m6A modification is increased during ferroptosis cell death in CRC cells

To investigate whether m6A modification is involved in ferroptosis during Erastin or RSL3 treatment, two well-established molecular compounds used as ferroptosis agonists. Firstly, we determined the IC50 of Erastin and RSL3 in inhibiting CRC cells, and found that Erastin and RSL3 have good inhibition in CRC cells at low concentration (Fig. S1a, b). Next, we explored the inhibition effect of Erastin and RSL3 on cell proliferation in CRC cells, and found that a significant inhibition on CRC cell proliferation by treating with either Erastin or RSL3 in a dose and time dependent manner (Fig. 1a, b and S1c, d). Lastly, we want to explore whether the inhibition effect of Erastin or RSL3 on CRC cell proliferation due to the ferroptotic death. Indeed, we found that the decreased cell proliferation could be prevented by simultaneously treating with iron chelator deferoxamine (DFO) or ferrostatin1 (Fer-1), not 3-Methyladenine (3-MA) and Z-VAD-fmk (Fig. 1c, d), suggesting that ferroptosis did occur during Erastin or RSL3 treatment in CRC cells.

To study whether ferroptosis poses effect on the m6A modification in CRC cells, we analyzed the effect of Erastin or RSL3 on m6A modification in CRC cells. Enzyme linked immunosorbent assay (ELISA) and m6A dot blotting assay both show that the m6A modification levels were increased in CRC cells by treating with either Erastin or RSL3 in a dose dependent manner (Fig. 1e, f and S1e, f). In addition, the increased m6A modification levels were prevented by treating with DFO or Fer-1 (Fig. 1g-j and S1g, h). Together, these findings suggested that Erastin or RSL3 treatment induce CRC ferroptotic cell death are associated with the increased m6A modification levels.

m6A modification upregulation upon ferroptosis requires FTO

The formation of m6A modification is a reversible process, which added by m6A “writers”, inducing methyltransferase like METTL3, METTL14, and WTAP, and removed by FTO and alkB homolog 5 (ALKBH5) [14]. To understand how m6A modification upregulation during ferroptosis, we analyzed the effects of ferroptotic damage on the expression of FTO, ALKBH5, METTL3, METTL14, YTHDF2, and WTAP. We found that the mRNA levels of FTO, WTAP, and YTHDF2 were decreased with the increasing the concentration of Erastin or RSL3 in CRC cells, while the mRNA levels of ALKBH5, METLL3, and METLL14 were not significantly affected by Erastin or RSL3 (Fig. S2a, b). However, we found that the protein levels of FTO, ALKBH5, METTL3, METTL14, WTAP, and YTHDF2 were all decreased in CRC cells by treating with either Erastin or RSL3 in a dose dependent manner (Fig. 2a, b and S2c, d). In addition, the decreased FTO, ALKBH5, METTL3, METTL14, and WTAP levels were prevented by treating CRC cells with DFO or Fer-1 (Fig. 2c, d and S2e, h). Combined with the results from that the increased m6A modification levels have occurred during Erastin or RSL3 treatment in CRC cell. Thus, we assumed that m6A demethylase may be a driver for the increased m6A modification levels during ferroptosis. Next, we explore the determined which m6A demethylase regulates m6A modification during ferroptotic death, and found that m6A modification levels were still increased in knockdown ALKBH5 cells treated with Erastin or RSL3, however, the increased m6A modification levels in CRC cells treated with Erastin or RSL3 were blocked with the knockdown of FTO (Fig. 2e, f and S2i, l). Furthermore, we found that knockdown of FTO decreased the levels of ALKBH5, METTL3, and METTL14 (Fig. S2m), which explained the decreased levels of ALKBH5, METTL3, METTL14, and WTAP were mediated by FTO during the CRC cells ferroptotic death. Collectively, these results demonstrated FTO is responsible for the m6A modification change during ferroptosis.

FTO mediates CRC cells ferroptosis

Given that FTO expression is downregulated during CRC ferroptotic cell death, we assumed that FTO may be a negative regulator of ferroptosis. Indeed, we found that knockdown FTO augmented MDA levels, and decreased the glutathione/glutathione (GSH/GSSG) ratio (Fig. 2g, h). Furthermore, targeting FTO by Rhein also decreased GSH/GSSG ratio and the increased MDA levels (Fig. S2n, o). While, the increased MDA levels and decreased GSH/GSSG ratio were not only prevented by treating with DFO or Fer-1 in the FTO knockdown cells (Fig. 2i-l), but also can be rescued by exogenous expression FTO (synonymous mutation, with a shRNA-resistant FTO) in FTO knockdown cells (Fig. S2p, q). In addition, the decreased GSH/GSSG ratio was also prevented by treating with glutathione (GSH) (Fig. S2r). Collectively, these findings suggested that FTO protects CRC cells from ferroptotic cell death.

Targeting FTO sensitizes CRC cells to ferroptosis

Our previous studies showed that FTO protects CRC cells from ferroptotic cell death, thus, we wonder whether pharmacological blocking of FTO would as a potential therapeutic approach in cancer with a ferroptotic cell death. We found that knockdown FTO increased the sensitivity of CRC cells to the Erastin or RSL3 treatment (Fig. 3a, b). Next, we hypothesized that a combination of FTO inhibitor and Erastin or RSL3 would show synergistic effects. Indeed, targeting FTO by Rhein increased the anticancer of Erastin or RSL3 in CRC cells (Fig. 3c, d and S3a, b). However, as DFO and Fer-1 have been marketed as ferroptosis inhibitor, which against the inhibition of targeting FTO on cell proliferation with a ferroptotic signature (Fig. 3e and S3c). Thus, these results indicated that a key role of FTO expression or activity in inhibiting the anticancer activity with a ferroptosis signature.

As the expression of METTL3, METTL14, and ALKBH5 were decreased under Erastin or RSL3 treatment mediated by FTO, we also examined the response of knockdown METTL3, METTL14, and ALKBH5 to Erastin or RSL3 treatment in CRC cells. The results showed that knockdown of METTL3 or METTL14 decreased the sensitivity of CRC cells to the Erastin or RSL3 treatment (Fig. S3d, e). While, knockdown of ALKBH5 increased the sensitivity of CRC cells to the Erastin or RSL3 treatment (Fig. S3f, g). These results suggested that METTL3, METTL14, and ALKBH5 are also involved the CRC ferroptosis induced by Erastin or RSL3.

We next sought to determine whether the genetic inhibition of FTO can enhance the in vivo anticancer activity to induce ferroptosis. We first examined the effect of FTO on tumor growth, then, we established a patient-derived xenograft (PDX) model, and found that knockdown FTO significantly decreased the tumor growth (Fig. S3h, i), tumor masses (Fig. S3j), and Ki67 expression, as well as with the increased ferroptosis biomarker 4-hydroxynonenal (4HNE) levels (Fig. S3k). Next, we used this PDX model, and explored the effect of inhibition of FTO on Erastin treatment. We found that Erastin has a slight inhibitory effect on the tumor growth in FTO knockdown control group (Fig. 3f, g). While the administration of Erastin at 15 mg/kg in FTO knockdown group significantly suppressed the xenograft tumor growth (Fig. 3f, g), tumor masses (Fig. 3h), and Ki67 expression (Fig. 3i). Interesting, the inhibition of FTO expression conferred Erastin therapy sensitivity that was associated with increased ferroptosis biomarker 4HNE levels (Fig. 3i). These preclinical animal studies support the hypothesis that targeting FTO significantly enhances the anticancer activity of Erastin in vivo.

FTO enhances the expression of SLC7A11 and GPX4 in CRC cells

To explore the molecular mechanism of FTO in regulating ferroptotic cell death, the RNA-sequencing (RNA-Seq) was performed in FTO knockdown and vector control cells. A total of 1, 655 significant differentially expressed genes were observed in FTO knockdown cells (Fig. S4a). What is more, heatmap analysis indicated a clear separation between the ferroptotic related gene expression profiles of FTO knockdown and vector control cells (Fig. S4b). As given that knockdown FTO enhances the anti-tumor effects of Erastin and RSL3, which targets SLC7A11 and GPX4, respectively. Thus, we next focus on the SLC7A11 and GPX4 as downstream of FTO in regulating ferroptosis. Validation studies showed that knockdown FTO significantly decreased mRNA levels of SLC7A11, while increased mRNA levels of GPX4 in CRC cells (Fig. 4a), these results are consistent with the RNA-seq data. Furthermore, we found that knockdown FTO significantly decreased protein levels of both SLC7A11 and GPX4 (Fig. 4b), while, a shRNA-resistant FTO can rescued the decreased protein levels of SLC7A11 and GPX4 in the FTO knockdown cells (Fig. 4c). Taken together, these results suggested that FTO regulates SLC7A11 and GPX4 expression.

FTO-mediated m6A modification enhances the expression of SLC7A11 mRNA

To explore the function of FTO in regulating SLC7A11 or GPX4 expression whether dependent on FTO m6A demethylase activity, we analyzed our published methylated RNA immunoprecipitation-m6A-sequencing (MeRIP-m6A-seq) data [18] and found that the m6A peaks were increased on SLC7A11 (Fig. 4d) and GPX4 mRNA (Fig. S4c) in FTO knockdown group. To validate the MeRIP-m6A-seq data, we applied the methylated RNA immunoprecipitation qPCR (MeRIP-qPCR) assay to confirm that SLC7A11 mRNA and GPX4 mRNA have m6A modification (Fig. S4d, e). Furthermore, we found that knockdown of FTO significantly increased the m6A modification levels on SLC7A11 mRNA and GPX4 mRNA in CRC cells (Fig. 4e). The m6A modification is a dynamically reversible process, added by methyltransferases (Writers: METTL3, METTL14) and removed by demethylases (Erasers: FTO, ALKBH5) [15]. Here, we generated METTL3 and METTL14 knockdown cells with the decreased total m6A levels (Fig. 4f) and found that METTL3 and METTL14 are responsible for the m6A modification on SLC7A11 mRNA and GPX4 mRNA (Fig. 4g). Besides, METTL3 can also rescued the decreased protein levels of SLC7A11 and GPX4 in the FTO knockdown cells (Fig. 4h). These results suggested that FTO regulates SLC7A11 and GPX4 expression mediated by m6A modification and can by reversed by its methyltransferase METTL3.

FTO enhances the expression of SLC7A11/GPX4 mediated by YTHDF2

As the m6A modification is recognized and bound by m6A-binding proteins (Readers), such as, YTH family, which play a specific role in control the fate of the methylated mRNA [15]. Next, we performed a RNA immunoprecipitation qPCR (RIP-qPCR) assays to screen for SLC7A11-related m6A readers and explore the direct interaction between the YTHDF1, YTHDF2, or YTHDF3 and SLC7A11 mRNA. RIP-qPCR assays showed that YTHDF2 and YTHDF3 bound to SLC7A11 mRNA (Fig. S4f), while only YTHDF2 bound to GPX4 mRNA (Fig. S4g). Next, we explore the exactly sites on SLC7A11 mRNA were recognized by YTHDF2 or by YTHDF3, and found there are have 11 potential m6A modifications by SRAMP analysis (www.cuilab.cn/sramp) (Fig. 4i). First, we validated that m6A modification site 3 and site 11 were predominantly m6A modification (Fig. 4j). Next, we found that the m6A modification site 3 and site 11 were recognized and bound by YTHDF2 not YTHDF3 (Fig. 4k). Lastly, we performed a streptavidin RNA pull-down assay further verified that YTHDF2 not YTHDF3 predominantly bound to the site 3 and site 11 on SLC7A11 mRNA (Fig. 4l). As YTHDF2 mediates the degradation of mRNA [24]. Thus, we next focus on the function of YTHDF2 in regulating SLC7A11 mRNA stability. Indeed, we found that the SLC7A11 mRNA stability was markedly decreased upon FTO knockdown in CRC cells (Fig. 4m). While, the decreased SLC7A11 mRNA stability and SLC7A11 protein levels were rescued by knockdown YTHDF2 in FTO loss cells (Fig. 4n, o). Taken together, our data suggested that FTO enhances the expression of SLC7A11 in a m6A modification mediated by YTHDF2.

Next, we explore the exactly sites on GPX4 mRNA were recognized by YTHDF2, and found there are 3 potential m6A modifications by SRAMP analysis (www.cuilab.cn/sramp) (Fig. 4p). Firstly, we validated that m6A modification site 2 was predominantly m6A modification (Fig. 4q). Secondly, we found that the m6A modification site 2 on GPX4 mRNA was recognized and bound by YTHDF2 (Fig. 4r). Our previously study showed that YTHDF2 promote 6-phosphogluconate dehydrogenase (6PGD) mRNA translation [25]. Then, we constructed pmirGLO-GPX4-site 2 m6A wild type and site 2 m6A mutant plasmids. The reporter mRNA translation assay showed that YTHDF2 promotes the GPX4 mRNA m6A wild type translation not m6A mutant (Fig. S4j). Thus, our data suggested that FTO enhances the expression of GPX4 in a m6A modification mediated by YTHDF2, which facilitates GPX4 mRNA translation.

As DFO and Fer-1 can rescue the decreased FTO expression under Erastin or RSL3 treated condition. Thus, we wonder whether DFO and Fer-1 also rescue the decreased SLC7A11 and GPX4 expression under Erastin or RSL3 treated condition. Indeed, the decreased SLC7A11 and GPX4 expression under Erastin or RSL3 treated condition were rescued by treating cells with DFO and Fer-1 (Fig. S4h,i). These results suggested that the expression of GPX4 and SLC7A11 were decreased by Erastin or RSL3 treatment mediated by FTO in a YTHDF2-denpendent manner.

FTO regulates ferroptosis and cell proliferation via SLC7A11/GPX4

To determine whether FTO regulates ferroptosis and cell proliferation mediated by regulating SLC7A11/GPX4 expression. We then forced expression of SLC7A11 or GPX4 into FTO knockdown cells. Firstly, we found that the exogenous expression of SLC7A11 in FTO knockdown cells rescued the increased MDA level and the decreased GSH/GSSG ratio (Fig. 5a, b and S5a). Similarly, the exogenous expression of GPX4, in FTO knockdown cells rescued the increased MDA level and the decreased GSH/GSSG ratio (Fig. 5c, d and S5b). Lastly, the decreased cell proliferation in FTO knockdown cells was rescued by forced expression of SLC7A11 or GPX4 (Fig. 5e, f and S5c). Taken together, these data suggested that FTO promotes cell proliferation via inhibiting ferroptotic cell death mediated by SLC7A11/GPX4.

As METTL3 and METTL14 are responsible for the m6A modification on SLC7A11/GPX4 mRNA, and the m6A modification on SLC7A11 mRNA is recognized by YTHDF2. Thus, we explore the role of METTL3, METL14, and YTHDF2 in regulating ferroptosis under the FTO loss condition. We found that knockdown METTL3 or METTL14 decreased product of MDA (Fig. S5d), and knockdown METTL3 not METTL14 increased the glutathione/glutathione (GSH/GSSG) ratio (Fig. S5e). Then, we further knockdown the expression of METTL3 or YTHDF2 in FTO knockdown cells, and found that knockdown of METTL3 or YTHDF2 rescued the increased MDA level in FTO loss condition (Fig. 5g, h). Lastly, knockdown of METTL3 or YTHDF2 also rescued the decreased GSH/GSSG ratio (Fig. 5i, j). While, when knockdown METTL3 or YTHDF2, the regulation of Erastin or RSL3 on MDA and GSH/GSSG ratio were abolished (Fig. S5f, g). These results suggested that FTO regulates ferroptosis by control SLC7A11 mRNA mediated by its m6A modification.

Discovery of a novel and potent FTO inhibitor Mupirocin

To identify potential FTO inhibitors, which suppress CRC tumor growth by inducing ferroptosis, we conducted a structure-based virtual screening of a small molecule library consisting of 1680 bioactive compounds based on the FTO crystal structure (PDB code: 3LFM). Firstly, 34 candidate compounds were appeared, when we set the the LibDOCK Score > 160. Then, 7 candidate compounds showed the highest docking scores with FTO by improving docking results (Fig. 6a and S6a). To validate the molecular docking results. Firstly, we performed in vivo demethylation assay. Then we assessed their efficacy on inhibition of FTO’s m6A demethylase activity, and identified three compounds (Mupirocin, Vitexin-4-O-glucoside, and Luteolin 7-O-glucuronide) have the best efficacy activity on inhibition of FTO’s m6A demethylase activity in vivo (Fig. 6b). In addition, cell-free m6A demethylase assays showed that Mupirocin and Luteolin 7-O-glucuronide exert effects on inhibition of FTO’s demethylase activity in vitro (Fig. 6c, d and S6b). As Mupirocin has best activity against FTO both in vivo and in vitro. Especially, it has not been reported to exert any anti-tumor effect in CRC, thus, we decided to focus on Mupirocin as novel inhibitor of FTO for further studies.

The molecular docking study based on the crystal structure of FTO (PDB code: 3LFM) suggested that Mupirocin fits in a catalytic pocket [26], surrounded by residues including R96, Y108, H231, and E234 of FTO (Fig. 6e). To examine whether Mupirocin binds to and inhibits FTO, we then performed Surface Plasmon Resonance (SPR) assays to determine affinity between Mupirocin and FTO protein, with a Kd at 3.63 × 10− 6 mol/L (Fig. 6f). Additionally, an in vitro thermal shift assay showed that the stability of the purified FTO was increased under the increasing concentrations of Mupirocin relative to DMSO treated group (Fig. 6g). According to the docking poses of Mupirocin and FTO protein, residues R96, Y108, H231, and E234 are essential for the binding of FTO with Mupirocin. To further evaluate Mupirocin target engagement, we performed a cellular thermal shift assay (CETSA) to examine the direct interaction of FTO and Mupirocin in LoVo cells, which transfected with Flag tag FTO WT, H231A/D233A, and R96A. CETSA assay showed that Mupirocin could block temperature-induced of wild-type (WT) FTO, but not that of mutant FTO H231A/D233A or FTO R96A (Fig. S6c). In summary, these results suggested that Mupirocin directly binds to the FTO.

To further analyze the interaction of Mupirocin with FTO, we synthesized a Mupirocin probe (Fig. S6d). The cell-free m6A demethylase assay also showed that Mupirocin probe exerts directly inhibitory effects on FTO’s demethylase activity in vitro (Fig. S6e, f). Furthermore, the anti-tumor activity showed that Mupirocin probe showed a comparable IC50 value with that of Mupirocin (Fig. S6g). Moreover, the in situ pull-down assay was performed and showed that Mupirocin could bind with FTO directly in HCT8 and LoVo cells (Fig. 6h). In all, such data confirmed that FTO binds directly with Mupirocin in the native cellular environment, and the mutated amino acids are essential for their interactions.

Mupirocin treatments modulate the signaling pathways of FTO

To explore the role and molecular mechanisms underlying effects of Mupirocin by targeting FTO, RNA-Seq was performed in Mupirocin treated samples and control groups. A total of 2,347 significant differentially expressed genes were observed in Mupirocin treated samples (Fig. S6h). Through global gene set enrichment analysis (GSEA) [27], we identified a set of up-regulated or down-regulated pathways upon Mupirocin treatment, or FTO knockdown. Notably, among the up-regulated or down-regulated pathways, Mupirocin treatment, and FTO knockdown groups shared the majority of their enriched signaling pathways, such as G2M checkpoint, E2F target, apoptosis, MYC target V1, and ferroptosis (Fig. S6i). What is more, heatmap analysis indicated a clear separation between the ferroptotic related gene expression profiles of Mupirocin treatment and vector control cells as FTO knockdown RNA-Seq (Fig. S6j). Taken together, these results suggested that a novel and potent FTO inhibitor Mupirocin has similar effect as FTO in regulating CRC function.

Mupirocin triggers ferroptosis and suppresses tumor growth in CRC

The above results suggested that Mupirocin serve as a novel and potent FTO inhibitor. Therefore, we next assumed whether Mupirocin regulates CRC ferroptosis and tumor growth through targeting FTO. Firstly, it was found that the Mupirocin treatment resulted in an increased MDA levels (Fig. 6i), and the decreased GSH/GSSG ratio (Fig. 6j). While, the effect of Mupirocin on MDA levels and GSH/GSSG ratio were blocked by knockdown of FTO (Fig. 6k, l). Lastly, western blotting showed that Mupirocin treatment resulted in decreasing the expression of SLC7A11 and GPX4 protein (Fig. S6k). Again, the effect of Mupirocin on the expression of SLC7A11 and GPX4 protein were also blocked by knocking down FTO (Fig. 6m). These results suggested that Mupirocin treatment could trigger ferroptosis in CRC cells by targeting FTO.

To examine the effects of Mupirocin on CRC cell proliferation, CRC cells were treated with various concentrations of Mupirocin, and the effect of Mupirocin on cell proliferation was determined. Cell proliferation and colony formation assays showed that Mupirocin exerted a strong inhibition efficacy on CRC cell proliferation in a time and dose-dependent manner (Fig. 6n and S6l). Again, a similar inhibitory effect was also observed in the CRC patient-derived organoid (PDO) model with indicated doses of Mupirocin (Fig. 6o and S6m). These data demonstrated that Mupirocin has a significant inhibitory effect on CRC cell proliferation in vitro.

Next, the anti-tumor effects of Mupirocin were determined in vivo using a PDX mouse model. Then, 25 and 50 mg/kg of Mupirocin was administrated intraperitoneal (ip) every two days into PDX mouse model. Mupirocin significantly decreased the tumor growth (Fig. 6p, q), tumor masses (Fig. 6r), Ki67 expression, GPX4, and SLC7A11 levels, as well as induced the expression of ferroptosis biomarker 4HNE (Fig. S6n). There was no significant difference in body weight between the drug treatment group and the control group (Fig. S6o). These data suggested that Mupirocin significantly suppresses excessive tumor growth in vivo by targeting FTO through inducing ferroptosis.

Mupirocin enhances the anti-tumor effects of Erastin and RSL3

Our findings demonstrated that Mupirocin regulates ferroptosis by targeting FTO, thus, we wonder whether targeting FTO by Mupirocin would also enhance the therapeutic efficacy of Erastin or RSL3, respectively. Firstly, we compared the sensitivity of CRC cells to Erastin or RSL3 treatment with or without Mupirocin and found that Mupirocin treatment increased the sensitivity of CRC cells to Erastin or RSL3 treatment based on cell number count and colony formation assay (Fig. S7a, d). Secondly, we further tested the synergistic effect of Mupirocin and Erastin or RSL3 by the Median-effect Method described by Chou and Talalay. Synergism (CI < 1) was observed between the two agents (Mupirocin and Erastin or RSL3) (Fig. 7a, b). While, the synergistic inhibition effect of Mupirocin and RSL3 (or Erastin) on cell proliferation were abolished in FTO knockdown group (Fig. S7e). These results suggested that Mupirocin and RSL3 (or Erastin) have synergistic inhibition on cell proliferation in a FTO-dependent manner.

To further evaluate its therapeutic effects of Mupirocin in combination with Erastin or RSL3, we established a PDX mouse model. As shown in Fig. 7c-h, the Mupirocin in combination with Erastin or RSL3 therapy significantly restrained tumor growth compared to the monotherapy and control group. There was no significant difference in body weight between the drug treatment group and the control group (Fig. S7f, g). Moreover, the PDX tumors IHC staining showed that Ki67 staining was reduced in the combination group (Fig. 7i, j). In addition, FTO, SLC7A11, and GPX4 expression were decreased, as well as the 4NHE levels were increased after Mupirocin and Erastin or RSL3 treatment (Fig. 7i, j). These implied that ferroptotic death occurred in the tumors of the combined drug group, resulting in an inhibition of tumor growth. Taken together, these results suggested that targeting FTO by Mupirocin enhances the anti-tumor effects of Erastin or RSL3 in CRC xenografts.

SLC7A11 and GPX4 expression is positively correlated with FTO in patients with colorectal cancer

Our previous study showed that FTO expression was upregulated in CRC tissues [18]. Here, in order to explore the correlation among FTO, SLC7A11, and GPX4 expression. We used another CRC tissues micro-array to determine the expression of FTO, SLC7A11, and GPX4 by IHC staining analysis. Our findings show that FTO, SLC7A11, and GPX4 protein are also highly expressed in patient tumor tissues than in normal tissues, respectively (Fig. S8a-f). What’s more, SLC7A11 and GPX4 expression positively correlated with FTO in CRC tissues (Fig. S8g, h). These data suggested that FTO, SLC7A11, and GPX4 expression are elevated in CRC tissues, and the expression of SLC7A11/GPX4 and FTO are positively correlated in CRC cancer.