m6A demethylase FTO suppresses pancreatic cancer tumorigenesis by demethylating PJA2 and inhibiting Wnt signaling.

Pancreatic cancer is the deadliest malignancy of the digestive system and is the seventh most common cause of cancer-related deaths worldwide. The incidence and mortality of pancreatic cancer continue to increase, and its 5-year survival rate remains the lowest among all cancers. N6-methyladenine (m6A) is the most abundant reversible RNA modification in various eukaryotic messenger and long noncoding RNAs and plays crucial roles in the occurrence and development of cancers. However, the role of m6A in pancreatic cancer remains unclear. The present study aimed to explore the role of m6A and its regulators in pancreatic cancer and assess its underlying molecular mechanism associated with pancreatic cancer cell proliferation, invasion, and metastasis. Reduced expression of the m6A demethylase, fat mass and obesity-associated protein (FTO), was responsible for the high levels of m6A RNA modification in pancreatic cancer. Moreover, FTO demethylated the m6A modification of praja ring finger ubiquitin ligase 2 (PJA2), thereby reducing its mRNA decay, suppressing Wnt signaling, and ultimately restraining the proliferation, invasion, and metastasis of pancreatic cancer cells. Altogether, this study describes new, potential molecular therapeutic targets for pancreatic cancer that could pave the way to improve patient outcome.

Introduction

Pancreatic cancer, as the deadliest malignancy of the digestive system, is the fourth leading cause of cancer-related deaths in the United States and the seventh most common cause of cancer-related deaths worldwide., Clinical advances and lifestyle changes have contributed toward a gradual decrease in the incidence and mortality of many tumors in recent decades. In contrast, the incidence and death rate of pancreatic cancer have been continuously increasing. With a 3- to 6-month median survival time, pancreatic cancer has the lowest 5-year survival rate among all cancers (9% in the United States from 2009 to 2015, <5% worldwide). Due to the location of the pancreas, pancreatic cancer is often asymptomatic before metastasis. Therefore, more than 80% of patients have late-stage cancer when first diagnosed, which deprives them of the most effective treatment (surgical resection). Moreover, 85% of patients experience recurrence even after resection. Therefore, a better understanding of the molecular mechanisms that contribute toward the occurrence and development of pancreatic cancer is urgently needed.

N6-methyladenine (m6A) RNA modification, as the most important modification of eukaryotic messenger and long noncoding RNAs, was first discovered in the purified poly(A) RNA fraction of Novikoff hepatoma cells, being found in abundance near stop codons, splicing sites, and the 3′ untranslated region., m6A is dynamically regulated by m6A methyltransferases, demethylases, and specific recognition proteins, namely “writers,” “erasers,” and “readers.” m6A writers, such as methyltransferase-like (METTL)3, METTL14, Wilms tumor 1-associated protein (WTAP), RNA-binding protein 15/15B (RBM15/15B), and vir-like m6A methyltransferase-associated protein (VIRMA), catalyze m6A formation. m6A erasers, including fat mass and obesity-associated protein (FTO) and its homolog AlkB family member 5 (ALKBH5), selectively remove the methylated modification from the target RNAs. m6A readers, such as YT521-B homology (YTH) domain-containing protein,, insulin-like growth factor 2 mRNA-binding protein 1/2/3, eukaryotic initiation factor 3, and heterogeneous nuclear ribonucleoproteins A2/B1, can specifically recognize m6A modification and function in the regulation of almost every step of the RNA life cycle, including mRNA transcription, splicing, nuclear export, localization, translation, stability, and degradation. There are two main mechanisms by which m6A RNA modification participates in epigenetic regulation. After m6A readers binding, m6A modification alters the post-transcriptional regulation of m6A-containing RNA. For example, the m6A-modified structure of mRNA combined with the m6A reader YTH domain family (YTHDF)1 promotes the efficient translation of the corresponding mRNA. In contrast, another m6A reader, YTHDF2, enhances the decay of the m6A-modified RNA. Furthermore, the m6A modification can change the conformation of the corresponding RNA. Studies have shown that in RNA unpaired base sites, RNAs with m6A modification have stronger stacking ability than those without m6A modification, which promotes the stability of the surrounding RNA structure or the folding of adjacent RNAs., The local structural changes of RNA caused by the m6A modification can affect the binding of RNA-binding protein to the corresponding RNA to form an “m6A switch,” thereby altering their selective splicing and expression of the corresponding transcript.

Recently, several studies have provided abundant evidence that m6A RNA modification plays vital roles in cancer proliferation, migration, metastasis, and resistance to chemotherapy and radiotherapy.27, 28, 29, 30 For example, m6A modification expression is downregulated in endometrial cancer, hepatocellular carcinoma, and other cancers,, whereas it is upregulated in gastric cancer. Lately, a systematic analysis of the genetic alterations and clinical relevance of m6A regulators across 33 cancers showed that m6A regulators have widespread genetic alterations, and the levels of these regulators are significantly associated with the activity of cancer hallmark-related pathways, ultimately impacting on the overall patient survival time. In addition, it has been shown that m6A is related to drug response and may be the epigenetic driving force of chemoresistance in leukemic cells, in which after treatment with tyrosine kinase inhibitors (TKIs), m6A hypomethylation response and FTO levels are increased to achieve resistance to TKI. Moreover, studies have shown that m6A modification is involved in tumor immunotherapy. Increased expression of FTO in melanoma promotes tumor growth by reducing m6A levels of programmed cell death protein 1 (PD-1), C-X-C chemokine receptor type 4 (CXCR4), and SRY-box transcription factor (TF) 10 (SOX10) and inhibiting YTHDF2-mediated degradation of these RNAs. In addition, FTO knockout in melanoma cells promotes an anti-PD-1 response in mice. This study aimed to clarify the function and regulation of m6A RNA modification in pancreatic cancer, as well as its underlying molecular mechanisms, contributing toward pancreatic cancer cell proliferation, invasion, and metastasis.

Results

m6A RNA modification is upregulated in pancreatic cancer

To explore the potential role of m6A RNA modification in pancreatic cancer, m6A differences between pancreatic cancer tissues and paired adjacent normal tissues were evaluated. First, the m6A levels in the total RNA of pancreatic adenocarcinoma (PAAD) and corresponding para-tumor tissues were measured. After detection using the EpiQuik m6A RNA Methylation Quantification Kit, the m6A levels were found to be significantly upregulated in PAAD tissues compared with their corresponding para-tumor tissues (Figures 1A and 1B). Moreover, m6A levels of total RNA in human pancreatic cancer cell lines (SW1990, AsPC-1, PANC-1, BxPC-3, and Capan-2) were found to be significantly higher than those in the total RNA of the pancreatic duct epithelial cell line HPDE6-C7. m6A levels in pancreatic cancer cell lines in a descending order were BxPC-3, PANC-1, AsPC-1, Capan-2, and SW1990 (Figure 1C). Taken together, these data indicate that m6A RNA modification is upregulated in pancreatic cancer.

Figure 1

Relative N6-methyladenine (m6A) RNA modification in total RNA is upregulated in pancreatic cancer (PC)

(A and B) m6A content in total RNA in 16 pairs of pancreatic adenocarcinoma (PAAD) and corresponding para-tumor tissues. Except for patients 3 and 8, the relative m6A RNA modification of PAAD tissues was significantly higher than that of the corresponding para-tumor tissues. (C) Relative m6A RNA modification was higher in PC cell lines. Human PC cell lines (SW1990, AsPC-1, PANC-1, BxPC-3, and Capan-2) and a human immortalized pancreatic duct epithelial cell line (HPDE6-C7) were subjected to RNA m6A detection. Tumor, PAAD tissue. (B) ∗p < 0.05 by paired Student’s t test; (C) ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001 by unpaired Student’s t test. Data are represented as mean ± standard error.

Reduced FTO expression is responsible for high m6A RNA modification levels in pancreatic cancer

It is widely known that the m6A RNA modification is mainly regulated by m6A methyltransferases (writers) and demethylases (erasers) through a dynamic and reversible process. Therefore, the disturbance of m6A RNA modification in pancreatic cancer might be caused by the abnormal expression of such m6A writers and erasers. To verify this assumption, quantitative real-time polymerase chain reaction (PCR) was performed to assess the expression of eight m6A writers: METTL3, METTL14, WTAP, RBMX, RBM15, RBM15B, VIRMA, and ZC3H13, and two m6A erasers: FTO and ALKBH5, on 16 pairs of fresh frozen PAAD and corresponding para-tumor tissues from patients who underwent surgical tumor resection at Shengjing Hospital of China Medical University. Only FTO and ALKBH5 levels were found to be significantly different between the paired samples, and the difference of FTO expression was more obvious (Figures 2A and 2B; for FTO: p < 0.0001; for ALKBH5: p = 0.0211). Further analysis of the FTO and ALKBH5 expression patterns based on publicly available The Cancer Genome Atlas (TCGA) data revealed that their expression was significantly positively correlated in PAAD tissues (Spearman coefficient: 0.30, p = 5.47e−5; Pearson coefficient: 0.32, p = 1.51e−5; Figure 2C). Previous studies have shown that ALKBH5 is downregulated in pancreatic cancer and acts as a tumor suppressor by regulating multiple targets.36, 37, 38 In addition, Kaplan-Meier survival analysis and log-rank test were performed to evaluate the impact of FTO on the survival of 177 PAAD patients. Patients with high FTO expression (N = 106) were found to have significantly longer overall survival (OS) time than low FTO-expressing patients (N = 71) (hazard ratio [HR] = 0.65; log-rank test p = 0.039; Figure 2D). The impact of FTO expression on the OS of PAAD patients at different cancer stages was also assessed (Figures S1A and S1B). As sample number in stages III and IV was very low for meaningful analysis, only stages I and II cases were evaluated. PAAD patients in stage I with high FTO expression (N = 14) showed significantly longer OS time than low FTO-expressing patients (N = 7), whereas no significant difference in stage II patients was observed. Hence, based on these findings and considering the catalytic ability of FTO and ALKBH5 and their co-expression relationship in pancreatic cancer, FTO was selected for further analysis as the candidate molecule for m6A modification disorder in pancreatic cancer.

Figure 2

FTO is downregulated in PC

(A) mRNA levels of 10 main m6A regulators (METTL3, METTL14, WTAP, RBMX, RBM15, RBM15B, VIRMA, ZC3H13, FTO, and ALKBH5) were evaluated in 16 pairs of fresh frozen PAAD and corresponding para-tumor tissues. (B and E) FTO mRNA and protein expression in 16 pairs of fresh frozen PAAD and corresponding para-tumor tissues. (C) cBioPortal analysis showed the co-expression relationship between FTO and ALKBH5 in PAAD. Spearman coefficient: 0.30 (p = 5.466e−5); Pearson coefficient: 0.32 (p = 1.507e−5). (D) Kaplan-Meier plot and log-rank test on the survival of 177 patients with PAAD from in The Cancer Genome Atlas (TCGA). Low FTO expression was associated with poorer overall survival in PAAD patients (hazard ratio [HR] = 0.65; log-rank test p = 0.039). GAPDH was used as internal control in western blotting (WB) and mRNA expression analyses. Para-tumor or P, corresponding para-tumor tissue; T, PAAD tissue. (A) ∗p < 0.05; ∗∗∗p < 0.001 by paired Student’s t test.

Western blotting (WB) analysis of FTO levels in 16 pairs of fresh frozen PAAD and corresponding para-tumor tissues led to results similar to those of quantitative real-time PCR. Except for patients 3 and 8, FTO levels in PAAD tissues were significantly reduced compared with the corresponding para-tumor tissues (Figure 2E). Furthermore, the relative mRNA and protein levels of FTO in five human pancreatic cancer cell lines were significantly lower than those in HPDE6-C7 cells (Figures 3A and 3B). Knockdown of FTO in SW1990 cells using small interfering (si)RNAs, including siFTO-1, siFTO-2, and siFTO-3, further showed that FTO depletion markedly increased m6A levels in the total RNA. Moreover, transfection of BxPC-3 cells with overexpressing plasmids further showed that overexpressing wild-type FTO (FTO-WT) significantly decreased m6A levels in the total RNA. However, almost no significant difference was observed in cells overexpressing mutant FTO (FTO-mut) plasmids (H231A/D233A point mutations, which led to complete loss of m6A demethylation activity of FTO) (Figures 3C–3H). Taken together, these results demonstrate that reduced expression of FTO is responsible for the high levels of m6A RNA modification in pancreatic cancer.

Figure 3

FTO acts as m6A demethylase in pancreatic cell lines

(A and B) Quantitative real-time PCR and WB assays showed reduced FTO levels in five PC cell lines. The mRNA and protein levels of FTO were (C and E) significantly reduced in SW1990 cells after small interfering (si)RNAs (siFTO-1, siFTO-2, and siFTO-3) transfection, whereas were (D and F) significantly increased in BxPC-3 cells after transfection with wild-type FTO (FTO-WT) and mutant FTO (FTO-mut) overexpressing plasmids. (G) FTO knockdown significantly upregulated the m6A level in total RNA of SW1990 cells. (H) FTO-WT overexpression significantly decreased the m6A level in total RNA of BXPC-3 cells. No significant changes were observed in the m6A levels upon FTO-mut overexpression. GAPDH was used as internal control in the quantitative real-time PCR and WB assays. NC, negative control. (A–H) ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001; ns, no statistical significance by unpaired Student’s t test.

Reduced expression of FTO is associated with poor survival of PAAD patients

To explore the clinicopathological role of FTO in pancreatic cancer, immunohistochemistry (IHC) examination of paraffin specimens of 50 pairs of PAAD tissues and corresponding para-tumor tissues from the Shengjing Hospital of China Medical University was performed. FTO was found to be mainly expressed in the nucleus, and its expression in PAAD tissues of 44 patients (88%) was lower than that in the corresponding para-tumor tissues (p < 0.0001; Figures 4A and 4B). With the use of the average expression of FTO in PAAD tissues as a threshold value, the patients were divided into high (N = 19) and low (N = 31) expression groups. The results showed that FTO expression in PAAD tissues was not significantly correlated with age, gender, pancreatitis status, diabetes status, smoking habits, drinking habits, preoperative obstructive jaundice, CA199 level, tumor size or location, perineural invasion, or tumor grade, whereas it was closely correlated with individual cancer stages and nodal metastasis status (Table 1). Next, the patients were divided into four groups according to the American Joint Committee on Cancer Clinical Staging Manual (8th Edition): stages I (N = 22), II (N = 21), III (N = 5), and IV (N = 2), showing that the later the cancer stage, the lower the FTO expression (Figure 4C). Further analysis based on the lymph nodal metastasis status of the patients showed that the FTO levels of the N0 group (N = 35) were higher than those of the N1 group (N = 9), which was in turn higher than those of the N2 group (N = 6), indicating that FTO was significantly downregulated in PAAD tissues with lymph node metastasis (Figure 4D). Log-rank test analysis also confirmed that the median survival time of patients in the FTO high expression group was significantly higher than that of the FTO low expression group (48.67 versus 16.90 months, p = 0.0474; Figure 4E). Altogether, these results show that PAAD patients with reduced FTO expression have poor survival outcome.

Figure 4

Decreased FTO expression in PC is related with clinicopathological characteristics and prognosis of patients

(A) Immunohistochemistry (IHC) staining of 50 pairs of formalin-fixed paraffin-embedded (FFPE) samples showed abundant FTO-positive cells in para-tumor tissues but fewer FTO-positive cells in PAAD tissues. (B) FTO IHC staining quantification. The staining was scored according to the following scale: from 0 (no staining) to 3 (high staining). FTO expression in PAAD was based on (C) individual patient cancer stages and (D) nodal metastasis status. (E) The Kaplan-Meier survival curve and log-rank test were used to assess the relationship between FTO expression and overall survival in 50 PAAD patients. Red and blue represent the FTO high and low expression groups, respectively. (B) ∗∗∗p < 0.001 by paired Student’s t test. (C and D) ∗p < 0.05; ∗∗∗p < 0.001 by unpaired Student’s t test.

Table 1

Relationship between FTO expression and clinicopathological parameters in 50 patients with pancreatic adenocarcinoma

Characteristic	Number of patients	Expression of FTO		p valuea
		Low (n = 31)	High (n = 19)
Age
<60	19	12	7	0.9439
≥60	31	19	12
Gender
Male	27	15	12	0.3039
Female	23	16	7
Pancreatitis status
With pancreatitis	7	5	2	0.9301
Without pancreatitis	43	26	17
Diabetes status
With diabetes	15	8	7	0.509
Without diabetes	35	23	12
Smoking habits
Smoker	21	12	9	0.3953
Non-smoker	29	19	10
Drinking habits
Drinker	11	9	2	0.1854
Non-drinker	39	22	17
Obstructive jaundice
With obstructive jaundice	19	11	8	0.4948
Without obstructive jaundice	31	20	11
CA199 (U/mL)
<37	9	6	3	0.3861
37–1,000	36	23	13
>1,000	5	2	3
Tumor location
Head	29	18	11	0.6717
Body/tail	21	13	8
Tumor size
<2 cm	4	1	3	0.1685
2–4 cm	29	17	12
>4 cm	17	13	4
Perineural invasion
With perineural invasion	22	14	8	0.491
Without perineural invasion	28	17	11
Tumor grade
1	15	9	6	0.804
2	22	15	7
3	13	7	6
Individual cancer stages
I	22	7	15	0.0004
II	21	18	3
III	5	4	1
IV	2	2	0
Nodal metastasis status
N0	35	16	19	<0.0001
N1	9	9	0
N2	6	6	0

p < 0.05 indicates a significant relationship among the variables.

FTO suppresses pancreatic cancer malignancy

To identify the function of FTO in tumor proliferation and metastatic capacity, siRNAs were used to knock down FTO in BxPC-3 and SW1990 cells. FTO knockdown was confirmed by quantitative real-time PCR and WB. Incucyte Zoom Live Cell Imaging System and clonogenic cell survival assays were used to analyze cell proliferation. Knockdown of FTO was found to markedly induce proliferation of both BxPC-3 and SW1990 cells. Similar results were observed with the Cell Counting Kit-8 (CCK8) assays (Figures 5A and 5B; Figures S2A and S2B). Wound healing migration and Transwell invasion assays further revealed that FTO knockdown significantly increased the migration and invasion ability of BxPC-3 and SW1990 cells (Figures 5C and 5D; Figures S2C and S2D). Thus, lack of FTO is associated with enhanced proliferation, metastasis, and invasion of pancreatic cancer cells.

Figure 5

Decreased FTO expression promotes proliferation, migration, and invasion of PC cell BxPC-3 in vitro

(A and B) Incucyte Zoom Live Cell Imaging System and clonogenic cell survival assays were used to detect the proliferation of BxPC-3 cells after siRNA knockdown (siFTO-1, siFTO-2, and siFTO-3) or overexpression of FTO. Simultaneously, Cell Counting Kit-8 (CCK8) assay was performed to assess the proliferation potential of these cells at 48 h. (C) Wound healing migration assays revealed the effect of FTO on the migration ability of BxPC-3 cells. The gap was measured at 0, 24, 48, 72, and 96 h after the scratch was performed. (D) Transwell invasion assays revealed the effect of FTO on the invasion ability of BxPC-3 cells. ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001.

FTO-WT and FTO-mut plasmids were used to overexpress FTO in BxPC-3 and SW1990 cells. FTO overexpression was confirmed by quantitative real-time PCR and WB. Incucyte Zoom Live Cell Imaging results and clonogenic cell survival assays showed that FTO-WT overexpression significantly suppressed cell proliferation of BxPC-3 and SW1990 cells. Similar results were observed in the CCK8 assays (Figures 5A and 5B; Figures S2A and S2B). Moreover, FTO-WT cells showed an obvious delay in migration in the scratch wound healing assays (Figure 5C; Figure S2C) and significantly suppressed invasiveness in the Transwell assays (Figure 5D; Figure S2D). Almost no statistically significant difference between FTO-mut and negative control (NC) groups was observed (Figure 5; Figure S2). Collectively, these findings demonstrate that overexpression of FTO-WT inhibits the proliferation, metastatic. and invasion potential of pancreatic cancer cells.

FTO enhances praja ring finger ubiquitin ligase 2 (PJA2) mRNA stability in an m6A-YTHDF2-dependent manner

To investigate the molecular mechanism of FTO and identify its downstream targets in pancreatic cancer, LinkedOmics, UALCAN, and cBioPortal databases were used to explore FTO co-expressing genes in PAAD tissues. In each analysis, the top 20 genes significantly positively or negatively correlated with FTO in PAAD tissues were selected, and their intersections were analyzed separately. Overall, five (AKT3, PJA2, PLEKHM3, ZYG11B, and WDR47) and three (NUDT8, YDJC, and RASSF7) genes were found to be positively and negatively related to FTO, respectively (Figures 6A and 6B). The statistical analysis and p values used in the three analyses are detailed in Tables S1–S3. mRNA expression analysis of these eight candidate genes in BxPC-3 and SW1990 cells lacking or overexpressing FTO was then performed. Overall, knockdown of FTO significantly decreased PJA2 expression, whereas FTO-WT overexpression significantly increased PJA2 levels, with almost no statistically significant difference between FTO-mut and NC groups. Moreover, WDR47 expression in the siFTO-3-transfected BxPC-3 cell was significantly higher than that in the NC group (p = 0.0037), similar to what was observed in SW1990 cells transfected with siFTO-2 (p = 0.0318). In addition, PLEKHM3 expression was significantly lower in siFTO-3 SW1990 cells than that in the NC group (p = 0.0026). The differences between other groups were not statistically significant (Figures 6C and 6D). These findings were also confirmed by WB (Figures 6E and 6F), further showing that, in pancreatic cancer cells, the expression of PJA2 follows the same trend as FTO. Next, the co-expression relationship between FTO and PJA2 in pancreatic cancer was investigated through the cBioPortal, which revealed that the expression of these genes was highly correlated in pancreatic cancer (Spearman coefficient: 0.74, p = 2.10e−32; Pearson coefficient: 0.75, p = 7.98e−34; Figure 6G). Analysis of the relative PJA2 m6A levels in BxPC-3 and SW1990 cells lacking or overexpressing FTO further showed that knockdown of FTO markedly increased the relative PJA2 m6A level, whereas overexpression of FTO-WT had the opposite effect. However, there was almost no statistically significant difference between the FTO-mut and NC groups (Figures 6H and 6I). Therefore, PJA2 was determined to be a downstream target of FTO in pancreatic cancer.

Figure 6

FTO enhanced PJA2 stability in an m6A-YTHDF2-dependent manner in PC cells

(A and B) Venn diagrams illustrated overlap in differentially bioinformatics analyses of the top 20 genes significantly positively or negatively related to FTO in PC. (C–F) Quantitative real-time PCR was performed to detect the mRNA expression of eight candidate genes after knockdown and overexpression of FTO in BxPC-3 and SW1990 cells. Only the PJA2 expression pattern, by both quantitative real-time PCR and WB, was consistent with that of FTO. (G) cBioPortal analysis displayed the co-expression relationship between FTO and PJA2 in PAAD tissues. Spearman coefficient: 0.74 (p = 2.10e−32); Pearson coefficient: 0.75 (p = 7.98e−34). (H and I) The relative PJA2 m6A level was detected in BxPC-3 and SW1990 cells after knockdown and overexpression of FTO, respectively. (J–M) Quantitative real-time PCR and WB were performed to detect the mRNA and protein expression of PJA2 after YTHDF1 and YTHDF2 knockdown by siRNA in BxPC-3 and SW1990 cells. (N and O) PJA2 stability was detected after knocking down YTHDF2, FTO by siRNA, and overexpression of FTO-WT and FTO-mut in BxPC-3 and SW1990 cells. ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001.

Next, the mechanism by which m6A RNA modification could regulate PJA2 expression in pancreatic cancer was evaluated. YTHDF1 and YTHDF2 are known to be the main m6A readers, playing important roles in the metabolic process of RNA by promoting the translation or decay, respectively, of m6A methylated transcripts., Given the aforementioned results showing that m6A RNA modification inhibits PJA2 expression in pancreatic cancer cells, it was reasonable to believe that PJA2 transcripts could be targets of YTHDF2. Consistent with this hypothesis, knockdown of YTHDF2 by siRNA in BxPC-3 and SW1990 cells increased the mRNA and protein expression of PJA2, whereas YTHDF1 knockdown had no significant effect (Figures 6J–6M). Furthermore, YTHDF2 knockdown significantly reduced the PJA2 decay rate (Figures 6N and 6O). Taken together, reduced FTO expression can upregulate the relative PJA2 m6A levels, which will in turn induce the degradation of PJA2 mediated by YTHDF2, thereby suppressing PJA2 mRNA and protein expression in pancreatic cancer cells.

PJA2 acts as a tumor suppressor in pancreatic cancer

PJA2 is an E3 ubiquitin ligase containing a RING domain that is widely expressed in most mammalian tissues. Kanomata et al. showed that high PJA2 expression is positively related to the prognosis of breast cancer patients. Furthermore, Cantara et al. reported that PJA2 is significantly overexpressed in differentiated thyroid cancer and that its expression is negatively correlated with the malignant phenotype of thyroid cancer. However, the role of PJA2 in pancreatic cancer remains unclear. Thus, Kaplan-Meier analysis and log-rank test were performed based on TCGA data to analyze the survival outcome of 177 PAAD patients according to PJA2 expression. The median survival time of 70 and 107 patients with high and low PJA2 expression, respectively, was not significantly different (20.47 versus 19.87 months). Nonetheless, patients with high PJA2 expression tended to have longer survival times (HR = 0.68, p = 0.081; Figure 7A). The impact of PJA2 expression on the OS of PAAD patients at different cancer stages has also been explored (Figures S1C and S1D). Moreover, the PJA2 mRNA and protein levels in SW1990, AsPC-1, PANC-1, BxPC-3, Capan-2, and HPDE6-C7 cell lines were found to be similar to those of FTO. That is to say, PJA2 levels in pancreatic cancer cell lines were also significantly lower than those in the non-cancerous HPDE6-C7 cells (Figures 7B and 7C). Knockdown of PJA2 in BxPC-3 and SW1990 cells using siRNAs (siPJA2-429, siPJA2-122, and siPJA2-100) was confirmed by quantitative real-time PCR and WB (Figure 7D; Figure S3A). Incucyte Zoom Live Cell Imaging System, clonogenic cell survival, and CCK8 assays demonstrated that reduced PJA2 expression significantly promoted proliferation of both BxPC-3 and SW1990 cells (Figures 7E and 7F; Figures S3B and S3C). Wound healing migration and Transwell invasion assays further revealed that PJA2 knockdown significantly increased the migration and invasion ability of pancreatic cancer cells (Figures 7G and 7H; Figures S3D and S3E). Taken together, knockdown of PJA2 promotes proliferation and enhances the metastatic and invasion potential of pancreatic cancer cells.

Figure 7

PJA2 acts as a tumor suppressor in PC

(A) A Kaplan-Meier plot and log-rank test were used to analyze the survival data of 177 patients with PAAD from TCGA. Red and black represent the PJA2 high (N = 70) and low expression groups (N = 107), respectively (HR = 0.68; log-rank test p = 0.081). (B and C) Quantitative real-time PCR and WB assays showed the decreased PJA2 levels in five PC cell lines. (D) WB and quantitative real-time PCR confirmed that PJA2 expression was significantly reduced after siRNAs (siPJA2-429, siPJA2-122, and siPJA2-100) transfection in BxPC-3 cells. (E–H) Decreased PJA2 expression promoted the proliferation, migration, and invasion of BxPC-3 cells in vitro. (E and F) Incucyte Zoom Live Cell Imaging and clonogenic cell survival assays showed the proliferation of BxPC-3 cells after PJA2 knockdown. Simultaneously, the CCK8 kit was used to assess the proliferation potential of BxPC-3 cells after PJA2 knockdown at 48 h. (G) Wound healing migration assays revealed the effect of PJA2 on the migration ability of BxPC-3 cells. The gap was measured at 0, 24, 48, 72, and 96 h after the scratches were performed. (H) Transwell invasion assays revealed the effect of PJA2 on the invasion ability of BxPC-3 cells. ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001.

PJA2 knockdown significantly alleviates the FTO inhibitory effect on pancreatic cancer growth and metastasis

To investigate whether PJA2 mediated the inhibitory effect of FTO on pancreatic cancer growth and metastasis, BxPC-3 cells stably overexpressing FTO-WT were co-transfected with siPJA2-429 to knock down PJA2 expression. Overall, suppression of PJA2 significantly alleviated the inhibitory effects mediated by FTO on cell proliferation (Figures 8A and 8B), as well as on cell migration and invasion (Figures 8C and 8D). These findings suggest that FTO-dependent inhibitory effects on pancreatic cancer growth and metastasis are mediated by PJA2.

Figure 8

FTO-dependent demethylation of m6A modification regulates the Wnt signaling pathway through PJA2 in PC

(A and B) Incucyte Zoom Live Cell Imaging and clonogenic cell survival assays showed the cell proliferation of BxPC-3 cells in different conditions. Simultaneously, the CCK8 assay was used to assess the proliferation potential of BxPC-3 cells in different groups at 48 h. (C and D) Wound healing migration and Transwell invasion assays displayed the migration and invasion ability of BxPC-3 cells in different groups. (E) Quantitative real-time PCR and (F) WB were performed to examine the expression of β-catenin (CTNNB1), Wnt5a (WNT5A), LEF1, GSK-3β (GSK3B), AXIN1, and WIF1 in FTO-WT, FTO-WT + siCtrl (transfection of NC siRNA [si-control]), and FTO-WT + siPJA2-429 cells. vector, empty vector. ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001.

FTO-dependent demethylation of m6A modification in pancreatic cancer regulates Wnt signaling via PJA2

A previous study reported that PJA2 inhibits the activity of the Wnt signaling pathway in mouse E14 embryonic stem cells. Thus, the interaction between PJA2 and the Wnt pathway was evaluated in pancreatic cancer cells. The mRNA expression of key proteins of the Wnt pathway, including β-catenin (CTNNB1), Wnt5a (WNT5A), LEF1, GSK-3β (GSK3B), AXIN1, and WIF1 was assessed in FTO-WT, FTO-WT + siCtrl (transfection of NC siRNA [si-control]), and FTO-WT + siPJA2-429 groups. In BxPC-3 cells, overexpression of FTO-WT significantly decreased the expression of CTNNB1, WNT5A, and LEF1, whereas increased AXIN1 and WIF1. However, there was almost no statistically significant difference in the expression of GSK3B. Moreover, PJA2 knockdown significantly alleviated these FTO-mediated effects (Figure 8E). These findings were further confirmed by WB, with exception of GSK-3β, which was found to be more phosphorylated (Ser9 is the phosphorylation site; pS9-GSK-3β) upon FTO overexpression, an effect that was reversed by PJA2 knockdown (Figure 8F). Thus, FTO acts as a tumor suppressor in pancreatic cancer via PJA2-mediated inhibition of the Wnt signaling pathway.

Discussion

m6A is the most abundant internal modification of mammalian mRNAs and is involved in almost every step of the RNA life cycle. Abnormal m6A patterns may contribute to RNA dysfunction and consequent deregulated gene expression, physiological dysfunction, and even cancer. Studies have shown that the expression of m6A modification is often associated with different expression disorders in different cancers,,, suggesting that m6A modification plays various roles among cancers. The main reasons for this are the following: first, humans have more than 7,600 mRNAs and 300 noncoding RNAs harboring the m6A modification, which means that the target gene with the m6A modification may be either a proto-oncogene or a tumor suppressor gene. Therefore, when the m6A levels change, it may alter the expression of proto-oncogenes as well as that of tumor suppressor genes. Second, m6A methyltransferases and demethylases, which dynamically regulate the levels of m6A modification, are differently abnormally expressed among different cancers. The overall activity of these two types of enzymes determines the overall levels of the m6A modification in the body. Third, the m6A modification can only work after being recognized by m6A readers. Various m6A readers were discovered so far and are still being explored. After combined with different m6A readers, the m6A modification can participate in almost all aspects of the RNA life cycle including RNA transcription, splicing, structural changes, localization, nuclear transport, translation, and degradation., Therefore, even when m6A modification expression is maintained, its different binding affinity toward different m6A readers may also lead to a different final fate of the corresponding RNA. Thus, specific molecular mechanisms are involved in the imbalance of the m6A RNA modification, and the differential expressions of its regulators in different cancers are extremely complicated. An in-depth study of m6A will not only help clarify the molecular mechanism of m6A RNA modification in cancer theoretically but also provide new strategies and potential targets for clinical diagnosis and treatments of cancer. The present study demonstrated that the m6A RNA modification is upregulated in both pancreatic cancer tissue and cell lines, mainly due to reduced FTO expression. FTO acts as a tumor suppressor and is associated with good survival of PAAD patients. Moreover, FTO enhances PJA2 stability in an m6A-YTHDF2-dependent manner, which promotes the inhibitory effect of PJA2 on Wnt signaling, thereby restraining the proliferation, invasion, and metastasis of pancreatic cancer cells.

Recent studies showed that m6A modification regulators are abnormally expressed in pancreatic cancer and play an important role in tumor occurrence and development. Taketo et al. reported that pancreatic cancer cells lacking m6A methylase METTL3 have higher sensitivity to gemcitabine, 5-fluorouracil, cisplatin, and radiation therapy. In turn, the m6A reader YTHDF2 promotes pancreatic cancer cells’ proliferation via the AKT/GSK3β/cyclin D1 pathway and suppresses their epithelial-mesenchymal transition (EMT) process via YAP signaling. To date, only two m6A demethylases have been identified in humans and mice, namely FTO and ALKBH5, which are both α-ketoglutarate and ferrous ion dependent. In recent years, ALKBH5 was reported to be a tumor suppressor in pancreatic cancer through a variety of pathways. Indeed, He et al. proved that ALKBH5 inhibits pancreatic cancer cell motility by demethylating the m6A modification of the long noncoding RNA KCNK15-AS1. Moreover, Guo et al. showed that ALKBH5 upregulates PER1 in an m6A-YTHDF2-dependent manner, leading to the reactivation of the ATM-CHK2-P53/CDC25C signaling pathway, thereby inhibiting the growth of pancreatic cancer cells. Tang et al. also showed that ALKBH5 restrained the proliferation, migration, and invasion of pancreatic cancer cells, while improving their sensitivity to chemotherapy by reducing WIF1 m6A modification and inhibiting the Wnt pathway. Nevertheless, the molecular mechanism of FTO in pancreatic cancer remains unclear. The present study revealed that reduced expression of FTO is a key factor leading to m6A RNA modification disorder in pancreatic cancer. The mRNA and protein expression of FTO is downregulated in both pancreatic cancer tissue and cell lines. Moreover, reduced FTO expression was found to be significantly related to poor survival of PAAD patients. Genetically knockdown of FTO remarkably promoted pancreatic cancer malignancy, whereas overexpression of FTO-WT markedly suppressed pancreatic cancer malignancy.

PJA2 is a RING-H2 E3 ubiquitin ligase that is reported to be a critical regulator of several kinase signaling pathways., Recently, emerging evidence revealed that PJA2 is a tumor suppressor in various cancers. Nonetheless, Rinaldi et al. demonstrated that PJA2 attenuates ERK1/2 phosphorylation, inhibits cell proliferation, and supports embryonic stem cell pluripotency via ubiquitylation of the kinase suppressor of Ras 1. However, Hedrick et al. found that PJA2 plays a key role on the regulation of cyclic AMP (cAMP)-dependent protein kinase A activity and promotes cancer cells’ survival. It was reported that PJA2-fer tyrosine kinase mRNA chimeras were associated with poor survival in non-small cell lung cancer. Therefore, PJA2 may play a dual role in different cancers. The present study shows that reduced expression of FTO decreases demethylating m6A modification of PJA2, thereby promoting the combination of YTHDF2 and m6A modification of PJA2 and consequent mRNA decay and expression suppression. Furthermore, knockdown of PJA2 significantly promotes the proliferation, metastasis, and invasiveness of pancreatic cancer cells. In addition, we have explored the methylation status of FTO and PJA2 promoter regions in TCGA with the tools applied by UALCAN, MethSurv, cBioPortal, and some other online tools. However, the results showed there were no statistical variations between the two genes’ methylation status and patients’ age, gender, ethnicity, metastasis, tumor stage, and survival. TFs were also predicted with the promoter sequences of FTO and PJA2 in TRANSFAC, and we found that some TFs such as SP1, nuclear factor κB (NF-κB), and Oct-1 had potential binding sites with FTO and PJA2, which meant they might promote transcription of FTO and PJA2. Nevertheless, there is still much work to do to explore the possible macroscopic regulation mechanisms of FTO and PJA2 including but not limited to their promoters.

The Wnt signaling pathway plays an important role on cell differentiation, polarization, and migration during development. It was demonstrated that Wnt signals are closely related to the occurrence and development of pancreatic cancer, with activation of the Wnt pathway promoting the proliferation, invasion, and metastasis and inhibiting the apoptosis of pancreatic cancer cells., Wnt signals can derive from the canonical β-catenin-dependent or the non-canonical Wnt signaling pathways. Bo et al. showed that Wnt5a, the most important Wnt family protein, is upregulated in pancreatic cancer tissues (81.3% [109/134]), promoting EMT and metastasis through the canonical β-catenin-dependent pathway. In recent years, accumulating evidence suggests that m6A RNA modification plays a crucial part in the occurrence and development of various cancers through the regulation of the Wnt pathway. Zhang et al. showed that m6A suppression promotes the proliferation and invasiveness of gastric cancer cells by activating Wnt and phosphatidylinositol 3-kinase (PI3K)-Akt signals. Furthermore, Zhang et al. revealed that FTO demethylated m6A modification of HOXB13, thereby preventing its mRNA decay, activating the Wnt pathway and promoting metastasis and invasion of endometrial cancer cells. Miao et al. also showed that METTL3 promotes osteosarcoma cell progression by increasing LEF1 m6A modification and activating the β-catenin/Wnt pathway. Moreover, a study showed that ALKBH5 can act as a tumor suppressor in pancreatic cancer by inhibiting the Wnt signals. The present study demonstrated that overexpression of FTO-WT significantly decreases the expression of β-catenin, Wnt5a, and LEF1, whereas it increases the expression of pS9-GSK-3β, AXIN1, and WIF1. Furthermore, knockdown of PJA2 significantly reversed these FTO-induced effects. Taken together, these results suggest that FTO acts as a tumor suppressor by preventing PJA2 decay through YTHDF2 and inhibiting the Wnt signal pathway in pancreatic cancer.

The in-depth development of m6A research technology in the past few years has supported a better understanding of the role, underlying mechanisms, and regulatory processes of the m6A modification in various cancers. However, currently available m6A-sequencing methods at the transcriptome level usually require a high total RNA starting amount (usually >20 μg); thus, it is difficult to perform m6A-sequencing detection on a large number of patients. Therefore, there is an urgent need to greatly improve the current m6A-sequencing technology while using smaller amounts of RNA and achieving a higher resolution. With the improved m6A-sequencing technology, it becomes possible to use precious primary tumor or blood samples from patients and even limited primary cancer stem cells for m6A modification analysis. Additionally, the m6A modification map of certain specific transcripts or transcript loci can be used as biomarkers for early cancer diagnosis, classification, prognostic prediction, and risk grading. Although this study has provided abundant evidence of the relevance of m6A and its demethylase FTO on pancreatic cancer malignancy, the underlying molecular mechanism still warrants additional investigations. Moreover, how m6A modification changes at the transcriptome or even at a single-base level in pancreatic cancer tissue still needs further investigations. Furthermore, it remains unclear how PJA2 inhibits the Wnt pathway in pancreatic cancer, whether it could act as an ubiquitin ligase in this process and which could be its targets.

Herein, the expression, function, and molecular mechanism of m6A RNA modification in pancreatic cancer were investigated. m6A modification was found to be upregulated in pancreatic cancer due to low FTO expression. This m6A demethyltransferase acted as a tumor suppressor in pancreatic cancer, with reduced FTO expression being associated with poor survival of PAAD patients. A detailed mechanistic analysis revealed that FTO enhances PJA2 stability in an m6A-YTHDF2-dependent manner. In summary, low FTO expression triggers the upregulation of m6A levels in PJA2, promoting its degradation via YTHDF2, which will ultimately lead to low PJA2 expression and consequent activation of the Wnt pathway. Altogether, these intricate mechanisms lead to the enhanced proliferation, invasion, and metastasis of pancreatic cancer cells (Figure 9). These findings shed light on the role and function of m6A RNA modification in pancreatic cancer. Furthermore, the described FTO/PJA2/Wnt pathway can represent a potential target for the diagnosis, prognosis, and treatment of patients with pancreatic cancer.

Figure 9

m6A demethylase FTO suppresses PC tumorigenesis by demethylating PJA2 and inhibiting Wnt signaling

In PC, reduced expression of FTO results in the upregulation of m6A levels of PJA2, promoting its degradation by YTHDF2, with consequent suppression of PJA2 expression and enhanced activation of the Wnt pathway. This FTO-mediated effect will promote the proliferation, invasion, and metastasis of PC cells.

Materials and methods

Clinical samples

All of the pancreatic cancer tissues and their corresponding adjacent normal tissues were obtained from patients who underwent resection at the Shengjing Hospital of China Medical University. None of these patients received any preoperative chemotherapy or radiotherapy. The protocol was approved by the Ethics Committee of the hospital, and written, informed consent was obtained from all patients. All samples were histologically diagnosed with PAAD by pathologic examination. The detailed clinicopathologic characteristics of patients are summarized in Table 1. The collection of follow-up data was carried out completely, and OS was defined as the time interval from the date of surgery to the date of death or the end-point of follow-up (December 30, 2020). The fresh frozen samples were obtained within 10 min after tumor excision and immediately stored at −80°C until use in experiments. The fresh frozen and archival formalin-fixed paraffin-embedded (FFPE) samples were collected from Biobank, Shengjing Hospital of China Medical University.

Cell lines, reagents, and culture conditions

Human pancreatic cancer cell lines SW1990 (American Type Culture Collection [ATCC] CRL-2172), AsPC-1 (ATCC CRL-1682), PANC-1 (ATCC CRL-1469), BxPC-3 (ATCC CRL-1687), and Capan-2 (ATCC HTB-80) were purchased from ATCC (Manassas, VA, USA). Human immortalized pancreatic duct epithelial cell line HPDE6-C7 was purchased from Biotechnology Company. All cell lines were authenticated by short tandem repeat profiling before receipt, tested for free from mycoplasma infection, and propagated for more than 6 months after resuscitation. AsPC-1, BXPC-3, and HPDE6-C7 cells were maintained in RPMI-1640 medium. PANC-1 was cultured in DMEM medium. SW1990 cell was cultured in L-15 medium. Capan-2 cell was cultured in ATCC-formulated McCoy’s 5a Medium Modified (catalog number [no.] 30-2007). All of the mediums were supplemented with 10% fetal bovine serum (FBS) and 1% penicillin/streptomycin. All cell lines except SW1990 cell were grown in an atmosphere of 5% CO2 and 99% relative humidity at 37°C. SW1990 cell was grown in an atmosphere without CO2.

Antibodies, siRNAs, and plasmid transfection

The primary antibodies were purchased from commercial sources, and information about them is given in Table S4. All of the siRNAs were ordered from GENERAL BIOL (Chuzhou, China). Sequences for all of the siRNAs were shown in Materials and methods S1. FTO-WT plasmid and FTO-mut plasmid were purchased from GeneCopoeia (Rockville, MD, USA). FTO-mut plasmid was constructed by using Stratagenes QuikChange II site-directed mutagenesis kit, and the mutant sequences were verified. Information about plasmid constructions are given in Materials and methods S1. Cells were seeded into 6-well plates at 50% confluence. After 24 h culture, siRNAs or plasmids were transfected into cells by using jetPRIME agent (Polyplus Transfection, Illkirsch, France) according to the manufacturer’s protocols.

RNA isolation and reverse transcription

Total RNA was isolated from cells and tissues by using RNAiso Plus reagent (TaKaRa, Beijing, China) according to the manufacturer’s instructions. Synergy H1 microplate reader (BioTek Instruments, Winooski, VT, USA) was used to determine RNA purity and to quantify RNA concentration. Reverse Transcription Kit (Takara Biomedical Technology, Beijing, China) was used to perform reverse transcription.

Quantitative real-time PCR and gene-specific m6A qPCR

Quantitative real-time PCR was used to assess the relative abundance of mRNA. Quantitative real-time PCR primers were obtained from Sangon Biotech (Shanghai, China), and the sequences are listed in Table S5. According to the manufacturer’s instructions, quantitative real-time PCR was performed using the SYBR Prime-Script RT-PCR Kit (Takara Biomedical Technology, Beijing, China) with a 7500 Fast Real-Time PCR System (Roche, Mannheim, Germany). Glyceraldehyde 3-phosphate dehydrogenase (GAPDH; for mRNA expression level) and hypoxanthine guanine phosphoribosyl transferase 1 (HPRT1; for relative m6A expression level) were used as internal controls to normalize qPCR results. Relative gene-expression levels were measured using cycle threshold (CT) values in the ΔΔCT calculation. Each experiment was performed in triplicates and repeated three times.

RNA m6A quantification

The m6A expression level of total RNA was measured by EpiQuik m6A RNA Methylation Quantification Kit (EpigenTek, Farmingdale, NY, USA) according to the manufacturer’s instructions. Briefly, a high-efficiency RNA binding solution was used to bind total RNA (200 ng) to the bottom of assay wells. Capture and detection antibody solutions were then added to assay wells separately in a suitable diluted concentration following the manufacturer’s instructions. Synergy H1 microplate reader (BioTek Instruments, Winooski, VT, USA) was used to detect optical density (OD) value of each well at a wavelength of 450 nm. Then, the m6A level was estimated by the following equation:where Sample OD is the OD value of sample; NC OD is the OD value of NC; PC OD is the OD value of positive control; S is the amount (nanograms) of sample RNA; and P is the amount (nanograms) of the positive control RNA.

WB analysis

WB analysis was performed as previously described. Briefly, protein was extracted from frozen tissues or cells by using radioimmunoprecipitation assay (RIPA) lysis (Beyotime, Beijing, China) supplemented with a protease inhibitor cocktail (Roche, Shanghai, China) at 4°C for 30 min. Then, samples were centrifuged at 12,000 rpm at 4°C for 10 min, and the supernatants were retained as protein lysates. The protein lysates were mixed with ¼ vol of 5× sodium dodecyl sulfate sample buffer and boiled for 10 min. For immunoblotting, 60 μg of protein lysates was separated by 10%–12% sodium dodecyl sulfate-polyacrylamide gel electrophoresis, transferred to polyvinylidene difluoride membranes (Solarbio, Beijing, China), blocked with 5% (w/v) skim milk in Tris-buffered saline-Tween 20 (TBST) for 2 h at room temperature, incubated with primary antibodies overnight at 4°C, and then followed by incubation with secondary antibody for 2 h at room temperature. Enhanced chemiluminescence chromogenic substrate (Millipore, Billerica, MA, USA) was used to visualize the bands, and the intensity of the bands was quantified by Image Lab software (Bio-Rad, Hercules, CA, USA).

IHC

FFPE samples were processed with the UltraSensitive SP IHC Kit (Maxim, Fuzhou, China) according to the manufacturer’s protocol. Heat-mediated antigen retrieval was performed with Tris/EDTA buffer (pH 9.0). After quenching endogenous peroxidase activity and blocking the non-specific binding, the samples were incubated in a humidity chamber with primary antibody. Then, the biotinylated secondary antibody, horseradish peroxidase streptavidin, and diaminobenzidine were used successively as the detection reagents. Finally, the samples were counterstained with hematoxylin and dehydrated through graded alcohols and xylene. All samples were assigned a score based on staining intensity (0 = no staining; 1 = low positive; 2 = medium positive; 3 = high positive). Images were processed with ImageJ.

Cell proliferation assays

Cell proliferation was assessed by using the CCK8 kit (Bimake, Houston, TX, USA) according to the manufacturer’s instruction. Briefly, the cells were seeded in 96-well plates at a density of 8 × 103 cells per well. After 48 h culture, cells were treated with 10 μL CCK8 kit to assess the proliferation potential. The cells were incubated at 37°C for another 2 h and then read at 450 nm with Synergy H1 microplate reader (BioTek Instruments, Winooski, VT, USA).

Incucyte Live Cell Imaging System

Imaging was performed using the Incucyte Zoom Live Cell Imaging System from Essen Bioscience (Ann Arbor, MI, USA). Cell confluence (percentage) was calculated using Incucyte Zoom software by phase-contrast images. Cells were scanned every 4 h from 0 to 96 h after treatment.

Clonogenic cell survival assays

Clonogenic cell survival assays that we described previously were used to evaluate the proliferative ability of pancreatic cancer cells in different conditions. Briefly, after different treatments, cells were counted. Every 200 cells were seeded in a 35-mm dish at 37°C under 5% CO2 conditions and incubated for enough time to allow macroscopic colony formation. Colonies were fixed with 4% paraformaldehyde for 20 min and then stained with Wright-Giemsa stain for 5 to 8 min. The number of colonies formed in each group was counted, and colonies containing approximately 50 viable cells were considered representative of clonogenic cells. The clonogenic fraction was calculated by the following equation:

Wound healing migration assays

Cells were seeded in 96-well plates at a density of 4.0 × 104 cells per well and cultured in complete medium. 24 h after seeding, scratches were performed in the center of confluent monolayer cells using the Essen Woundmaker (Ann Arbor, MI, USA). Then, cell migration into the wound was visualized by the Incucyte Zoom Live Cell Imaging System from Essen Bioscience (Ann Arbor, MI, USA). Wound confluence (percentage), relative wound density (percentage), and wound width (micrometers) were calculated using Incucyte Zoom software by phase-contrast images. Cells were scanned every 4 h from 0 to 96 h after treatment.

Transwell cell invasion assays

Cells were seeded at a density of 1 × 105 cells/100 μL in serum-free RPMI-1640 or L-15 medium into the upper chambers (pore size, 8 μm; Corning), which were precoated with Matrigel Basement Membrane Matrix (Solarbio, Beijing, China). Each lower chamber was filled with 600 μL of 20% FBS medium. After incubation for 24 h, cells embedded in the membranes were fixed with 95% ethyl alcohol for 20 min, stained with 0.1% Giemsa Stain solution for 30 min, and then washed three times with PBS. Cells were counted in three randomly selected fields under an optical microscope (×200 magnification). All experiments were conducted in triplicate.

RNA stability assays

Cells were seeded in 6-well plates at 50% confluency. After 24 h culture, cells were transfected with corresponding siRNAs or plasmids. 16 h after transfection, actinomycin D was added to 5 mg/mL at 6 h, 3 h, and 0 h before trypsinization and collection. The total RNA was purified by the RNeasy kit with an additional DNase-I digestion step on the column. Quantitative real-time PCR was conducted to quantify the relative levels of target mRNA. The degradation rate of target mRNA (k) was estimated by the following equation:where t is the transcription inhibition time, and Nt and N0 are the RNA quantities at time t and time 0. The mRNA half-life time (t1/2) was estimated by the following equation:

cBioPortal analysis

The cBioPortal for Cancer Genomics (http://www.cbioportal.org/) is an open web-based platform for exploring, visualizing, and analyzing multidimensional cancer genomics data. We initially used the cBioPortal to analyze the correlation expression of FTO and ALKBH5 (as well as FTO and PJA2) in PAAD tissues from TCGA database. To identify the downstream targets of FTO in pancreatic cancer, we used cBioPortal, UALCAN (http://ualcan.path.uab.edu/), and LinkedOmics (http://www.linkedomics.org/login.php) to explore FTO co-expressing genes in PAAD tissues. In addition, we also used cBioPortal, UALCAN, MethSurv (https://biit.cs.ut.ee/methsurv/), and some other online tools to explore the methylation status of FTO and PJA2 promoter regions in TCGA.

Kaplan-Meier Plotter

The Kaplan-Meier Plotter is a user-friendly, open web-based platform to assess the effect of 54k genes on survival of 21 types of tumors from GEO, European Genome-Phenome Archive (EGA), and TCGA. We used Kaplan-Meier Plotter and chose “Auto select best cutoff” to explore the survival analysis of PAAD patients with high and low FTO (as well as PJA2) expression. When we chose Auto select best cutoff, all possible cutoff values between the lower and upper quartiles are computed, and the best performing threshold is used as a cutoff. Cutoff value used in our analysis for FTO was 1,007; for PJA2, it was 2,987. In addition, we have also explored the impact of FTO and PJA2 expression on the OS of PAAD patients in different stages separately. Sample number in stages III and IV is too low for meaningful analysis, so we could only obtain the analysis in stages I and II. Cutoff value for FTO in stage I PAAD patients was 1,057; in stage II, it was 1,271. Cutoff value for PJA2 in stage I PAAD patients was 2,221; in stage II, it was 3,025.

Statistical analysis

Results were presented as mean ± standard error of the mean or standard deviation for at least three independent biological replicates. Student’s t tests were used for continuous variables. Kaplan-Meier analysis and log-rank test were used to evaluate the differences in patients’ survival. For statistical correlation, Pearson’s and Spearman’s correlation coefficient was used according to requirements. Statistical analyses were performed utilizing the statistical software in GraphPad Prism version 8.0. All statistic information in this study was shown in Materials and methods S2. Data were considered statistically significant as follows: ∗p < 0.05; ∗∗p < 0.01; and ∗∗∗p < 0.001.