N6-Methyladenosine modification activates the serine synthesis pathway to mediate therapeutic resistance in liver cancer

Abstract

Metabolic adaptation serves as a significant driving force for cancer growth and poses a substantial obstacle for cancer therapies. Herein, we unraveled the role of m6A-mediated serine synthesis pathway (SSP) regulation in both hepatocellular carcinoma (HCC) development and therapeutic resistance. We demonstrated that treatment of highly specific m6A inhibitor (STM2457) effectively inhibited HCC cell line growth and suppressed spontaneous HCC formation in mice driven by liver-specific Tp53 knockout and Myc overexpression. Using GLORI-seq, we delineated a single-base-resolution m6A landscape in human HCC cell lines. Interestingly, we identified three core enzymes in the SSP (PHGDH, PSAT1, and PSPH) as novel targets of METTL3-mediated m6A modification. In these SSP genes, m6A modification recruited m6A reader IGF2BP3 to stabilize their mRNA transcripts, thereby enhancing their mRNA and protein expression in HCC cells. Most importantly, our GLORI-seq data revealed that sorafenib-resistant HCC cells elevated m6A modification in SSP genes to promote protein expression and antioxidant production. STM2457 treatment attenuated the serine synthesis pathway, induced oxidative stress, and sensitized HCC cells to sorafenib and lenvatinib treatments. In conclusion, our findings suggest that targeting m6A could be a potential therapeutic strategy for HCC treatment.

Keywords: IGF2BP3; METTL3; SSP pathway; STM2457; TKI treatment; liver cancer; m6A; m6A modification.

Graphical abstract

Figure 1

m6A inhibitor STM2457 suppressed HCC cell growth (A) Representative crystal violet staining image of Hep3B, Huh7, MHCC97L, HepG2, and MHCC97H HCC cell lines and MIHA immortalized hepatocyte cells treated with increasing doses of STM2457 for 5 days. (B) GI₅₀ of STM2457 inhibitor with indicated doses in HCC cell lines for 6 days. (C) Treatment of 20 μM STM2457 for 72 h suppressed poly(A) RNA m6A/A ratio detected by LC-MS. Mean ± SEM (n = 2) with two-tailed Student’s t test. (D) STM2457 treatment reduced HCC cell proliferation rate. Mean ± SEM (n = 3) with two-tailed Student’s t test. (E) Treatment of 20 μM STM2457 suppressed HCC migration in the transwell migration assay (24 h for Huh7 and 48 h for Hep3B and Hepa1-6 cells). Mean ± SEM (n = 3) with two-tailed Student’s t test. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.

Figure 2

m6A inhibitor STM2457 suppressed Tp53^KO/MycOE-induced HCC tumor development (A) Overexpression of MELLT3 was associated with TP53 mutation and Myc overexpression in The Cancer Genome Atlas cohort. (B) Mettl3 and Mettl14 were significantly upregulated in mouse HCC tumor induced by HDTVi-based liver-specific overexpression of Myc and Tp53 knockout (Tp53^KO/MycOE) (n = 12). (C) Mice weights over the STM2457 (50 mg/kg) treatment period. Mean ± SEM (n = 7). (D) STM2457 (50 mg/mL) 5 days on/2 days off treatment regimen for 3 weeks suppressed HDTVi-based liver-specific Tp53^KO/MycOE-induced spontaneous mouse HCC development. Mean ± SEM (n = 13) with two-tailed Student’s t test. (E) STM2457 extended the survival time of mice in the HDTVi-based Tp53^KO/MycOE model. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.

Figure 3

GLORI-seq showed that m6A is enriched in the SSP in HCC (A) Metagene profile plot of m6A profile in MHCC97L cells showed that m6A enriched at the 3′ untranslated region (3′ UTR). (B) Motif analysis by MEME showed that GLORI-seq-identified m6A sites have a canonical m6A sequence (DRAC, where D = G/A/T, R = A/G). (C) Histogram plot of frequency of m6A sites in single genes showed that most genes have no more than 25 m6A sites. (D) Boxplot of log2 fold change of RNA-seq FPKM value (STM2457/DMSO) for gene clusters with different numbers of m6A sites; low (1–2 m6A), medium (3–5 m6A), high (6–10 m6A), and very high (>10 m6A). (E) Venn diagram showed that many genes overlapped in RNA-seq downregulated genes and 1 or 2 m6A site marked genes. (F) KEGG enrichment analysis of the overlapping genes in (E). (G) Boxplot shows global downregulation of m6A methylation level after treatment with STM2457. Two-tailed Student’s t test. (H) Scatterplot showing downregulation of m6A level at single-site level. (I) Track image of read depth and m6A sites in DMSO- and ST2457-treated HCC cells. STM2457 treatment abolished m6A sites in the SSP. (J) MeRIP-RT-PCR showed that STM2457 treatment suppressed m6A methylation in SSP genes. Mean ± SEM (n = 3) with two-tailed Student’s t test. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.

Figure 4

m6A depletion inhibited SSP expression (A and B) KEGG pathway enrichment analysis showed that STM2457 treatment downregulated glycine, serine, and threonine metabolism pathway in MHCC97L RNA-seq (A) and Huh7 proteomic mass spectrometry profiling (B). (C) Gene set enrichment analysis (GSEA) showed downregulation of glycine, serine, and threonine metabolism pathway of Huh7 mass spectrometry results. (D) RT-qPCR analysis showed that the SSP is downregulated. Mean ± SEM (n = 3) with two-tailed Student’s t test. (E) Western blot showed that the SSP is downregulated with STM2457 treatment. (F) Western blot showed that knockdown of METTL3 suppressed the SSP. (G) RNA stability assay with RT-qPCR showed that STM2457 treatment suppressed the RNA stability of the SSP in HepG2 cells. Mean ± SEM (n = 3) with two-tailed Student’s t test. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.

Figure 5

IGF2BP3-mediated m6A stabilization of the SSP (A) RT-qPCR showed that knockdown of IGF2BP3 suppressed the SSP in Huh7 and MHCC97L cells. Mean ± SEM (n = 3) with two-tailed Student’s t test. (B) Western blot showed that shIGF2BP3 reduced SSP expression in Huh7 cells. (C) KEGG pathway enrichment analysis showed downregulation of glycine, serine, and threonine metabolism pathways in HepG2 shIGF2BP3 RNA-seq (ENCODE: ENCFF952DGB). (D) IGF2BP3 eCLIP peak track showed IGF2BP3 enrichment at SSP genes in HepG2 cells (ENCODE: ENCSR993OLA). (E) IGF2BP3 RIP-seq signal showed that SSP genes are among the top candidates of IGF2BP3 binding transcripts in 293T cells (GSE90639). (F) IGF2BP3 RIP-RT-qPCR showed that 72 h 20 μM STM2457 treatment suppressed IGF2BP3 binding of SSP mRNA in MHCC97L cells. Mean ± SEM (n = 3) with two-tailed Student’s t test. (G) RNA stability RT-qPCR using actinomycin D showed that knockdown of IGF2BP3 destabilizes SSP mRNAs in HepG2 cells. Mean ± SEM (n = 3) with two-tailed Student’s t test. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.

Figure 6

Reduced rate of serine synthesis in m6A-depleted HCC induces ROS stress and apoptosis (A) Simplified diagram show that three key enzymes PHGDH, PSAT1, and PSPH are involved in the SSP. (B) STM2457 treatment (20 μM and 96 h) inhibited serine synthesis and caused phosphoglyceric acid (3PG) accumulation, 3-phosphoserine (3PS), α-ketoglutarate (α-KG), and serine reduction. Mean ± SEM (n = 2) with two-tailed Student’s t test. (C) STM2457 (20 μM) treatment in MHCC97L cells induced ROS stress in a time-dependent manner. Mean ± SEM (n = 3) with two-tailed Student’s t test. (D) STM2457 (20 μM) treatment in MHCC97L cells induced apoptosis in a time-dependent manner. Mean ± SEM (n = 3) with two-tailed Student’s t test. (E) Adding back 2 mM antioxidant N-acetylcysteine (NAC) or 2 mM α-KG for 48 h rescues STM2457-induced apoptosis (5 days, 20 μM). Mean ± SEM (n = 3) with two-tailed Student’s t test. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.

Figure 7

m6A stabilizes the SSP in sorafenib-resistant cells (A and B) Average m6A violin plot of GLORI-seq in DMSO and sorafenib-resistant (7 μM for 7 days) MHCC97L cells showed that global m6A level remains largely unchanged upon sorafenib treatment. (C) Venn diagram of overlap between m6A sites identified in DMSO and sorafenib-resistant cells. (D) Venn diagram showing overlap of sorafenib upregulated genes (fold change >2) and sorafenib exclusive m6A sites. (E) Scatterplot of expression difference between sorafenib and DMSO. The color indicates that the m6A site is exclusive to sorafenib (red) or DMSO (blue). (F) STM2457 treatment blocks the upregulation of PHGDH in sorafenib resistance cells (6 days sorafenib treatment). Mean ± SEM (n = 3) with two-tailed Student’s t test. (G and H) 20 μM STM2457 treatment suppressed the SSP (G) mRNA and (H) protein level under 6 μM sorafenib treatment for 5 days. Mean ± SEM (n = 3) with two-tailed Student’s t test. (I) GLORI-seq showed increased m6A sites numbers at the SSP in sorafenib-resistant MHCC97L cells. (J) TKI treatment upregulated METTL3 and SSP proteins in MHCC97L cells. (K) STM2457 treatment suppressed sorafenib-mediated mRNA stability in MHCC97L (7 days, 7 μM sorafenib and 20 μM STM2457). Mean ± SEM (n = 3) with two-tailed Student’s t test. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.

Figure 8

STM2457 sensitizes HCC cells to sorafenib by inducing oxidative stress (A) STM2457 treatment induced apoptosis and further enhanced the apoptotic rate in sorafenib- or lenvatinib-treated MHCC97L HCC cells. Mean ± SEM (n = 3) with two-tailed Student’s t test. (B) STM2457 and sorafenib/lenvatinib alone triggers ROS stress, and combination treatment further enhanced the ROS level in MHCC97L HCC cells. Mean ± SEM (n = 3) with two-tailed Student’s t test. (C) Proliferation curve of HCC cells treatment (Huh7, Hepa1-6, and MHCC97L) with STM2457 and sorafenib/lenvatinib. Mean ± SEM (n = 3) with two-tailed Student’s t test. The combination treatment group has the slowest proliferation rate in all HCC cells tested. (D–G) STM2457 and sorafenib inhibited MHCC97L subcutaneous implanted HCC tumor development as demonstrated by growth curve, tumor volume, and tumor size. Combination treatment further suppressed the tumor size (blue dotted line, control; green dotted line, sorafenib; orange dotted line, STM2457; and the red dotted line is combo treatment). Mean ± SEM (n = 8) with two-tailed Student’s t test. Scale bar, 1 cm. (H) Mice weight after receiving indicated drug treatment. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.