Screening of transcription factors associated with NSCLC

The significance of open chromatin-accessible regions containing crucial genomic elements for transcription factor (TF) binding and gene regulation has been acknowledged [22, 23]. The assay for transposase-accessible chromatin followed by sequencing (ATAC-seq) and Deoxyribonuclease I (DNase I)-hypersensitive site sequencing (DNase-seq) has been widely used to measuring open regions of chromatin [24, 25]. To search for TFs associated with NSCLC, we initially conducted an analysis of chromatin accessibility profiling in human normal lung tissues, as well as LUAD and LUSC tissues, utilizing publicly available data (ATAC-seq, DNase-seq). A substantial number of accessible peaks were found near the transcription start site (TSS) in LUAD, LUSC, and normal lung tissues (Fig. 1A), indicating a propensity for binding to transcription factors. Additionally, compared to normal lung tissues, we found 33,215 and 45,630 differentially accessible peaks in LUAD and LUSC tissues, respectively (fold change >|3.5|, false discovery rate FDR < 0.01) (Fig. 1B). Among these, 10.06% and 9.76% of the differential accessible peaks were located in the promoter region (2 kb region upstream and downstream), respectively (Fig. 1C). Subsequently, these differential accessible peaks of promoter region were annotated to the nearest gene. After removing duplicate genes, we obtained 2810 and 6604 genes in LAUD and LUSC, respectively (Fig. 1C).

Fig. 1

Screening of transcription factors associated with NSCLC. A Line plot shows the chromatin accessibility in NSCLC and normal tissue around the TSS of the nearest genes. B Volcano plot showing differential expressed peaks between NSCLC and normal tissue. C Relative distribution of gene coding regions, introns, exons, and upstream and downstream regions. D Differential gene expression analysis of NSCLC. E–F Point density plots indicating the correlation between promoter chromatin accessible regions and gene expression in LUAD (E) and LUSC (F). G Venn diagram showing the overlapping number of differential genes. H-I GO biological process (H) and KEGG pathway (I) analysis of overlapping differential genes. J Overlapping TF motif of LUAD and LUSC

Given the complexity of gene expression regulation, it is necessary to explore biological questions from different perspectives. Therefore, multi-omics analysis is becoming increasingly important. We examined the differentially expressed genes (DEGs) in LUAD, LUSC, and para-tumor samples from the TCGA RNA-seq database. Comparative analysis revealed 4,674 DEGs in LUAD and 3,490 DEGs in LUSC when compared to the para-tumor samples (Fig. 1D). Subsequently, we assessed the correlation between gene expression and chromatin accessibility. Significantly positive correlations were observed between gene expression and promoter accessibility in LUAD (correlation coefficient r = 0.21, p value = 0.02) and LUSC (r = 0.22, p value = 4.94e-05) (Fig. 1E–F). By overlapping the differentially accessible promoters identified by ATAC-seq and the DEGs from RNA-seq, we obtained a total of 576 and 1,085 overlapping genes in LUAD and LUSC, respectively (Fig. 1G).

To study the role of these overlapping genes in the progression of NSCLC, Gene Ontology (GO) analysis was performed. The analysis revealed enriched categories in fundamental biological processes for the 576 overlapping genes in LUAD, such as cell adhesion, cell differentiation, and synaptic membrane adhesion (Fig. 1H). Likewise, for the 1085 overlapping genes in LUSC, 65 functional terms in the biological process category were found, including cell migration, cytoskeleton organization, and cell adhesion. In addition, KEGG enrichment chart of overlapping genes in LUAD and LUSC were conducted (Fig. 1I). The analysis of LUAD identified 6 enriched functional clusters, including transcriptional mis-regulation in cancer, cell adhesion molecules, cell cycle, PI3K-AKT signaling pathway, hematopoietic cell lineage, and protein digestion and absorption. The analysis of LUSC revealed 6 enriched functional clusters, including axon guidance, ECM-receptor interaction, focal adhesion, cell adhesion molecules, protein digestion and absorption, and PI3K-AKT signaling pathway. These findings provide evidence of the critical roles played by overlapping genes in NSCLC tumorigenesis and metastasis. Furthermore, we performed an analysis of possible transcription factor (TF) motifs in the overlapping genes using the de novo TF motif discovery software HOMER [20]. We identified a total of 33 and 31 TF motif candidates enriched at the promoter regions of overlapping differential genes in LUAD and LUSC, respectively. Notably, only one TF (NFIC) was found to be common between LUAD and LUSC (Fig. 1J). These results suggest that NFIC may play a role in regulating the development and progression of NSCLC.

NFIC overexpression inhibits the malignant phenotypes of NSCLC cells by inactivating the PI3K/AKT pathway

In order to verify the above hypothesis, we first analyzed the expression pattern of NFIC in LUAD and LUSC using TCGA and GEO (GSE81089) data. The results demonstrated a significant downregulation of NFIC in tumor tissues compared to normal tissues (Fig. 2A). This downregulation was further confirmed through qRT-PCR (Fig. 2B) and immunohistochemistry (IHC) analysis (Fig. 2C). Additionally, we utilized a receiver operating characteristic (ROC) curve to assess the diagnostic potential of NFIC as a biomarker for NSCLC. Figure 2D showed that NFIC has an area under the ROC curve (AUC) of 0.7289, suggesting its ability to distinguish between NSCLC and normal tissue with good diagnostic efficiency. Meanwhile, we analyzed the RNA-seq data of NSCLC cell lines (A549, H460) and human normal lung epithelial cell lines (HBE) available in the GEO database. The results revealed a significant downregulation of NFIC in A549 and H460 cells compared to HBE cells (Fig. 2E). Consistent with the GEO database, qRT-PCR and western blot analysis further confirmed the decline of NFIC in A549 and H460 cells compared to HBE cells (Fig. 2F–G).

Fig. 2

Analysis of NFIC expression. A NFIC expression in NSCLC tissues and adjacent normal tissues from the TCGA and GSE81089 datasets. B-C NFIC expression levels in tumor tissues were detected by qRT-PCR (B) and immunohistochemical (C). D ROC curve analysis of the NFIC gene. E Differentially expressed genes in A549 and H460 cells compared to HBE cells. F-G NFIC mRNA expression and protein expression patterns in NSCLC cells and HBE cells were measured by qRT-PCR (F) and western blot analysis (G), respectively. Bar = mean ± SD. **P < 0.01, ***P < 0.001

To further explore the function of NFIC in NSCLC, we transfected NFIC overexpression (oe-NFIC) vector into NSCLC (A549, H460) cells. In the qRT-PCR assay, the expression of NFIC was found to be significantly upregulated in A549 and H460 cells (Fig. 3A). The overexpression efficiency of NFIC was verified by western blot assay (Fig. 3B). Moreover, the overexpression of NFIC significantly suppressed the proliferation of A549 and H460 cells as determined via Cell Counting Kit-8 (CCK8) (Fig. 3C, D) and EdU staining (Fig. 3E). We used flow cytometry to analyze cell cycle progression, the data showed that NFIC overexpression caused a dramatic decrease in S-phase and accumulation in G1 phase of A549 and H460 cells (Fig. 3F), and NFIC overexpression markedly reduced colony formation in both A549 and H460 cells (Fig. 3G). Furthermore, the results of the wound healing assays showed that NFIC overexpression led to decreased cell migration (Fig. 3H). Transwell assays showed that the number of migrated and invaded cells decreased in NFIC overexpressing A549 (Fig. 3I) and H460 (Fig. 3J) cells compared to control cells. Subsequently, in vivo experimental results showed that compared to control group, the NFIC overexpression groups displayed smaller tumors and slower tumor growth (Fig. 3K–M).

Fig. 3

NFIC overexpression inhibits proliferation, migration, and invasion of NSCLC cells. A qRT-PCR analysis of NFIC overexpression efficiency in A549 and H460 cells. B Western blot analysis for NFIC, PI3K, p-PI3K, AKT, and p-AKT protein expression. C-D Proliferation of A549 (C) and H460 (D) cells following NFIC overexpression was determined using CCK8 assays. E EdU assays in A549 and H460 cell lines; scale bars = 100 µm. F Analyses of A549 and H460 cell cycle distributions by flow cytometry. G. Colony formation assays. H Wound-healing assays in A549 and H460 cell lines; scale bars = 100 µm. I-J Transwell migration and matrigel invasion assays for A549 (I) and H460 (J) cells. K The effect of NFIC overexpression on NSCLC subcutaneous xenografts in vivo. L-M The tumor volume (L) and weight (M) of tumors xenografted in nude mice. Bar = mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001

Previously, overlapping differential genes between chromatin property and RNA-seq data enriched in PI3K/AKT signaling pathway (Fig. 1I). To determine whether the overexpression of NFIC regulates the PI3K/AKT signaling pathway, the effects of NFIC overexpression on PI3K/AKT signaling pathway were investigated by western blot. The results showed that the overexpression of NFIC decreased the phosphorylation level of PI3K and AKT (Fig. 3B). These results suggested that the overexpression of NFIC suppressed the malignant phenotypes of NSCLC cells by inactivating the PI3K/AKT pathway.

NFIC negatively regulates METTL3 expression in A549 and H460 cells

Transcription factors, as key regulators of gene transcription, often affect the occurrence and development of cancer through regulating the transcription process of target genes. Therefore, it is necessary to further explore the target genes of the transcription factor NFIC. Recently, M6A modification has emerged as one of the most popular fields in cancer research [26], and a large number of studies have shown that m6A-related genes have been associated with NSCLC [10, 11, 27]. We found that NFIC overexpression resulted in a downregulation of global m6A modification level in both A549 and H460 cells compared to the control (Fig. 4A). Additionally, in the above-mentioned chromatin accessibility data (ATAC-seq, DNase-seq) and RNA-seq data, the promoter region chromatin accessibility of m6A-related genes (METTL3, FTO, IGF2BP3, HNRNPC, HNRNPA2B1) is increased in NSCLC (Fig. 4B), and these genes exhibited significant differential expression levels in NSCLC tissues compared to normal lung tissues (Fig. 4C). Moreover, as depicted in Fig. 4D, E, METTL3, FTO, IGF2BP3, HNRNPC, and HNRNPA2B1 demonstrated significant associations with prognosis in LUAD and LUSC patients (logrank p values < 0.05). Subsequently, we investigated the expression levels of these m6A-related genes in NSCLC (A549, H460) and HBE cells. The results indicated significant expression differences for METTL3, FTO, and IGF2BP3 between the A549 and H460 cells relative to HBE cells (Fig. 4F). Further analysis demonstrated that FTO and IGF2BP3 mRNA did not exhibit a significant change with NFIC overexpression compared to the control group in A549 and H460 cells (Fig. 4G). However, overexpression of NFIC significantly decreased METTL3 expression (Fig. 4H, I), suggesting that NFIC negatively regulates the expression of METTL3 in NSCLC cells.

Fig. 4

NFIC negatively regulates METTL3 expression in A549 and H460 cells. A The total m6A level of A549 and H460 cells after METTL3 overexpression. B Integrative Genomics Viewer tracks displaying chromatin accessibility read distributions in m6A-related genes. C M6A-related genes expression in NSCLC tissues and normal tissues form the TCGA and GSE81089 datasets. D-E Forest map of m6A-related genes on survival analysis in LUAD (D) and LUSC (E). F METTL3, FTO, IGF2BP3, HNRNPC, and HNRNPA2B1 mRNA expression in A549 and H460 cells. G The expression of FTO and IGF2BP3 mRNA with NFIC overexpression in A549 and H460 cells. H-I qRT-PCR (H) and western blot (I) analysis of the expression of METTL3 with NFIC overexpression in A549 and H460 cells. Bar = mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001

NFIC overexpression delayed the progression of NSCLC by downregulating METTL3 expression

Based on the findings obtained, it was observed that METTL3 exhibited a significantly high expression and a strong correlation with NFIC in NSCLC. To further substantiate these results, the expression of METTL3 was assessed using qRT-PCR. As shown in Fig. 5A, METTL3 was upregulated in NSCLC tissues compared with the adjacent noncancerous tissues (Normal). Moreover, the diagnostic potential of METTL3 as a biomarker for NSCLC was evaluated using a ROC curve. The area under the AUC was determined to be 0.9467 (Fig. 5B), indicating its ability to effectively discern between NSCLC and normal tissue with good diagnostic efficiency. Additionally, an investigation into the gene expression correlation between METTL3 and NFIC was carried out, revealing a negative association between the expression of NFIC and METTL3 in NSCLC tissues (r =−0.5453, p value = 0.035) (Fig. 5C).

Fig. 5

Knockdown of METTL3 inhibited the proliferation, migration, and invasion of NSCLC cells. A METTL3 expression levels in tumor tissues were detected by qRT-PCR. B ROC curve analysis of the METTL3 gene. C. METTL3 and NFIC correlation in NSCLC tissues. D-F qRT-PCR (D-E) and western blot (F) analysis of METTL3 knockdown efficiency in A549 and H460 cells. G Proliferation of A549 and H460 cells following METTL3 knockdown was determined using CCK8 assays. H Colony formation assay was performed in A549 and H460 cells after knockdown of METTL3. I-L Transwell migration assays (I-J) and invasion assays (K-L). Bar = mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001, compared to si-NC group; ##P < 0.01, compared to sh-NC group

To investigate the impact of METTL3 on the biological functions of NSCLC cells, a transfection experiment was conducted using siRNA-METTL3, shRNA-METTL3, siRNA-control, and shRNA-control in A549 and H460 cells. After 24 h of transfection, qRT-PCR analysis (Fig. 5D, E) and western blot analysis (Fig. 5F) were performed to assess the expression pattern of METTL3 in A549 and H460 cells. Remarkably, both siRNA-METTL3 and shRNA-METTL3 transfection resulted in a significant reduction in METTL3 expression when compared to the control groups. The proliferation of A549 and H460 cells was found to be significantly inhibited upon suppression of METTL3, as demonstrated by the CCK-8 assay (Fig. 5G). Consistently, colony formation assays revealed a notable decrease in the number of cell colonies in A549 and H460 cells upon knockdown of METTL3 (Fig. 5H). Additionally, as depicted in Fig. 5I–L, silencing METTL3 expression led to a substantial decrease in the migratory and invasive abilities of A549 and H460 cells. These findings suggest that knockdown of METTL3 effectively restrains the progression of NSCLC.

Next, we further investigated whether NFIC regulated the progression of NSCLC by downregulating METTL3 expression. The plasmid for overexpressing METTL3 was transfected into A549 and H460 cells with NFIC overexpression. The results revealed that overexpression of METTL3 reversed the inhibitory effect of NFIC overexpression on NSCLC cell proliferation, colony formation, migration, and invasion (Fig. 6A–E). Furthermore, we used the JASPAR database to predict the binding sites of NFIC on the METTL3 promoter. The analysis indicated a potential binding site of NFIC at the 1553–1569 region upstream of the METTL3 TSS (Fig. 6F). Subsequently, ChIP -qPCR and dual-luciferase reporter assay were performed to further verify the results. The ChIP-qPCR results showed a relative enrichment of NFIC at the METTL3 promoter (Fig. 6G), and dual-luciferase reporter assays showed that overexpression of NFIC decreased the activity of luciferase with wild-type METTL3 but not mutated METTL3 in A549 (Fig. 6H) and H460 (Fig. 6I) cells. These results suggested that the NFIC directly regulated METTL3 expression by binding to the promoter region of METTL3, thereby inhibiting the malignant phenotype of NSCLC cells.

Fig. 6

NFIC delayed NSCLC progression via the regulation of METTL3. A-B CCK8 assays for A549 (A) and H460 (B) cells transfected with overexpression plasmid METTL3 alone or overexpression plasmids of both METTL3 and NFIC (METTL3 + NFIC). C Colony formation assays for A549 and H460 cells. D-E Transwell migration (D) and invasion (E) assays for overexpression plasmid METTL3 alone or overexpression plasmids METTL3 + NFIC transfected A549 and H460 cells. F Prediction results of the binding of NFIC at the site of upstream the TSS of METTL3. G ChIP-qPCR detected NFIC binding to METTL3 promoter region in A549 and H460 cells. H-I Assessment of METTL3 promoter activity after NFIC overexpression in A549 (H) and H460 (I) cells via dual‐luciferase reporter assay. *P < 0.05, **P < 0.01, ***P < 0.001, compared to oe-NC group; #P < 0.05, ##P < 0.01, ###P < 0.001, compared to overexpression METTL3 (oe-METTL3) group

METTL3 positively regulates KAT2A mRNA via m6A modification

METTL3 have been studied to regulate cancer progression by regulating target genes through m6A modification [28]. Therefore, we conducted further analysis to identify the target genes that can be regulated by METTL3 through m6A modification in NSCLC. First, weighted correlation network analysis (WGCNA) analysis was performed using prognosis-related m6A regulators and differentially expressed genes of TCGA-LUAD and TCGA-LUSC. The co-expression modules, generated from the scale-free network, were visualized using dynamic tree cutting (Fig. 7A, B). Subsequently, the 11 and 12 modules marked were identified in LUAD and LUSC, respectively. Among them, the pink and purple module were significantly positively (r > 0.5, p value < 0.01) correlated with the METTL3 in LUAD and LUSC, respectively (Fig. 7C, D). Additionally, we calculated the Pearson’s correlation coefficients between each module and found that each module demonstrated independent validation (Fig. 7E, F). The correlation between the gene significance (GS) and module membership (MM) in the pink (LUAD) and purple (LUSC) module were evaluated. The correlation was significant in the pink (r = 0.84, p value = 3.4e-38) and purple (r = 0.89, p value = 7e-36) module (Fig. 7G, H). We also identified 57 overlapping genes in the pink module (LUAD) and purple module (LUSC), and further identified 5 hub overlapping genes using cytoHubba from Cytoscape (https://cytoscape.org/) (Fig. 7I).

Fig. 7

WGCNA of the m6A-related genes. A-B Hierarchical clustering tree in LUAD (A) and LUSC (B). C-D The correlation between the gene module and prognosis-related m6A regulators in LUAD (C) and LUSC (D). E–F Clustering module hub genes by hierarchical structure and heatmap of the adjacencies in the hub gene network in LUAD (E) and LUSC (F). G-H Scatter plot of the GS for the grade vs. the MM in the pink (G, LUAD) and purple (H, LUSC) module. I The hub genes of protein–protein interaction (PPI) network of overlapping genes

Next, the expression of hub overlapping genes was examined via qRT-PCR in NSCLC cells (A549, H460) and HBE cells. The results showed that among the tested genes, only KAT2A mRNA expression displayed significant differences in A549 and H460 cells (Fig. 8A). Furthermore, western blot analysis revealed that KAT2A was significantly upregulated in A549 and H460 cells compared to HBE cells (Fig. 8B). Similarly, the TCGA and GSE81089 datasets demonstrated higher expression of KAT2A in NSCLC tissue samples when compared to adjacent normal tissues (Fig. 8C). These observations were further validated using qRT-PCR (Fig. 8D) and immunohistochemistry (Fig. 8E) in NSCLC tissues. To assess the diagnostic potential of KAT2A as a biomarker for NSCLC, an ROC curve was employed. As shown in Fig. 8F, the area under the ROC curve (AUC) was 0.9156 (p value < 0.001), suggesting that KAT2A can effectively distinguish between NSCLC and normal tissue with high diagnostic accuracy. Furthermore, the correlation between METTL3 and KAT2A expression in NSCLC tissues was analyzed. The results demonstrated a positive association between the expression of KAT2A and METTL3 in NSCLC tissues (r = 0.5732, p value = 0.0255) (Fig. 8G). Meanwhile, qRT-PCR (Fig. 8H) and western blot (Fig. 8B) analysis showed that the expression of KAT2A in NSCLC cells was lowered by METTL3 silencing, whereas overexpression of METTL3 had the opposite results (Fig. 8I–B).

Fig. 8

METTL3 positively regulates KAT2A mRNA via m6A modification. A qRT-PCR detection of MSH5, ATAT1, ATAD3B, POLG2, and KAT2A mRNA expression. B Western blot results. C The relative expression of KAT2A in NSCLC from the TCGA and GSE81089 dataset. D qRT-PCR validation of KAT2A expression levels on cDNA microarrays. E Representative immunohistochemical staining for KAT2A (scale bar: 200 or 50 μm). F ROC curve analysis of the KAT2A gene. G Correlations between the expressions of METTL3 and KAT2A in NSCLC tissues. H-I qRT-PCR analysis of the expression of KAT2A with METTL3 knockdown (H) and overexpression (I) in A549 and H460 cells. J The mRNA density coverage of differential m6A peaks. K Display of MeRIP-seq read distributions in KAT2A using Integrative Genomics Viewer. L The mRNA density coverage of differential m6A peaks between METTL3-depleted A549 cells and control cells. M–N The total m6A level of A549 and H460 cells after METTL3 knockdown (M) and overexpression (N). O Schematic photo of 3’UTR-WT, 3’UTR-mutant in KAT2A mRNA (left). MeRIP‐qPCR results of KAT2A m6A modification levels in A549 and H460 cells (right). P Dual-luciferase assay result. Bar = mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001

Based on the published MeRIP-seq data (GSE198288) of LUAD samples and tumor-adjacent normal samples, we observed the highest number of differential m6A peaks between LUAD samples and tumor-adjacent normal samples in the stop codon and 3’UTR (Fig. 8J), and the m6A occupancy of KAT2A was higher in LUAD tissues (Fig. 8K). Moreover, we re-analyzed the MeRIP-seq data (GSE55572) from METTL3-depleted A549 cells. The results show a decreased m6A occupancy of KAT2A from METTL3-depleted A549 cells compared to control (Fig. 8K), and the largest number of differential m6A peaks between METTL3-depleted A549 cells and control were found in the stop codon and 3’UTR (Fig. 8L). To explore this relationship further, we assessed the m6A levels of total RNAs in A549 and H460 cells. METTL3 knockdown resulted in a downregulation of global m6A modification level in both A549 and H460 cells compared to the control (Fig. 8M), while METTL3 overexpression increased global m6A modification level (Fig. 8N). Additionally, we used the SRAMP online tool [29] to predict potential m6A sites in KAT2A. The analysis revealed two significant m6A sites in the 3’UTR of KAT2A mRNA (Fig. 8O). To validate these predictions, we conducted MeRIP-qPCR and dual-luciferase reporter assays. MeRIP-qPCR showed that METTL3 knockdown significantly reduced m6A levels of fragments associated with the predicted site (Fig. 8O), and the dual-luciferase reporter assays confirmed that METTL3 knockdown decreased the activity of luciferase with wild-type KAT2A, but not with mutated KAT2A (Fig. 8P). These results imply that METTL3 regulated KAT2A expression by methylating the m6A site in 3’UTR of KAT2A mRNA in NSCLC cells.

METTL3-mediated m6A mRNA modification of KAT2A mRNA promotes NSCLC progression

Considering the METTL3 positively regulates KAT2A mRNA via m6A modification. Thus, we further investigated whether METTL3 promotes the progression of NSCLC via mediating KAT2A mRNA m6A modification. First, we investigated the effects of KAT2A on the progression of NSCLC. A549 and H460 cells were transfected with si-NC, si-KAT2A, sh-NC, or sh-KAT2A. qRT-PCR and western blot analyses revealed that transfection with siRNA-KAT2A or shRNA-KAT2A significantly decreased KAT2A expression compared to controls (Additional file 3: Fig. S1A-S1C). Cell proliferation, colony formation assays, cell migration, and invasion assays, revealed that KAT2A knockdown inhibited the malignant phenotype of A549 and H460 cells in vitro (Additional file 3: Fig. S1D-S1K). Next, rescue experiments show that KAT2A knockdown largely suppressed the promoting effect of overexpression of METTL3 on the malignant phenotype of A549 and H460 cells (Fig. 9A–G). In addition, in vivo experiments demonstrated that the overexpression of METTL3 significantly increased tumor size and weight compared to control groups, but this effect was inhibited by KAT2A knockdown (Fig. 9H–I). IHC staining revealed that overexpression of METTL3 increased the expression of KAT2A and ki67, which was counteracted by knockdown of KAT2A (Fig. 9J–K). Additionally, western blot analysis demonstrated that the METTL3 overexpression enhanced the phosphorylation of PI3K and AKT, which was recovered by KAT2A knockdown (Fig. 9L), suggesting that KAT2A knockdown impairs the activation of PI3K/AKT signaling pathway induced by overexpression of METTL3.

Fig. 9

METTL3-mediated m6A mRNA modification of KAT2A mRNA promotes NSCLC progression. A-C CCK8 assays for A549 (A) and H460 (B) cells transfected with overexpression plasmid METTL3 alone or both oe-METTL3 and sh-NFIC. D Cell colony formation assays. E Scratch assay experiments on A549 and H460 cells. F-G Transwell migration (F) and invasion (G) assays. H-I The tumor volume and weight of tumors xenografted in nude mice. J-K IHC staining results and IHC score (H-score). L Western blot results. Bar = mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001, compared to oe-NC + sh-NC group; #P < 0.05, ##P < 0.01, ###P < 0.001, compared to overexpression METTL3 (oe-METTL3 + sh-NC) group

NFIC negatively regulates the expression of KAT2A by directly binding to the KAT2A promotor region

Based on the above data, NFIC modulates NSCLC progression by indirect regulation of KAT2A mRNA through METTL3-mediated m6A modification. Interestingly, significant differences in the chromatin features of KAT2A were identified between NSCLC and normal tissues (Fig. 1A). Therefore, we investigated whether NFIC binds to the KAT2A promoter region and directly regulates its transcription. The JASPAR online tool identified four potential NFIC binding sites at positions 101–1137 upstream of the KAT2A transcription start site (Fig. 10A). In order to validate these predicted binding sites, specific primers were designed to amplify the four sites, which were combined into a region of less than 200 base pairs, and ChIP-qPCR was performed to confirm the results. The ChIP-qPCR results showed a relative enrichment of NFIC at the METTL3 promoter (2#) (Fig. 10B). In order to further confirm the binding sites, the potential NFIC binding sites in the KAT2A promoter were mutated as shown in Fig. 10A. Dual-luciferase reporter assays demonstrated that overexpression of NFIC led to a decrease in luciferase activity with the wild-type KAT2A, but not with the mutated KAT2A, in A549 (Fig. 10C) and H460 cells (Fig. 10D). Additionally, the expression of KAT2A in the NFIC overexpression group were significantly decreased compared to the control group in A549 and H460 cells (Fig. 10E-F). These results suggested that the NFIC directly regulated KAT2A expression via binding to the promoter region of KAT2A in NSCLC cells.

Fig. 10

NFIC negatively regulates the expression of KAT2A by directly binding to KAT2A promotor region. A Prediction results of the binding of NFIC at the site of upstream the TSS of KAT2A. B ChIP-qPCR detected NFIC binding to KAT2A promoter region. C-D Assessment of KAT2A promoter activity after NFIC overexpression in A549 (C) and H460 (D) cells via dual‐luciferase reporter assay. E–F qRT‒PCR (E) and western blot (F) measurement of KAT2A expression in A549 and H460 cells. G An illustration of the molecular and functional mechanisms of NFIC in NSCLC cells. Bar = mean ± SD. **P < 0.01, ***P < 0.001