Cell culture

The human thyroid cancer cell line 8305 C and cancer-associated fibroblasts (CAFs) were purchased from Procell Life Science & Technology. Cells were cultured in DMEM (Gibco) supplemented with 10% fetal bovine serum (ExCell Bio) and 1% penicillin-streptomycin (Beyotime) at 37 °C in a 5% CO₂ incubator. Cells were passaged at approximately 80% confluence using 0.25% trypsin and maintained under standard culture conditions. For long-term storage, cells were frozen in freezing medium and stored in liquid nitrogen.

Establishment of stable knockdown cell lines

Short hairpin RNAs (shRNAs) targeting ATF4, FGFR4, and a negative control (sh‑NC) were designed and cloned into the pLKO.1 vector (Sangon Biotech). Lentivirus packaging was performed by co-transfecting HEK293T cells with the shRNA plasmid and packaging plasmids using Lipofectamine 3000 (Thermo Fisher Scientific). Viral supernatants were collected 48 h post‑transfection, filtered through 0.45 μm filters, and stored at − 80 °C until use. 8305 C cells were infected with lentivirus and selected with puromycin (Sigma‑Aldrich) for 7 days. Knockdown efficiency was confirmed by RT-qPCR and Western blot.

RNA sequencing and bioinformatic analysis

To investigate transcriptomic alterations associated with ATF4/FGFR4 silencing, total RNA was extracted from sh-ATF4, sh-FGFR4, and sh-ATF4 + sh-FGFR4 8305 C cells, with three biological replicates per group, using the FastPure RNA Kit (Vazyme) according to the manufacturer’s instructions. RNA quality and concentration were assessed before library preparation. Strand-specific RNA-seq libraries were constructed and sequenced by LC-Bio Technologies (Hangzhou) Co., Ltd. on an Illumina NovaSeq 6000 platform using a paired-end 150-bp sequencing strategy. An average of approximately 41.47 million clean reads was obtained per sample.

Raw sequencing reads were assessed using FastQC v0.10.1 and processed with Cutadapt to remove adapter sequences and low-quality reads. Clean reads were aligned to the human reference genome using HISAT2 v2.2.1, and transcript assembly and gene-level quantification were performed using StringTie v2.1.6. Pairwise differential-expression analyses were conducted using DESeq2 v1.22.2, whereas the overall comparison among the three knockdown groups was performed using edgeR v3.22.5. For pairwise comparisons, genes with an absolute log2 fold change ≥ 1 and a nominal P value < 0.05 were considered differentially expressed. For the multiple-group comparison, genes with a nominal P value < 0.05 were considered differentially expressed, without applying a fold-change cutoff. Benjamini–Hochberg-adjusted q values were also calculated and reported in the differential-expression output tables.

Differentially expressed genes identified in the multiple-group comparison were subjected to Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses. Gene- and transcript-expression distributions, principal component analysis, hierarchical clustering, and enrichment plots were generated to summarize and visualize the transcriptomic data. Sequencing and read-mapping quality-control metrics are provided in Supplementary Table S4.

Cell proliferation assay

Cell proliferation was measured using the EdU Apollo kit (RiboBio). Cells were seeded in 96-well plates and incubated with 50 µM EdU for 2 h. After fixation with 4% paraformaldehyde, cells were permeabilized, stained following the manufacturer’s instructions, and counterstained with DAPI. Fluorescent images were captured using an EVOS M5000 microscope (Thermo Fisher), and the percentage of EdU-positive cells was quantified with ImageJ software.

Wound-healing assay

Cells were seeded in 6-well plates and grown to confluence. Linear scratches were made using a sterile 200 µL pipette tip, and cells were washed twice with PBS to remove debris. Cells were then incubated in serum-free medium. Images were taken at 0 h and 48 h, and wound closure was quantified.

Transwell invasion assay

Transwell chambers (8 μm; Corning) were pre-coated with Matrigel. Cells (2 × 10⁵ cells/mL) in serum-free medium were added to the upper chamber, while medium containing 10% FBS was placed in the lower chamber. After 24 h, non-invading cells were removed, and invading cells on the lower surface were fixed, stained with crystal violet, and counted in three random fields under a microscope.

CAF-conditioned medium treatment

CAFs were cultured to 80% confluence, and the supernatant was collected, centrifuged, and filtered. 8305 C cells from the sh-NC, sh-ATF4, sh-FGFR4, or sh-ATF4 + sh-FGFR4 groups were then cultured in this CAF-conditioned medium for 48 h before subsequent assays, including proliferation, migration, invasion, apoptosis, molecular expression analysis, immunofluorescence, and ELISA.

Apoptosis assay

Apoptosis was detected using the Annexin V-FITC/PI kit (Beyotime). Treated cells were collected, washed, and stained with Annexin V-FITC and PI for 15 min in the dark. Samples were analyzed by flow cytometry (Attune NxT, Thermo Fisher).

RNA extraction and RT-qPCR

Total RNA was extracted using the FastPure RNA Kit (Vazyme) and reverse-transcribed into cDNA using HiScript III (Vazyme). RT-qPCR was performed on a CFX96 Touch (Bio-Rad) using SYBR Green mix (Vazyme). Primers were as follows: ATF4: Forward 5’-GGGATCGGGAAAGCGTAGTC-3’, Reverse 5’-TCTTGGTTCCTGCCACGTTT-3’. FGFR4: Forward 5’-GAGGAGGACCCCACATGGA-3’, Reverse 5’-AGGGATGAGCTTGACTTGCC-3’. GAPDH: Forward 5’-AATGGGCAGCCGTTAGGAAA-3’, Reverse 5’-GCGCCCAATACGACCAAATC-3’. Relative expression was calculated using the 2⁻ΔΔCt method.

Western blotting

Total protein was extracted from cells using RIPA lysis buffer (Beyotime) containing PMSF (Beyotime), and protein concentrations were determined using a BCA Protein Assay Kit (NCM Biotech). Equal amounts of protein were separated by SDS-PAGE and transferred onto PVDF membranes (Sigma-Aldrich). After blocking with 5% skim milk, membranes were incubated overnight at 4 °C with primary antibodies against ATF4 (Proteintech, 81798-1-RR, 1:2,000) and FGFR4 (Proteintech, 81069-1-RR, 1:8,000), followed by HRP-conjugated goat anti-rabbit IgG H&L secondary antibody (Bioss, bs-0296G-HRP, 1:20,000). Protein bands were visualized using an ECL substrate (NCM Biotech, P2100) and quantified with ImageJ.

Immunofluorescence staining of cultured cells

Treated 8305 C cells were fixed with 4% paraformaldehyde, permeabilized with 0.25% Triton X-100, and blocked with goat serum. Cells were incubated overnight at 4 °C with anti-ATF4 (Proteintech, 60035-1-Ig, 1:200) and anti-FGFR4 (Proteintech, 11098-1-AP, 1:200), followed by CoraLite Plus 555-conjugated goat anti-mouse IgG (Proteintech, RGAM003, 1:800) and CoraLite 488-conjugated goat anti-rabbit IgG (Proteintech, SA00013-2, 1:800). Nuclei were counterstained with DAPI, and images were captured under identical settings and quantified using ImageJ.

Enzyme-linked immunosorbent assay

Culture supernatants were collected after CAF-conditioned medium treatment and centrifuged to remove debris. IL-6 and TNF-α levels were measured using commercial ELISA kits (Hnybio) according to the manufacturers’ instructions. Absorbance was read at 450 nm, and cytokine concentrations were calculated from standard curves.

In vivo xenograft model

Four-week-old male BALB/c nude mice were randomly divided into four groups. After pre-culture in CAF-conditioned medium, 5 × 10⁶ 8305 C cells from each group (sh-NC, sh-ATF4, sh-FGFR4, or sh-ATF4 + sh-FGFR4) were injected subcutaneously into the flank of each mouse. After four weeks, mice were euthanized, tumors were excised, weighed, and processed for histological analysis. All animal experiments were approved by the Animal Ethical and Welfare Committee (AEWC) of Guangzhou Miles Biosciences (Approval No. MIS2026006) and were performed in accordance with institutional guidelines for the care and use of laboratory animals.

Histological, immunohistochemical, and immunofluorescence analyses

Tumor tissues were fixed in 4% paraformaldehyde, paraffin-embedded, and sectioned at 4 μm. H&E staining was performed for morphological evaluation. For immunohistochemistry, sections were incubated with antibodies against ATF4 (Abcam, ab184909, 1:200) and FGFR4 (Abcam, ab41948, 1:200), followed by a ready-to-use secondary antibody reagent (ZSGB-BIO) and DAB visualization kit (ZSGB-BIO). For immunofluorescence, sections were incubated with antibodies against α-SMA (Abcam, ab7817, 1:200), vimentin (Abcam, ab92547, 1:200), and FAP (R&D Systems, MAB3715, 1:200), followed by Alexa Fluor 555 goat anti-mouse IgG (Thermo Fisher, A-21422, 1:500), Alexa Fluor 488 goat anti-rabbit IgG (Thermo Fisher, A-11008, 1:500), and Alexa Fluor 594 donkey anti-rat IgG (Thermo Fisher, A-21209, 1:500). Nuclei were counterstained with DAPI and images were analyzed using ImageJ.

TUNEL assay

Apoptotic cells in tumor sections were detected using a TUNEL Apoptosis Detection Kit (Beyotime) according to the manufacturer’s instructions. Briefly, paraffin sections were deparaffinized, rehydrated, incubated with TUNEL reaction mixture at 37 °C for 60 min in the dark, and observed under a fluorescence microscope. TUNEL-positive cells were quantified in three random fields per section.

Statistical analysis

All data were expressed as mean ± SD from at least three independent experiments. Comparisons between two groups were performed using Student’s t-test, and multiple groups were analyzed by one-way ANOVA with Tukey’s post hoc test. A P value < 0.05 was considered statistically significant.