KRAS mutations may cause platinum drug resistance by changing ALKBH5 protein modifications in lung cancer

Despite the widespread occurrence of KRAS constitutively active mutations in lung cancers (24–26), the association between these mutations and platinum resistance in NSCLC has not been fully investigated. KRAS G12C (41%), KRAS G12V (22%), KRAS G12D (12%), and KRAS G12A (9.3%) represent the most commonly observed mutations in KRAS within lung cancers (7, 27). We first established BEAS-2B cells derived from normal bronchial epithelium, stably expressing vector, KRAS constitutively active form (KRAS G12V), or a KRAS enzymatic mutant (KRAS S17N) and treated these cells with either DMSO or cisplatin. As shown in Figure 1A, the overexpression of constitutively active KRAS (KRAS G12V) but not KRAS enzymatic mutant (KRAS S17N) led to an increase in phosphorylated ERK and JNK protein levels in BEAS-2B cells. Notably, cisplatin treatment activates ERK/JNK signaling, and this activation can be further enhanced by the overexpression of KRAS G12V (Figure 1A). Meanwhile, cisplatin exposure significantly induced DNA damage in BEAS-2B cells, as evidenced by an increased expression of phosphorylated γH₂AX, a sensitive marker of DNA damage (Figure 1A). Strikingly, KRAS-G12V significantly bolstered the resistance of BEAS-2B cells to cisplatin-induced DNA damage (Figure 1A). Next, we treated KRAS WT NSCLC cells, including NCI-H522 and NCI-H292, KRAS G12C-mutant NSCLC cells, such as NCI-H23 and NCI-H2122, and KRAS G12A-mutant NSCLC cells, such as NCI-H1573 and NCI-H2009, with either DMSO or cisplatin. Consistently, ERK/JNK signaling was more significantly activated, resulting in lower DNA damage in response to the chemotherapeutic drug in KRAS-mutant NSCLC cell lines, including NCI-H23, NCI-H2122, NCI-H1573, and NCI-H2009 as compared with KRAS WT lung cancer cell lines such as NCI-H522 and NCI-H292 (Figure 1B). Precise single-cell DNA damage analysis using the alkaline comet assay revealed that KRAS WT NSCLC cells exhibit greater sensitivity to cisplatin-induced DNA damage compared with KRAS mutant lung cancer cells (Figure 1C and Supplemental Figure 1A; supplemental material available online with this article; https://doi.org/10.1172/JCI185149DS1↗). Additionally, cisplatin treatment markedly induced apoptosis in NCI-H522, whereas it had a marginal effect on apoptosis of NCI-H23 cells (Figure 1D and Supplemental Figure 1B). We next examined colony forming ability of these cells. As shown in Figure 1E and Supplemental Figure 1C, NCI-H522 (KRAS WT) gave rise to fewer colonies than NCI-H23 (KRAS G12C) when the cells were treated with cisplatin. Collectively, these results suggest a positive correlation between KRAS constitutively active mutations and platinum resistance in NSCLC cells.

To rigorously investigate whether KRAS mutations confer platinum resistance in lung cancer cells, we adopted 2 approaches: overexpressing a constitutively active KRAS mutant in NCI-H522 (KRAS WT) and knocking down KRAS in NCI-H23 (KRAS G12C) cells. KRAS G12V overexpression markedly inhibited cisplatin-induced DNA damage and cell apoptosis in NCI-H522 cells (Figure 1, F and G, and Supplemental Figure 1D). Conversely, KRAS knockdown (KD) greatly enhanced cisplatin-induced DNA damage and cell apoptosis in NCI-H23 cells (Figure 1, H and I, and Supplemental Figure 1, E and F). In addition to the platinum-based drugs, paclitaxel (PTX) is also a frequently used chemotherapeutic drug in lung cancer treatment (28–31). Therefore, we next examined whether KRAS mutants induce PTX resistance in lung cancer cells. As shown in Figure 1, J–O, and Supplemental Figure 1G, KRAS-mutant NSCLC cells including H23 and H1573 and KRAS-WT NSCLCs including H522 and H292 are responsive to PTX treatment while KRAS KD did not increase the sensitivity of H23 and H1573 NSCLC cells to PTX treatment. However, we also observed that ERK/JNK signaling is highly activated in KRAS-mutant cells, exhibiting lower levels of DNA damage compared with KRAS WT cells when treated with other DNA damage reagents, such as doxorubicin and etoposide (Supplemental Figure 2, A and B). Taken together, these results provide compelling evidence that KRAS constitutively active mutations specifically confer platinum resistance, as well as other DNA damage inducers, but not PTX in NSCLC cells.

ATP-binding cassette (ABC) transporters are the largest and oldest membrane proteins in humans, which pump out various toxic compounds from the cells. The major cause of multidrug resistance (MDR) and chemotherapeutic failure is believed to be the efflux of toxic drugs mediated by ABC transporters (32–34). Therefore, we next examined whether KRAS mutant-mediated platinum resistance is possibly mediated by ABC transporters. As shown in Supplemental Figure 2, C–E, the expression of ABC transporters including ABCB1, ABCG2, and ABCC1 is comparable in KRAS-WT and mutant NSCLC cells. Additionally, KRAS KD did not affect the expression of ABC transporters in KRAS-mutant lung cancer cells (Supplemental Figure 2, F–H). Together, these data suggest that KRAS-mutant–mediated NSCLC platinum resistance is not attributed to the dysregulation of ABC transporters.

Our previously published study demonstrated that mammalian cells activate ERK/JNK signaling to induce m⁶A methylation in DNA repair-related genes. This process safeguards the genomic stability by regulating ALKBH5 PTMs in response to oxidative stress (35). The ERK/JNK signaling pathway can be activated by ROS stress and oncogenes such as KRAS (24, 36–38). To examine whether the KRAS mutant regulates PTMs of ALKBH5, we established BEAS-2B cells, stably expressing vector, constitutively active KRAS mutant (KRAS G12V), and KRAS enzymatic mutant (KRAS S17N). Denaturing immunoprecipitation (IP) analysis of ALKBH5 revealed that expression of constitutively active KRAS significantly induced endogenous ALKBH5 phosphorylation and SUMOylation (Figure 2A and Supplemental Figure 3A). Consistently, inhibition of KRAS G12C by sotorasib, or ERK by PD0325901, markedly reduced both phosphorylation and SUMOylation of ALKBH5 in NCI-H23 cells. These findings suggest that ALKBH5 PTMs, including phosphorylation and SUMOylation, are driven by KRAS/ERK signaling (Supplemental Figure 3, B and C). In addition, both ALKBH5 phosphorylation-deficient mutant S325A and ALKBH5 SUMOylation-deficient mutant ALKBH5 K86R/K321R significantly reduced KRAS G12V–induced ALKBH5 phosphorylation and SUMOylation (Figure 2, B and C, and Supplemental Figure 3, D and E). These findings suggest that the constitutively active KRAS mutant induces ALKBH5 phosphorylation at serine 325 (S³²⁵), and SUMOylation at lysines 86 (K⁸⁶) and 321 (K³²¹). Our previous study suggests that ALKBH5 phosphorylation triggers its SUMOylation, which in turn inhibits its m⁶A demethylase activity (35). Therefore, we checked whether the constitutively active KRAS mutant induces global mRNA m⁶A modification. Consistently, ectopic expression of KRAS G12V but not KRAS S17N markedly increased global mRNA m⁶A methylation in BEAS-2B cells (Figure 2D). To further determine the effect of the KRAS constitutively mutant on mRNA m⁶A methylation transcriptome wide, we performed m⁶A-Seq analyses. We observed that KRAS G12V overexpression led to 1,542 m⁶A peak alterations in total among transcripts (log₂ fold change [log₂FC] > 0.3 or log₂FC< –0.3, P < 0.05). Consistent with previous studies (13, 39, 40), the identified m⁶A peaks are located in sequences containing the canonical m⁶A methylation consensus motif RRACH (R = G or A; H = A, C, or U; where A is converted to m⁶A) (Supplemental Figure 3F). In line with the m⁶A level determined by dot blot, m⁶A-Seq results revealed that the majority of m⁶A peaks are upregulated upon KRAS G12V expression. Overall, 1,259 peaks were upregulated and 283 peaks were downregulated (Figure 2, E and F, and Supplemental Figure 3G). Additionally, Gene Ontology (GO) analysis of 1,259 m⁶A peaks that were significantly upregulated upon KRAS G12V overexpression showed that these peaks are enriched in the genes involved in pathways including RAS and MAPK signaling pathways, platinum drug resistance, and nucleotide excision repair (NER) (Figure 2G). Platinum-based drugs serve as antitumor drugs mainly by facilitating cancer cells DNA damage through inducing crosslink formation between purine nucleotides (22, 41, 42). m⁶A-seq analysis suggests that the constitutively active KRAS mutant overexpression led to an m⁶A increase in the genes associated with NER, suggesting an important role of the NER pathway in KRAS-mutant–mediated platinum resistance in lung cancer cells. Consistently, cisplatin-induced m⁶A increase in KRAS-mutant NCI-H23 cells was significantly higher as compared with KRAS WT NCI-H522 cells (Figure 2H). Meanwhile, inhibition of KRAS G12C or ERK effectively blocked cisplatin-induced m⁶A methylation in KRAS G12C-mutant NCI-H23 cells, suggesting that activation of KRAS/ERK signaling is responsible for the increased m⁶A methylation observed following cisplatin treatment (Supplemental Figure 3, H and I). Additionally, blocking mRNA m⁶A increase by expression of either ALKBH5 S325A or ALKBH5 K86/321R significantly sensitized NCI-H23 cells to cisplatin-induced DNA damage (Figure 2, I and J). Conversely, overexpression of the ALKBH5 phosphorylation-mimic mutant ALKBH5 S325D in KRAS WT H522 cells significantly increased their resistance to cisplatin (Figure 2K). Collectively, these results suggest that the KRAS mutant regulates global mRNA m⁶A methylation by modulating ALKBH5 PTMs. Moreover, KRAS-mutant–driven platinum resistance in NSCLC correlates with KRAS-mutant–induced ALKBH5 PTMs.

Based on the aforementioned observations, we conducted a comparison of cisplatin-induced ALKBH5 PTMs between KRAS WT NCI-H522 and KRAS-mutant NCI-H23 cells. Notably, phosphorylation of ERK and JNK, as well as phosphorylation and SUMOylation of ALKBH5, were more significantly induced by cisplatin in NCI-H23 cells as compared with NCI-H522 cells (Figure 2L and Supplemental Figure 3J). In contrast, the levels of cisplatin-induced γH2A.X in NCI-H522 cells were considerably higher than those in NCI-H23 cells (Supplemental Figure 3J). These results indicate that the KRAS mutant promotes chemoresistance in lung cancer cells, a phenomenon correlated with the upregulation of ERK/JNK signaling as well as increased ALKBH5 phosphorylation and SUMOylation. To further confirm that the KRAS mutant confers drug resistance via ALKBH5 SUMOylation in NSCLC cells, we inhibited ALKBH5 SUMOylation in both NCI-H522 and NCI-H23 cells by knocking down SUMO E2 UBC9. The results showed that KRAS-mutant NCI-H23 cells are more sensitive to UBC9 depletion as compared with KRAS WT NCI-H522 cells and inhibition of ALKBH5 SUMOylation markedly enhances cisplatin-induced DNA damage and cell apoptosis in KRAS-mutant cells (Figure 3, A–G). Together, these findings strongly suggest that cisplatin-induced ALKBH5 PTMs play important roles in drug resistance conferred by KRAS mutants.

To further explore the molecular mechanism underlying KRAS-mutant–mediated platinum resistance in lung cancer, we performed RNA-Seq analysis in control and KRAS G12V–expressing NCI-H522 cells. As shown in Figure 4A, KRAS G12V led to significant alterations in gene expression, with 429 and 283 genes upregulated and downregulated, respectively (log₂FC > 0.3 or log₂FC< –0.3, P < 0.05). GO analysis of those 712 differentially expressed genes induced by KRAS G12V revealed that the downstream target genes of the KRAS mutant are enriched in pathways involved in RAS and MAPK signaling pathways and platinum resistance, as well as pathways in cancer (Figure 4B). Additionally, gene set enrichment analysis (GSEA) analyses revealed that the downstream target genes of the KRAS mutant are enriched in pathways involved in RAS signaling and DNA damage repair (Figure 4, C and D). By integrative analysis of RNA-Seq and m⁶A-Seq data, 105 genes were differentially expressed with an upregulation of m⁶A methylation level upon KRAS G12V expression. GO analysis of these genes revealed that those genes are also enriched in the pathways involved in the activation of RAS and MAPK signaling as well as platinum resistance (Figure 4, F and G). Among these genes, DDB2 and XPC stood out due to their important roles in multiple pathways that regulate the NER and platinum resistance (22, 23) (Figure 4, D, F, and G). Notably, both the m⁶A methylation and expressions of DDB2 and XPC are significantly induced by KRAS G12V (Figure 4, A, E, H, and I). Consistent with the RNA-Seq and m⁶A-Seq results, both the transcription and mRNA m⁶A methylation levels of DDB2 and XPC were markedly increased by KRAS G12V, as determined by quantitative reverse transcriptase PCR (qRT-PCR) and the methylated RNA IP (MeRIP) followed by RT-PCR analyses, respectively. METTL3 KD, which blocks KRAS G12V–induced m⁶A methylation, significantly inhibited DDB2 and XPC expression (Figure 4, J–M), suggesting that the KRAS mutant regulates DDB2 and XPC mRNA expression in an m⁶A-dependent manner. Notably, KRAS G12V–induced upregulation of DDB2 and XPC was reversed by overexpression of a SUMOylation-deficient mutant ALKBH5 but not WT ALKBH5, suggesting that KRAS mutant regulates DDB2 and XPC expression through ALKBH5 SUMOylation (Supplemental Figure 4, A and B). We next investigated whether the KRAS-mutant drove platinum resistance, at least partially through the induction of DDB2 and XPC expression. As illustrated in Supplemental Figure 4, C and D, cisplatin treatment significantly induced the expression of DDB2 and XPC; and KRAS G12V further augmented cisplatin-induced expression of these genes in BEAS-2B cells (Supplemental Figure 4, C and D). In addition, cisplatin significantly induced expression of DDB2 and XPC in KRAS WT NCI-H522 cells (Supplemental Figure 4, E and F). Notably, the induction of expression of these genes was more significant in KRAS-mutant NCI-H23 cells compared with NCI-H522 cells (Supplemental Figure 4, E and F). These results suggest that the enhanced NER pathway with upregulation of DDB2 and XPC likely contributes to the resistance to chemotherapeutic drug in KRAS G12C-mutant NCI-H23. Collectively, these results indicate that the KRAS mutant induces chemoresistance possibly by facilitating the expression of NER-related genes, including DDB2 and XPC, in an m⁶A-dependent manner in NSCLC cells.

We next investigated the interplay between cisplatin-induced gene expression and the elevated m⁶A methylation levels of DDB2 and XPC. As depicted in Figure 5, A and B, either expression of a SUMOylation-deficient mutant ALKBH5 or METTL3 KD by 2 specific shRNAs effectively blocked the cisplatin-induced m⁶A methylation increase of DDB2 and XPC, leading to a downregulation of both genes in both NCI-H522 and NCI-H23 cells (Figure 5, C and D). Increased m⁶A methylation levels of DDB2 and XPC in KRAS-mutant NCI-H23 resulted in the prolonged half-lives of DDB2 and XPC mRNA compared with KRAS WT NCI-H522 cells. Cisplatin treatment markedly enhanced the stability of DDB2 and XPC mRNA in KRAS-mutant NCI-H23 cells compared with KRAS WT NCI-H522 cells. Notably, the prolonged half-lives of DDB2 and XPC mRNA induced by cisplatin in NCI-H522 and NCI-H23 cells were entirely reversed by expression of the SUMOylation-deficient ALKBH5 or by METTL3 KD (Figure 5, E–H). Similarly, either pharmacological inhibition of KRAS G12C or ERK completely reversed the prolonged mRNA half-lives of DDB2 and XPC in KRAS G12C harboring H23 cells (Supplemental Figure 4, G–J). Thus, these results suggest that cisplatin-induced m⁶A methylation of DDB2 and XPC leads to stabilization of their mRNA, which can be further augmented by the KRAS mutant in NSCLC cells.

Next, we aimed to uncover the mechanism underlying KRAS/ERK/ALKBH5 PTMs/ DDB2 and XPC signaling axis-mediated platinum resistance in NSCLC cells. Given that both DDB2 and XPC are key components of NER machinery, we sought to determine whether the NER pathway is involved in KRAS mutation–driven platinum resistance in lung cancer. Consistent with previous studies (23, 43), KD of either DDB2 or XPC significantly reduced NER activity in NCI-H23 cells (Supplemental Figure 5, A–D). Notably, NER activity was significantly higher in KRAS-mutant NSCLC cells compared with KRAS WT lung cancer cells (Supplemental Figure 5, E and F), suggesting a positive correlation between KRAS mutations and NER activity in NSCLC cells. Additionally, KRAS G12V overexpression significantly enhanced NER activity in KRAS WT H522 cells (Supplemental Figure 5, G and H). Conversely, NER activity in KRAS-mutant H23 cells was significantly inhibited by KRAS KD (Supplemental Figure 5, I and J). Together, these data provide compelling evidence that KRAS mutations positively regulate NER activity in NSCLC cells. Moreover, as shown in Supplemental Figure 6, A–F, KD of either DDB2 or XPC significantly sensitized KRAS-mutant H23 cells to cisplatin-induced DNA damage. Furthermore, KRAS G12V overexpression–induced H522 cisplatin resistance was completely blocked by KD of either DDB2 or XPC (Figure 5, I–L). Collectively, these results suggest that DDB2 and XPC play key roles in KRAS mutation-driven platinum resistance in NSCLC cells and that KRAS mutations confer drug resistance by enhancing NER activity.

RNA m⁶A methylation is dynamically regulated by m⁶A writer, of which the major catalytic subunit is METTL3, and erasers, including ALKBH5 and FTO (13, 35). Therefore, we investigated whether KRAS mutation–driven platinum resistance involves the regulation of FTO or METTL3 expression. Interestingly, KRAS G12V overexpression did not affect the protein levels of FTO or its PTMs, including phosphorylation and SUMOylation (Supplemental Figure 7A). Similarly, cisplatin resistance of KRAS-mutant H23 cells could not be overcome by FTO overexpression (Supplemental Figure 7B). Consistently, neither the cisplatin-induced expression nor the m⁶A methylation of DDB2 and XPC was restored by FTO overexpression (Supplemental Figure 7, C–F), suggesting that DDB2 and XPC, as functional mediators of KRAS mutations, are specific downstream targets of ALKBH5. Moreover, KRAS G12V overexpression significantly upregulated METTL3 expression (Supplemental Figure 7, G and H). However, both KRAS G12V– and cisplatin-induced METTL3 expression were completely reversed by overexpression of a SUMOylation-deficient mutant ALKBH5 (Supplemental Figure 7, H and I), indicating that KRAS mutations induce METTL3 expression by regulating ALKBH5 SUMOylation. Collectively, these findings suggest that KRAS-mutant–driven platinum resistance in NSCLC cells is mediated directly through the regulations of ALKBH5 SUMOylation. Furthermore, DDB2 and XPC, identified as functional mediators of KRAS mutants, are specific downstream targets of ALKBH5.

To further determine whether KRAS mutation confers NSCLC drug resistance through the KRAS/ERK/JNK/ALKBH5 PTMs/m⁶A/DDB2 and XPC/NER signaling axis in vivo, we carried out xenograft experiments with NSCLC cells. As shown in Figure 6, A–C, KRAS-mutant NCI-H23 cells were more resistant to cisplatin treatment compared with KRAS WT NCI-H522 in vivo. Notably, ectopic expression of SUMOylation-deficient mutant ALKBH5 (SD-ALKBH5) substantially sensitized NCI-H23 cells to cisplatin treatment in vivo. Consistent with previously published studies (44, 45), the toxic effect of cisplatin treatment was minimal in our experimental settings, as evidenced by the stable mouse weights and unaltered xenograft growth (Supplemental Figure 7J). In addition, ERK/JNK signaling was significantly more activated, resulting in lower levels of DNA damage in KRAS-mutant NCI-H23 cells in the xenograft model with cisplatin treatment compared with KRAS WT NCI-H522 xenografts (Figure 6D). Expression of SUMOylation-deficient mutant ALKBH5 (SD-ALKBH5) substantially facilitated cisplatin-induced DNA damage in NCI-H23 xenografts (Figure 6D). Consistently, ALKBH5 PTMs, including phosphorylation and SUMOylation, are significantly more pronounced in response to cisplatin treatment in KRAS-mutant H23 cells compared with KRAS WT H522 cells in vivo (Figure 6D). Moreover, global mRNA m⁶A methylation levels were induced more significantly in NCI-H23 xenografts by cisplatin treatment as compared with NCI-H522 xenografts (Figure 6E). SUMOylation-deficient mutant ALKBH5 (SD-ALKBH5) overexpression completely blocked cisplatin-induced mRNA m⁶A methylation in NCI-H23 xenografts (Figure 6E). More importantly, cisplatin treatment significantly induced m⁶A methylation of DDB2 and XPC in KRAS WT NCI-H522 xenografts (Figure 6, F and G). The induction of m⁶A methylation levels of these genes was even more pronounced in KRAS-mutant NCI-H23 xenografts (Figure 6, F and G). Importantly, the cisplatin-induced m⁶A methylation of DDB2 and XPC genes in KRAS-mutant NCI-H23 xenografts was blocked by overexpression of the SUMOylation-deficient mutant ALKBH5 (SD-ALKBH5) (Figure 6, F and G). Consistently, the expression levels of DDB2 and XPC were higher in NCI-H23 xenografts than in NCI-H522 xenografts with cisplatin treatment (Figure 6, H and I). Overexpression of SUMOylation-deficient mutant ALKBH5 (SD-ALKBH5) blocked cisplatin-induced upregulation of DDB2 and XPC in NCI-H23 xenografts (Figure 6, H and I). Collectively, these results indicate that the KRAS mutant promotes platinum resistance in NSCLC cells in vivo by hijacking ALKBH5 PTM–mediated DNA repair pathways.

To investigate whether METTL3 KD exerts a similar rescue phenotype as ectopic expression of SD-ALKBH5, we established stable lines of NCI-H522 and NCI-H23 cells expressing scramble control or METTL3-specific shRNAs. As shown in Supplemental Figure 8 A and B, METTL3 was significantly KD by both specific shRNAs. METTL3 KD exhibited greater sensitivity to cisplatin-induced DNA damage and cell apoptosis in KRAS-mutant NCI-H23 cells compared with KRAS WT NCI-H522 cells (Figure 7A and Supplemental Figure 8, C–E). To assess the potential therapeutic application of targeting METTL3 in KRAS-mutant NSCLC cells, we employed a small molecule, STM2457, which potently and selectively inhibited METTL3 enzymatic activity in a recent study (46). Consistently, METTL3 inhibition by STM2457 markedly inhibited global mRNA m⁶A methylation in KRAS-mutant NCI-H23 cells (Figure 7B). Similar to METTL3 KD, METTL3 inhibition significantly sensitized NCI-H23 cells to cisplatin-induced DNA damage (Figure 7, C and D). Notably, γ-H2AX levels were increased upon METTL3 inhibition in NSCLC cells. METTL3 inhibition reduced m⁶A methylation in NER-related genes, such as DDB2 and XPC, resulting in their mRNA decay and subsequent suppression of NER activity. Additionally, METTL3 inhibition significantly enhanced the cisplatin-mediated suppression of the colony-forming ability of KRAS-mutant NCI-H23 cells (Figure 7E and Supplemental Figure 8F), and it markedly increased the sensitivity of NCI-H23 cells to cisplatin treatment in vivo (Figure 7, F–H). Meanwhile, in vivo METTL3 inhibition using STM2457 demonstrated minimal toxicity. Over a 40-day monitoring period, STM2457 injection didn’t cause acute mortality or substantial body weight loss in mice, nor did it visibly affect the morphology of major organs. Collectively, these results suggest that METTL3 is a promising and safe target for sensitizing KRAS-mutant NSCLC to cisplatin treatment.

To further determine whether the aforementioned observations also exist in the primary lung cancer cells from patients, we collected 3 pairs of platinum-based chemotherapeutic primary lung adenocarcinoma tissues, both KRAS WT and mutant, from patients at University of Florida Shands Hospital. As shown in Figure 8A, the ERK/JNK signaling is more significantly activated, resulting in lower levels of DNA damage in KRAS-mutant lung cancer cells compared with KRAS WT lung cancer cells from patients. Consistently, ALKBH5 PTMs, including phosphorylation and SUMOylation, are much more abundant in KRAS-mutant primary lung cancer cells compared with KRAS WT cells (Figure 8A). Moreover, qRT-PCR analyses showed that both the DDB2 and XPC genes were expressed at much higher levels in primary KRAS-mutant lung cancer cells compared with primary KRAS WT lung cancer cells (Figure 8, B and C). In addition, the m⁶A methylation levels of DDB2 and XPC transcripts were also higher in primary KRAS-mutant lung cancer cells compared with primary KRAS WT lung cancer cells (Figure 8, D and E). These findings suggest that the identified KRAS/ERK/JNK/ALKBH5 PTMs/m⁶A/DDB2 and XPC/NER signaling axis are also active in primary lung cancer cells from patients. KRAS mutants confer platinum resistance at least partially through the posttranscriptional regulation of DDB2 and XPC in an m⁶A-dependent manner, thereby facilitating the nucleotide excision of the crosslinked purine nucleotides induced by platinum-based chemotherapy drugs.

In all NSCLC triple transgenic NSG-SGM3 (NSGS) mouse xenograft studies, both male and female mice were used. Sex was not considered as a biological variable in the statistical analyses. The NSGS mice used for NSCLC xenograft studies were purchased from The Jackson laboratory.

Both the normal epithelial cells BEAS-2B, WT KRAS harboring NCI-H522 and NCI-H2087, and KRAS mutant NSCLC cells including NCI-H23, NCI-H2122, NCI-H1573, and NCI-H2009 were provided in house. The BEAS-2B cells were cultured in the BEGM Bronchial Epithelial Cell Growth Medium BulletKit (Lonza, catalog CC-3170). For routine maintenance, all the NSCLC cells were cultured at 37°C with 5% CO₂ in RPMI-1640 containing 10% FBS and 1% penicillin/streptomycin.

pCDH-Flag-KRAS G12V and pCDH-Flag-KRAS S17N were subcloned from plasmids provided in house. The pCDH-Strep-ALKBH5-HA expression plasmid was generated by cloning the corresponding coding sequence into the pCDH-Strep vector. All the pCDH-Strep-HA-ALKBH5 K/R (lysine to arginine) or S/A (serine to alanine) mutants were derived from pCDH-Strep-HA-ALKBH5 by site-directed mutagenesis. Information about antibodies used in this study is provided in. Supplemental Table 1

For the lung cancer cell drug resistance analysis, the cells were treated with DMSO or 20 μM cisplatin for 24 hours. For the rescue experiment by METTL3 inhibition, the indicated cells were treated with 10 μM STM2457 for 24 hours. For KRAS G12C inhibition, NCI-H23 cells were treated with 0.1 μM sotorasib for 3 hours. For ERK inhibition, NCI-H23 cells were treated with 1 μM PD0325901 for 3 hours.

The Western blot and co-IP analyses were performed according to standard protocols as described previously with minor changes (65), by using the antibodies as indicated. For examining SUMO-modified proteins, cells were lysed in denaturing buffer (50 mM Tris-HCl pH7.5, 150 mM NaCl, 4% SDS, 1mM EDTA, 8% glycerol, 50mM NaF, 1 mM DTT, 1mM phenylmethylsulfonyl fluoride (PMSF), and protease inhibitors) supplemented with 20 mM N-ethylmaleimide (NEM) and heated at 90°C for 10 minutes. For the following IP assays, the lysates were further diluted to 0.1% SDS and immunoprecipitated with antibodies against target proteins at 4°C overnight. SUMO-modified proteins were then tested by Western blotting.

The alkaline comet analyses were performed with the Comet Assay kit (R&D Systems, catalog 4250-050-K) according to the manufacturer’s instructions with minor changes. Briefly, we combined cells at 0.5 million per mL with molten low melting agarose (LMA) gel at a ratio of 1:10 (v/v) and immediately pipetted 80 μL onto the comet slice and placed it at 4°C for 30 minutes in the dark. We immersed slice into 4°C lysis buffer for 1.5 hours. Next, we immersed the slice in alkaline unwinding solution (200 mM NaOH, 1 mM EDTA, pH>13) for 20 minutes at room temperature. Finally, electrophoresis was performed in alkaline electrophoresis solution and the comet slices were stained with SYBR gold dye at room temperature for 30 minutes. The tail length was calculated by ImageJ software (NIH).

KD of target genes by shRNAs was done as described previously (65). Scramble sequence and all the shRNAs against target genes were inserted into the pLKO.1 vector. The sequences for shRNAs are listed in Supplemental Table 2. For qRT-PCR analysis, total RNA was extracted from various cells as indicated and reverse transcribed by using kits purchased from Thermo Fisher. The primer sequences used in the qRT-PCR are listed in Supplemental Table 2.

0.5 × 10⁶ of the indicated cells were treated with DMSO or 20 μM cisplatin for 24 hours. After that, all the cells were collected and washed with ice-cold PBS and 1× annexin V binding buffer, respectively. Then the cells were stained by 2.5 μL anti–annexin V antibody and 1μM DAPI (final concentration) in the dark and on ice for 30 minutes. After that, the cells were subjected to flow cytometry analysis.

Briefly, 2 million lung adenocarcinoma cells were subcutaneously injected into 2 flanks of each NSGS mouse. And 5 mg/kg cisplatin alone or together with 30 mg/kg STM2457 was given i.p. every 3 days when tumor volume reached approximately 100 mm³. Tumor volume and mouse weight measurements were taken every 4 days and 7 days, respectively. Tumor volume was calculated according to the formula [ D × (d²) ] /2, where D represents the large diameter of the tumor and d represents the small diameter of the tumor. Animals were individually monitored throughout the experiment.

mRNA m⁶A methylation was analyzed by dot-blot assays according to our published procedures with minor changes (13, 35). Briefly, total RNA was extracted using TRIzol reagent (Thermo Fisher, catalog 15596018), and mRNAs were separated using the Dynabeads mRNA Purification Kit (Thermo Fisher, catalog 61006). The mRNAs were denatured at 95°C for 5 minutes, followed by chilling on ice directly. Next, 400 ng mRNAs were spotted to positively charged nylon (GE healthcare), air-dried for 5 minutes, and crosslinked using a 245 nm UV cross linker. The membranes were blocked in 5% nonfat milk plus 1% BSA in PBST for 2 hours and then incubated with anti-m⁶A antibodies at 4°C overnight. After 3 times washing with PBST, the membranes were incubated with Alexa Fluor 680 goat anti-rabbit IgG secondary antibodies at room temperature for 1 hour. Membranes were subsequently scanned using image studio. Methylene blue staining was used as a loading control to make sure equal amounts of mRNAs were used for dot-blot analysis.

m⁶A RNA IP (MeRIP) analyses were performed according to the published paper (66). The primer sequences used in the qRT-PCR are listed in Supplemental Table 1.

All the indicated cells were treated with 5 μg per mL actinomycin D and collected at indicated time points. The total RNA was extracted by TRIzol reagent and subjected to qRT-PCR analysis. The primer sequences used in qRT-PCR are listed in. Supplemental Table 2

Total RNAs were extracted from NCI-H522 cells stably expressing empty vector and KRAS G12V by TRIzol reagent (Thermo Fisher, catalog 15596018). 10 μg of total RNAs were fragmented with RNA fragmentation buffer (Thermo Fisher, catalog AM8740), and 1 μg of RNA fragments were kept for RNA-Seq analysis. 9 μg Of RNA fragments were used for IP enrichment by using anti-m⁶A antibody (Synaptic Systems, catalog 202 003), namely for the m⁶A-seq analysis. Both the 1 μg of RNA fragments saved for the RNA-Seq analysis and the m⁶A antibody–enriched RNA fragments for m⁶A-seq analysis were rRNA depleted by using the rRNA Depletion Kit (NEB, catalog E6310L). Then the rRNA-depleted RNA fragments were used in the sequence library construction by using the NEBNext Ultra II Directional RNA Library Prep Kit for Illumina (NEB, catalog E7760L). Finally, cDNA libraries purified by using AMPure beads (Beckman Coulter, catalog A63881) were submitted to the next-generation sequencing service at the core facility of University of Florida for sequencing. All libraries were processed on a NovaSeq S4 2X150 platform (Illumina) with a paired-end 150-base pair read length, and 50 × 10⁶ reads per sample was required.

For bulk RNA-Seq analysis, bulk RNA-Seq raw sequencing reads were aligned to the human genome, hg38, and sequencing quality and alignment rate were examined using Nextflow pipeline (nf-core/rnaseq 3.12) (67). Gene expression was quantified at the gene level using Salmon. RNA-Seq libraries were then normalized using median of ratios method, and genes were tested for differential expression between the empty vector and KRAS G12V–overexpressed samples with DESeq2, version 1.36 (68). The Wald test (69) was employed to identify differentially expressed genes. For visualization, pheatmap, version 1.0.12, was used for showing differential expression between samples. The gene set enrichment test was performed using clusterProfiler, version 4.7.1 (70). KEGG (71) and Reactome (72) databases were used in GSEA. To control the false positive rate, multiple testing correction was applied using the Benjamini-Hochberg method to adjust the P values obtained from both the differential expression analysis and GSEA. We set a significance threshold of adjusted P value at 0.05 to control the false discovery rate (73).

m⁶A-seq raw reads were trimmed using Trim Galore, version 0.6.10. FastQC, version 0.12, was used to examine the sequencing reads quality, and low-quality reads were removed. Raw reads were aligned to human reference genome hg38, Hisat2, version 2.2.1 (74). Peaks were called using Macs2, version 2.2.7.1 (75). m⁶A-Seq libraries were normalized to RNA-Seq libraries using DiffBind, version 3.8.4 (76). Differential analysis between empty vector and KRAS G12V–overexpressed samples was performed using DESeq2, version 1.38.3 (68). For visualization, metagene plot was generated using Guitar, version 2.14.0 (77). Motif analysis was performed using homer (78). To control the false positive rate, multiple testing correction was applied using the Benjamini-Hochberg method to adjust the P values obtained from both the differential expression analysis and GSEA. We set a significance threshold of adjusted P value at 0.05 to control the false discovery rate (73).

Results are presented as mean ± SD. Statistical analysis was calculated with 2-tailed Student’s t test or with ordinary 1-way ANOVA with Dunnett’s multiple-comparison test using GraphPad Prism 9 software. The colony-forming assay, qRT- PCR, and cell culture experiments were done with 3 technical replicates and repeated at least 3 times. P values equal to or less than 0.05 were considered statistically significant.

All the animal studies were approved by the mouse core facility at the University of Florida.

The raw and processed RNA-Seq and m⁶A-Seq data have been deposited into the NCBI’s Gene Expression Omnibus (GEO GSE268671↗). Values for all data points in graphs are reported in the Supporting Data Values file.

In-Press Preview

Electronic publication

The raw and processed RNA-Seq and m⁶A-Seq data have been deposited into the NCBI’s Gene Expression Omnibus (GEO GSE268671↗). Values for all data points in graphs are reported in the Supporting Data Values file.