci.nih.gov/tcga/), including 426 ovarian cancers sample and 88 normal ovary sample matched GTEx data. As Shown in Fig. 1A, B, ARNTL expression in tumor tissue was significantly lower compared with normal tissue (P < 0.01). Moreover, the protein levels of ARNTL (Fig. 1C) was significantly lower in ovarian cancer tissues than in normal ovary tissues based on HPA database. Furthermore, in the CCLE database, the expression level of ARNTL across the examined ovarian cancer cell lines is consistently lower than in normal ovarian epithelial cells (Table S1). Additionally, real-time qPCR validated the consistent expression trend in SKOV3 cell line with CCLE database (Figure S1A), compared with HOSEC as normal control. Next, we assessed the prognostic value of ARNTL in ovarian cancer patients, we found ovarian cancer patients with lower expression levels of ARNTL showed worse overall survival in Kaplan-Meier Plotter (Figure S1B), and public dataset TCGA (Figure S1C), GSE147641 (Figure S1D), GSE30161 (Figure S1E), GSE23554 (Figure S1F). Taken together, these results indicate that ARNTL is downregulated in ovarian cancer and suggest a potential association with tumor prognosis.

Fig. 1.

ARNTL is downregulated in human ovarian cancers and cell construction. A, The expression level of ARNTL from ovary cancer (red) and normal ovary (green). B, The expression level of ARNTL from ovary cancer (red) and Match TCGA normal and GTEx data normal ovary (blue). C, Validation of ARNTL from the HPA database. D-E, WB detection of ARNTL protein expression in ARNTL cells and Vehicle cells., (F-G), Immunoblots detection of ARNTL protein expression in sh-ARNTL cells and sh-Vehicle cells. H-I, Real-time qPCR detection of ARNTL gene expression in indicated cells

ARNTL inhibits ovarian cancer cell proliferation, invasion and metastasis in vitro and in vivo

To investigate the impact of ARNTL on ovarian cancer progression, we constructed SKOV3 cells stably expressing ARNTL (hereafter referred to as ARNTL); as well as stably knockdown of ARNTL in SKOV3 cells (hereafter referred to as sh-ARNTL). The Western blot results confirmed that both genetic modifications achieved sustained and stable expression levels (Fig. 1D-G and Figure S2A-D). Given that the core circadian feedback loop is initiated by ARNTL-CLOCK heterodimer and repressed through PER/CRY-mediated negative feedback, we examined these three regulatory nodes to comprehensively assess the impact of ARNTL perturbation on the entire circadian network. Therefore, we further examined the expression changes of these core circadian components (Fig. 1H-I). Furthermore, we observed an obvious effect of ARNTL on tumor proliferation was found (Fig. 2A and Fig S3A). Subsequent wound-healing assay suggested that the number of ARNTL overexpression cells migrating into the wound area was much lower than control cells (Fig. 2B). Contrary, loss of ARNTL increased wound closure in ARNTL knockdown cells (Figure S3B). Consistent with this observation, overexpression of ARNTL significantly reduced invasion of ovarian cancer cells (Fig. 2C). In contrast, loss of ARNTL promoted cell migration (Figure S3C). Accordingly, overexpression of ARNTL resulted in elevated expression levels of circadian rhythm such as CLOCK, NPAS2, and the down-regulation of PER1, PER3, TIMELESS, CRY1-2 (Fig. 2D). Conversely, knockdown of ARNTL led to opposite effects (Figure S3D). Consistently, overexpression of ARNTL resulted in elevated expression levels of genes related to cell proliferation and migration such as CDH1, RB1, BRCA1, ATM, PTEN, TP53, MLH1 and the down-regulation of MYC, MMP9, ZEB1 (Fig. 2E). Conversely, knockdown of ARNTL led to opposite effects (Figure S3E). Taken together, these results confirm that ARNTL reduces proliferation, migration and invasion of ovarian cancers cells in vitro.

Fig. 2.

ARNTL inhibits ovarian cancer cell invasion and metastasis in vitro and in vivo. A, Relative growth rates of indicated cells determined by CCK8 assay. Values are mean ± SD (two-way ANOVA). B, Representative images for wound healing of indicated cells (left). Quantification of migration rate (right). Values are mean ± SD. C, Representative images for Transwell invasion assays of indicated cells (top), scale bar: 100 μm. Relative numbers of invasion cells (down). Values are mean ± SD. D, Real-time qPCR analysis for genes related to circadian rhythm in indicated cells. Values are mean ± SD. E, Real-time qPCR analysis for genes related to cell proliferation and migration in indicated cells. Values are mean ± SD. F-G, Representative images for dynamic detection of tumor volume growth in ARNTL and Vehicle cells after tumor formation. Values are mean ± SD. H, Body weight change curve of mice after ARNTL and Vehicle cells tumorigenesis

The effect of ARNTL on ovarian cancer proliferation was then investigated in vivo. We assessed the effect of ARNTL in tumor formation in athymic BALB/c nude mice by subcutaneously injecting SKOV3 cells. Subsequently, the tumor growth of the mice was observed and measured every 5 days (Fig. 2F). In consistence with the results in vitro, overexpression of ARNTL inhibited the proliferation colonization of SKOV3 cells. Furthermore, results from the tumor volume of the ARNTL group was significantly smaller than that of the Vehicle group, indicating that overexpression of ARNTL has a significant inhibitory effect on the cancer development ability of SKOV3 cells in vivo (Fig. 2G). It is worth mentioning that no significant difference in body weight was observed between the two groups of mice (Fig. 2H). Meanwhile, Real-time qPCR was employed to evaluate circadian rhythm genes and cell proliferation and migration genes expression in mice tumor. We observed an obvious effect of ARNTL on mice tumor in circadian rhythm regulation was found (Figure S4A). Consistently, overexpression of ARNTL resulted in elevated expression levels of genes inhibit cell proliferation and migration, such as TP53, RB1, BRCA1, BRCA2, and the down-regulation of genes promote cell proliferation and migration, such as Ki67, PCNA, MYC, KRAS (Figure S4B). In contrast, the loss of ARNTL led to opposite effects (Figure S3F-H and Figure S4A, B). Altogether, these results suggest that ARNTL inhibits cell proliferation migration and invasion of ovarian cancer cells in vitro and in vivo.

Integrative analysis of ARNTL-directed remodeling of the extracellular matrix and cancer-associated fibroblast subtypes to suppress ovarian cancer invasiveness in vitro

To elucidate the mechanism underlying ARNTL-mediated ovarian cancer aggressiveness biological processes, RNA-seq was performed (Fig. 3A and Figure S5A), yielding a list of 1879 and 1349 genes that were up- and down-regulated in ARNTL cells vs. Vehicle cells (Fig. 3B) and a list of 433 and 169 genes that were up- and down-regulated in sh-ARNTL cells vs. sh-Vehicle cells (Figure S5B) .We found that ARNTL have a profound effect on the circadian rhythm genes (Fig. 3C) and cell proliferation and migration genes (Fig. 3D) transcriptional level of indicated cells in SKOV3 cell. Then, we acquired the a list of 289 genes in shared differentially expressed genes of ARNTL cells vs. Vehicle cells and sh-ARNTL cells vs. sh-Vehicle cells (Fig. 3E). Next, we investigated the signaling pathway associated with ovarian cancer aggressiveness biological processes using GO and KEGG analysis (Fig. 3F and Figure S5C and Table S2-3). Meanwhile, we performed a hierarchical classification of the enriched, significant pathways and found that the ARNTL-regulated DEGs were predominantly enriched in circadian depression, cytotoxicity action, cytokine-cytokine receptor interaction, Focal adhesion and ECM-receptor interaction related signaling pathways (Figure S5D). Notably, the results revealed that all of these pathways converge on the regulation of fibroblasts, which is closely associated with the occurrence and development of tumors. Furthermore, differentially expressed genes were mapped to the reference fibroblast-related pathways in the KEGG database to probe ARNTL-related intracellular signaling in an unbiased fashion. The KEGG pathway analysis revealed that cellular signaling pathways involved in the feedback regulation of Fibroblast, such as Wnt pathway, VEGF pathway, TGF-β pathway, cytokine-cytokine receptor interaction, ECM-receptor interaction, Focal adhesion and PI3K-Akt pathway (Fig. 3G). Collectively, our findings establish ARNTL as a key transcriptional regulator that orchestrates fibroblast-associated signaling networks in ovarian cancer.

Fig. 3.

ARNTL modulate ovarian cancer cells by fibroblast related signaling pathway (A), Whole transcriptomes were subjected to MDS on expressed genes to assess sample diversity and relatedness in ARNTL (red) and Vehicle (black). B, Volcano plots represent DEGs between two group comparisons (ARNTL: Vehicle). Red dots indicate upregulation in DEG (Log₂FoldChange ≥ 1; P value ≤ 0.05), and blue dots indicate down regulation (Log₂Foldchange≤-1; P value ≤ 0.05). C, The transcriptional level of circadian rhythm genes in indicated cells. D, The transcriptional level of cell proliferation and migration genes in indicated cells. E, Venn diagram overlapping distribution of DEGs betwen ARNTL vs. Vehicle and sh-ARNTL vs. sh-Vehicle. F, Gene ontology (GO) assignment of the top 5 biological processes (BP), Cellular Component (CC) and Molecular Function (FF) associated with Co-DEGs (ranked by P value). See also Table S2. G, Interactome network shows that ARNTL was largely involved in regulating fibroblast-associated signaling networks

To systematically analyze the regulatory effect of ARNTL on the microenvironment of ovarian cancer, we performed MCP counter algorithm on RNA-seq expression matrix and quantified the infiltration abundance of various immune cells and stromal cells in tumor tissue using linear regression model. Notably, the results revealed that overexpression of ARNTL resulted in a lower abundance of fibroblasts, which is closely related to the biological characteristics of ARNTL overexpression inhibiting the proliferation and invasion of ovarian cancer cells (Fig. 4A). In contrast, loss of ARNTL led to opposite effects (Fig. 4B). Next, we have reclassified fibroblasts into the following groups based on their unique gene expression profiles (Table S4), mainly including antigen-presenting CAFs (apCAFs), dividing CAFs (dCAFs), inflammatory CAFs (iCAFs), Tumor-like CAFs (tCAFs), and Vascular CAFs (vCAFs) (Fig. 4C, D). Specifically, overexpression of ARNTL is significantly correlated with enrichment of CAFs features in tumor suppression. On the one hand, it significantly downregulates the pro-cancer subpopulation: apCAFs that present antigens but induce T cell tolerance, iCAFs that secrete IL-6/IL-8/CXCL1/2 to amplify inflammation and reshape ECM, and vCAFs that drive angiogenesis through the VEGF/PDGF axis; On the other hand, synchronous upregulation of tumor suppressor subgroups: dCAFs that trigger cell cycle arrest and DNA damage response, and tCAFs that activate cell apoptosis and aging programs (Fig. 4C). In Contrast, loss of ARNTL led to opposite effects (Fig. 4D). Therefore, ARNTL synergistically inhibits the proliferation and invasion of ovarian cancer cells by suppressing the pro-cancer CAFs and activating the anti-cancer CAFs. Furthermore, multi-algorithm validation further substantiates the regulatory role of ARNTL in CAFs infiltration. Leveraging four deconvolution algorithms: TIDE (Fig. 4E), MCPCOUNTER (Fig. 4F), EPIC (Fig. 4G), and xCell (Fig. 4H), we assessed the association between ARNTL expression and fibroblast abundance across public ovarian cancer cohorts. All four methods consistently revealed a significant positive correlation (P < 0.05). These population-level data support that ARNTL modulates CAFs infiltration and thereby contributes to ovarian cancer initiation and progression. Taken together, functional in vitro assays further demonstrate that the coordinated regulation among distinct CAFs subpopulations drives proliferation, invasion, and immune evasion, thereby serving as a central mechanism underlying ovarian cancer initiation and progression.

Fig. 4.

Integrated Analysis of ARNTL regulation on stromal components and cancer-associated fibroblast subtypes in ovarian cancer cells. A-B, The MCP counter algorithm quantitatively deconvolves transcriptome data to estimate the absolute abundance of infiltrating immune and stromal populations in indicated cells. C-D, The composition ratio and infiltration abundance of different fibroblast subpopulations in indicated cells. Multi algorithm validation confirmed the correlation between ARNTL and CAFs infiltration in TIDE(E), MCPCOUNTER(F), EPIC(G), and xCell(H)

Comprehensive analysis of ARNTL regulation on transcriptional regulation in ovarian cancer and tumor-associated fibroblast subtypes in vivo

To delineate the biological processes governed by ARNTL-mediated transcriptional reprogramming in vivo, we performed genome-wide RNA-seq on SKOV3-derived xenograft tumors established by subcutaneous injection in athymic BALB/c nude mice. Multidimensional scaling (MDS) of whole-transcriptome data revealed pronounced segregation of ARNTL-overexpressing tumors from vehicle controls, underscoring distinct transcriptional trajectories that persist until transcriptional shutdown (Fig. 5A, C). FPKM profiling confirmed these findings, showing negligible similarity between ARNTL and vehicle samples (Fig. 5B). Consistently, ARNTL depletion (sh-ARNTL) generated a transcriptome markedly divergent from that of sh-Vehicle tumors (Fig. 5D). Next, pathway-act-network analysis of ARNTL versus Vehicle tumors (Fig. 5E) revealed a coordinated regulatory network centered on apoptosis, cell cycle, chemokine signaling, cytokine-cytokine receptor interactions, focal adhesion, and NOD-like receptor pathways, collectively restraining cell proliferation, invasion, and cancer-associated fibroblast activation. Meanwhile, pathway-act-network analysis between sh-ARNTL and sh-Vehicle tumors (Fig. 5F) reveals that dysregulated signals converge on chemokine signaling, cytokine-cytokine receptor interactions, focal adhesion, and the NOD-like receptor axis, jointly potentiating cell proliferation, invasion, and cancer-associated fibroblast activation. Therefore, we performed MCP counting algorithm on the RNA-seq expression matrix of tumor tissue and quantified the infiltration abundance of various immune cells and stromal cells in tumor tissue using a linear regression model. It is worth noting that consistent with the results obtained from in vitro cells, the overexpression of ARNTL leads to a decrease in the abundance of fibroblasts, which is closely related to the biological characteristics of ARNTL overexpression inhibiting the proliferation and invasion of ovarian cancer cells (Figure S6A, B). Next, we will reclassify fibroblasts based on their unique gene expression profiles. Overexpression of ARNTL reduces apCAFs, iCAFs, and vCAFs, thereby inhibiting their pro-cancer functions (immune escape, inflammation, and angiogenesis). At the same time, the increase of dCAFs and tCAFs activates apoptosis and aging programs, synergistically inhibiting the proliferation and invasion of ovarian cancer (Fig. 5G). In contrast, the loss of ARNTL leads to the opposite effect (Fig. 5H). In summary, both in vivo and in vitro results are consistent, and in vivo experiments also indicate that the coordinated regulation between different CAFs subgroups drives proliferation, invasion, and immune evasion, thus becoming the core mechanism for the occurrence and development of ovarian cancer.

Fig. 5.

Comprehensive analysis of ARNTL regulation on transcriptional regulation in ovarian cancer and CAFs subtypes. A, Whole transcriptomes were subjected to MDS on expressed genes to assess sample diversity and relatedness in ARNTL(red) and Vehicle(black). B, Hierarchical clustering of quantitative genes expression profiling for ARNTL cells and vehicle cells. C, Whole transcriptomes were subjected to MDS on expressed genes to assess sample diversity and relatedness in sh-ARNTL(red) and sh-Vehicle(black). D, Hierarchical clustering of quantitative genes expression profiling for sh-ARNTL cells and sh-vehicle cells. E, Pathway-Act-Network analysis for ARNTL versus Vehicle. A node represents a signaling pathway. The node color is correlated with pathway expression pattern. Red indicates that the signaling pathway is activated, while green indicates that the signaling pathway is suppressed. Yellow indicates that the genes included in the corresponding signaling pathway are both upregulated and downregulated. Lines represent interactive relationship between signaling pathways. The direction of the arrow is from the cascade source to the target. F, Pathway-Act-Network analysis for sh-ARNTL versus sh-Vehicle. G-H, The composition ratio and infiltration abundance of different fibroblast subpopulations in indicated group

Single-cell RNA Sequencing analysis of cancer-associated fibroblasts dynamics during ARNTL overexpression

To more accurately confirm the regulatory role of ARNTL in CAFs, we performed single-cell RNA sequencing on fibroblasts from tumor tissues of mice subcutaneously inoculated with the model. After stringent quality control and normalization, unsupervised graph-based clustering was applied to the gene-expression matrix (Fig. 6A, Figure S7). At a resolution of 0.8, 17 robust clusters were identified (Fig. 6B). Each cluster was then annotated according to canonical marker-gene signatures, ultimately delineating five functionally distinct CAF subpopulations (Fig. 6C). Among these five subgroups, inflammatory CAFs (iCAFs) constitute the dominant invasive-CAF population (n = 5867) and are defined by high expression of CD34, complement C3, CXCL12, and IL-6 (Fig. 6D). These cells secrete pro-inflammatory cytokines, stimulate angiogenesis, remodel the extracellular matrix, and foster an immunosuppressive, pro-metastatic niche. GO terms emphasizing angiogenesis and cell adhesion (Figure S8A), with pathway analysis reveals enrichment of TNF-α and PI3K-AKT signaling (Figure S8B). Notably, iCAFs abundance is sharply reduced in SKOV3-ARNTL tumors, underscoring ARNTL’s tumor-suppressive function (Fig. 6G, H). Antigen-presenting CAFs (apCAFs) were defined by high expression of MHCII-related genes (CD74, CXCL4) (Fig. 6D). Pathway analysis revealed up-regulated MAPK and PI3K-Akt signaling (Figure S8D), while GO terms pointed to enhanced cell migration and circadian control (Figure S8C), consistent with the proliferative/migratory phenotype. ARNTL overexpression selectively attenuated apCAFs subtype, aligning with its tumor-suppressive role. Tumor-like CAFs (tCAFs, n = 594) were characterized by high expression of proliferation-, migration- and metastasis-associated genes (CDH2, SNAIL, TWIST, PGK1) and stress-response genes (ENO1, GAPDH) (Fig. 6D). GO and pathway analyses revealed activated cell-migration programs concomitant with p53-mediated apoptosis and senescence pathways, indicative of late-stage differentiation (Figure S8E, F). ARNTL overexpression significantly expanded this subtype, suggesting that it may exert anti-tumor effects through the pro-apoptotic and senescence properties of tCAFs. Dividing CAFs (dCAFs, n = 220) exhibited high expression of mitotic markers (TUBA1B, MKI67) (Fig. 6D). GO and pathway analyses revealed robust activation of cell-cycle, cell-division, DNA-replication, and DNA-repair pathways, confirming their actively proliferating phenotype (Figure S8G, H). Vascular CAFs (vCAFs) were identified as a discrete cluster marked by high MCAM (CD146) expression. GO and pathway analyses confirmed exclusive enrichment for angiogenesis-related programs (Figure S8I, J). Consistent with their pro-tumoral role, vCAFs were markedly reduced in SKOV3-ARNTL tumors, underscoring ARNTL’s anti-angiogenic and tumor-suppressive activity and highlighting its translational potential in ovarian cancer therapy. In addition, CytoTrace and pseudotime analyses revealed ARNTL-driven heterogeneity in CAFs differentiation. Stemness ranked tCAFs> iCAFs> dCAFs≈vCAFs≈apCAFs (Figure S9A, B). Trajectory mapping showed iCAFs and tCAFs spanning the entire pseudotime axis, whereas apCAFs, dCAFs and vCAFs accumulated later; ARNTL overexpression selectively reduced iCAFs and tCAFs abundance (Figure S9C-G). Gene expression at the early branch point (Figure S9H) coupled with pathway analysis demonstrated that iCAFs/tCAFs activate PI3K-Akt, MAPK and Wnt signaling to promote proliferation and invasion; their depletion in SKOV3-ARNTL tumors confirms ARNTL’s tumor-suppressive role in CAFs-mediated ovarian cancer progression. In sum, CAFs critically orchestrate ovarian cancer progression. Consistent with previous in vitro and in vivo analysis results, single-cell profiling showed that SKOV3 tumors harbored CAFs with up-regulated pro-migratory pathways and attenuated apoptosis relative to SKOV3-ARNTL tumors (Fig. 6I-J), underscoring ARNTL as a putative tumor-suppressor gene.

Fig. 6.

scRNA-seq Analysis of CAFs Dynamics During ARNTL overexpression (A), The UMAP Plot for Vehicle and ARNTL in CAFs; (B), The Cluster Cell types analysis; (C), Cellular phenotypes analysis; (D), Bubble Chart of the top five differentially expressed genes for each cell type in scRNA-seq data of all cells (n = 16,704); (E), Composition of cell quantity in Vehicle and ARNTL; (F), Composition of cell type in Vehicle and ARNTL; (G-H), Percent comparison among various cell subsets in in Vehicle and ARNTL; (I), Volcano plot of differentially expressed genes in CAFs between Vehicle and ARNTL; (J), GO functional enrichment analysis of differentially expressed genes in CAFs

ARNTL disrupts the feedback loop between cancer cells and cancer-associated fibroblasts

Leveraging ARNTL’s established role in sculpting CAFs heterogeneity and function within ovarian cancer, we hypothesize that enforced ARNTL expression in tumor cells can abrogate the phenotypic transition of normal fibroblasts (NFs) into cancer-associated fibroblasts (CAFs). NFs were isolated from ovariectomy specimens obtained from three patient, and their purity was confirmed using standard fibroblast biomarkers, including flat spindle like cell morphology, vimentin positive, and α-SMA negative expression (Figure S10A). Next, an indirect co-culture system was used to investigate tumor-stroma interactions. In this system, fibroblasts were seeded in the bottom well of the plate while ovarian cancer cells were plated on the top of the insert chamber (Fig. 7A). As shown in Fig. 7B-E, co-culture with ARNTL-overexpressing ovarian cancer cells resulted in decreased α-SMA expression in fibroblasts, a marker for activated fibroblasts. Meanwhile, RT-qPCR analysis revealed that fibroblasts co-cultured with ARNTL-overexpressing ovarian cancer cells exhibited reduced activation, as evidenced by decreased mRNA levels of PDGFR and FAP (Fig. 7D). Conversely, fibroblasts co-culture with ARNTL-depleted ovarian cancer cells displayed enhanced activatoion(Figure S10B-E). Similarly, fibroblasts treated with conditioned medium from ARNTL-overexpressing ovarian cancer cells exhibited reduced expression of α-SMA, FAP and PDGFR (Fig. 7F-H), whereas conditioned medium from ARNTL-depleted ovarian cancer cells produced the opposite effects (Figure S10F-H). Collectively, these data indicate that ARNTL knockdown in ovarian cancer cells promotes fibroblast activation within the tumor microenvironment.

Fig. 7.

ARNTL inhibits the positive feedback regulation between cancer cells and CAFs. A, Graphic scheme for the indirect co-culture system. B-C, Immunoblot analysis for amounts of α-SMA in fibroblasts treated with indicated co-culture conditions for 48 h. D, Real-time qPCR analysis for activation of fibroblasts treated in indicated co-culture systems for 48 h. Values are mean ± SEM. E, Representative images of immunofluorescence assay for α-SMA (green) or nucleus (DAPI) in fibroblast treated in indicated conditions for 48 h, scale bar: 400 μm. F-G, Immunoblot analysis for α-SMA activation in fibroblasts treated with indicated conditioned co-culture media for 48 h. H, Real-time qPCR analysis for activation of fibroblasts treated with indicated conditioned media for 48 h. Values are mean ± SEM. I, Representative images of invasion cells (top), scale bar: 100 μm. Relative numbers of invasion cells (down). Values are mean ± SD. J, Graphic scheme for indirect conditioned media system. K, Representative images of colony formation assays of ARNTL and Vehicle for indirect conditioned media system. L, Representative images for Transwell invasion assays of SKOV3 cells after incubation with indicated conditioned media for 48 h(top), scale bar: 100 μm. Relative numbers of invasion cells (down). Values are mean ± SD

Given that activated CAFs can modulate cancer progression via secretion of growth factors, cytokines, and extracelluar matrix components among others [35], we next sought to assess the feedback influence of these activated fibroblasts on tumor cells. As showed, fibroblasts either co-cultured with or exposed to conditioned medium from ARNTL-overexpressing ovarian cancer cells displayed decreased levels of tumor-promoting cytokines SDF-1 and PDGF (Fig. 7D, H), whereas those exposed to ARNTL-depleted ovarian cancer cells showed increased expression of these tumor-promoting cytokines (Figure S10D, H). Additionally, fibroblasts co-cultured with ARNTL-overexpressing ovarian cancer cells showed attenuated invasion property, which may limit their contribution to tissue fibrosis and tumor metastasis (Fig. 7I). Conversely, co-culture with ARNTL-depleted ovarian cancer cells increased fibroblast invasive capacity and promoted these pathological processes (Figure S10I). Next, we investigated the reciprocal influence of fibroblasts on ovarian cancer cells by assessing fibroblast-mediated changes in cancer cell proliferation and invasive capacity. Notably, colony formation and transwell invasion assays showed that fibroblast deactivated by ARNTL-overexpressing ovarian cancer cells significantly diminished the proliferation and invasive capacity of ovarian cancer cells (Fig. 7J-L), whereas fibroblast activated by ARNTL-depletd ovarian cancer cells improved these malignant behaviors (Figure S10J, K). Furthermore, to elucidate the role of specific cytokines in the ARNTL-CAF axis, we performed functional rescue experiments. Supplementation with recombinant SDF-1 in co-culture systems restored the invasive capacity and proliferation of ARNTL-overexpressing cells, as evidenced by increased invasion and larger colony formation (Figures S11A-B). Conversely, treatment with the SDF-1 inhibitor NOX-A12 attenuated the invasive and proliferative capacities of ARNTL-depleted cells, resulting in reduced invasion and smaller, sparser colonies (Figures S11C-D). These bidirectional rescue experiments establish the functional importance of SDF-1 in mediating the ARNTL-CAF regulatory axis. Taken together, these data demonstrate that fibroblasts deactivated by tumor cells with ARNTL overexpression diminish ovarian cancer progression and metastasis through negative feedback.

Discussion

This study found that the circadian transcription factor ARNTL is a key inhibitory node in the progression of ovarian cancer. Bioinformatics and functional experiments have shown that overexpression of ARNTL can significantly inhibit the proliferation, invasion, and metastasis of tumor cells, while its absence promotes malignant progression by regulating the tumor microenvironment. Collectively, these findings establish ARNTL as a novel progression-restricting regulator and provide mechanistic rationale for circadian-targeted combination therapies in ovarian cancer.

Although our current study demonstrates that ARNTL is lowly expressed in ovarian cancer and that this downregulation promotes tumor progression, ARNTL expression and function in ovarian cancer are modulated by multidimensional factors. At the environmental level, circadian disruption (e.g., long-term shift work) impairs ARNTL’s rhythmic amplitude, thereby increasing cancer susceptibility [36]. From a physiological perspective, aging further dampens circadian oscillations, leading to the loss of ARNTL cycling and elevated oxidative stress, which may drive the rising incidence in older women [37]. On the genetic and molecular fronts, the intronic SNP rs4757151 acts as an eQTL to modulate ARNTL expression and heighten disease risk [38]. Meanwhile, somatic mutations in core clock genes (such as CLOCK, PER3, and CRY1) can reduce ARNTL-CLOCK transcriptional activity by over 50%, potentially weakening its tumor-suppressive efficacy [39]. Given the context-dependent dynamic plasticity of ARNTL, this multifactorial regulation may confer it with functional duality, enabling it to switch roles between a tumor suppressor and an oncogenic promoter. Future research should integrate environmental exposure, aging, germline variants, and somatic mutations into a comprehensive experimental and clinical framework to elucidate its precise role in ovarian cancer progression and therapeutic response.

Tumor proliferation, invasion and metastasis constitute a multi-step cascade whose success is dictated by sustained and dynamic crosstalk between malignant cells and the host microenvironment [40, 41]. These interactions not only endow cancer cells with adaptive phenotypes, but also trigger functional aberrancies in stromal components, ultimately promoting in situ growth and distant metastasis and colonization of tumor cells [42, 43]. Among the diverse stromal populations within the tumor microenvironment (TME), CAFs represent the most abundant and functionally pivotal constituents [44–46]. CAFs serve as the principal architects of the tumor microenvironment, orchestrating tumor initiation and progression through their precise regulation of the extracellular matrix and signaling networks [47, 48]. In the tumor microenvironment, CAFs originate from reprogrammed NFs and orchestrate tumor progression and metastasis through a spectrum of secreted cytokines [49]. In our study, we found that the activated signaling in ARNTL-overexpresion cells play a critical role in regulating fibroblasts. We further detected that overexpression of ARNTL in tumor cells could inhibit the transition of NFs into CAFs. In summary, our findings provide new directions for the clinical diagnosis and treatment of ovarian cancer, indicating that ARNTL may regulate tumor progression and metastasis by reprogramming CAFs in ovarian cancer.

In addition, tumor associated fibroblasts (CAFs) play a dual role in tumor progression, which is due to the functional heterogeneity of their secreted factors and reshaped extracellular matrix (ECM) [50]. Recent single-cell RNA sequencing (scRNA-seq) advancements have categorized CAFs into distinct functional subpopulations, such as apCAFs, iCAFs, and tCAFs, which possess divergent pro- or anti-tumorigenic properties [51–53]. In the present study, scRNA-seq analysis of mouse ovarian cancer tissues revealed that ARNTL overexpression significantly suppresses pro-tumorigenic subtypes (apCAF, iCAF, vCAF) while enriching anti-tumorigenic cohorts (tCAF, dCAF). These findings, consistent with observation by Lena Cords et al., demonstrate that ARNTL exerts tumor-suppressive effects through modulation of CAF subpopulation dynamic and functional balance [53]. Although our study provides a preliminary outline of the landscape in which ARNTL regulates CAF heterogeneity, there are still certain limitations. For example, the inference of subpopulation abundance based on deconvolution algorithm is still a computational estimate and may not fully restore its actual spatial structure and in situ function. The causal relationship between transcriptome signals and biological flux still needs to be further confirmed through in vitro functional testing (such as Annexin V/PI, SASP analysis) and in vivo lineage tracing experiments. However, as a discovery-phase investigation, our study provides a comprehensive landscape of ARNTL-associated CAF heterogeneity and establishes high-priority hypotheses regarding circadian clock regulation of stromal remodeling and tumor microenvironment evolution. In summary, our study provides key preliminary clues for regulating tumor matrix remodeling through the circadian rhythm mechanism.

In conclusion, ARNTL depletion correlates with reduced tumor-suppressive CAF features and enhanced malignant phenotypes (proliferation, migration, invasion) in ovarian cancer, suggesting that ARNTL loss promotes tumor progression and metastasis through microenvironment reprogramming. Clinically, low ARNTL expression is strongly associated with poor tumor differentiation and adverse prognosis, establishing ARNTL defiency as a novel biomarker for risk stratification and a critical indicator for personalized therapeutic intervention.

Abbreviations

ARNTL

Aryl hydrocarbon receptor nuclear translocator like

CAFs

Cancer-associated fibroblasts

ECM

Extracellular Matrix

DEGs

Differentially Expressed Genes

NFs

Normal Fibroblasts

TCGA

The Cancer Genome Atlas

GTEx

Genotype-Tissue Expression project

CCLE

Ancer Cell Line Encyclopedia

TME

Tumor Microenvironment

GO

Gene Ontology

KEGG

Kyoto Encyclopedia of Genes and Genomes

MDS

Multidimensional scaling

CM

Conditioned Medium

MCP

Microenvironment Cell Populations (counter)

EPIC

Estimation of Proportions of Immune and Cancer cells

TIDE

Tumor Immune Dysfunction and Exclusion

Authors’ contributions

Yanna Zhang and Yan Chen conceived and supervised the whole project. Yanna Zhang, Jiaoya Lin, Nanxi Liu, Wanpei Chen, and Lu Zhang performed the data curation and analysis. Yanna Zhang wrote the manuscript, Yan Chen and Hong Yuan participated in the manuscript editing and discussion.

Funding

This investigation was supported by National Natural Science Foundation (32200957and 82003113), Sichuan Science and Technology Program (2024NSFSC1889) and Sichuan Provincial People’s Hospital Foundation (25QMPY004).

Data availability

The datasets used and/or analyzed in this study are available from the corresponding author on reasonable request. The dataset generated in the current study is available in the National Genomics Data Center. The submission of relevant data is as follows: [subHRA020913](https:/ngdc.cncb.ac.cn/gsa-human/submit/hra/subHRA020913/detail) ，subCRA049992，subCRA050909.

Declarations

Ethics approval and consent to participate

The research involving animals has been approved by the Animal Ethics Committee of West China Hospital of Sichuan University. All procedures followed the AVMA Guidelines for the Euthanasia of Animals (2020) and were approved by the Institutional Animal Care and Use Committee (protocol No. IACUC20220301146).

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.