Data mining analysis of miR‐638 and key genes interaction in cisplatin resistant triple‐negative breast cancer

Cisplatin is one of the chemotherapy for the treatment of triple‐negative breast cancer (TNBC), but its effectiveness is limited because of the phenomenon of chemoresistance. miR‐638 was shown to regulate chemoresistance; however, it has never been validated in the cisplatin‐resistant tumor from patients. This present study aimed to identify the key gene regulatory networks of miR‐638 and evaluate the potential role of the miR‐638 and its targets as potential prognosis biomarkers for cisplatin‐resistance triple‐negative breast cancer patients. The miR‐638 target was obtained from the miRecords database while the mRNA of chemoresistance biomarker candidate was obtained from the GSE18864 of GEO database, which is mRNA of cisplatin‐resistance TNBC patients. CCND1 and FZD7 are potential candidates for cisplatin chemoresistance biomarkers in patients with TNBC. Moreover, a Kaplan‐Meier survival plot showed that breast cancer patients with low mRNA levels of FZD7 had significantly worse overall survival than those in higher mRNA expression group. Taken together, miR‐638 plays a role in cisplatin resistance mechanism through a mechanism involving its target gene CCND1 and FZD7. Overall, miR‐638, CCND1, and FZD7 are candidates for cisplatin biomarker resistance in TNBC.


Introduction
Triplenegative breast cancer (TNBC) occurs in about 20% of cases of breast cancer and is associated with the risk of relapse and poor prognosis (Kuo et al. 2017). Cis platin is a chemotherapy drug, which is used for the treat ment of triplenegative breast cancer, but its effectiveness has not been maximized due to the problem of chemoresis tance (Hu et al. 2015). Chemoresistance is a phenomenon when cancer cells become insensitive to chemotherapy and are classified into intrinsic and acquired resistance (Ji et al. 2019). The TNBC is an aggressive subtype that usually evolves chemoresistance (Kim et al. 2018). One of the biomarkers for predicting chemoresistance and prognosis is miRNA ), a small noncoding RNA con sisting of 2122 nucleotides that negatively target mRNA, and thus suppresses the expression of its target genes (Orso et al. 2019).
miR638 is one of the miRNAs that has been exten sively investigated in the development of cancer (Li et al. 2011; Lin et al. 2015; Wei et al. 2017). It acts as either a tumor suppressor gene or oncogene. miR638 possesses a tumor suppressor gene by inducing apoptosis and inhibit ing cell proliferation, invasion, and migration (Shen et al. 2017). In osteosarcoma, miR638 promotes apoptosis by suppressing cyclin D1, phospholipase D1 (PLD1) and vas cular endothelial growth factor (VEGF) (Xue et al. 2019). miR638 directly targets HOXA9 and suppresses the ex pression of Wnt/betacateninregulated oncogenes cyclin D1 and CMYC (Zheng et al. 2018). On the other hand, miR638 acts as an oncogene. miR638 promotes metas tasis and prevents cell death in melanoma cells (Bhat tacharya et al. 2015). It induces cell proliferation, migra tion, and invasion in oesophagal squamous cell carcinoma and breast cancer cells by targeting DACT3, a key regula tor of Wnt/betacatenin signaling (Ren et al. 2017).
miR638 also regulates chemoresistance in cancer cells. Increasing expression of miR638 after chemother apy in nonsmall cell lung cancer patients is correlated with better survival . It also enhances the efficacy of bleomycin and cisplatin in K562 leukemic cells (He et al. 2016). In MDAMB231 cells, miR638 regulates cell migration and sensitivity to cisplatin (Tan et al. 2014). Nevertheless, no study has been conducted on the regulation of miR638 and its regulatory network in cisplatinresistant TNBC using patient samples.
Over the past few years, bioinformatics has grown and provided new methods for the prediction of drugtarget genes using multiplatform analysis ). Computational approaches have been used to mine and in tegrate data in public databases to provide researchers with accurate and fast information in the field of biomedicine and drug discovery (Pandika 2018). In this study, several databases were used, including GEO, TargetScan, ON COMINE, KMPlotter, STRING, and cBioportal to iden tify the interactions between miR638 and its target genes in patients with cisplatinresistance TNBC.
In this study, we utilize a bioinformatics approach with data mining analysis to identify key gene regulatory net works of miR638 and evaluate the potential role of the miR638 and its targets as potential prognosis biomark ers for cisplatinresistance triplenegative breast cancer patients. The target of miR638 was predicted using miRecords database. Gene expression profile of cisplatin resistant breast cancer was obtained from GEO datasets. We also performed validation using KM Plot and ON COMINE and identified genetic alterations among target genes in cBioportal database.

Data collection and processing
Microarray data were obtained from GSE18864, which contains twentyeight women with triplenegative breast cancer stage II or III, which received four cycles of cis platin. Patient age ranged from 29 to 69 years at diagno sis. Fourteen patients were considered a good response, and fourteen patients were considered as a poor response based on MillerPayne score (Silver et al. 2010). Data pro cessing was conducted using GEO2R, an online tool for GEO data analysis based on the R programming language (https://www.ncbi.nlm.nih.gov/geo/geo2r/). Differential expression genes (DEGs) between cisplatin sensitive and resistant cells/tissues were screened. Adjusted P value <0.05 and IFCI >1.5 were used to select significant DEGs, as described in a previous study (Zhao et al. 2018).

Validation of target genes in cisplatin-resistant and sensitive breast cancer cells
Confirmation of the reliability of the target genes in cis platin sensitive and resistant breast cancer cells was con ducted using ONCOMINE (https://www.oncomine.org), a cancer microarray database and webbased datamining platform (Rhodes et al. 2004). Briefly, the expression level of SRGAP1, HIC2, CCND1, SAP30BP, and FZD7 among cisplatin resistance breast cancer studies were re trieved from ONCOMINE.

Analysis of genetic alterations among target genes
The genetic alterations of target genes were analyzed from breast cancer studies using cBioPortal (http://www.cbiopo rtal.org) (Cerami et al. 2012; Gao et al. 2013. Screened target genes (SRGAP1, HIC2, CCND1, SAP30BP, and FZD7) were subjected to genetic alterations analysis in all breast cancer studies. The breast cancer study with the highest genetic alterations was chosen for mutual exclusiv ity of the screened target genes with p<0.05 was selected as the cutoff value.

Identification of miR-638 target genes and miR-638-target gene regulatory network
A total of 254 and 206 genes were extracted from miRecords and GSE18864, respectively (Figure 1a). A Venn diagram generated five DEGs from miRecords and GSE18864, including SRGAP1, HIC2, CCND1, SAP30BP, and FZD7. A miR638 target gene regulatory network was constructed (Figure 1b). Interaction between miR638 and its target genes in target sites was analyzed by TargetScan ( Figure 2).

Kaplan Meier survival analysis
Kaplan Meier plot for overall survival of breast cancer pa tients showed that patients with the high miR638 level had significantly worse overall survival than those in the low expression level group (p= 0.021) (Figure 1c). The overall survival was also obtained according to the low and high expression levels of each target gene ( Figure 3). The re sults showed that patients with the high mRNA level of SR GAP1 (p=0.13), CCND1 (p=0.18), and FZD7 (p=0.046) ( Figure 3) have better survival than patients with the low  mRNA level. Moreover, patients with the high mRNA level of HIC2 (p=0.81) and SAP30BP (p=0.26) have worse survival than those with the low mRNA level.

Validation of target genes in cisplatin-resistant and sensitive breast cancer cells
ONCOMINE was used to confirm the reliability of the tar get genes in cisplatin sensitivity (Figure 4). A study using cell lines showed the downregulation of HIC2 in cisplatin resistance breast cancer cells (Lee et al. 2007). Another study showed a similar level of CCND1 among cisplatin resistant and cisplatinsensitive breast cancer cells (Gar nett et al. 2012). A study showed the downregulation of SAP30BP cisplatinresistance breast cancer cells (Gyorffy et al. 2006). Moreover, a study using cell lines showed a similar expression level of FZD7 among cisplatin sensitive and cisplatinresistance breast cancer cells (Gar nett et al. 2012). No study was found in ONCOMINE re lated to SRGAP1 and cisplatin resistance in breast cancer.

Discussion
This present study aimed to identify the key gene regula tory networks of miR638 and evaluate the potential role of the miR638 and its targets as potential prognosis biomark ers for cisplatinresistant TNBC patients. Understanding the relevance of miRNA and its mRNA target is very im portant to elucidate the mechanism of gene transcription and cellular pathophysiology. In addition, understanding the mechanism of resistance is very important for diagno sis and treatment in TNBC patients because this subtype is hard to treat. In this present study, five genes were identified from miRecords and GSE18864. Based on Kaplan Meier over all survival (Figure 3) and validation of target genes in cisplatinresistant and sensitive breast cancer cells with ONCOMINE (Figure 4), two potential biomarkers were identified, which are CCND1 and FZD7. Genetic al terations analysis among samples from the MBC Project (Lefebvre et al. 2016) showed genetic alterations of CCND1 and FZD7 in 35% and 0.6% of samples, respec tively ( Figure 5). Thus, CCDN1 and FZD7 are the poten tial key genes in cisplatinresistant TNBC.
The two biomarker candidates, CCND1 and FZD7, are extensively studied for its regulation in cancer develop ment. CCND1 encodes cyclin D1, which plays a role in cell cycle progression in the G1S phase transition (Seiler et al. 2014). Cyclin D1 plays a role in the process of cell proliferation and growth regulation, DNA repair, cell migration, and a prognostic and predictive marker in dif ferent types of cancer (RamosGarcia et al. 2017). Cy clin D1 is frequently overexpressed in human cancers, in cluding breast cancer (Maia et al. 2016), cervical cancer (Xu et al. 2016), and nonsmall cell lung cancer (Baykara et al. 2017). Cytoplasmic level of cyclin D1 is used for a biomarker of early diagnosis in breast cancer (Ullah Shah et al. 2015), as well as for biomarkers of invasiveness in endometrial, breast, prostate and colon cancer (Fuste et al. 2016). However, high expression of Cyclin D1 has a posi tive correlation with the beneficial effect of chemotherapy in metastatic bladder cancer (Seiler et al. 2014).
Genetic alterations study with cBioportal revealed al terations of CCND1 in 35% of patient samples, with am plification as the highest alterations. Previous studies demonstrated that amplification in CCND1 is an early event in the development of a breast cancer stem cells (Burandt et al. 2016), and mutations in CCND1 is associ ated with increased risk of breast cancer (Soleimani et al. 2017). A previous study demonstrated that overexpression of CCND1 has occurred through amplification, transloca tion, or posttranscriptional regulation . CCND1 gene amplification is a molecular key alteration in breast cancer and was suggested to predict resistance to endocrine therapy (Kilker et al. 2004). Taken together, gene amplification of CCND1 possibly plays an essential role in cisplatinresistant TNBC. This mechanism needs to be explored further.
The results of this present study revealed that CCND1 is downregulated in cisplatinresistant TNBC. A previous study demonstrated that targeting CCND1 with miR503 leads to the induction of G0/G1 cell cycle arrest and re duction of cell proliferation in breast cancer (Long et al. 2015). Another study showed that downregulation of cyclin D1 inhibits proliferation and colony formation in SKOV3 ovarian cancer the cells . In ad dition, inhibition of proliferation in human ovarian cancer cells by cisplatin is correlated with inhibition of CCND1 expression (Dai et al. 2016). miR638 targets CCND1 and thus inactivates PI3K/Akt pathwayregulated cell growth in Sertoli cells (Hu et al. 2017). Therefore, further studies on the role of CCND1 in TNBC resistance mechanism to cisplatin are needed.
FZD7 encodes frizzled homolog 7 (Yang et al. 2011) and plays an important role as a membrane receptor in Wnt/βcatenin signaling in cancer cells (Xie et al. 2018). Wnt signaling is activated in TNBC (King et al. 2012). The expression of FZD7 is upregulated in patients with breast cancer compared to normal tissues (Jia et al. 2018). In addition, inhibition of FZD7 with interfering RNA (King et al. 2012; Yang et al. 2011 or monoclonal anti body (Zarei et al. 2018) could reduce cell proliferation in TNBC. Recently, FZD7 is targeted by miR638, leading to inhibition of Wnt signaling in glioma progression (Chen and Duan 2018). Therefore, it is necessary to further in vestigate the role of FZD7 in the mechanism of cisplatin resistance in TNBC. Further, in vitro and in vivo studies need to be done on the mechanism of miR638 regulat ing cisplatin resistance in TNBC, as well as how miR638 regulates its target gene.

Conclusions
Our study provides an integrated data mining analysis of the cisplatin resistance association between miR638 with the overall survival of breast cancer patients. miR638 plays a role in cisplatin resistance mechanism through a mechanism involving its target genes CCND1 and FZD7. This present study also identifies miR638 and its target genes (CCND1 and FZD7) as a key gene and the potential biomarker of cisplatin resistance in TNBC. However, fur ther in vitro and in vivo validation is needed to develop the target gene as a biomarker.

Authors' contributions
AH-conception and design of the study, acquisition, analysis and interpretation of data, drafting and revising the article and final approval of the version to be pub lished, HP-acquisition and analysis of data, drafting the article and final approval of the version to be published. All authors read and approved the final version of the manuscript.