A miR‐15a related polymorphism affects NSCLC prognosis via altering ERCC1 repair to platinum‐based chemotherapy

Abstract Platinum‐based chemotherapy is regarded as a preferential curative‐intent option for non‐small cell lung cancer (NSCLC), while the acquired drug resistance has become a major obstacle that limits its clinical application. Since the repair efficiency of tumour cells to platinum‐DNA adducts plays a crucial role in chemotherapy resistance, we aimed to explore whether several meaningful polymorphisms of DNA repair genes were associated with the benefits of platinum‐based chemotherapy in NSCLC patients. Firstly, six single nucleotide polymorphisms (SNPs) located in the 3'untranslated region (3'UTR) of three DNA repair genes were detected in 246 NSCLC patients receiving platinum‐based chemotherapy and analysed the correlation of these candidate SNPs with the overall survival. Cox proportional hazard model showed that NSCLC patients carrying ERCC1 rs3212986 AA genotype had a shorter overall survival compared to those with CC. Mechanistically, we performed tumour chemosensitivity assay to observe the convincing linkage of rs3212986 polymorphism with ERCC1 expression and cisplatin sensitivity. The subsequent in vitro experiments identified that rs3212986 polymorphism altered the post‐transcriptional regulation of ERCC1 via affecting the binding of miR‐15a, and further changed the sensitivity to platinum analogue. It reminded that patients carrying ERCC1 rs3212986 CC homozygote were expected to respond better to platinum‐based chemotherapy due to a lower expression of ERCC1. Compared with previous studies, our current comprehensive study suggested that rs3212986, a 3'UTR polymorphism in ERCC1, might have clinical relevance in predicting the prognosis of NSCLC patients receiving platinum‐based chemotherapy.


| INTRODUC TI ON
Non-small cell lung cancer (NSCLC) remains the leading cause of cancer-related mortality worldwide, 1 and its incidence and mortality in China have increased rapidly in the last two decades. 2 Approximately 50% of NSCLC patients undergoing adjuvant chemotherapy will relapse within 5 years, 3 because only a minority of them responded better to the standard treatments including chemotherapy. 4 Accordingly, there is an urgent need to identify reliable prognostic biomarkers to assist in developing personalized therapies.
Although platinum-based chemotherapy has been determined as a standard first-line treatment for advanced NSCLC, the frequent drug resistance remains a major obstacle that limits the therapeutic efficacy in clinic. 5 Platinum-DNA adduct accumulation is a determinant step for the cytotoxicity of platinum-based antitumor agents which lead to the destabilization of double helix, blocking replication and inhibiting transcription. 6 However, DNA repair capacity, which varies widely among individuals, plays a fundamental role in timely removing DNA adducts. 7 In general, a high DNA repair capacity of tumour cells is a warning of potential chemotherapy resistance of NSCLC patients. 8 As known, there are at least several DNA repair systems in the human body, such as nucleotide excision repair (NER), base excision repair (BER) and mismatch repair (MMR). 9 Among them, NER is the most significant and flexible one for excising platinum-DNA adducts. 10 The main repair steps of NER include the separation of double helix at lesion sites, the excision of the lesioncontaining single stranded DNA fragment, the synthesis of a new DNA fragment to replace the gap and the ligation of the remaining single-stranded nick. 11 Therefore, the NER pathway plays a pivotal role in repairing platinum-DNA adducts and may affect the sensitivity of individuals to platinum chemotherapy potentially. 12,13 The excision repair cross-complementation group 1 (ERCC1) is a known gene as a rate-limiting enzyme in the NER pathway, 14 and emerging evidence suggested that ERCC1 expression acted as a promising biomarker to predict the prognosis of multiple cancers. 15 ERCC1 overexpression showed a certain clinical resistance to platinum-based chemotherapy in ovarian, cervical, colorectal carcinomas and NSCLC. [16][17][18][19] Additionally, NSCLC patients with complete resection of ERCC1-negative tumours benefited more from cisplatin-based adjuvant chemotherapy than those with ERCC1 positive tumours. 20 In other words, a higher expression of ERCC1 may indicate a potential failure in chemotherapy due to the efficient repair of genetic damage in tumour cells induced by platinum analogues.
Although the clinical trial of customized chemotherapy relying on ERCC1 expression has been carried out. 21 Actually, it is often difficult to obtain sufficient tumour tissues to quantify the expression of ERCC1 clinically. Instead, the germline polymorphisms are easy to measure and constant over time and have gradually become promising biomarkers to predict clinical outcomes of NSCLC patients.
Single nucleotide polymorphisms (SNPs) in human DNA repair genes have been reported to modulate DNA repair capacity, which is generally considered as valid biomarkers to reflect the repair efficiency of chemical-induced DNA damage. 22 Notably, the variation of coding regions can affect the function of protein, while other variations of non-coding regions may affect gene expression and protein activity. Several studies revealed that the SNPs of rs1007616, rs735482 and rs3212986 located in the 3′ untranslated region (UTR) of ERCC1 could reduce the stability of its mRNA and further affect DNA repair capacity, 23,24 which reminded us that the post-transcriptional regulation of ERCC1 might play a crucial role in the development of lung cancer. 25,26 MicroRNAs (miRNAs) are a kind of endogenous non-coding RNA with the length of about 22 nucleotides, regulating the expression of genes by base paring with the 3'UTR of target mRNA generally. 27 As known, the miRNAs play a growing important role in various essential and important biological processes, such as cell development, proliferation differentiation, apoptosis, signal transduction, viral infection and so on. 28,29 Therefore, the relationship between the allele-specific alterations related to miRNAs and the sensitivity to platinum-based chemotherapy in NSCLC patients needs to be investigated.
In the present study, a survival analysis of NSCLC patients received platinum-based chemotherapy was firstly performed to evaluate the association of six candidate 3'UTR polymorphisms in DNA repair genes with the prognosis. Then, we found that ERCC1 rs3212986 polymorphism exhibited a more convincing linkage with the overall survival in the clinical investigation. Furthermore, NSCLC tumour tissue samples were detected to confirm the relationship between ERCC1 rs3212986 polymorphism and cisplatin sensitivity. Finally, a series of in vitro functional experiments were carried out to clarify whether the target polymorphism was causally linked with the sensitivity to platinum analogues via altering the post-transcriptional regulation of ERCC1 due to a certain miRNA.
Accordingly, our current comprehensive study contributes to identify the potential biomarkers in predicting the prognosis of NSCLC patients receiving curative-intent chemotherapy.

| Study subjects and sample collection
In this study, we recruited 246 patients who had been histologically confirmed with NSCLC and received curative-intent therapy including surgical resection and platinum-based chemotherapy at the Organization (Patient information was provided as a supplementary sheet in .xlsx format.). Besides, 5 ml blood samples were drawn from each participant and their fresh tumour tissues were collected during surgery, then immediately transferred into liquid nitrogen. Finally, our accumulation was stopped in February 2019 to guarantee a minimum follow-up time of 4 years and the detailed follow-up information was recorded. Overall survival (OS) was specified as the primary endpoint of this study, which was calculated from diagnosis to the last follow-up or death due to any cause. for 30 s. To realize a phenotype-blind genotyping process, the collection of data, the test of samples and the analysis of results were conducted by different experimenters, respectively. For rs1007616 and rs108621, thymine (T) was replaced with cytosine (C); cytosine (C) was replaced with adenine (A) in rs735482 and rs3212986; guanine (G) was replaced with adenine (A) in rs23362196; whereas in rs1052133 cytosine (C) was replaced with guanine (G). TaqMan SNP genotyping IDs were as follows: ERCC1 (rs3212986, C>A, assay ID is C_2532948_10; rs735482, A>C, assay ID is C_341729_10; rs2336219, G>A, assay ID is C_16204465_10), MLH3 (rs108621, T>C, assay ID is C_2178406_10) and hOGG1 (rs1052133, G>C, assay ID is C_3095552_1). Afterwards, the supernatant was carefully removed and 100 μl dimethyl sulfoxide was added to each well to discontinue the reaction.

| Tumour chemosensitivity assay
Besides, blue-violet formazan particles were dissolved for 10 min in the dark at 25°C. Finally, the absorbance at 490 nm was detected by quant universal microplate spectrophotometer (BioTek Instruments, Inc.), and the sensitivity of each patient to CDDP was evaluated.  2 −ΔΔCt method was adopted to quantify the relative expression of genes and GAPDH gene acted as an internal reference.

| Western blot
Tumour tissue homogenate was made by tissue homogenizer before it was lysed in RIPA buffer solution. After a series of extraction operations following the manufacturer's instructions, 30-60 μg of total protein was collected for the detection of ERCC1 expression was acquired. Western blot was performed strictly according to the standard protocol and finally the immunoreactive bands were developed with hypersensitive chemiluminescence reagents (Beyotime, China).

| Dual-luciferase reporter assay
Above all, bioinformatics analysis was conducted to acquire genomic information including the Targetscan database (http://www.targe tscan.org/) and RNAhybrid database (https://bibis erv.cebit ec.unibiele feld.de/rnahy brid), from which we got the binding fragment of miRNA in ERCC1 3'UTR, the GenBank database (https://www.ncbi. nlm.nih.gov/genba nk/), from which the upstream and downstream sequences of target genes were obtained and the Vector NTI software, by which relevant primers were favourably designed.

| Genotyping lung adenocarcinoma cell line A549
The genotype of rs3212986 polymorphism in A549 cells was detected and the allele-specific primers synthesized via standard phosphor amidite synthesis were as follows: F: 5'-CACGAGCCCTTCTTGAAA-3′

| Statistical analysis
All data were statistically processed with spss19.0 software (SPSS, Inc.) and graphpad prism6.0 software (GraphPad Software, Inc.). χ 2 test was used to analyse the association between the candidate polymorphisms and clinic-pathological characteristics. Kaplan-Meier curves and log-rank test were applied to assess the effect of candidate polymorphisms on overall survival (time between diagnosis and death or last follow-up). In addition, Cox proportional hazard model was adopted to further clarify the association of candidate polymorphisms with overall survival after a preliminary univariate regression analysis. Corresponding hazard ratios and 95% confident intervals (95% CIs) were estimated after adjusting for age, gender, stage, pack year of smoking and treatment regimens. A two-tailed p-value <0.05 was considered statistically significant.

| Association between the clinicopathologic characteristic and overall survival in NSCLC patients
This study incorporated 246 patients with NSCLC who received curative-intent therapy, including surgical resection and platinumbased chemotherapy. The average age was 59.3 years old, 62% of which are males. Most of them were adenocarcinomas (65.1%) and 78% were in early stage (I and II) (see Table S1). The median fol-  Table 1).
The univariate analysis showed that the patients carrying AA genotype of rs735482 had a significantly higher survival index than those with CC genotype ( Figure 1A; p < 0.05). Meanwhile, the median survival time of the patients carrying AA genotype of rs2336219 was apparently longer than that of those with GG genotype ( Figure 1B; p < 0.05). However, the multivariate Cox model did not show that rs735482 and rs2336219 had any effect on survival indices. Instead, the patients carrying AA genotype of rs3212986 were associated with a significantly shorter survival time in both univariate ( Figure 1C; p < 0.05) and multivariate analyses ( Figure 1D;

| ERCC1 rs3212986 AA genotype was linked with a higher expression of ERCC1 and a lower sensitivity to cisplatin
In order to verify the clinical relevance of ERCC1 rs3212986 polymorphism, the primary tumour cells from NSCLC patients were cultured and attacked with CDDP to evaluate the individual sensitivity to platinum analogues. The IC 50 values provided by the MTT assay showed a wide discrepancy of cells in response to cisplatin treatment. Consistent with previous reports, whether the high expression of ERCC1 mRNA or protein was positively correlated with the resistance of tumour cells to CDDP (Figure 2A,B). Furthermore, the mRNA and protein expression of ERCC1 in tumour tissues were compared between the AA and CC genotypes of rs3212986. The data from RT-qPCR and Western Blot, respectively, indicated that the average levels of ERCC1 mRNA and protein in patients carrying AA genotype was significantly higher than those with CC genotype ( Figure 2C,D). Therefore, we speculated that rs3212986 polymorphism was linked with a different sensitivity to cisplatin via altering the expression of ERCC1.

| MiR-15a was screened and verified as a target miRNA binding to rs3212986
We presumed that the presence of some miRNAs can affect the post-transcriptional regulation of ERCC1 and alter its expression by binding to 3'UTR. Accordingly, the potential miRNAs binding to ERCC1 rs3212986 were predicted by the Targetscan and RNAhybrid databases. Finally, two miRNAs including the miR-15a and miR-4298 were identified, and the minimum free energies were −40.80 Kcal/mol for miR-4298 and −25.80 Kcal/mol for miR-15a, respectively (see Table S2). The above results indicated that ERCC1 3'UTR contains potential target sequences for miR-15a and miR-4298, and rs3212986 located in the MRE (miRNA response element) of mature miR-15a ( Figure 3A). To evaluate the direct effects of miR-15a and miR-4298 on ERCC1 expression, we constructed luciferase reporter plasmids containing the WT or MUT 3'-UTR of ERCC1 (including the ERCC1 MUT and rs3212986 MUT), and then co-transfected them into HEK293T cells with miR-15a mimics, miR-4298 mimics or NC mimics, respectively. Interestingly, miR-15a significantly suppressed the luciferase activity in the ERCC1 WT group (p < 0.05) ( Figure 3B) but did not show any effect in the rs3212986 MUT group. However, a down-regulated luciferase activity was found in both ERCC1 WT (p < 0.05) and rs3212986 MUT groups (p < 0.05) after the transfection of miR-4298 ( Figure 3C). Besides, co-transfection of miRNA mimics and ERCC1 MUT reporter plasmids showed no effect on luciferase activity. Our data demonstrated that although the luciferase activity could be regulated by both miR-15a and miR-4298, only miR-15a-mediated regulation of ERCC1 expression was diminished when rs3212986 C allele was transferred to A allele.

| MiR-15a decreased ERCC1 expression in A549 cells carrying rs3212986 CC genotype
Sanger sequencing results of human lung adenocarcinoma cell line A549 was presented in Figure 4A

| ERCC1 expression was negatively linked with miR-15a in NSCLC tissues carrying rs3212986 CC genotype
To verify the correlation between the ERCC1 rs3212986 polymorphism and miR-15a, the expression of miR-15a was firstly detected in tumour tissues carrying different genotypes of rs3212986 and no difference was observed between the two genotypes ( Figure 5A), which suggested that the rs3212986 polymorphism had no effect on the expression of miR-15a. However, the expression of miR-15a was negatively correlated with the ERCC1 mRNA expression in tumour tissues carrying CC genotype, but not in AA one ( Figure 5B,C).
Considering the post-transcriptional regulation of miR-15a on ERCC1 mRNA, we further performed an immunohistochemical staining on clinical tumour tissues to determine the correlation between the expression of ERCC1 protein and miR-15a. As shown in Figure 5D,E in tumour tissues with CC genotype, the low expression of ERCC1 was generally observed in the high expression group of miR-15a (E), but not in AA genotype (D). Based on the analysis here, our data suggested that ERCC1 expression might be affected by rs3212986 polymorphism via the regulation of miR-15a, which helps to partially explain the higher sensitivity to platinum-based chemotherapeutics in NSCLC patients carrying rs3212986 CC genotype due to a lower expression level of ERCC1 mediated by miR-15a.

| DISCUSS ION
As known, the pharmacological role of platinum-based chemotherapeutics is to form platinum-DNA adducts, which will lead to the inhibition of proliferation in tumour cells. In this case, the DNA repair capacity in tumour cells has an immense impact on clinical efficacy in NSCLC patients who receive platinum-based chemotherapeutics. 30 with the sensitivity to platinum analogues via affecting the posttranscriptional regulation of ERCC1 and alter the DNA repair capacity of tumour cells. Moreover, increasing studies have been devoted to developing computational models to predict potential miRNAdisease associations. 54,55 Then, the accumulation of biological data, the combination between the depth computational prediction model and causal experimental verification could form an effective method to explore miRNA-disease associations.
Briefly, the basic process and main findings of this study were summarized and shown in Figure 6.

| CON CLUS ION
Rs3212986, a 3'UTR polymorphism in ERCC1, is linked with the sensitivity to platinum analogues via altering ERCC1 expression due to the binding of miR-15a. As a potential prognostic biomarker, the rs3212986 polymorphism is expected not only to help clinicians perform personalized chemotherapy based on different genetic backgrounds of patients (e.g. the patients with AA genotype of rs3212986 may not be suitable for platinum-based chemotherapy) but also to contribute to making more accurate prediction for the prognosis of NSCLC patients who receive platinum-based chemotherapy. Xiaobo Lu: Funding acquisition (lead); methodology (lead); project administration (lead); resources (lead); supervision (lead); writingreview and editing (equal).

ACK N OWLED G EM ENTS
The authors would like to acknowledge the National Natural Science Foundation of China for the financial support and the First Affiliated Hospital of China Medical University for the supply of clinical F I G U R E 5 ERCC1 expression was negatively correlated with miR-15a expression exclusively in tumor tissues carrying rs3212986 CC genotype. Expression of miR-15a in NSCLC tissues carrying different rs3212986 genotypes (A). Correlation between the expression of miR-15a and ERCC1 mRNA in NSCLC tissues carrying AA (B) and CC (C) genotypes. U6 was used as an internal control to normalize the expression of miR-15a. Correlation between ERCC1 protein and the expression of miR-15a in NSCLC tissues carrying AA (D) and CC (E) genotype.

F I G U R E 6
The basic process and main findings of the present study samples. This work was supported by the National Natural Science Foundation of China [No. 82173567, 81773470].

CO N FLI C T O F I NTE R E S T
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data used to support the findings of this study are included within the article.