Bioinformatics analysis

The online tools ENCORI (http://starbase.sysu.edu.cn/index.php), CircInteractome(https://circinteractome.nia.nih.gov/) and miRDB (http://www.mirdb.org/) were used for predicting potential interactions between circRNAs and miRNAs [20, 21]. miRDB, ENCORI, and TargetScan (http://www.targetscan.org/mamm_31/) were utilized to predict miRNA target genes. HCC cohort datasets from The Cancer Genome Atlas (TCGA) (https://www.cancer.gov/ccg/research/genome-sequencing/tcga), Gene Expression Profiling Interactive Analysis (GEPIA) (http://gepia.cancer-pku.cn/), and Gene Expression Omnibus (GEO) (https://www.ncbi.nlm.nih.gov/geo/) were analyzed to compare gene expression levels between tumor and normal tissues, and to assess gene correlations. Gene Set Enrichment Analysis (GSEA) was conducted using the R package “Cluster Profile” to investigate biological functions in control and sh-circPIAS1 HCC cells.

Reagents

The following commercially available antibodies were utilized: NUPR1 (15056-1-AP, Proteintech, 1:1000 for WB, and 1:400 for IHC), GAPDH (60004-1-Ig, Proteintech, 1:2000 for WB), FTH1 (4393T, Cell Signaling Technology, 1:1000 for WB and 1:400 for IHC), GPX4 (52455T, Cell Signaling Technology, 1:1000 for WB), SLC11A2 (15083T, Cell Signaling Technology, 1:1000 for WB), SLC7A11 (12691T, Cell Signaling Technology, 1:1000 for WB), KEAP1 (8047T, Cell Signaling Technology, 1:1000 for WB), NRF2 (12721T, Cell Signaling Technology, 1:1000 for WB), Ago2 (07-590, Merck Millipore), IgG rabbit (7074 S, Cell Signaling Technology), IgG mouse (12-371B, Merck Millipore). Small molecule inhibitors included ZZW-115 (HY-111,838, MedchemExpress), Z-VAD-FMK (S7023, Selleck), Necrostatin-1 (S8037, Selleck), Ferrostatin-1 (S7243, Selleck), and Liproxstatin-1 (S7699, Selleck). All other chemical reagents were obtained from Sigma-Aldrich, unless specified otherwise.

Tissue samples

Thirty-six paraffin-embedded HCC samples were obtained from Sun Yat-sen University Cancer Center (Guangzhou, China) survival prognosis analysis. To compare circPIAS1 levels in HCC versus non-tumor tissues, 8 pairs of fresh HCC samples and their adjacent non-cancerous counterparts were obtained from the Third Affiliated Hospital of Sun Yat-sen University. A surgical tumor resection was performed on each patient in the department of hepatobiliary surgery. The study was approved by the Institute Research Ethics Committee at the Sun Yat-sen University Cancer Center and the Third Affiliated Hospital of Sun Yat-sen University. Written informed consent was obtained from each patient.

Immunohistochemistry (IHC) analysis

Tissue samples were paraffin-embedded and cut into 5-µm sections. The slides were then heated at 65 °C for 120 min, followed by de-paraffinized, hydrated and antigen retrieval. Endogenous peroxidases were blocked with 3% peroxide for 15 min, then in 5% BSA for 1 h and incubated overnight at 4 °C with the indicated primary antibodies. Subsequently, anti-rabbit/mouse secondary antibodies (K5007, Dako) were applied and incubated for 60 min at 37 °C. Signals were revealed using freshly prepared DAB substrate solution (K5007, Dako) at room temperature for 5 min. Finally, the sections were counterstained with Mayer’s hematoxylin, dehydrated, and mounted. Staining results were captured using microscopy (DM4000B, Leica). Each section was evaluated by two independent pathologists blinded to the clinical status of patients and graded according to the positive staining intensity scores (no staining, 0; weak staining, 1; moderate staining, 2; strong staining, 3) and the expression extent scores (< 25%, 1; 25–50%, 2; 50–75%, 3; > 75%, 4).

CircPIAS1 RNAScope in situ hybridization assay

CircPIAS1 expression in HCC tissues was examined using RNA in situ hybridization with the BaseScope™ RED Kit 2 (Advanced Cell Diagnostics, USA), following the manufacturer’s protocol. A custom-designed probe specific to circPIAS1 was employed. Detection sensitivity for CircPIAS1 molecules was achieved at the single-copy level. Quantification of single-molecule signals was conducted on a cell-by-cell basis through manual counting. A probe targeting human PPIB mRNA (476,701, Advanced Cell Diagnostics) was used as the positive control, while a probe targeting Bacillus subtilis DapB mRNA (310,043, Advanced Cell Diagnostics) served as the negative control. Cells were considered positive if they displayed a visible red dot or cluster at 40× magnification under a microscope (DM4000B, Leica). Scoring criteria were as follows: circPIAS1 low, 0 ≤ positive cells < 3 per 20 cells; circPIAS1 high, positive cells ≥ 3 per 20 cells.

Cell culture

The human embryonic kidney cell line HEK-293T, several human HCC cell lines (Huh7, PLC/PRF/5, Hep3B, LM3, MHCC-97 H, MHCC-97 L, SNU449, SNU387, and HepG2), and two immortalized hepatic cell lines (MIHA and LO2) were sourced from the College of Life Sciences, Sun Yat-sen University and employed in this study. All cell lines were cultured in Dulbecco’s Modified Eagle’s Medium (DMEM, Invitrogen) supplemented with 10% fetal bovine serum (FBS, Gibco) at 37 °C and 5% CO2. Thawed from early passage stocks, the cells were regularly subcultured every 2 days. Bimonthly PCR assays were conducted to confirm the absence of mycoplasma contamination.

Reverse transcriptase PCR (RT-PCR) and quantitative real-time PCR (qPCR)

RNA was extracted using TRIzol (T9424, Sigma-Aldrich) and reverse transcribed with the GoScript system (A5001, Promega). RT-PCR was performed using PrimeSTAR Master Mix (R045A, Takara) according to the manufacturer’s instructions, including PCR control. Products were separated on a 2% agarose gel and visualized with GelRed (D0140, Beyotime). qPCR was conducted on a LightCycler 480 (Roche) with Platinum SYBR Green mix (11,744,500, Invitrogen). CircRNA, mRNA, and miRNA expression levels were normalized to GAPDH and U6, respectively, using the 2−ΔΔCt method for determining gene expression level [20]. Primer sequences are listed in Supplementary Table 1.

Plasmid construction, oligonucleotide synthesis and transfection

The pLO5-ciR plasmid (Geneseed, Guangzhou, China), containing the sequence of circPIAS1, was constructed and used to upregulate circPIAS1 expression. Two specific short hairpin RNAs (shRNAs) targeting the covalent closed junction of circPIAS1 were cloned into the pLKO.1 plasmid (Sigma-Aldrich). The coding sequence of NUPR1 with a C-terminus Flag tag was constructed in the pcDNA3.1(+) plasmid (Invitrogen). Correct constructs were confirmed by DNA sequencing. The oligonucleotides of the miR-455-3p mimic/inhibitor and controls were synthesized by RiboBio (Guangzhou, China). Plasmids and oligonucleotides were transfected using ViaFect Transfection Reagent (Promega) or Lipofectamine RNAiMax (Invitrogen), respectively, according to the manufacturer’s instructions. The shRNA sequences targeting the circPIAS1 covalent closed junction were as follows: sh-circPIAS1#1, 5’-TATTGATGGCATCAGACAACA-3’, and sh-circPIAS1#2, 5’-TTGATGGCATCAGACAACAGT-3’. The non-targeting shRNA sequence served as a negative control: shNC, 5’-CAACAAGATGAAGAGCACCAA-3’.

Lentivirus packaging and infection

Lentiviruses for circPIAS1 overexpression and knockdown were produced by co-transfecting constructed plasmids and the packaging plasmids psPAX2 and pMD2.G (Addgene) into 293T cells for 48 h. Culture supernatants containing lentivirus were collected, filtered, and concentrated. Lentiviruses expressing miR-455-3p (pHBLV-h-miR-455-3p-puro) or miR-455-3p inhibitor (pHBLV-h-sh-miR-455-3p-puro) were purchased from HanBio (Shanghai, China). HCC cells were infected with lentivirus in the presence of 8 µg/mL polybrene (Sigma-Aldrich). Infected cells were screened with 2 µg/mL puromycin (Merck) for 2 weeks, and successful establishment was confirmed by qPCR.

Luciferase reporter assay

Reporter plasmids containing the wild-type or mutant miR-455-3p putative binding site were constructed by inserting the circPIAS1 and 3’UTR of NUPR1 sequences into the pMIR-REPORT vector (Invitrogen). These constructs were then co-transfected with miR-455-3p mimics or inhibitors. The FTH1 promoter sequence was inserted into the pGL3-reporter vector (Promega). Luciferase activities of all reporter vectors were evaluated using the dual luciferase reporter assay system (Promega), with Renilla luciferase used for normalization.

RNase R and actinomycin D treatment

For RNase R treatment, 2 µg of total RNA was incubated with 5 U/µL RNase R (RNR07250, Lucigen, USA) at 37 °C for 30 min before reverse transcription. For actinomycin D treatment, HCC cells were cultured in six-well plates and treated with 5 µg/mL actinomycin D (D23070, Sigma-Aldrich) when they reached approximately 60% confluence. The cells were treated for the indicated time intervals. The expression levels of circPIAS1 and linear PIAS1 mRNA were analyzed using qPCR.

Western blotting

The cells were lysed with NETN buffer (20 mM Tris-HCl at pH 8.0, 100 mM NaCl, 1 mM EDTA, and 0.5% Nonidet P-40 (56,741, Sigma-Aldrich)) supplemented with protease and phosphatase inhibitors (Thermo Fisher Scientific). The lysate protein concentration was measured using the BCA protein assay kit (Pierce). After equalization, 10 µg of each protein sample was processed via SDS-PAGE, transferred to PVDF membranes, and blocked using 5% non-fat milk (232,100, BD Biosciences) diluted in 1× Tris-buffered saline supplemented with 0.5% Tween-20 (TBST). These membranes were incubated overnight at 4 °C with primary antibodies, followed by their HRP-labeled secondary counterparts (W4011 for rabbit and W4021 for mouse originated primary antibodies, Promega). The immunoreactive bands were visualized by the enhanced chemiluminescence (ECL). The primary antibodies were diluted with Antibody Dilution Buffer (P0023A, Beyotime) and the secondary antibodies were diluted in 1× TBST. GAPDH was used as the control. Band quantification was done via ImageJ.

CCK-8 and colony formation assays

Cell viability was evaluated using a CCK-8 kit (CK04, Dojindo). The cells were seeded in 5 replicates in a 96-well plate at a density of 1,000 cells and cultured with 100 µL DMEM containing 10% FBS per well. At the indicated time point, 10 µL of the CCK-8 solution was added to each well, and the cells were incubated for another 2 h at 37 °C. The OD value at 450 nm was then measured using a microplate reader (ELx800, BioTek). For colony formation, HCC cells (1,000/well) were placed in six-well plates. After 14 days, cells were fixed with 4% paraformaldehyde, tinted with 0.1% crystal violet (C6158, Sigma), and left to dry. Colonies, identified as groups of 50 or more cells, were counted under a light microscope. The procedure was conducted three times.

Wound healing assay

HCC cells were plated to full confluency on a 96-well dish, subjected to scratching, and then cultured under serum-free conditions for an additional 48 h. Images were captured at 0 and 48 h to document changes in wound width.

Transwell assay

HCC cells (50,000 cells/well) were seeded into the upper chamber of 24-well Transwell plates with 8 μm-pore size (3374, Corning, USA) without Matrigel coating, while the lower chamber was filled with DMEM containing 10% FBS as a chemoattractant. After incubating for 24 h, non-migrating cells on the upper side of the chamber were removed by scrubbing, and migrating cells on the lower side were fixed with 4% paraformaldehyde and stained with crystal violet. The number of migrating cells was assessed in six random microscope fields.

Measurement of Fe2+, lipid reactive oxygen species (ROS), and GSH levels

To evaluate lipid ROS and Fe2+ levels in HCC cells, C11-BODIPY581/591 (10 µM, GC40165, GLPBIO) or FerroOrange (5 µM, F374, DOJINDO) was introduced into the cell culture medium supernatant and incubated for 30 min in the dark, followed by PBS washing. The cells were then visualized using a fluorescence microscope (DMi8, Leica). GSH levels were determined using a GSH/GSSG Assay Kit (S0053, DOJINDO) following the manufacturer’s instructions.

RNA fluorescence in situ hybridization (FISH)

The FISH assay was conducted using the Fluorescent In Situ Hybridization Kit (H0101, GenePharma, China) according to the manufacturer’s instructions. A total of 30,000 cells were seeded into confocal dishes, fixed, and then exposed to Cy3-labelled miR-455-3p and FAM-labelled circPIAS1 probes (Servicebio, China). Cell nuclei were counterstained with DAPI. Fluorescence was excited and imaged using a confocal laser scanning microscope (DMi8, Leica).

RNA immunoprecipitation (RIP) assay

The RIP assay was conducted using the Magna RIP RNA-Binding Protein Immunoprecipitation Kit (17–700, Merck) according to the manufacturer’s instructions. In brief, the protein A/G magnetic beads were first incubated with specific antibodies and anti-IgG controls. Subsequently, cells were lysed using the kit’s lysis buffer and incubated overnight with protein A/G magnetic beads coated with antibodies. After washing, the lysates were digested with protease and RNase inhibitors for purification. The isolated RNA underwent qPCR evaluation.

RNA pulldown assay

The RNA pull-down procedure used the RNA pull-down kit (20,164, Thermo Scientific) according to the manufacturer’s instructions. Biotinylated probes for circPIAS1 and control sequences were custom-designed and obtained from Sangon Biotech (China). Approximately 1 × 107 cells were lysed in lysis buffer after washing with ice-cold PBS, followed by incubation at room temperature with 3 µg of biotinylated probes for 2 h. Subsequently, the biotin-coupled RNA complex was pulled down by incubating the cell lysates with streptavidin magnetic beads for an additional 4 h. The beads were then washed five times with lysis buffer, and bound miRNAs in the pull-down materials were extracted using Trizol reagent. A qPCR assay was conducted to analyze the bound miRNAs. The sequences of the biotin-labeled circPIAS1 and control probes are as follows: circPIAS1 probe: UCUCGAAAGCGCUGACUGUUGUCUGAUGCCAUCAAUAAUAAGGUGUUCAUAUGGAGCCUUCUU; control probe: UUGUACUACACAAAAGUACUG.

Separation of nuclear and cytoplasm fractions

RNA extraction from the nuclear and cytoplasmic fractions was carried out using the PARIS Kit (AM1921, Life Technologies) according to the manufacturer’s instructions. Following extraction, qPCR was employed to determine the relative RNA levels from each fraction. Nuclear control transcripts U3 and U6 were utilized, with GAPDH mRNA acting as the cytoplasmic marker.

Chromatin immunoprecipitation (ChIP) assay

ChIP assays were performed using the EZ-Magna ChIP A/G Kit (17-10086, Millipore) according to the manufacturer’s instructions. Briefly, HCC cells were cross-linked with 1% formaldehyde (F8775, Sigma-Aldrich) at room temperature for 10 min and then quenched with glycine. After washing, cells were lysed in the lysis buffer at 4 °C for 30 min and sonicated (Sonifier 450D, Branson) (50% amplitude, 10 s pulse, 30 s rest on ice, 4 cycles) to generate DNA fragments (200-1,000 bp in length). A total of 10 µg of protein-DNA complexes were immunoprecipitated with the indicated antibodies or isotype-matched IgG. The immunoprecipitated DNA was then purified and utilized for qPCR analysis. The ChIP-qPCR primer sequences are shown in Supplementary Table 1.

In vivo animal study

All mice (Charles River Laboratories, China) were handled in compliance with the Guide for the Care and Use of Laboratory Animals and approved by the Institutional Animal Care and Use Committee of the Third Affiliated Hospital of Sun Yat-sen University. They were housed in standard, specific pathogen-free conditions with ad libitum access to rodent laboratory chow and tap water, maintaining a temperature of 24 ± 1 °C, a humidity of 50 ± 10%, and a 12:12 h light/dark cycle. Food and water were freely available throughout the study.

For in vivo tumor growth monitoring, BALB/c-nude mice aged 3–4 weeks were subcutaneously injected with 1 × 106 HCC cells. Tumors were measured every two days, and on the final day, they were excised, weighed, and their volumes calculated using the formula: tumor volume = π/6 × large diameter × smaller diameter2.

To evaluate in vivo metastasis, BALB/c-nude mice aged 5–6 weeks were intravenously injected with 100 µL of HCC cell suspension (5 × 106/mL) through the tail vein. After 8 weeks, mice from each group were euthanized, and their lung tissues were collected for the evaluation of metastatic foci using standard histopathological methods.

To assess the treatment effect of ZZW-115, 5 × 105 HCC cells were subcutaneously injected into 4- to 6-week-old BALB/c-nude mice. When the tumor volumes averaged approximately 200 mm3, mice were randomly assigned to specified groups. Treatments were administered every five days per week, including: 0.5% DMSO in PBS (carrier), lenvatinib 30 mg/kg (oral), ZZW-115 5 mg/kg (i.p. injection), or a combination of lenvatinib 30 mg/kg and ZZW-115 5 mg/kg. Mouse weight and tumor volume were measured every five days. After 25 days, mice were euthanized, and subcutaneous tumors were excised for routine hematoxylin and eosin (H&E) staining and subsequent IHC analysis, following the manufacturer’s instructions.

Statistical analysis

Each experiment was performed thrice for technical consistency. Error bars, unless specified otherwise, denote standard deviation (SD). Data were analyzed using GraphPad Prism 6.0, with pairwise group comparisons conducted using a two-tailed, unpaired Student’s t-test, and multiple comparisons analyzed using one-way or two-way ANOVA. Statistical significance was represented as *p < 0.05, **p < 0.01, and ***p < 0.001, with ns indicating no significance.