Excessive miR-25-3p maturation via N6-methyladenosine stimulated by cigarette smoke promotes pancreatic cancer progression

N6-methyladenosine (m6A) modification is an important mechanism in miRNA processing and maturation, but the role of its aberrant regulation in human diseases remained unclear. Here, we demonstrate that oncogenic primary microRNA-25 (miR-25) in pancreatic duct epithelial cells can be excessively maturated by cigarette smoke condensate (CSC) via enhanced m6A modification that is mediated by NF-κB associated protein (NKAP). This modification is catalyzed by overexpressed methyltransferase-like 3 (METTL3) due to hypomethylation of the METTL3 promoter also caused by CSC. Mature miR-25, miR-25-3p, suppresses PH domain leucine-rich repeat protein phosphatase 2 (PHLPP2), resulting in the activation of oncogenic AKT-p70S6K signaling, which provokes malignant phenotypes of pancreatic cancer cells. High levels of miR-25-3p are detected in smokers and in pancreatic cancers tissues that are correlated with poor prognosis of pancreatic cancer patients. These results collectively indicate that cigarette smoke-induced miR-25-3p excessive maturation via m6A modification promotes the development and progression of pancreatic cancer.

METTL3 which, in turn, promotes its maturation by DROSHA, in pancreatic cancer. They go on showing that Increase in DROSHA processing augments miR-25 level, which is an oncogenic miRNA in pancreatic cancer because directly represses PHLPP2. Interestingly they further found that a novel m6A reader, NKAP, interacts with m6A modification presents on pri-miR-25 which induces DGCR8 and DROSHA recruitment on this primary miRNA. This study reports several novels, interesting findings, therefore I believe that, after some extra experimental work, it deserves to be published in Nature Communications.
Specifically: 1) In figure 1e the author reported that miR-25-3p is up-regulated in PDAC compared to normal samples in two independent cohorts. These results are not totally convincing because of the evident high variability in miR-25-3p expression across the samples. However, several miRNA expression profiling studies of miRNA levels in PDAC comparing to normal specimens have been previously reported. The authors should perform a metanalysis of these data to verify whether miR-25-3p is recurrently over-expressed in PDAC by exploiting these multiple datasets. Also, what is the relationship between miR-25-3p expression and PDAC grade?
2) In order to improve survival data (fig 1f-h), it would be interesting to also include analysis of publicly available datasets of miRNA expression in PDAC, such as the ones coming from TCGA data.
3) The authors indicate that cigarette smoke concentrate (CSC) increases miR-25 maturation through NFIC, METTL3 and NKAP, but they do not properly demonstrate this model. To do so, the authors should test whether CSC-dependent increase of miR-25-3p maturation can be rescued by siRNA treatment of NFIC, METTL3 and/or NKAP. 4) Methods lack information regarding correlation analyses of gene pair obtained by using TCGA or GTEx datasets and should be included in the revised version.
Reviewer #1 (Remarks to the Author): It has been well established that cigarette smoking is a risk factor for PDAC, but the underlying mechanism remains largely unknown. In this manuscript, Zhang et al propose a mechanistic link between CSC and PDAC by showing that CSC may cause the dysregulation of miR-25-3p via aberrant adenosine methylation (m 6 A) of its primary microRNA. They first conducted microarray analysis of the differential expression of microRNAs in an immortalized human pancreatic duct epithelial cell treated with CSC and found that miR-25-3p was the most dramatically upregulated miRNA upon exposure to CSC. They then found that the miR-25-3p levels were significantly higher in PDAC than in normal pancreatic tissues, which is significantly correlated with poor survival in PDAC patients. Furthermore, they detected higher serum miR-25-3p in smokers compared with nonsmokers. They found that CSC may cause demethylation in the METTL3 promoter, facilitating the binding of transcription factor NFIC to its regulatory region and consequently enhanced expression of METTL3 in CSC-exposed PDAC cells. They also investigated the mechanism by which m 6 A methylation promotes miR-25-3p maturation and found that NKAP may mediate DGCR8 recognition of pri-miR-25. They also identified PHLPP2 as the downstream target of miR-25-3p. High level of miR-25-3p suppresses PHLPP2, then activates the downstream ATK-p70S6K signaling pathway. Overall, this is a well-designed comprehensive study illustrating several novel findings, whereas the following issues need to be addressed by the authors. expression profiling studies of miRNA levels in PDAC comparing to normal specimens have been previously reported. The authors should perform a metanalysis of these data to verify whether miR-25-3p is recurrently over-expressed in PDAC by exploiting these multiple datasets. Also, what is the relationship between miR-25-3p expression and PDAC grade?
Response 1: As per the suggestion, we have analyzed the miRNA expression profiling in PDAC in Gene Expression Omnibus database (GSE24279, GSE25820 and GSE41369) and the results show that PDACs expressed higher miR-25-3p level than normal pancreatic tissues. We have added this result in revised manuscript ( Supplementary Fig. 1, page 5, lines 118 120). Besides, previous study reported by other group also showed that miR-25-3p was significantly up-regulated in pancreatic cancer compared with normal pancreatic tissue (Volinia et al, Proc. Natl. Acad. Sci. USA 23, 21522165, 2006, doi/10.1073. We have included this information in revised manuscript (Reference 36). These previous findings are consistent with our result shown in Fig. 1e.
We have analyzed the relationship between miR-25-3p expression levels and PDAC stage in our patients' cohorts but the results are negative. Two reasons might explain the negative result: first, although we have two independent patient groups in this study, the sample sizes (N=102 and N=73, respectively) are relatively small for further stratification analysis of grade or stage; second, to obtain tissue specimens for molecular analysis, 83.3% of patients in the present study were in early stage (I/II) and underwent surgical resection, which may have selection bias for the relationship analysis. Further studies with randomly selected larger sample size are needed to address this issue.
We also analyzed the TCGA data for differential cancer-normal expression of miR-25-3p and relationship between miR-25-3p expression levels and PDAC stage. Unfortunately, there are only 4 paired PDAC and normal tissue samples in TCGA database (Fig. 1a for Reviewer), which cannot provide conclusive result. However, we found an increasing trend for miR-25-3p levels in terms of increasing PDAC stages (Fig. 1b for Reviewer) although the differences were not statistically significant. Because this is primarily negative results, we did not include it in the manuscript. One previous study has reported that level of miR-25 gradually increased in multistep tumorigenesis of pancreatic cancer, ranging from normal, hyperplastic, angiogenic islets, to tumors, indicating a high correlation between miR-25 expression and PDAC stages (Olson et al. Genes Dev. 23, 21522165, 2009, doi/10.1101. We have added this reference in the revision (Reference 35).

Fig. 1 for Reviewer
The expression levels of miR-25-3p in PDAC from TCGA database. a Expression level of miR-25-3p was detected in 4 paired PDAC and normal tissue samples in TCGA database. b Shown were increasing trend for miR-25-3p levels in terms of increasing PDAC T and N stages.

Comment 2:
In order to improve survival data (fig 1f-h), it would be interesting to also include analysis of publicly available datasets of miRNA expression in PDAC, such as the ones coming from TCGA data.
Response 2: As per the suggestion, we have analyzed the TCGA database for the survival issue. No positive result is seen for overall survival of PDAC patients (Fig. 2 for Reviewer, left). We found that patients with high level of miR-25-3p had worse relapse-free survival compared with those with low level of miR-25-3p ( Fig. 2 for Reviewer, right), but this is not statistically significant. The reason for the inconsistency between our results and the TCGA data is not evident, probably due to differences in other molecular or genetic alterations between different races of patients in study. Because the result coming from TCGA data is basically negative, we did not include this information in the manuscript.

Fig. 2 for Reviewer
The association between miR-25-3p level and survival of PDAC patients from TCGA database.

Comment 3:
The authors indicate that cigarette smoke concentrate (CSC) increases miR-25 maturation through NFIC, METTL3 and NKAP, but they do not properly demonstrate this model. To do so, the authors should test whether CSC-dependent increase of miR-25-3p maturation can be rescued by siRNA treatment of NFIC, METTL3 and/or NKAP.

Response 3:
In the initial manuscript, we did test the effects of siRNA treatment of METTL3 or NKAP on CSC-induced miR-25-3p maturation and the results are positive (Fig. 2d for METTL3 and Fig. 4k for NKAP). Per the suggestion, we have performed additional experiments to test the effects of siRNA treatment of NFIC on CSC-induced miR-25-3p maturation and the result is also positive and is in line with our conclusion. We have included this new result in the revision (Page 7, lines 171 172; Supplementary Fig. 5f). We hope that the addition of this result has satisfied the comment.
Comment 4: Methods lack information regarding correlation analyses of gene pair obtained by using TCGA or GTEx datasets and should be included in the revised version.
Response 4: Thanks. Per the comment, we have added the information on the analysis methods in the public data processing section under Methods (Pages 25 26, lines 607 612) and the websites of corresponding database or platform are also included in the "URLs" part (Page 29).