Erastin (cat. no. HY-15763), RSL3 (cat. no. HY-100218A), ferrostatin-1 (cat. no. HY-100579), Z-VAD-FMK (cat. no. HY-16658B) and necrosulfonamide (cat. no. HY-100573) were purchased from MedChemExpress. Imidazole ketone erastin (cat. no. T5523) was purchased from Topscience. Primary antibodies, X cystine/glutamate antiporter (xCT; cat. no. T57046; 1:1,000; Abmart Pharmaceutical Technology Co., Ltd.), GPX4 (cat. no. T56959; 1:1,000; Abmart Pharmaceutical Technology Co., Ltd.), DGCR8 (cat. no. ET1609-32; 1:1,000; HuaBio), Ubiquitin (cat. no. T55965; 1:1,000; Abmart Pharmaceutical Technology Co., Ltd.) and β-actin (cat. no. P30002; 1:1,000; Abmart Pharmaceutical Technology Co., Ltd.). Secondary anti-rabbit antibodies were from Cell Signaling Technology, Inc. (cat. no. 7074S; 1:2,000).

RNA sequencing was provided by CloudSeq Biotech Inc. A total of six male HOTTIP KO mice and male WT mice were provided by Cyagen Biosciences. The mice were bred in a specified pathogen-free environment at temperature and humidity controlled (26°C and 50% humidity) environment with 12-h light/dark cycle, and served a standard diet. The mice were 4–6 weeks old and weighed 20–22 g and consisted of three HOTTIP KO mice and three WT mice. They were sacrificed and their bone marrow was isolated, total RNA was extracted using Triquick Reagent (cat. no. R1100; Beijing Solarbio Science & Technology Co., Ltd.). Total RNA was isolated and reverse transcribed into cDNA to generated an indexed Illumina library (Illumina, Inc.), including fragmentation (~300 bp), adapter ligation (unique dual indexes), and PCR amplification (15 cycles). Library quality was validated via Agilent 2100 Bioanalyzer (Agilent Technologies, Inc.; peak size: 280-320 bp) and quantified by Qubit Flex Fluorometer (Invitrogen; Thermo Fisher Scientific, Inc.), followed by sequencing on an Illumina Novaseq platform (Illumina, Inc.).

Human OS cell lines including MG63 and U2OS were provided by Lingnan Medical Research Center, Guangzhou University of Chinese Medicine (Guangzhou, China). Cell lines were authenticated by STR profiling (ATCC) and tested negative for mycoplasma. The two OS cell lines were cultured in Dulbecco's modified Eagle's medium (DMEM; Shanghai VivaCell Biosciences, Ltd.) with 10% fetal bovine serum (FBS; Shanghai ExCell Biology, Inc.) and 1% penicillin/streptomycin (Invitrogen; Thermo Fisher Scientific, Inc.). These cells were incubated and maintained in a humidified atmosphere at 37°C, 5% CO₂.

The OS cells were seeded into 96-well plates and cultured for 12 h. Then, cells were starved using 2% FBS for 12 h, the Fer-1 group was preincubated with 2 µM Fer-1 (MedChemExpress). Subsequently, cells were incubated with 10% FBS and induced by 10 µM Erastin (MedChemExpress) and 1 µM RSL3 (MedChemExpress) for 24 h. The cell viability was detected by Cell Counting Kit-8 (Beyotime Institute of Biotechnology) examination and the absorbance was measured at 450 nm using a Hybrid Multi-Mode Microplate Reader (Tecan Group, Ltd.). All the experiments were performed in triplicates.

The MG63 and U2OS cells were seeded into 6-well plate and treated with 10 µM Erastin and 2 µM Fer-1 at 37°C in a 5% CO₂ incubator for two weeks. At room temperature, the cells were fixed with 4% paraformaldehyde for half an hour and stained with crystal violet staining solution for half an hour. The images were captured using the ImmunoSpot analyzer (Cellular Technology Ltd.) and the colony numbers were counted by ImmunoSpot version 6.0 Academic system (Cellular Technology Ltd.).

The OS cells were seeded and ferroptosis was induced by Erastin and RSL3 at 37°C in a 5% CO₂ incubator for 24 h. The samples were collected for the ferroptosis examination. The level of intracellular iron was detected using an Iron Assay kit (Applygen Technologies, Inc.) according to the manufacturer's instructions. For intracellular ROS staining, the lipid ROS level was examined by flow cytometry with a 5 µM DHE fluorescence probe (Applygen Technologies, Inc.) in a visible spectrum of red range (emission maximum 610 nm). The levels of malondialdehyde and glutathione were evaluated by a lipid peroxidation (MDA) assay kit and a reduced glutathione (GSH) assay kit (Njjcbio).

Total RNA from 5×10⁵ cells was extracted using Animal Total RNA Isolation Kit (Chengdu Fuji Biotechnology Co., Ltd.) and it was reversely transcribed using PrimeScript RT Reagent Kit (Takara Bio, Inc.) following the manufacturer's instructions. The RT-qPCR examinations were conducted using Power Up TB Green Master Mix (Takara Bio, Inc.) on an ABI-QuantStudio 5 System (Thermo Fisher Scientific, Inc.). The cycling conditions were as follows: Initial denaturation at 95°C for 5 min, followed by 25–45 cycles of denaturation at 95°C for 15 sec, annealing at 60°C for 30 sec (depending on the target gene), and elongation at 72°C for 1 min, with a final extension at 72°C for 5 min. The reactions were stopped during the exponential phase to ensure accurate comparisons. The primer sequences are listed in Table I. GAPDH served as the endogenous control and fold changes were calculated using the relative quantification (2^−ΔΔCq) method (13). All the experiments were performed in triplicates.

Total protein was lysed using Radio Immunoprecipitation Assay (RIPA) buffer supplemented with protease and phosphatase inhibitor (Beyotime Institute of Biotechnology). The supernatant fraction was collected by centrifugation and quantified by BCA assay (Thermo Fisher Scientific, Inc.). Subsequently, an equal volume of protein mixture (30 µg) were separated by 10% SDS-PAGE and transferred to a PVDF membrane (MilliporeSigma). The membranes were blocked with 5% fat-free milk (Bio-Rad Laboratories, Inc.) for 1 h at room temperature and incubated at 4°C overnight with primary antibodies. Next, the membranes were incubated with secondary antibodies in dark at room temperature for 1 h. The expression levels were visualized by chemiluminescence (MilliporeSigma). Densitometric analysis was carried out using ImageJ software (ver. 1.46; National Institutes of Health).

Co-IP was conducted with a magnetic IP kit (cat. no. 88804; Thermo Fisher Scientific, Inc.). Briefly, 1×10⁷ cell lysates were gently rotated at 4°C overnight with anti-DGCR8 antibody, or normal rabbit IgG (Wuhan Servicebio Technology Co., Ltd.). Afterwards, pre-washed protein A/G magnetic beads were incubated with rotating at 4°C for 16 h. The beads were collected and washed three times. Western blotting was performed after eluting bound proteins with elution buffer.

RIP was performed as mentioned before (14) using RNA Binding Protein Immunoprecipitation Kit (cat. no. JKR23003; Wuhan Gene Create Biological Engineering Co., Ltd.). According to the manufacturer's protocol, the A/G protein magnetic beads were pre-incubated with anti-DGCR8 antibody or IgG (Wuhan Servicebio Technology Co., Ltd.) at 4°C overnight. The 1×10⁷ cells were washed with cold PBS twice and lysed with 200 µl lysis buffer. Then, the lysates were incubated with the pre-coated beads at 4°C for 16 h. The total RNA was extracted with TRIzol^® reagent (Invitrogen; Thermo Fisher Scientific, Inc.) and detected by RT-qPCR examination.

Stable HOTTIP overexpression or knockdown cell lines were constructed using 2nd lentiviral system as previously reported (15). Briefly, the 4 µg lentivirus was generated in 293T cells (The Cell Bank of Type Culture Collection of The Chinese Academy of Sciences) by co-transfecting with two packaging vectors (3 µg psPAX2; 1 µg pMD2G). Lentiviral particles were harvested 48 h post-transfection, followed by infection of OS cells for 24 h. Cells (30% confluence) were seeded in cell culture dishes and were infected with lentiviral plasmids at multiplicity of infection (MOI) of 10. After 37°C and screening of the cells for 7 days with medium containing 1 µg/ml puromycin (Beyotime Institute of Biotechnology), these virus particles infected OS cells and the stable cell lines were developed. The sequence of short hairpin (sh)HOTTIP were Sense: 5′-nnnnnGGCACTTTATATGCTGTAAnnnnnnnnnTTACAGCATATAAAGTGCCnnnnnn-3′; Anti-sense: 5′-nnnnnnnnnnGGCACTTTATATGCTGTAAnnnnnnnnnTTACAGCATATAAAGTGCCn-3′. Has-miR-214-3p mimics were designed and synthesized by Beijing Tsingke Biotech Co., Ltd. These oligoes were transfected in OS cells using Lipofectamine^® 3000 (Invitrogen; Thermo Fisher Scientific, Inc.). The sequence of miR214-3p were Sense: 5′-ACAGCAGGCACAGACAGGCAGU-3′; Anti-sense: 5′-ACUGCCUGUCUGUGCCUGCUGU-3′.

The GPX4 3′UTR-untranslated region (3′UTR) containing wild-type or mutated miR-214-3p binding sites were respectively synthesized and inserted into pmirGLO luciferase vector (Wuhan Gene Create Biological Engineering Co., Ltd.) and then co-transfected with miR-214-3p mimics into OS cells using Lipofectamine^® 3000 (Invitrogen; Thermo Fisher Scientific, Inc.) at 37°C for 48 h. The cell lysates were collected 48 h post transfection and Firefly and Renilla luciferase activities were detected by the Dual-Luciferase Reporter Assay kit (Promega, USA). The luciferase activity was normalized to Renilla luciferase activity.

OS cells were seeded in confocal petri dishes and cultured to 80% confluence. The dishes were then washed twice with phosphate-buffered saline (PBS), fixed with 200 µl 4% paraformaldehyde at room temperature for 15 min, and permeabilized with 200 µl 0.1% Triton X for 15 min at room temperature. After washing twice with PBS, the dishes were incubated with the anti-DGCR8 antibody with 1:200 dilution at 4°C overnight. The next day, the dishes were washed twice with PBS, then anti-rabbit IgG-Cy3 Fluor 570 (Beyotime Institute of Biotechnology) was added at 1:1,500 dilution and further incubated at room temperature for 1 h. The nucleus was then counterstained with DAPI (Shanghai Yeasen Biotechnology Co., Ltd.) at room temperature for 5 min and the images were captured using a Zeiss Axiophot 2 microscope (Zeiss AG). The magnification of all images was ×40.

A total of 20 female Balb/c-nude mice (3–4 weeks old, weight 10–13 g) were purchased from the Laboratory Animal Center, Southern Medical University. The mice were bred in a specified pathogen-free environment at temperature and humidity controlled (26°C; 50% humidity) environment with 12-h light/dark cycle and served a standard diet ad libitum. The animals were randomly assigned into four groups (n=5), two groups (Group1 and Group2) were inoculated with 1.5×10⁶/100 µl shHOTTIP infected OS cells through trans-tibia injected into the medullary cavity of the right tibia of mice; the other two groups (Group3 and Group4) were injected with equivalent shNC infected OS cells at the same site. Group1 and Group3 were intraperitoneally (i.p.) injected with 20 mg/kg IKE (Shandong Topscience Biotech Co., Ltd.) and the other two groups were treated with an equal volume of vehicle NaCl. The IKE or NaCl was added once every other day and tumor formation was monitored. The tumor volume was calculated using the formula V=0.5 × L × W² (L, length and W, width). The largest tumor diameter allowed in this experiment was <20 mm. At the end of the experiment, the mice were anesthetized with a dose of 100 mg/kg pentobarbital (intraperitoneal), followed by cervical dislocation. Mortality was confirmed by the cessation of breathing and heartbeat. And their tumor tissues were isolated. All animal experimental procedures were approved by the Ethics and Animal Research Committee of Southern Medical University (approval no. SMUL2022219; Guangzhou, Guangdong).

Tumor tissues were fixed in 10% neutral formalin fixative solution (BBI Solutions) embedded in paraffin at room temperature overnight, and sectioned at 3-µm, which were deparaffinized with xylene and then rehydrated with a succession of decreasing alcohol concentrations (100, 95, 85, 70 and 50% ethanol). Tissue sections were then placed in antigen retrieval buffer (pH 9.0; Wuhan Servicebio Technology Co., Ltd.) in a microwave oven on medium power for 10 min until boiling, then cooled for 8 min and switched to medium-low power for 8 min. After natural cooling, the slides were placed in PBS and washed three times on a destaining shaker. Subsequently, 3% hydrogen peroxide solution was added to the sample at room temperature for 25 min to block endogenous peroxidase, followed by blocking with 3% BSA (Sangon Biotech Co., Ltd.) at room temperature for 30 min. The histological sections were then incubated at 4°C for 4 h with Ki67 antibody (cat. no. GB111499; Wuhan Servicebio Technology Co., Ltd.) and GPX4 antibody (cat. no. T56959; Abmart Pharmaceutical Technology Co., Ltd.) with 1:100 dilution, and with a horseradish peroxidase-conjugated goat anti-rabbit IgG secondary antibody (1:200; cat. no. SA00001-2; Proteintech Group, Inc.) at 37°C for 30 min. Sections were then counterstained with 0.1% hematoxylin (Boster Biological Technology) at room temperature for 2 min and were stained with the chromogen DAB (Boster Biological Technology). Histological images were captured using a light microscope (cat. no. AE2000; Motic Incorporation, Ltd.), the representative images were taken with at ×40 magnification and the positive cells were quantified with ImageJ software (ver. 1.46; National Institutes of Health).

Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis was performed on differentially expressed genes using the enrichKEGG function from the clusterProfiler package in R software (ver. 3.20; http://www.genome.jp/kegg/kegg_ja.html). The result of 25 pathways were visualized using the Pathview package (ver. 3.20, R Core Team, 2024) (16). Gene Ontology (GO) analysis was performed on genes that were increased or decreased by >2-fold in the HOTTIP KO sample compared with the WT sample. The analysis was conducted using the gseGO function in the clusterProfiler package (ver. 4.14.3) in R software (ver. 4.4.1, R Core Team, 2024), based on the Gene Set Enrichment Analysis approach (https://www.gsea-msigdb.org/gsea/index.jsp). GO terms which had adjusted P-value <0.01 were considered markedly regulated GO. Up- or downregulated GO terms were featured by R software. Heatmaps and volcano plots were generated using TBtools (v 1.055; http://github.com/CJ-Chen/Tbtools/releases). Venn diagrams showing the intersection between target genes identified in microRNA databases (Targets can. http://www.targetscan.org/) and Ferroptosis Marker (FerrDb v2 database. http://www.zhounan.org/). The online bioinformatics programs, miRbase (https://www.mirbase.org/), Targets can and RNAhybrid (https://bibiserv) were applied to predict the binding site of miR-214-3p with GPX4.

Data were shown as mean ± SD from at least three independent experiments. Data were analyzed by two-tailed unpaired Student's t-test between two groups and by one-way ANOVA followed by the Tukey post hoc test for multiple comparisons, Bonferroni correction was applied where appropriate. Statistical analysis was carried out using GraphPad Prism software version 8 (Dotmatics). P<0.05 was considered to indicate a statistically significant difference.

HOTTIP has been reported to serve as an oncogene in OS and the emergence of ferroptosis provides a novel opportunity to clarify the underlying mechanism. To clarify the relationship between HOTTIP and ferroptosis, the transcriptome alternation between bone marrow derived from KO and wild-type (WT) mice was analyzed (Fig. 1A). The RNA-sequencing data indicated that the differently expressed gene mainly focused on 25 signaling pathways using the KEGG database (Fig. 1B). According to their relative expression level, the changed genes were divided into upregulation and downregulation groups (Fig. 1C). Gene ontology (GO) analysis showed a number of biological processes (BPs) were abnormally influenced such as lipid transport, cellular response to iron ion, cellular response to reactive oxygen species and regulation of lipid metabolic process (Fig. 1D and E). Furthermore, the related ferroptosis genes were predicted from Gene Set Enrichment Analysis (GSEA) website and it was found that Biological Oxidations gene set was clearly upregulated (Fig. 1F). Considering that OS is a type of primary malignant bone tumor and most OS come from bone marrow (17), it was therefore hypothesized that HOTTIP may be involved in ferroptosis of OS.

To determine the involvement of HOTTIP in ferroptosis, the present study successfully established a ferroptotic cell death model with OS cells. As shown in Fig. 2A and B, the cell viability was markedly suppressed after respectively treated with the canonical inducers of ferroptosis Erastin or RSL3. The potent ferroptosis inhibitor ferrostatin-1 (Fer-1) partially reversed these suppressive effects, whereas the apoptosis inhibitor Z-VAD-FMK and necroptosis inhibitor necrostatin-1 (Necro) appeared to possess no rescue abilities in Erastin or RSL3 treated OS cells (Fig. 2A and B). The colony formation assay also confirmed that Erastin clearly inhibited the numbers and sizes of colonies and Fer-1 markedly attenuated this suppressive effect (Fig. 2C). As a new programmed cell death, reactive oxygen species (ROS) and iron accumulation are two important characters of ferroptosis (18). The level of intracellular iron was determined using the Iron assay kit and the accumulation of total Fe was promoted by Erastin while Fer-1 markedly rescued this promotion (Fig. 2D and E). The flow cytometry assays were used to detect the intracellular ROS and the results displayed that ROS signal was increased by Erastin and this increase was rescued by Fer-1 in OS cells (Fig. 2F and G). GSH depletion and MDA accumulation are essential for ferroptosis. As expected, the intracellular level of MDA was increased whereas GSH level was suppressed by Erastin in MG63 and U2OS cells (Fig. 2H-K). These results indicated that Erastin successfully triggered the ferroptosis in OS cells, indicating that the ferroptotic cell death model was successfully constructed. Using this cell model, it was found that HOTTIP was markedly suppressed in Erastin-treated OS cells (Fig. 2L), suggesting that HOTTIP may participate in ferroptosis of OS cells.

To identify the real function of HOTTIP in ferroptosis, HOTTIP-overexpressing stable OS cells were established (Fig. S1A). According to cell viability assays, ectopic expression of HOTTIP clearly attenuated the suppressive cell viability triggered by Erastin or RSL3 (Fig. 3A). The accumulation of intracellular iron (Fig. 3B), intracellular ROS signal (Fig. 3C) and the expression level of MDA (Fig. 3D) were all suppressed while the GSH expression was promoted (Fig. 3E) in this HOTTIP-overexpressing cells. The canonical markers of ferroptosis including xCT and GPX4 were markedly promoted by HOTTIP overexpression in OS cells (Fig. 3F-H). On the other hand, the HOTTIP silencing stable OS cells were developed (Fig. S1B) and it was showed that silencing of HOTTIP markedly enhanced the suppressive effects of the Erastin or RSL3 on cell viability (Fig. 3I). Fer-1 successfully reversed this suppressed cell viability in OS cells. Further investigation showed that the accumulation of intracellular iron, intracellular ROS signal and the expression level of MDA were clearly promoted by HOTTIP knockdown (Fig. 3J-L). Conversely, GSH expression was markedly suppressed in HOTTIP silenced OS cells (Fig. 3M). The expression of xCT and GPX4 was suppressed in HOTTIP silenced OS cells with Erastin or RSL3 treatment (Fig. 3N-P). All these results revealed that HOTTIP inhibited the ferroptosis in OS cells.

To further confirm the in vivo function of HOTTIP in ferroptosis, an orthotopic intra-tibia tumor-bearing model was established using shHOTTIP infected OS cells. A canonical in vivo ferroptosis inducer, imidazole ketone erastin (IKE), was given intraperitoneally (i.p.) with 20 mg/kg once every other day. As expected, the IKE groups carried smaller burden when compared with control (NaCl) group and the shHOTTIP group displayed smaller tumor when compared with shNC group (Fig. 4A). Similar results were observed in tumor growth rate (Fig. 4B) and weight (Fig. 4C). The levels of Fe, MDA and GSH were examined in animal tissues and the similar results to the in vitro investigation were recorded (Fig. 4D-F). Further immunohistochemical staining indicated that the expressions of Ki-67 and GPX4 were markedly suppressed by IKE treatment and shHOTTOP group exhibited more sensitive to IKE treatment (Fig. 4G and H). These data suggested that HOTTIP knockdown facilitates the ferroptosis-based anti-tumor therapy in OS.

LncRNAs are recognized to exert their function through physically interacting with RNA binding proteins. DGCR8, a known RNA-binding protein, was predicted to directly bind HOTTIP and the RIP assay validated that DGCR8 successfully pulled down HOTTIP RNA transcript (Fig. 5A). Subsequent study demonstrated that overexpression of HOTTIP suppressed while its knockdown promoted DGCR8 expression at protein level (Fig. 5B and C). The immunofluorescence examination also confirmed that the expression of DGCR8 was suppressed by HOTTIP overexpression whereas it was promoted by HOTTIP knockdown (Fig. 5D). However, the mRNA expression level of DGCR8 was inconsistent with HOTTIP overexpression or knockdown (Fig. S2). It was therefore hypothesized that HOTTIP mediated DGCR8 expression via a post-transcriptional manner. In order to validate this hypothesis, cycloheximide (CHX) chase assays were conducted to monitor the stability of DGCR8 protein. A shorter half-life was observed in the HOTTIP-overexpressing OS cells (Fig. 5E and F). Given that the ubiquitin-proteasome pathway is the most significant signaling to mediate the protein stability (19), an inhibitor of proteasome MG-132 was added in the HOTTIP-overexpressing OS cells to examine the protein stability of DGCR8. As shown in Fig. 5G and H, MG-132 treatment markedly reversed the suppressive effect of DGCR8 driven by HOTTIP overexpression. Next DGCR8 antibody was utilized to pull down endogenous DGCR8 proteins and their modification examined by an ubiquitin antibody. Based on the results of Co-IP assays, the decreased DGCR8 ubiquitination was detected in HOTTIP silencing cells (Fig. 5I). This result proposed that destabilization of DGCR8 mediated by HOTTIP may depend on the proteasomal degradation. Collectively, the data demonstrated that HOTTIP physically interacted with DGCR8 and modulated its stability by promoting its ubiquitin-mediated degradation.

Notably, DGCR8 is the RNA-binding partner of the nuclease Drosha. The DGCR8/Drosha complex named by the microprocessor (recognized and cleaved pri-miRNAs) is essential for the processing of primary (pri-)miRNAs in the nucleus (20,21). To identify which miRNA biogenesis was influenced in the present study, miRNA sequencing was conducted using HOTTIP silencing OS cells. As shown in Fig. 6A, among the 10 promising candidates, hsa-miR-214-3p was the most upregulated. Subsequent relevant assays were conducted in the miR-214-3p transfected MG63 cell lines (Fig. S3A), cell viability examination showed that miR-214-3p mimics markedly inhibited cell viability while Fer-1 alleviated this suppressive ability (Fig. 6B). Further investigation of Fe accumulation (Fig. 6C), ROS signal (Fig. 6D) and MDA expression level (Fig. 6E) showed that they were promoted by this miR-214-3p mimics and Fer-1 treatment successfully reversed this promotional effects. In term of GSH expression, the converse results were observed in miR-214-3p transfected cells and partially rescued by Fer-1 treatment (Fig. 6F).

It is known that miRNAs trigger their function via suppressing their targets expression. To find the targets of miR-214-3p, an online program was searched to predict the potential candidates and GPX4 was identified as the most promising one (Fig. S4). Using the luciferase activity assays, it was found that miR-214-3p mimics markedly suppressed the luciferase activity of the GPX4-3′UTR-WT reporter but the mutagenesis of these binding sites successfully abolished this suppressive effect (Fig. 6G). Further investigation showed that miR-214-3p mimics suppressed GPX4 expression while the suppressive effects were partially reversed by Fer-1 (Fig. 6H and I).

To discover whether miR-214-3p participated in the HOTTIP mediated ferroptosis, ectopic miR-214-3p mimics were applied in the HOTTIP overexpressing cells to examine the rescue effects on ferroptosis (Fig. S3B). The cell viability examination revealed that ectopic expression of miR-214-3p successfully inhibited the viability in OS cells and HOTTIP overexpression reversed this suppressive effect (Fig. 6J). The similar rescued effects of miR-214-3p mimics on Fe accumulation (Fig. 6K), ROS signal (Fig. 6L), MDA (Fig. 6M) and GSH expression (Fig. 6N) and GPX4 expression (Fig. 6O and P) were also detected in the HOTTIP overexpressing OS cells.