MicroRNA let-7g acts as tumor suppressor and predictive biomarker for chemoresistance in human epithelial ovarian cancer

Remarkable deregulation of microRNAs has been demonstrated in epithelial ovarian cancer (EOC). In particular, some of the let-7 miRNA family members have been proposed as tumor suppressors. Here, we explored the functional roles of let-7g in EOC. The ectopic overexpression of let-7g in OVCAR3 and HEY-A8 EOC cells induced i) a down-regulation of c-Myc and cyclin-D2 thus promoting cell cycle arrest, ii) a reduction of Vimentin, Snail and Slug thus counteracting the progression of epithelial to mesenchymal transition, iii) a chemosensitization to cis-platinum treatment. Next, analysis of human EOC tissues revealed that let-7g expression was significantly reduced in tumor tissue specimens of patients with EOC compared to their non-tumor counterparts (p = 0.0002). Notably, low let-7g tissue levels were significantly associated with acquired chemoresistance of patients with late-stage of EOC (n = 17, p = 0.03194). This finding was further validated in the serum samples collected from the same cohort of patients (n = 17, p = 0.003). To conclude, we demonstrate that let-7g acts as tumor suppressor and might be used to disable EOC tumor progression and chemoresistance to cis-platinum-based chemotherapy. Furthermore, we propose that decreased expression of let-7g could serve as a tissue and serum biomarker able to predict the chemo-resistant features of EOC patients.

development of chemoresistance, by acting either as oncomiR or tumor-suppressor miRNAs 7,8 . It has been widely demonstrated that the deregulation of microRNAs, both in tissue and blood, is a hallmark of each cancer cell type [8][9][10][11] . In blood, miRNAs are highly stable since they are packed in exosomes or vesicles, or bound to proteins/lipoproteins, thus representing a reliable tool to evaluate cancer development and to monitor cancer progression 8-11 . In ovarian cancer, different miRNAs have been found aberrantly expressed [12][13][14] . miR-200 family members are down-regulated and play a critical role in the control of epithelial-to-mesenchymal transition (EMT); miR-34a, one of the best described p53-regulated miRNA, contributes to tumor suppression by inhibiting cellular proliferation and survival; down-regulation of let-7a, let-7e, let-7f, some of let-7 family members, is associated with aggressive behaviour of tumor and represents potential markers of invasion and metastasis in EOC [12][13][14] .
The let-7 miRNA family is composed of 13 members, located on 9 different chromosomes, with overlapping or distinct functions 15,16 . Since let-7 expression is reduced in almost all human cancers and this reduction is correlated with poor prognosis, they are largely described as tumor suppressors [17][18][19][20][21][22][23][24][25] . Let-7, indeed, negatively regulate proteins with oncogenic potential such as RAS 17,18,20 , HMGA2 [20][21][22][23][24] , c-Myc 17,18,20 , cyclin-D2 17,18 thus repressing cancer development, differentiation and progression. However, it has been reported that at least some of the let-7 family members may act as oncogenes. Among others, Brueckner et al. showed that overexpression of let-7a is associated with enhanced tumor phenotype in human cancers 26 . Aberrant expression of let-7 members has been also correlated to the acquisition of resistance to chemotherapeutic drugs [27][28][29][30] . In 2008, a discrimination of EOC patients in completely responding or non completely responding to cis-platinum-based chemotherapy has been proposed based on let-7i expression profile; however, the let-7i targets responsible for this effect have not been discovered 27 . More recently, Xiao M et al. showed that let-7e improves the response of EOC cells exposed to cis-platinum through the modulation of DNA double strand break repair mechanisms 29 . Therefore, the extremely variable behaviour of the different let-7 family members in tumors originating from different tissues prompts to better characterize the function of individual let-7 microRNAs in specific cancer types.
In this study, we demonstrated that let-7g acts as tumor suppressor in epithelial ovarian cancer. Indeed, ectopic overexpression of let-7g in OVCAR3 and HEY-A8 EOC cells induced cell cycle arrest, slowed down the progression of EMT and significantly improved cell response to cis-platinum treatment. Furthermore, we showed that let-7g amounts were significantly reduced in both tissue and plasma samples of a cohort of chemotherapy-resistant high grade serous ovarian cancer (HGSC) patients.
Overall, our results indicate that let-7g is a potential biomarker to predict chemotherapy response in ovarian cancer patients and suggest that the administration of let-7g mimics, alone or in combination with other chemotherapies, might disable tumor progression.
Let-7g suppressess EMT and migration ability in OVCAR3 and HEY-A8 cells. The metastatic potential of OVCAR3 and HEY-A8 cells, in response to the overexpression of let-7g, was investigated in terms of EMT and migration ability. As shown by immunofluorescence assays in Fig. 4a,b, the transient transfection of let-7g mimic for 12 h is accompanied by the decrease of the mesenchymal marker Vimentin and the EMT-related transcription factors Snail and Slug in OVCAR3 let-7g mimic and HEY-A8 let-7g mimic compared to their relative control cells. These data were corroborated by the Western Blot analysis of the same EMT markers in Fig. 5a. The migration ability was next assessed by a triplicate set of independent wound healing assays. A representative image with www.nature.com/scientificreports www.nature.com/scientificreports/ relative densitometry (Fig. 5b) shows that OVCAR3 let-7g mimic and HEY-A8 let-7g mimic cells possess, at 12 h, a reduced migratory capability consistent with a slow down of epithelial to mesenchymal transition.
Enhanced let-7g expression increases the sensitivity of EOC cells to cis-platinum treatment. Taken all together the above results strongly suggest that let-7g acts as tumor suppressor in the EOC cell lines. Thus, to explore whether this function could also affect the response to chemotherapy, OVCAR3 let-7g mimic and HEY-A8 let-7g mimic cells as well as their relative controls were exposed to serial concentrations of cis-platinum (OVCAR3: 6 µM, 12 µM, 25 µM, 50 µM; HEY-A8: 100 µM, 250 µM, 500 µM, 1000 µM. The overexpression of the let-7g significantly increased cell sensitivity to cis-platinum treatment in both the EOC cell lines for the majority of drug concentrations (cis-platinum concentration are expressed as log [µM]). Indeed, as shown in Fig. 6a, OVCAR3 let-7g mimic exhibited a log EC 50 of 0.85 ± 0.00037 compared to OVCAR3 Negative Control which had a log EC 50 of 1.12 ± 0.039 (p < 0.001) and compared to OVCAR3 WT which had a log EC 50 of 1.149 ± 0.036, (p < 0.0001); similarly, as shown in Fig. 6b, HEY-A8 let-7g mimic exhibited a log EC 50 of 2.497 ± 0.0367 compared to HEY-A8 Negative Control which had a log EC 50 of 2.761 ± 0.0373 (p < 0.001) and compared to HEY-A8 WT which had a log EC 50 of 2.764 ± 0.0466 (p < 0.001). EC 50 were calculated by using GraphPad Prism ® version 5.01 and analyzed by Sidak test.
Let-7g tissue levels correlate with chemoresistance in advanced EOC patients. To further validate the tumor suppressive role of let-7g in EOC, the microRNA levels in tumor tissues obtained from 17 late-stage patients (stage III and IV) with well-documented clinical and oncological data were analyzed. At 5 years, the overall survival (OS) was 45.5% and the disease-free survival (DFS) 58.8%; the recurrence rate was 52.9%. Among the 17 patients, 9 were defined as resistant while 8 as sensitive to chemotherapy. Patient characteristics are listed in Table 1.
First, absolute TaqMan analysis of let-7g expression levels was performed in EOC tumor tissues (n = 10) and matched adjacent non-tumor tissues when available (n = 10) (see representative immunoistochemical images in Fig. 7a). As shown in the box plot in Fig. 5b, statistical Kruskal-Wallis test highlighted that let-7g was significantly down-regulated in tumor tissues compared to their non-tumor counterparts (p = 0.0002). We also observed that, within the tumor tissue specimens, let-7g amount was significantly lower in the patients with resistance to chemotherapy compared to chemo-sensitive ones (p = 0.03194) (Fig. 7c). Receiver operating characteristics (ROC) curve analysis was performed to evaluate the accuracy of tissue let-7g to distinguish chemoresistant EOC patients from www.nature.com/scientificreports www.nature.com/scientificreports/ those who were sensitive to chemotherapy. An area under the curve (AUC) of 0.931 had a high discriminatory power (95% CI: 0.792-1.000, p < 0.05) (Fig. 7d).
Low levels of serum let-7g is associated with chemoresistance in patients with advanced EOC. Driven by the above results, we further examined let-7g expression levels in serum samples collected from the same cohort of patients at the time of diagnosis. The correlation analysis between the level of serum let-7g and the response to chemotherapy highlighted that let-7g serum amount was significantly lower in patients with resistance to chemotherapy compared to chemo-sensitive ones (p = 0.003) (left in Fig. 8). ROC curve analysis showed that AUC of 0.799 had a high discriminatory power (95% CI: 0.563-1.000, p < 0.05) (right in Fig. 8).

Discussion
Ovarian cancer usually presents in advanced stages with a bad prognosis and a high mortality rate [1][2][3][4] . Primary debulking surgery has been the standard of care in this disease for a long period. More recently, the administration of a platinum/taxol-based neoadjuvant chemotherapy followed by surgical cytoreduction has proven to be a more promising treatment strategy for the management of advanced EOC patients 31 . However, ovarian cancer is one of the most referred cancer types developing drug resistance that, therefore, dramatically reduces the www.nature.com/scientificreports www.nature.com/scientificreports/ neoadjuvant chemotherapy benefits [1][2][3][4] . The identification of reliable biomarkers enabling clinicians to distinguish the chemoresistant from chemosensitive subjects, among stage III and IV EOC patients, is one of the main goals of translational medicine.
In the recent years, the enormous efforts in the molecular characterization of EOC pathogenesis have led to the identification of a subset of miRNAs which might represent novel strategies for early detection, diagnosis, and treatment of this disease [12][13][14] , although for most of them the prospective validation is still pending.
MicroRNAs Let-7 family, that insist on multiple cellular pathways mainly connected with neoplastic transformation, are frequently found down-regulated or even lost in many cancers, hence the definition of tumor suppressor miRNAs [15][16][17] . Let-7 microRNAs belong to a family composed by 14 highly homologous members, whose expression is often co-regulated. However, it has been demonstrated that the relationship between let-7 family www.nature.com/scientificreports www.nature.com/scientificreports/ and neoplastic transformation is a complex phenomenon in which individual members may either show a different expression level within the same cancer cell either play distinct roles in different types of malignancies [15][16][17] .
In this work, we investigated the functional role of a single member of let-7 family, let-7g, on epithelial ovarian cancer. Taken all together experimental assays performed on OVCAR3 and HEY-A8 EOC cell lines identify microRNA let-7g as a tumor suppressor able to consistently disable many of the cancer hallmarks. First, we found that let-7g overexpression induced a significant reduction of cancer cell growth. This effect was associated with a partial arrest in G 0 /G 1 cell cycle phase in both OVCAR3 and HEY-A8 cells. It has been widely reported that let-7 microRNA family universally interferes with cell growth through directly targeting several oncogenic proteins involved in cell cycle progression including c-Myc, CDC25A, CDK6 and cyclin D2 16,17,32 . Consistent with these data, we found that the overexpression of let-7g remarkably down-regulated c-Myc and cyclin-D2 in OVCAR3 and HEY-A8 cells. The role of let-7 family in the regulation of programmed cell death appears, instead, context-dependent due to its ability to act on different target genes. Geng, L et al. showed that in human colorectal cancer cells let-7 inhibits apoptosis by decreasing the expression of Fas 33 while Zhang, H et al. demonstrated, in CRC cell lines, that let-7 promotes apoptosis by targeting some members of anti-apoptotic Bcl-2 protein family 34 . In our experimental model, let-7g overexpression differently affects apoptosis in the two cell lines: in OVCAR3 cells, the miRNA promoted an increase in the apoptotic rate due to its preferential activity on Bcl-2 while in the apoptotic resistant HEY-A8 cells let-7g mimic preferentially targeted FAS, decreasing its amount. These data provide further confirmation of the context-specific activity of let-7 also in ovarian cancer cell lines.
Next, we demonstrated that in EOC cell lines let-7g overexpression cut back the epithelial to mesenchymal transition (EMT) process and reduced the cell motility. These activities were accompanied by a break-down of the mesenchymal marker Vimentin and by a diminished expression of the two EMT-related transcription factors Snail and Slug. The identification of let-7g as suppressor of EMT in ovarian cancer raises some interesting questions. To the best of our knowledge none of the three proteins has been described as direct target of let-7 family. Hence, we performed target prediction analyses by PicTar, miRTar and TargetScan bioinformatic prediction softwares that showed the lack of any regions of complementarity between Vimentin, Snail and Slug mRNAs and let-7g (data not shown), thus suggesting the existence of indirect regulatory mechanisms. Interestingly, let-7g www.nature.com/scientificreports www.nature.com/scientificreports/ overexpression in OVCAR3 and HEY-A8 cells was accompanied by the up-regulation of other miRNAs involved in EMT control and ovarian cancer pathogenesis, namely let-7b, -e and -i and miR-200b (see Supplementary  Fig. S1) 35,36 . The dissection of the metabolic routes connecting let-7g with this repertoire of EMT-related miRNAs require further and deeper future studies.
The tumor suppressive effects of let-7g in EOC cell lines is further confirmed by the enhanced sensitization of both OVCAR3 and HEY-A8 cells exposed to cis-platinum treatment. In ovarian cancer, c-Myc has been reported as central hub in the cis-platinum resistance, so that its specific targeting has been proposed as possible strategy to overcome chemoresistance 37 . The diminished expression of c-Myc upon let-7g mimic transfection in both cell types might constitute one of the molecular ground of this phenomenon.
Our in vitro data, indicating the tumor suppressive role of let-7g in EOC, were supported by the analysis of let-7g expression levels ex vivo. Recently, Yang N et al. have reported let-7i as an important miRNA involved in the chemotherapy response of EOC patients, thus proposing it as putative prognostic biomarker 27 . Our data support and expand these results; in fact, although the small cohort size of 17 EOC patients, we show that another member of let-7 family, let-7g, is able to discriminate, both in tissue specimens and serum samples, patients with risk to develop resistance from those who completely respond to platinum-taxane chemotherapy.
Overall, this study demonstrates that let-7g acts as tumor suppressor by reducing epithelial ovarian cancer cell aggressiveness. The in vitro analyses add Vimentin, Snail and Slug to the repertoire of the molecules involved in let-7g downstream networks. Furthermore, the here presented ex vivo analyses, albeit needing to be confirmed in larger patient cohorts in future studies, suggest that tissue let-7g detection and potentially the quantification of its circulating levels, might be useful to guide decision making on usage of neo-adjuvant chemotherapy in patients with advanced epithelial ovarian cancer.   www.nature.com/scientificreports www.nature.com/scientificreports/ Patients and specimens. We selected a group of 17 patients with High Grade Serous Ovarian Cancer (HGSC) who were treated at the Unit of Gynaecologic Oncology, Magna Graecia University, Germaneto, and Pugliese-Ciaccio Hospital, Catanzaro, Italy, between April 2013 and March 2016. Data were retrieved from the charts, collected and tabulated. Tissue and serum samples of patients were retrieved from our bio-bank to perform analysis of miRNA let-7g expression.
Procedures carried out in this study were in accordance with the guidelines of the Helsinki Declaration on human experimentation and good clinical practice (CGP). Approval by the "Pugliese-Ciaccio" institutional review board (IRB number: AOPC12404) was obtained before starting patient's enrollment. Furthermore, an informed consent was obtained from all patients before processing their data from the time of hospitalization, even if data did not include any personal identifying information. Inclusion criteria were as follow: availability of clinical data and biological samples; stage IIIc-IV HGSC surgically staged. Patients with previous or concurrent cancer located in other sites, known genetic susceptibility to gynecologic or non-gynecologic cancers (BRCA1-2 carriers, associated polyposis conditions (APC), Fanconi syndrome) or positive family anamnesis for ovarian and/or breast cancer were excluded. Biological samples consist in surgical tissue and serum aliquots. Specifically, tumoral and healthy surgical samples consist in tissues fixed in 4% paraformaldehyde and subsequently embedded in paraffin.
Absolute TaqMan Analysis was also used to determine the expression of let-7g in the 17 tumor tissue specimens and serum samples. Briefly, starting from a sample of known template concentration, a 5-point 10 fold serial standard curve was prepared, and the concentration of all other samples were calculated by simple interpolation of each threshold cycle (Ct) into this standard curve. MiRNA expression data are reported as log quantity (ng) and represent the mean of three independent technical replicates.