METTL1 Mediated m 7 G Modication Regulates Ferroptosis and Chemotherapy Resistance via Involvement of pri-miR-26a/FTH1 Axis in Osteosarcoma

N7-Methyladenosine (m7G), one of the most prevalent internal modications in mammalian RNAs, makes a great contribution to many bioprocesses. However, its functions and underlying mechanisms in the occurrence and development of osteosarcoma remain elusive. The expression of METTL1 and its correlation with clinicopathological features using tissue microarrays and the Cancer Genome Atlas (TCGA) dataset.Targets of METTL1 on osteosarcoma were identied by AlkAniline-seq, RNA immunoprecipitation (RIP), quantitative real-time PCR, western blot and luciferase assays. The effects of METTL1 on the biological characteristics of osteosarcoma cells were investigated on the basis of gain- and loss-of-function analyses. Ferroptosis were analyzed by lipid peroxidation assay, iron assay and ransmission electron microscopy (TEM). Subcutaneous models further uncovered the role of METTL1 in tumour growth. Here, we identied METTL1, an m7G methyltransferase, is low expressed in tissues and plays an antitumor role in Mechanistically, METTL1 enhances cell ferroptosis by targeting pri-miR-26a and promoting its mature through m7G methylation, which could target FTH1 mRNA and eliminate FTH1 translation eciency mediated by METTL1. FTH1 is the main component of ferritin is crucial for iron and the inhibition lipid METTL1 overexpression signicantly improved the ecacy of doxorubicin and cisplatin via the increasement of ferroptosis. Collectively, our studies demonstrate the critical role of METTL1 in the development and chemosensitivity of osteosarcoma, as featured by inducing lipid peroxidation and promoting ferroptosis of osteosarcoma cells, and uncover a previously unrecognized signaling axis involving METTL1/pri-miR-26a/FTH1 in osteosarcoma. Moreover, our study highlights the functional importance of the m 7 G modication mediated by METTL1 in osteosarcoma and provides profound insights into the molecular mechanisms underlying tumorigenesis in osteosarcoma. In addition, given the functional importance of METTL1 in leukemogenesis and drug response, targeting METTL1 signaling by selective promotors may represent a promising therapeutic strategy to treat osteosarcoma, especially in METTL1-low patients.

and differentiation [12][13][14] and post-ischemic angiogenesis 15 . Besides, previous researchs have observed that the dysregulation of METTL1 mediated m 7 G modi cation is functionally correlated with the development and prognosis of colon, liver, breast cancer, and intrahepatic cholangiocarcinoma [16][17][18][19][20] . Given the known crucial role of METTTL1 in tumors, characterizing the effects involved in osteosarcoma and clarifying the underlying mechanism has become a focus of intensive study.
Ferroptosis is an iron-dependent form of regulated cell death driven by excessive lipid peroxidation, of which morphology and mechanism is distinct from cell apoptosis 21 . Generally, glutathione (GSH)-dependent selenoenzyme glutathione peroxidase 4 (GPX4) is deemed essential for preventing ferroptosis via reducing toxic lipid hydroperoxides to nontoxic lipid alcohol 22,23 .
Emerging studies have proved that ferroptosis plays an extremely critical antitumor role in diverse cancers, including pancreatic cancer 24 , gastric cancer 25 , glioblastoma 26 and so on. Moreover, triggering ferroptosis enhanced drug-resistant cancer cells sensitivity to chemotherapeutic drugs such as cisplatin, temozolomide, and doxorubicin [27][28][29] , indicating that induction of ferroptosis can potentially be a novel and effective strategy in cancer treatment.

Materials And Methods
Cell lines and cell culture supplemented with 10% FBS (Biological Industries, Israel). All cell lines were maintained in an incubator at 37℃ in an atmosphere containing 5% CO 2 . All cells were repeatedly screened for mycoplasma and maintained in culture for < 6 months after receipt.

Cell transfection
Cells were transfected to knock down the expression of METTL1 and AMO-miR-26a-5p using Lipofectamine TM 3000 Transfection Reagent (Cat# L3000-015; Invitrogen, California, USA). METTL1, FTH1, FTH1-mut plasmid and miR-26a-5p mimic were purchased from Gene Pharma (China) to increase the expression of the plasmid. Brie y, for siRNAs, 2 × 10 5 cells were seeded in a 6-well plate and they will be 70% con uent at the time of transfection 6 µL Lipofectamine TM 3000 reagent and 20 nM siRNA (Gene Pharma, Shanghai, China) were diluted in 125 µL Opti-MEM (Cat# 31985-070; gibco, Grand Island, USA) medium respectively. After Mix and incubating for 2 min separately, these two regents were then mixed and incubated for another 10 min and then added to cells. For plasmids, cells were transfected with 500 ng plasmid using Lipofectamine TM 3000 Transfection Reagent according to the manufacturers' protocols. Subsequent experimental measurements were performed 24 h after transfection.
The sequences used are as following:

Invasion assay
Matrigel invasion assay was performed using 24 well plates inserted by 24 mm Transwell® chambers (Corning #3412, USA) precoated with Matrigel (BD Biosciences, San Jose, CA). 5 × 10 4 cells were resuspended in 200 µL Serum-free medium, seeded into the upper chamber, and medium containing 10% FBS was added into the bottom chamber subsequently. After incubation at Migration assay Cells were plated into 6 well plates at a density of 2.5 × 10 5 cells/mL. When the con uence of cells reached 70%, a wound was created by scraping the cells with a 200 µL pipette tip. Cells were washed with phosphate-buffered saline (PBS) and then transfected with siRNAs or plasmid. Images were captured at 0 h, 24 h, and 48 h after wounding with standard light microscopy (ECLIPSE TS100, Nikon, Japan). The wound area was measured using ImageJ software (National Institutes of Health, USA).
Alkaline hydrolysis and aniline cleavage sequencing (AlkAniline-seq) The m 7 G AlkAniline-Seq sequencing service was provided by CLOUDSEQ (Shanghai cloud-seq biomart, http://www.cloudseq.com.cn). Alkaline hydrolysis of poly (A) -enriched mRNA fragments were used. Cells were incubated with thermosensitive phosphatase (NewEnglandBiolabs, Inc. USA) to dephosphorylate RNA sheet pairs, followed by incubation and lysis in 1 M aniline. RNA libraries were constructed using NEBNext® MultiplexSmallRNALibraryPrepSetforIllumina ® (NewEnglandBiolabs, Inc., USA) according to the supplier's instructions, and library quality control and quanti cation were performed using the BioAnalyzer2100 system (Agilent Technologies, USA). High-throughput sequencing was performed using an IlluminaHiSeq sequencer.
Quantitative real-time polymerase chain reaction (qRT-PCR) Total RNA was extracted using TRIZOL reagent (Life Technologies Corporation) followed the manufacturer's protocol. 500 ng total RNA was reverse-transcribed into 10 µL cDNA using High Capacity cDNA Reverse Transcription Kit (Cat#00676299; Thermo Fisher Scienti c, Waltham, USA). qRT-PCR analysis was needed 1 µL cDNA, 1 µL forward primer, 1 µL reverse primer, and SYBR Green PCR Master (Cat#31598800; Roche) by a 7500 Fast Real-Time instrument (Applied Biosystems, Foster City, USA). Gene expression was normalized to endogenous GAPDH. The miRNA ampli ed transcript level was normalized to U6. Primer sequences can be found in Additional le le: Table S1.
Ethynyl-2-deoxyuridine (EdU) staining assay EdU Apollo DNA in vitro kit (Ribobio, Guangzhou, China) was used to detect cell proliferation. Cells were plated into 24 well plates (NEST, Hong Kong, China) at a density of 2 × 10 5 . Brie y, cells were xed with 4% paraformaldehyde (m/v) for 30 min and followed by incubation of 30 µM/mL EdU at 37℃ for 90 min. After permeabilized in 0.5% Triton X-100, the Apollo staining solution was added into the cell culture medium for 30 min in the dark. Finally, the cells were incubated with 20 µg/mL 4′,6diamidino-2-phenylindole (DAPI) for 10 min. The EdU index (%) was the average ratio of the number of EdU positive cells over total cells in ve randomly selected areas under the confocal laser scanning microscope (FV10i, Olympus, Tokyo, Japan).
Tissue microarrays (TMAs) and Immunohistochemistry (IHC) analysis Osteosarcoma tissue microarrays were purchased from the Bioaitech Company (Xi'an, China), comprised of 11 normal bone tissues, 70 malignant osteosarcoma cores. The slide was baked at 60℃ for 30 min and then followed by antigen retrieval with tris-EDTA buffer (pH 9.0), medium heat for 10 min to boil, cease-re for 5 min, and washed with PBS for 5 min × 3 times.
Endogenous peroxidase was blocked with 3% H 2 O 2 -methanol at RT for 25 min and washed with PBS for 5 min × 3 times. The sections were added to normal non-immune animal serum at RT for 10 min and then removed the serum and added different primary antibodies FTH1 (#4393, CST), 4-HNE (bs6313R, Bioss), Ki67 (27309-1-AP, Protein tech), METTL1 (14994-1-AP, protein tech) at 4℃ overnight. Then it was washed with 0.1% tween-20 PBS for 5 min × 3 times. Biotin-labeled sheep anti-mouse/rabbit IgG was added and incubated in a 37℃ wet box for 30 min followed by washing with 0.1 % tween-20 PBS for 5 min × 3 times. DAB working solution was incubated for 5 min and stopped by distilled water washing. After hematoxylin re-staining, washing and differentiation, the slide returned to be blue with full washing followed with regular dehydration transparent and being sealed by neutral gum. The percentage of positive cells were counted in 5 (×400) highpower elds (upper, lower, left, right, and middle) under the microscope, and the mean values were then calculated.
RNA immunoprecipitation (RIP)-qPCR RIP assay was carried out in 143B cells using Magna RIP Kit (17-700, Millipore, MA) following the manufacturer's instructions. In brief, a su cient number of 143B cells (more than 2 × 10 7 cells per sample) were lysed by RIP lysis buffer, magnetic beads precoated with 5 µg m 7 G antibody or mouse IgG (Millipore) were incubated with su cient cell lysates at 4℃ overnight. The mixture was digested with proteinase K before the immunoprecipitated RNAs were extracted, puri ed and subjected to qPCR. The RNA levels were normalized to the input RNA levels (10%).

RNA stability assay
Cells were cultured in 6 well plates and transfected with desired constructs as described above. After 24 h transfection, cells were treated with actinomycin D (Act D, 10 µg/mL, Cat# GC16866, GLPBIO) for 0 h, 3 h, 6 h and 9 h before collection. Total RNAs were isolated for qRT-PCR analysis.

Iron assay
Cells were cultured in a 6 well plate, after 24 h transfection, cell precipitation was collected, the intracellular ferrous iron level using an iron assay kit from ScienCell (8448, California, USA). The absorbance was nally measured at 590 nm.

Luciferase reporter assay
The luciferase reporter assay was performed with the Dual-Luciferase (Promega E2920) according to the manufacturer's protocol. HEK-293T cells seeded in 24 well plates were transfected with a SV40-re y-Luciferase-MCS fused the wild-type FTH1, miR-26a-5p mimics or NC were co-transfected. Fire y luciferase activity was normalized to Renilla luciferase activity to re ect transfection e ciency.

Lipid peroxidation assay
The relative lipid peroxidation level in cells was assessed using the Image-iT® Lipid Peroxidation Kit (molecular probes, C10445). Cells were treated with 5 µM C11-BODIPY for 30 min harvested, washed twice with PBS and resuspended in 500 µL PBS. Oxidation of the polyunsaturated butadienyl portion of the dye results in a shift of the uorescence emission peak from 590 nm to 510 nm.
Xenograft tumorigenesis model BABL/c female nude mice were purchased from Beijing Vital River Laboratory Animal Technology Limited Company (Beijing, China) and randomized into two groups. All animal experiments were performed in accordance with protocols approved by the Animal Care and Use Committee of Harbin Medical University. Mice were injected subcutaneously with 5 × 10 6 143B cells to form subcutaneous xenografts. Mouse tumor growth was monitored by measuring tumor length as well as width. Mice weights in each group were recorded accordingly. When tumors grew to a volume of 200 mm 3 , the mice were divided randomly into six groups (n = 5/group) and treated with normal saline, 1 mg/kg Dox (Doxorubicin hydrochloride, MCE, HY15142), 1 mg/kg Fer-1 (Ferrostatin-1, MCE, HY100579, ferroptosis inhibitor), vehicle by daily intraperitoneal administration, body weights of mice in each group during treatment were also recorded accordingly, the animals were sacri ced four weeks after 143B cell transplantation. Xenograft sampling was performed, and tumor volume was calculated by the standard formula: (length × width 2 )/2. Before xenograft sampling, the xenografts were anesthetized with iso urane gas, scanned with an IVIS in vivo imaging system (IVIS LuminaIII, MA, USA), and the total photon radiation at each tumor site was quanti ed using Living Image software. Mice were euthanized for xenograft sampling.

Transmission electron microscopy
After the cells were pretreated accordingly, the cells were mixed and centrifuged for 250 revolutions for 5 min. Discard the culture medium, add 1 mL PBS culture medium, and resuspend the cells. Next, the samples were horizontally centrifuged at 3000 rpm for 20 min. After the completion of centrifugation, PBS was discarded, and 2.5% glutaraldehyde was added to x the cells. Cells were dehydrated with an ethanol concentration gradient and embedded in resin. They were then sectioned with an ultramicrotome, stained with uranyl acetate-lead citrate double staining, and nally observed under a transmission electron microscope.

Western Blot
Cell samples were washed using PBS, and then the total proteins were extracted using 1×RIPA lysis buffer (Beyotime, Shanghai, China). The lysates were fully crushed by ultrasonic and cleared by high-speed centrifugation at 13,500g for 15 min. The extracted total proteins were quanti ed by BCA Protein Assay Kit (Beyotime, Shanghai, China), then equal amounts of proteins were separated by SDS-PAGE and transferred to NC membrane. After the membranes were blocked with 5% w/v non-fat milk for 1h, the proteins were probed at 4℃ overnight using different primary antibodies METTL1 (14994-1-AP, proteintech), FTH1 (#4393, CST), GPX4 (A13309, Abclonal), GAPDH (AC002, Abclonal), β-Tubulin (AC021, Abclonal). On the next day, the membranes were incubated with a secondary antibody (RS23910, ImmunoWay) for 1 h at room temperature. Results were detected using Odessey Clex (LI-COR, America), followed by further analysis.

Hematoxylin and Eosin Staining (H&E) staining
The H&E staining Kit (G1120, Solarbio) was used for H&E staining, para n sections were stained with hematoxylin for 5 min and differentiated with differentiation solution for 30 s, followed by soaking into tap water for 15 min. And then the sections were stained with eosin, followed by dehydrated with gradient alcohol and cleared with xylene. Images were acquired using a Leica microscope and Leica image software.

Statistical analysis
All experiment results were at least repeated three times and expressed as means ± SEM. Statistical analyses were performed using GraphPad Prism 8.0 and Student's T-test was used for two-group comparisons. p < 0.05 was considered statistically signi cant. *P < 0.05; **P < 0.01; ***P < 0.001.

Elevated METTL1 expression correlates with a good prognosis of patients with osteosarcoma
We initially used immunohistochemistry (IHC) assays to detected METTL1 protein expression levels in osteosarcoma tissues from 71 patients with osteosarcoma from tissue microarrays (TMAs) and 20 normal bone tissues ( Figure 1A and Additional le 1: Figure S1). METTL1 expression was signi cantly lower in osteosarcoma tissues than that in normal bone tissues, which was further decreased with the higher clinical stage ( Figure 1B). Kaplan-Meier analysis using data derived from the TCGA cohort also revealed that high expression of the METTL1 was associated with a good prognosis in patients with sarcoma ( Figure 1C).

FTH1 is the target gene of METTL1 identi ed by AlkAniline-Seq
To identify potential targets of METTL1 the m 7 G methylation levels of which are regulated by METTL1, we conducted alkaline hydrolysis and aniline cleavage sequencing (AlkAniline-Seq) in 143B cells, a human osteosarcoma cell line ( Figure 2A).
AlkAniline-Seq data revealed that 45,048 m 7 G methylated sites ≥ 2 folds decrease in METTL1 silencing group compared with those in negative control (NC) group, and of those, 2,047 m 7 G methylated sites counts were greater than or equal to 50, which had 150 m 7 G methylated sites counts greater than or equal to 100 in NC group ( Figure 2B). We listed the top 10 genes with the most signi cant changes in methylation sites (Table 1), among those m 7 A abundances of ferritin heavy chain 1 (FTH1), myosin ID (MYO1D), and Disks Large Homolog 2 (DLG2) were the most markedly decreased upon METTL1 knockdown ( Figure 2C).
Considering that METTL1 has been proved to promote m 7 G modi cation of mRNA as a methyltransferase and to facilitate RNA synthesis 30 , we hypothesized that FTH1, MYO1D, and DLG2 are potential targets of METTL1. Furthermore, METTL1 knockdown in 143B cells resulted in a signi cant downregulation of FTH1, MYO1D, and DLG2 mRNA, while overexpression of METTL1 led to a notable upregulation of FTH1 and MYO1D mRNA, except DLG2 ( Figure 2D and 2E). However, FTH1 exhibited the most consistent decreased m 7 G level and mRNA level in METTL1 knockdown 143B cells versus that of the NC group. We next performed gene-speci c m 7 G qPCR assays for FTH1 and con rmed the m 7 G level decrease in FTH1 transcript, demonstrating the reliability of our transcriptome-wide AlkAniline-Seq data ( Figure 2F). FTH1 has ferroxidase activity and is involved in iron accumulation which is important for iron homeostasis 31 . Kaplan-Meier analysis exhibited that sarcoma patients with low expression of the FTH1 have a longer survival time ( Figure 2G). To further explore the potential mechanism of METTL1 regulating FTH1 mRNA expression, we treated NC or METTL1 knockdown 143B cells with the transcription inhibitor actinomycin D (Act D) and detected the half-life of FTH1 transcripts. As expected, METTL1 silencing indeed resulted in a remarkable decrease in the half-life of FTH1 transcripts (3.85 to 14.84 h, Figure 2H).  11 .
Surprisingly, forced expression METTL1 in 143B cells signi cantly reduced FTH1 protein levels ( Figure 3A), indicating that there may be other underlying mechanisms for FTH1 post-transcriptional regulation. AlkAniline-Seq dates showed that METTL1 knockdown also has led to a noticeable reduction of m 7 G methylated sites in pri-miRNA, while 4,278 m 7 G methylated sites decreased 2 folds compared to the NC group, and 65 of these sites decreased counts greater than or equal to 50 in the NC group ( Figure 3B). Notably, among the top 20 genes with the most signi cant changes, pri-miR-26a and pri-miR-98 were highly conserved ( Table 2). In addition, METTL1 knockdown in 143B cells reduced the expression of pri-miR-26a, while overexpression of METTL1 resulted in a signi cant upregulation of pri-miR-26a ( Figure 3C). However, the expression of pri-miR-98 was too low to be detected (data not shown) and METTL1 had no signi cant effect on the expression of miR-98 ( Figure 3D). Of note, METTL1 positively regulated the mature of miR-26a-5p indicating that METTL1 might regulate the miRNA biogenesis by m 7 G modi cation ( Figure 3D). To establish the cause of the decrease in FTH1 protein levels,anti-miR-26a-5p oligonucleotide (AMO) or miR-26a-5p mimics were transfected in 143B cells. This revealed a signi cant increase or decrease in the expression of FTH1 in miR-26a-5p AMO or mimics groups respectively, at protein levels rather than mRNA levels compared to the respective control cells ( Figure 3E and 3F). It indicated that miR-26a-5p might target FTH1 and regulate FTH1 translation. To further verify whether FTH1 is the target of miR-26a-5p, HEK-293T cells were co-transfected with miR-26a-5p mimics or NC mimics and FTH1 luciferase vectors. Dual-luciferase reporter assay showed that the luciferase activity of the FTH1 mRNA was markedly reduced in the group transfected with miR-26a-5p mimics ( Figure 3G). Collectively, METTL1 both increases the m 7 G modi cation and the expression of FTH1 mRNA and pri-miR-26a, and further substantially promotes miR-26a-5p mature which could bind to FTH1 mRNA and eliminate the improvement of FTH1 translation e ciency ( Figure 3H).

METTL1 suppresses cell viability of human osteosarcoma cells via inducing ferroptosis
We then performed gain-and loss-of-function studies to elucidate the pathological role of METTL1 in 143B and U2OS cells, two human osteosarcoma cell lines, and the transfection e ciency was detected by western-blot (Additional le 1: Figure S2A and S2B). The cell proliferation was analyzed using CCK-8 assay, EdU staining, and clone formation. All these results showed that 143B and U2OS cell proliferative capability was dramatically decreased following METTL1 overexpression ( Figure 4A-4C). The cell invasion and migration rate were also obviously abrogated by forced expression of METTL1 through transwell assay and wound-healing assay ( Figure 4D and 4E). Conversely, knockdown of METTL1 in 143B cells slightly promoted cell proliferation and invasion ability (Additional le 1: Figure S3A and S3B).
Given that METTL1 regulates m 7 G methylation modi cation of FTH1, which is crucial for labile iron pool stabilization, we next elucidated the function of METLL1 in ferroptosis. We used C11BODIPY staining to detected lipid peroxidation in 143B and U2OS cells after transfected with or without METTL1 plasmids. C11BODIPY staining showed obvious lipid peroxidation in response to METTL1 overexpression ( Figure 5A). Ferrostatin-1 (Fer-1), a synthetic antioxidant, is known to be a speci c ferroptosis inhibitor inhibiting lipid peroxidation. Consistently, 4 h pretreatment with Fer-1 clearly abrogated METTL1-induced lipid peroxidation ( Figure 5A). The morphological characteristics of ferroptosis in transmission electron microscopy (TEM) include dense and small mitochondria, increased membrane density, and cristae degeneration 21 . To identify these morphological features, we performed TEM evaluation in 143B cells. Dense and shrunken mitochondria were remarkably apparent in the METTL1 overexpression group compared to the control group, which could be reversed by Fer-1 treatment ( Figure 5B). Similarly, METTL1 overexpression also dramatically increased iron accumulation, while Fer-1 could abolish this phenomenon ( Figure 5C).
Nevertheless, there was no noticeable change of the protein level of glutathione peroxidase 4 (GPX4), an intrinsic negative regulator for lipid peroxidation and ferroptosis, after METTL1 overexpression both in 143B and U2OS cells (Additional le 1: Figure S4). Moreover, METTL1 overexpression further intensi ed the inhibition effect of ferroptosis inducer RAS-selective lethality 3 (RSL3) on cell viability ( Figure 5D). In particular, Fer-1 signi cantly alleviated METTL1 overexpression induced cell proliferation inhibition as measured by EdU staining assay in 143B and U2OS cells, respectively ( Figure 5E). The involvement of ferroptosis in METTL1 induced cell viability inhibition was further supported by the negligible effects of the pan-caspase inhibitor Z-VAD-FMK and the necroptosis inhibitor necrostatin-1 (Nec-1) treatment on cell proliferation inhibition caused by METTL1 overexpression (Additional le 1: Figure S5). Taken together, these data suggests that ferroptosis is predominantly responsible for the inhibition of cell viability caused by METTL1 overexpression.

Fer-1 eliminates tumor growth inhibition resulted by METTL1 in vivo
To assess METTL1 gain of function, we constructed stably METTL1 overexpression and control 143B luciferase cell line via lentivirus transfection and used it to establish animal xenograft model ( Figure 6A). METTL1 overexpression group mice were randomly divided into 2 groups, one of which was given Fer-1 by intraperitoneal injection on day 10 ( Figure 6A), the time we exactly observed tumor formation and average radiance values of both the METTL1 group and prepared for the Fer-1 group were slightly lower than the control group (Additional le 1: Figure S6). The average radiance value detected by in vivo imaging, tumor weight, and tumor volume indicated noticeable retardation of tumor growth resulted by METTL1 overexpression, which was gradually and dramatically reversed by Fer-1 ( Figure 6B-6E). Similarly, these treatment effects were further supported by hematoxylin and eosin (H&E) staining as well as Ki67 immuno-staining on tissue sections from the differently treated groups ( Figure 6F). Interestingly, IHC staining of lipid peroxidation marker 4-hydroxynonenal (4-HNE) levels showed that METTL1 could considerably induce the production of 4-HNE, while Fer-1 treatment signi cantly decreased the expression of 4-HNE ( Figure 6F).

miR-26a-5p is involvement in the ferroptosis and cell viability inhibition modulated by METTL1
Considering the m 7 G methylation of miR-26a-5p modi ed by METTL1, we then asked whether miR-26a-5p affects METTL1mediated ferroptosis and antitumor function. TEM of bronchial epithelial cells was examined in 143B cells. Apparently, accumulation of mitochondria with increased membrane density was presented in the METTL1 overexpression group, which was barely detected in the co-transfected group containing METTL1 plasmids and miR-26a-5p AMO ( Figure 7A). Furthermore, miR-26a-5p silencing also remarkably eliminated the excessive iron levels and lipid peroxidation production caused by METTL1 overexpression (Figure 7B and 7C). Given that miR-26a-5p could target FTH1 and regulated its translation ( Figure 3E), we examined the translation regulation of the FTH1 association between METTL1 and miR-26a-5p. Of note, the miR-26a-5p inhibitor signi cantly enhanced the protein level of FTH1 reduced by METTL1 overexpression. Consistently, miR-26a-5p silencing both reversed cell proliferation inhibition of 143B and U2OS cells regulated by METTL1 overexpression ( Figure 7E). Taken together, these results indicated that the involvement of miR-26a-5p in METTL1-mediated ferroptosis and cell viability inhibition is at least partially via post-transcriptional regulation of FTH1.

METTL1-mediated ferroptosis increases sensitivity to chemotherapy both in vitro and in vivo
The above results prompted us to ask whether METTL1-mediated ferroptosis and antitumor function are su cient to cause antidrug resistance. To this end, we analyzed the expression of METTL1 in human osteosarcoma cell lines treated with chemotherapy agents both doxorubicin (Dox) and cisplatin (Cis). Obviously, these drugs signi cantly decreased the mRNA expression of METTL1 in the 143B and U2 OS cell lines as well as protein levels ( Figure 8A and 8B). To explore the potential role of METTL1 in osteosarcoma cells sensitivity to chemotherapy, we transfected METTL1 plasmids into 143B and U2OS cells with 24 h treatments of Dox or Cis. We found that METTL1 overexpression in these cells rendered them noticeably more sensitive to Dox and Cis-induced cell injury associated with proliferation inhibition (Figure 8C-8F). Accumulating evidence has demonstrated the roles of Dox or Cis in ferroptosis 32,33 . Interestingly, METTL1 overexpression could exacerbate the lipid peroxidation production and iron accumulation led by Dox and Cis treatment, which can be partially prevented with Fer-1 treatment ( Figure  8G-8J).
To further explored the function of METTL1 on chemosensitivity of osteosarcoma cells in vivo, the BALB/c nude mice were subcutaneously transplanted with 143B tumor cells transfected with lentivirus. Mice were then treated with Dox beginning at day 10 ( Figure 9A). Tumors derived from the mice cotreated with empty lentivirus and Dox grew more slowly compared with those derived from empty lentivirus transfected, which could be further alleviated by lentivirus-mediated METTL1 overexpression, both in weight and volume ( Figure 9B-9E). Consistent with these results, H&E staining and Ki67 immuno-staining further demonstrated that METTL1 overexpression promoted chemosensitivity of osteosarcoma bearing nude mice ( Figure 9F).
Moreover, ectopic expression of METTL1 along with Dox treatment dramatically enhanced 4-HNE production indicating that METTL1-mediated ferroptosis is involved in the chemotherapy resistance inhibition ( Figure 9F). has been reported to participate in the physiological process of brain injury 36 , Parkinson's disease 31 , and cardiomyopathy 35 .

Discussion
Moreover, degradation of FTH1 exhibited antitumor activity in leukemia by reducing lipid peroxides content and further restoring cell ferroptosis 37 . However, whether FTH1-mediated ferroptosis is involved in osteosarcoma progression is still largely unknown.
As an m 7 G RNA methyltransferase, METTL1 enhances the m 7 G level of FTH1, which in turn leads to the upregulation at the RNA level via promoting its stabilization. Our study differs from the previous observation that METTL1-mediated m 7 G modi cation in mammalian mRNA increases translation e ciency 11,30 , whereas our dates indicate that overexpression of METTL1 signi cantly decreases the FTH1 protein level. A possible explanation for this apparent discrepancy is that METTL1 is known to have abundant cellular regulatory targets.
Interestingly, by further analyzing of AlkAniline-Seq dates and validation studies, we revealed that METTL1 de ciency dramatically reduces the mature of miR-26a-5p via m 7 G methylation, which has been reported to act as a tumor suppressor in osteosarcoma 38,39 . METTL1-mediated m 7 G position has been identi ed to be essential for let-7e stem-loop equilibrium and thereby promotes its processing, which can further selectively regulate the translation of oncogenic mRNAs in hepatoma carcinoma cells 10 . Consistently, our data showed miR-26a-5p combines with FTH1 transcript promoting FTH1 translation.

Consent for publication
Not applicable.
Availability of data and materials Not applicable.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.