Hypoxia blocks ferroptosis of hepatocellular carcinoma via suppression of METTL14 triggered YTHDF2‐dependent silencing of SLC7A11

Abstract Residue hepatocellular carcinoma (HCC) cells enduring hypoxic environment triggered by interventional embolization obtain more malignant potential with little clarified mechanism. The N6‐methyladenosine (m6A) biological activity plays essential roles in diverse physiological processes. However, its role under hypoxic condition remains largely unexplored. RT‐qPCR and Western blot were used to evaluate METTL14 expression in hypoxic HCC cells. MDA assay and electronic microscopy photography were used to evaluate ferroptosis. The correlation between SLC7A11 and METTL14 was conducted by bioinformatical analysis. Flow cytometry was used to verify the effect of SLC7A11 on ROS production. Cell counting kit‐8 assay was performed to detect cells proliferation ability. Hypoxia triggered suppression of METTL14 in a HIF‐1α–dependent manner potently abrogated ferroptosis of HCC cells. Mechanistic investigation identified SLC7A11 was a direct target of METTL14. Both in vitro and in vivo assay demonstrated that METTL14 induced m6A modification at 5’UTR of SLC7A11 mRNA, which in turn underwent degradation relied on the YTHDF2‐dependent pathway. Importantly, ectopic expression of SLC7A11 strongly blocked METTL14‐induced tumour‐suppressive effect in hypoxic HCC. Our investigations lay the emphasis on the hypoxia‐regulated ferroptosis in HCC cells and identify the HIF‐1α /METTL14/YTHDF2/SLC7A11 axis as a potential therapeutic target for the HCC interventional embolization treatment.

photography were used to evaluate ferroptosis. The correlation between SLC7A11 and METTL14 was conducted by bioinformatical analysis. Flow cytometry was used to verify the effect of SLC7A11 on ROS production. Cell counting kit-8 assay was performed to detect cells proliferation ability. Hypoxia triggered suppression of METTL14 in a HIF-1α-dependent manner potently abrogated ferroptosis of HCC cells. Mechanistic investigation identified SLC7A11 was a direct target of METTL14.
Both in vitro and in vivo assay demonstrated that METTL14 induced m 6 A modification at 5'UTR of SLC7A11 mRNA, which in turn underwent degradation relied on the YTHDF2-dependent pathway. Importantly, ectopic expression of SLC7A11 strongly blocked METTL14-induced tumour-suppressive effect in hypoxic HCC. Our investigations lay the emphasis on the hypoxia-regulated ferroptosis in HCC cells and identify the HIF-1α /METTL14/YTHDF2/SLC7A11 axis as a potential therapeutic target for the HCC interventional embolization treatment.

| INTRODUC TI ON
Liver cancer is one of the most common and aggressive tumours that causes 841,000 new cases and 782,000 deaths each year in the world. 1  Interventional therapy has been widely applied to patients with unresectable HCC, whereas the hypoxic condition caused by interventional embolization could benefit HCC cells a lot for proliferation and metastasis, [6][7][8] and the underlying mechanism remains unclear.
Therefore, exploring the molecular mechanisms by which hypoxic HCC develops is critical to advance future therapeutic strategies.
Strikingly, it has been demonstrated that the aberrant epigenetic changes can result in profound disruption of gene expression to facilitate HCC initiation and progression. 9  [10][11][12][13][14] In terms of this complex, METTL3 is the catalytic subunit that binds to the methyl donor S-adenosyl methionine (SAM) and catalyses methyl group transfer, whereas METTL14 is responsible for m 6 A deposition by stabilizing METTL3 conformation and recognizing substrate RNAs. 15,16 The m 6 A modification sites, which are frequently enriched in the 3' untranslated region (3' UTR) and coding sequence (CDS), with a particularly high enrichment around the stop codon area contain a classical consensus sequence DRACH (D = G, A, or U; R = G or A; H = A, C, or U). 17,18 This reversible catalytic process is carried on by m 6 A eraser fat mass and obesity-associated protein (FTO) and alkB homologue 5 (ALKBH5). 19,20 m 6 A-modified RNA can be identified by m 6 A reader proteins, including YT521-B homology (YTH) domain family proteins (YTHDF1-YTHDF3, YTHDC1 and YTHDC2). 17,21,22 Among them, YTHDF2 is the first identified and well-studied m 6 A reader protein which targets m 6 A via its C-terminal YTD domain and the N terminal domain is responsible for pushing the mRNA to the processing bodies for further degradation. 22 Recent studies suggest that m 6 A is involved in diverse physiological processes. 23

| Transfection and stable cell lines
We manipulated transfection and stable cell lines referring to the previous study.

| Quantitative real-time PCR
The quantitative real-time PCR is carried out as described before. 36 According to the manufacturer's instructions, total RNA was extracted
Subsequently, an anti-rabbit secondary antibody was used for 2 h at 22°C (about room temperature). The protein bands were visualized using a chemiluminescence ECL kit (Tanon).

| Wound healing assay
The wound healing assay is followed with the previous work. 39 Briefly, in order to evaluate the migration ability of cells, wound healing assays were performed. NAC and/or pcDNA3.1-SLC7A11-His(+)

| ROS detection
Our ROS detection was validated by Flow cytometer similar to a previous work. 40

| Dual-Luciferase reporter assay
The dual-luciferase reporter assay was measured like the previous study. 39 Plasmids were constructed in advance. Briefly, proce- wide-type and mutated promoter sequence of SLC7A11 were provided in supplemental materials (S1).

| Methylated RNA immunoprecipitation (MeRIP)
MeRIP was followed by the manufacturer's instruction (Pierce™ quantified. PolyA + RNA was fragmented to approximately 100 nt long fragments. Before proceeding to m 6 A-IP, RNA fragmentation was ensured by using a bioanalyzer. Afterwards, first-strand cDNA synthesis was performed first. qPCR was performed as described.

| RNA decay assay
In the RNA decay assay, we mainly followed the method used by Qing et. al. 41 Briefly, Huh7 and HCCLM3 cells upon METTL14 or YTHDF2 were treated with 3 mg/ml actinomycin D (A9415, Sigma-Aldrich).
In order to see the RNA stability of SLC7A11, cells were treated as described in method and harvested at the certain time points (8 h, 4 h, 2 h and 0 h). In addition then, the total RNA was extracted, with the DNase-I digesting to eliminate genomic DNA. Afterwards, the collected RNA was reversed transcribed into cDNA followed by RT-qPCR detection. The primer of SLC7A11 used in RT-qPCR has been described above: SLC7A11 (F: 5'-GTCTGGAGAAACAGCCAAGG-3'; R: 5'-CGGAGTTCCTCGAATAGCTG-3').

| MDA Assay
The MDA assay was gone through in the light of the previous study. 42 To test the lipid peroxidation, we use Lipid Peroxidation (MDA) Assay Kit (Cat. No. MAK085, Sigma-Aldrich). The procedure was followed by the manufacturer's instruction.

| Electron microscopy photography
The electron microscopy photography was referred to Caroline's study. 43 To obtain the location of mitochondria in the cells, electron

| Animal study
The BALB/C nude mice were obtained from Shanghai SLAC Laboratory Animal Co., Ltd. We performed the nude mice experiment referring to the previous work. 44 Briefly, for the subcutaneous im-

| H&E staining and Immunohistochemistry
The H&E staining and Immunohistochemistry was performed as the previous study. 45 Briefly, xenograft samples were fixed in 4% paraformaldehyde (Sigma-Aldrich, DK-2860) overnight and embedded in paraffin before sections of 6 µm were cut. Antigen retrieval was performed at

| Statistical analysis
In this study, GraphPad Prism 8.00 (GraphPad Software) was performed for analyses. Student's t test was used for statistical analysis.
A p value less than 0.05 was considered statistically significant. Data are presented as the means ± SEM.

| The levels of METTL14, ROS and lipid peroxidation were limited by HIF-1α in hypoxia condition
To determine the role of METTL14 and RNA m 6 A modification under hypoxia, we first examined the expression of METTL14 in Huh7 and HCCLM3 cell lines with 1% O 2 (Hypoxic condition).
Our data confirmed that compared to control group (Normoxia), hypoxia efficiently decreased the METTL14 and increased the HIF-1α expression in the two HCC cell lines ( Figure 1A).  Figure 1D,E). What's more, MDA assay revealed that compared with that in control group, MDA contents were higher in HIF-1α knockdown group (p < 0.01) ( Figure 1F). Additionally, it is known that ferroptosis could induce typical morphological changes characterized by condensed and disrupted mitochondria. 46,47 Consistently, we did observe that HIF-1α knockdown dramatically induced smaller and denser mitochondria with damaged membrane compared to that in control group ( Figure 1G). In summary, our data indicated hypoxia downregulated METTL14 in the HIF-1α-dependent manner and knockdown of HIF-1α induced ferroptosis in HCC cells.

| METTL14 negatively regulates SLC7A11 expression in HCC
To explore the potential mechanism by which METTL14 regulated ferroptosis, the expression pattern of METTL14 and SLC7A11 which is a core member of system x c − that mediates ferroptosis was analysed with multiple databases. Firstly, the expression pat- negatively regulates SLC7A11 expression.

| METTL14 triggers m 6 A methylation at 5'UTR of SLC7A11 mRNA in HCC
In order to verify the specific relationship between SLC7A11 and METTL14, we first checked the RNA Base v2.0 (http://www.sysu. edu.cn) and found that there were several potential m 6 A sites within It has been demonstrated the R298P mutation greatly reduces METTL14 methylation activity. 15 We therefore established stable METTL14-R298P expression Huh7 and HCCLM3 cell lines ( Figure 3B,C & Figure S1A,S1B).
To identify if the m 6 A modification of SLC7A11 was mediated by METTL14, we first detected the total m 6 A level in negative control group and stable METTL14 overexpression as well as METTL14-R298P groups through m 6 A dot blot assay. As expected, m 6 A levels were substantially increased with the overexpression of METTL14 but were reduced by R298P mutation in two HCC cell lines ( Figure 3D & Figure S1C).
To explore the essence of m 6 A modification on SLC7A11, luciferase reporter assays were conducted with a wild-type (WT) and a mutant (MUT) plasmid ( Figure 3E). For mutant reporter, cytosine bases (C) were designed to replace the adenosine bases (A) in several predicted m 6 A sites to block the effect of m 6 A methylation, while wild-type reporter contained intact m 6 A sites.  Figure 3N).
In summary, these data suggested that METTL14 performed a tumour-suppressive function via targeting SLC7A11 in an m 6 Adependent manner in HCC.

| METTL14-induced SLC7A11 mRNA decay is m 6 A-YTHDF2-dependent
It is essential to figure out the readers of SLC7A11, since m 6 Amodified mRNA transcripts depend on the reader proteins to functionally participate in biological processes. Previously, we demonstrated METTL14 overexpression remarkably downregulated SLC7A11 mRNA in both Huh7 and HCCLM3 cell lines ( Figure 2J).
Next, we tested if m 6 A modification affects the mRNA stability of SLC7A11. qPCR showed that overexpression of METTL14 significantly enhanced SLC7A11 mRNA degradation in the presence of Actinomycin D ( Figure 4A,4B).
YTHDF2 (YTH domain family 2), a recognized m 6 A reader protein, has been demonstrated to regulate the mRNA stability. Strikingly, we found knockdown of YTHDF2 substantially increased SLC7A11 expression at both mRNA and protein levels ( Figure 4C), which indicated the potential role of YTHDF2 in SLC7A11 modulation. Furthermore, luciferase reporter assays were performed with plasmid containing the wild-type (WT) or mutant (MUT) SLC7A11 as described previously. As expected, knockdown of YTHDF2 dramatically increased the luciferase activity of wide-type group, but had little effect on the mutant group ( Figure 4D). Taken together, these data indicated that m 6 A -YTHDF2 conducts METTL14-induced SLC7A11 mRNA degradation.

| Knockdown of SLC7A11 stimulates ferroptosis and exhibits an anti-tumour effect in HCC
To address if SLC7A11 inhibition can mimic the tumour-suppressive function of METTL14, we first detected the effect of SLC7A11 knockdown on ferroptosis induction. As Figure 5A,B shown, shSLC7A11 strongly stimulated ROS production, while the ROS scavenger N-acetyl-L-cysteine (NAC) dramatically blocked the shSLC7A11induced ROS accumulation as well as EMT reversion ( Figure 5C,D) in both Huh7 and HCCLM3 cells. Moreover, NAC treatment potently abrogated the shSLC7A11 induced mitochondria shrinking detected by electronic microscopy. Compared to the shSLC7A11 group, the mitochondria in shSLC7A11 plus NAC group exhibited relatively more intact membrane and larger size ( Figure 5E). In addition, the same tendency was observed in vivo by nude mice. Compared to the control group, downregulation of SLC7A11 significantly inhibited tumour growth ( Figure 5F,G). The H&E staining of tumour sections was indicated as tumour structures ( Figure 5H). Furthermore, immunohistochemistry analysis showed that the shSLC7A11 group had a higher expression of COX2 ( Figure 5I). These data verified the anti-tumour effect by inhibiting of SLC7A11 in HCC cells.

METTL14-induced tumour-suppressive effect under hypoxia in HCC
We further examined whether inhibition of SLC7A11 contributes to the anti-tumour effects of METTL14. As Figure 6A,B shown, under the hypoxic environment, wide-type METTL14 but not the METTL14-R298P mutant strongly inhibited SLC7A11 Ferroptosis which is a newly discovered way of programmed cell death has been demonstrated to be related to iron metabolism and reactive oxygen species (ROS). 46 The function of hypoxia-induced HIF-1α has been widely investigated in HCC progression. 60

ACK N OWLED G M ENTS
We thank all members in the laboratory for critical reading and discussion.

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.