Down-regulation of miRNA-148a and miRNA-625-3p in colorectal cancer is associated with tumor budding

Background MiRNAs are often deregulated in colorectal cancer and might function as tumor suppressors or as oncogenes. They participate in controlling key signaling pathways involved in proliferation, invasion and apoptosis and may serve as prognostic and predictive markers. In this study we aimed to evaluate the role of miRNA-148a and miRNA-625-3p in metastatic colorectal cancer. Methods Fifty-four patients with a first-time diagnosed CRC receiving FOLFOX ± Bevacizumab were involved in the study. Tumor samples underwent routine pathology examination including evaluation for tumor budding and KRAS. MiRNA-148a and miRNA-625-3p expression analysis was done by RT-PCR. Associations between expression of both miRNAs and clinico-pathological factors, treatment outcomes and survival were analyzed. Results Both miRNA-148a and miRNA-625-3p were down-regulated in the tumors compared to normal colonic mucosa. Significantly lower expression of both miRNAs was noticed in tumors with budding phenomenon compared to tumors without it (median values of miRNA-148a were 0.314 and 0.753 respectively, p = 0.011, and 0.404 and 0.620 respectively for miRNA-625-3p, p = 0.036). Significantly lower expression of miRNA-625-3p was detected in rectal tumors, compared to tumors in the colon (median 0.390 and 0.665 respectively, p = 0.037). Progression free survival was significantly lower in patients with high miRNA-148a expression (6 and 9 months respectively, p = 0.033), but there were no significant differences in PFS for miRNA-625-3p and in overall survival for both miRNAs. Conclusions There was a significant relationship between low miRNA-148a and miRNA-625-3p expression and tumor budding, which is thought to represent epithelial-mesenchymal transition. Both studied miRNAs may be associated with a more aggressive phenotype and could be the potential prognostic and predictive biomarkers in CRC. Further investigation is needed to confirm miRNAs involvement in EMT, and their prognostic and predictive value. Electronic supplementary material The online version of this article (10.1186/s12885-017-3575-z) contains supplementary material, which is available to authorized users.


Background
Colorectal cancer (CRC) is one of the most frequently occurring cancers worldwide, accounting for 1.2 million new diagnoses and 600.000 deaths every year [1]. There is a tendency of decreasing morbidity and mortality, which could be associated with the development of new effective anticancer agents and cancer prophylactics.
Identification of novel biomarkers, based on molecular changes caused by the disease, could help to improve those results. In combination with pathological and clinical evaluation, recent biomarker investigations involve genomic, proteomic and transcriptomic research.
Currently, prognosis and treatment choice is based mainly on the information regarding tumor stage. Only a few tumor markers (RAS and BRAF mutation, mismatch repair status) have been implemented into clinical practice [2,3]. Recent evidence suggests, that tumor budding, described as the presence of individual cells and small clusters of tumor cells at the invasive front of the carcinoma, could serve as an additional prognostic factor in stage II CRC [4,5], but there is little information regarding its significance in the advanced disease. The down-regulation of epithelial and the up-regulation of mesenchymal markers suggests that tumor budding is a morphological expression of epithelial mesenchymal transition (EMT), which is associated with tumor invasiveness, formation of metastases and resistance to chemotherapy [6,7].
MicroRNAs (miRNAs) are a class of small (18-25 nucleotides in length), single-stranded noncoding RNAs. They negatively regulate the expression of target genes by binding to 3'UTR segments of messenger RNAs and are aberrantly expressed in most human cancers, including CRC, in which they may function as tumor suppressors or as oncogenes [8]. This posttranscriptional regulation plays a crucial role in controlling key signaling pathways of CRC (RAS/RAF/MAPK, AKT/MTOR and others), though miRNAs are actively involved in proliferation, invasion, angiogenesis, apoptosis and other biological processes [8][9][10].
We have focused on miRNA-148a and miRNA-625-3p because of the growing evidence of their importance in carcinogenesis, invasion and progression of CRC and because both of these miRNAs are associated with resistance to oxaliplatin based regimens [17][18][19]. Recent data shows that these miRNAs potentially regulate EMT through their targets such as MET/SNAIL, WNT signaling pathways, E-cadherin, N-cadherin, fibronectin and others [20][21][22].

Patients
Fifty-four patients with first-time diagnosed metastatic colorectal cancer participated in a prospective observational study conducted in the National Cancer Institute (Lithuania) in 2011-2014. The median age was 63 years (range 44-76). The male-female gender ratio of the group was 52% and 48%, respectively. 64% of the tumors were located in the colon and 36% in the rectum. The histological tumor type in 80% of the cases was adenocarcinoma, and in 20%mucinous adenocarcinoma.
98% of the tumors were medium grade. All the patients had metastases in the liver, and for 38% of the patients it was the only site where metastases were detected. In 91% of the cases, the primary tumor was removed before starting chemotherapy. 25% of the patients had undergone resection or radiofrequency ablation of liver metastases and it was performed after 2 months of chemotherapy. The patients received FOLFOX4 chemotherapy with or without Bevacizumab until disease progression or an unacceptable toxicity, according to the institutional guidelines. The median number of chemotherapy cycles was 8. Bevacizumab was administered to 53% of the patients. Treatment efficacy was evaluated every 2 months by a CT (computer tomography) scan according to RECIST 1.1 criteria. After completing the treatment, patients were followed up for progression or survival every 3 months.
The study has been approved by the Regional Biomedical Research Ethics Committee and performed in accordance with the Helsinki declaration. All patients signed the informed consent form before entering the study. Tumor samples were analyzed in the National Pathology Center (Lithuania) and the Scientific Research Center of the National Cancer Institute (Lithuania), and testers were blinded to treatment allocation and outcomes.

Pathology examination
A routine pathology examination, including KRAS testing, was performed. Additionally, samples were evaluated for tumor budding. All the pathology testing was performed by blinded pathologist before starting the treatment under study.

Tumor budding examination
Optimal tumor block for tumor budding evaluation was selected after reviewing H&E stained slides. Then H&E and pan-cytokeratin AE1/AE3 stained sections were assessed using the 10-HPF method. Areas of highest budding density were identified at low power field (×4-×10) and then tumor buds were counted under high magnification (×40). The cut-off value for determining a high degree of budding was considered 10 budding cells per high power field [23].

MiRNA extraction and analysis
Formalin-fixed paraffin embedded tissue samples of primary tumors (ensuring the presence of at least 50% of tumour cells in the tumor) and adjacent normal mucosa were prepared by the pathologist. 4-5 sections of 5 μm thickness were obtained and processed for MiRNAanalysis. MiRNA from the samples was extracted using "miRNeasy FFPE Kit" (QIAGEN, Germany). The concentration of miRNA was measured by "NanoDrop2000C" spectrophotometer (Thermo Fisher Scientific,USA). MiRNA expression analysis was done by the RT-PCR method using "TaqMan® MicroRNA Reverse Transkription Kit" (Life Technologies, USA) and "TaqMan® Universal Master Mix II, no UNG" (Life technologies, USA). Each sample was examined in triplicate. Changes in miRNA expression were calculated using the 2 -ΔΔCt method.

Bioinformatics
In silico miRNR target analysis was performed using microT-CDS algorithm and miRTarBase to extract theoretically predicted and experimentally validated targetgenes, respectively [24,25]. KEGG pathway enrichment analysis of obtained miRNA targets was completed using WebGestalt online tool kit (http://www.webgestalt.org/option.php) as described previously [26]. P values were evaluated using Hypergeometric distribution and adjusted with Benjamini and Hochberg posthoc test. Genes involved in epithelial-mesenchymal transition (EMT) were retrieved form dbEMT database. Genes potentially regulated by miR-148a were further visualized using Cytoscape ver. 3.4.0.

Statistical analysis
Descriptive statistics were used to describe demographic characteristics. The non-parametric Wilcoxon test was used to evaluate the differences between the two data sets. The differences between the two qualitative data groups were evaluated by the Chi-square test. Risk factors for PFS and OS were assessed by Cox regression analysis. Survival trends were evaluated by the Kaplan-Meier method. A log-rank test was used to evaluate the difference between Kaplan-Meier curves. Progression free survival (PFS) was calculated as the timespan from the first day of treatment to the first date of disease progression, the day of a confirmed new tumor or death. Overall survival (OS) was calculated as the time from the first day of treatment to death. If during the last visit to the clinician there was no evidence of disease progression or a new tumor, the date was confirmed as censored. A two-tailed p-value less than 0.05 was considered to be significant. Statistical analysis was performed using the Statistical Analysis System (SAS) package version 9.2.

Results
Both miRNA-148a and miRNA-625-3p were downregulated in the tumors compared to normal adjacent mucosa. The median values of miRNA-148a were 0.527 (range 0.005-4.485) and 2.94 (range 1.826-4.091) respectively, p < 0.001, and for miRNA-625-3p -0.492 (range 0.021-4.289) and 5.584 (4.595-6.314) respectively, p < 0.001 (Fig. 1). The patients were divided into high and low expression groups based on the median value. There were no significant differences in the groups according to the age, gender, number and location of metastases or KRAS mutations (Table 1). A significantly higher number of tumors in the colon had a higher expression of miRNA-625-3p (p = 0.024), but there were no association between miRNA-148a expression and tumor location. There was a trend toward lower miRNA-148a and miRNA-625-3p expression in tumors with a budding phenomenon (both p = 0.051).
Further analysis revealed a significantly lower expression of both miRNAs in tumors with budding phenomenon compared to tumors without it (median values of miRNA-148awere 0.314 and 0.753 respectively, p = 0.011; and of miRNA-625-3p were 0.404 and 0.620 respectively, p = 0.0360 ( Fig. 2a and b). A lower expression of both miRNAs was also noticed in tumors of the rectum, compared to tumors in the colon (median values of miRNA-148a were 0.412 and 0.624 respectively, p = 0.098; and of miRNA-625-3p were 0.390 and 0.665 respectively, p = 0.037). The difference was significant only for miRNA-625-3p ( Fig. 2c and d).
The Univariate Cox regression analysis revealed that a higher expression of miRNA-148a was associated with a shorter PFS. No other significant factors were found in this analysis.
Inorder to elucidate the biological processes which could be potentially regulated by miR-148a and miR-625-3p, miRNA target analysis was performed. The analysis identified 502 and 98 genes theoretically predicted as miR-148a and miR-625-3p targets (Fig. 4a). However, only 11 of them were determined as regulated in common between the semiRNAs (Fig. 4b). In addition, the analysis revealed that 29 target genes were associated with miR-148a with high experimental confidence. Further, to investigate any relations between miR-148a/miR-625-3p and epithelial-mesenchymal transition (EMT), the list of all known EMT genes retrieved from dbEMT was compared with the list of obtained target genes. Venn diagram analysis indicated that 13 and 6 EMT genes could be potentially regulated by miR-148a and miR-625-3p, respectively (Fig. 4c). Interestingly, among them key players of EMT ZEB1 and TGFB2 were predicted targets of miR-148a, and TGFBR1 -miR-625-3p. The list of EMT genes potentially regulated by these two miRNAs is depicted in Additional file 1: Table S1. Next, KEGG pathway enrichment analysis indicated 7 KEGG categories enrichedin miR-148a target-genes (Additional file 2: Table S2). Among them FoxO (p = 0.012) and PI3K-Akt (p = 0.0187) signaling pathways were the most significantly enriched functional categories. Inaddition, the pathway enrichment data revealed that categories associated with cell adhession including focal adhesion, regulation of actin cytoskeleton and ECM-receptor interaction were significantly enriched in miR-148a targetgenes. miR-148a and target gene networks related to FoxO/PI3K-Akt (27 target-genes) and cell adhesion (21 target-genes) functional groups are visualized in Fig. 4d. However, KEGG analysis did not obtain any significant pathway enrichment of miR-625-3p target-genes (data is not shonwn).

Discussion
Multiple data affirm that miRNAs are involved in mechanisms of excessive growth, resistance to apoptosis, angiogenesis, invasion, and metastasis [27,28]. In this study, we have investigated miRNA-148a and miRNA-625-3p expression, due to the growing evidence of their importance in carcinogenesis and progression of CRC as well as in prediction of treatment efficacy.
MiRNA-148a down-regulation has been detected in gastric, breast, non-small cell lung cancer and other tumors [20,29,30] compared to normal tissues. Several trials reported down-regulation of miRNA-148a in CRC cells and it's association with more advanced disease and poor prognosis [18,[31][32][33].
H. L. Tsai et al. [32] found a relationship between early relapse of CRC patients in the II and III stages that underwent curative resection and low miRNA-148a expression in the tumor (OR 5.221; 95% CI, 2.068-13.174, p < 0.0001) and serum (p = 0.045). Lower miRNA-148a expression in the tumor was also associated with shorter DFS (p = 0.0006) and a worse OS (p = 0.0156). Investigators demonstrated that miRNA-148a inhibits cell proliferation (possibly through the mechanism of arresting the cell cycle during phase 2) and migration.
M. Takahashi et al. [18] reported that low miRNA-148a expression is associated with advanced CRC stages (III/IV versus normal mucosa, p < 0.0001) and is an independent prognostic factor of PFS for stage III patients (HR 1.83, 95% CI 1.12-2.99, p = 0.017) and OS for stage IV patients (HR 1.93, 95% CI 1.15-3.23, p = 0.014). In addition, they reported that miRNA-148a expression may be regulated by promoter methylation, and that there is an independent association between miRNA-148a methylation and poor survival of stage IV CRC patients.
H. Huang et al. [34] reported that MiRNA-148a negatively regulates the expression of MMP7, which is involved in tumor cell invasion and associated with advanced stages and poor clinical outcome of CRC.
H. Zhang et al. [35], however, did not show any significant differences between miRNA-148a expression in tumor and normal tissue, as well as no correlation with patients' clinical and pathological characteristics. It was confirmed that BCL-2 is a target of miRNA-148a, revealing its proapoptotic potential.
Also there is evidence regarding the importance in cancer development of other targets of miRNA-148a, such as DNMT1inhibition of which increases expression of multiple genes by epigenetic changes (reduced methylation) [36]; ERBB3involved in angiogenesis [37]; CDC25Bcell cycle regulator [38]; and IGF-IR, AKT, MAPK/ERKcell growth [39]. By silencing these and potentially other target genes, miRNA-148a inhibits cell growth, proliferation, angiogenesis, invasion, andsusceptibility to apoptosis.
In the present study, we determined the deregulation of miRNA-148a in tumor tissue, compared to normal colonic mucosa. There was a trend toward higher miRNA-148a expression in mucinous adenocarcinoma compared to adenocarcinoma (p = 0.068), but there were no associations with other major clinical and pathological features (such as age, gender, tumor location and size, lymph node involvement, number or site of metastases, KRAS mutation status). These findings correspond to the results of the H. Zhang study [35], but in both studies, the results might have been influenced by the small sample size.
In our study, miRNA-148a expression was lower in cancers with tumor budding (median values accordingly 0.314 and 0.753, p = 0.011), which has been found by other studies to be associated with a more invasive phenotype and distant metastases [40] and determined as prognostic factor independent from CRC stage [3,4].
Tumor budding indicates the presence of individual cells and small clusters of tumor cells at the invasive front of the carcinoma and is thought to represent epithelial-mesenchymal transition (EMT) [3,4]. EMT is mainly regulated by SNAIL, TWIST and ZEB transcription factors and characterized by down-regulation of epithelial markers (E-cadherin, claudin etc.) and upregulation of mesenchymal markers (Fibronectin, Ncadherin, MMPs etc.) [40]. The correlation between EMT and tumor budding is based on similarities in characteristicslow membranous level of E-cadherin [41], upregulation of fibronectin and vimetin [42], activation of TGF-β [43] and other pathways [7].
Studies demonstrated that miRNA-148a suppresses EMT and the expression of several genes overexpressed in EMT [20,21]. J.P. Zhang et al. [21] reported that miR-148a may promote the expression of an epithelial marker (E-cadherin), reduce the levels of mesenchymal markers (N-cadherin, fibronectin or vimentin), and negatively regulate MET/SNAIL signaling and in turn inhibit the EMT and metastasis of hepatoma cells. Z. Qin et al. [22] recently suggested that miR-148a, miR-505 and miR-1207-5p might be induced by growth MiRNA-625-3p is another miRNAwith growing evidence of involvement in development and progression of different types of cancer [44][45][46][47]. M. Wang et al. [45] have reported that miRNA-625-3p silences the expression of integrin-linked kinase (ILK), the activation of which results in oncogenic transformation to invasive and metastatic phenotypes. In concordance with this, an inverse relationship has been established between downregulated miRNA-625-3p expression in gastric cancer and lymph node metastasis. In recent studies, downregulation of miRNA-625-3p was associated with more advanced Tstages and metastases of esophageal cancer. In addition, it was demonstrated that miRNA-625-3p negatively regulates Sox2, which promotes tumor invasion and metastasis by stimulating EMT via regulation of WNT/β-catenin, and tumor growth via AKT/MTOR signaling pathways [45,47].
In our study, we found a significantly lower miRNA-625-3p expression in tumors with budding phenomenon (0.404 and 0.620 respectfully, p = 0.036). This finding indicates potential miRNA-625-3p involvement in EMT, which is also supported by earlier reported data about this miRNAs role in WNT/β-catenin and SCAI/E-cadherin/MMP9 pathways.
Taking in to account the wide spectrum of miRNA targets, they have been actively investigated as potential predictive markers. One of the main chemotherapy agents for treating CRCoxaliplatin -induces cell apoptosis by causing DNA interstrand cross-links, [49]. Bcl2 as target of miRNA-148a provides the rationale for using miRNA-148aas a predictive marker of the response to oxaliplatin. M. Takahashi et al. [18] reported a significant association between low miRNA-148a expression and a worse response to oxaliplatin-based therapy (p = 0.006), a worse OS (median 16.1 vs. 25.6 month, p = 0.024), and a trend toward a worse PFS (median 8.1 vs. 10.1 month, p = 0.16) in metastatic CRC patients. In contrast, J.B. Kjersem et al. [11] reported an association between high miRNA-148a expression in the plasma of metastatic CRC patients receiving oxaliplatin-based chemotherapy and a decrease in PFS (HR 1.3; 95% CI 1.1-1.6, p = 0.007). M. H. Rasmussen et al. [19] have reported that high miRNA-625-3p expression is a negative predictive factor for response to first line oxaliplatin-based chemotherapy in metastatic CRC (OR 6.25, 95% CI 1.8-21.0). Current evidence suggests that the investigation of both miRNAs mentioned, could be beneficial in predicting the efficacy of oxaliplatin. Our results did not confirm this statementneither miRNA-148a expression nor miRNA-625-3p expression was associated with response rates in studied patients. High miRNA-148a expression was associated with worse PFS (6 month and 9 month respectively, p = 0.033), but not OS (18 month and 24 month respectively, p = 0.199). MiRNA-625-3p did not influence PFS or OS. Small sample size of the study limited the statistical power to identify differences in PFS and OS.
To our knowledge, our study is the first that determined an association between low miRNA-148a and miRNA-625-3p expression and tumor budding. The small sample size is the major limitation of our study and our findings need to be validated in an independent cohort. However, our results allow us to propose an interesting hypothesis regarding the possible role of the tested RNAs in the tumor budding process which is likely associated with EMT.
Our bioinformatics analysis revealed that 19 EMT related genes including key regulators of EMT -ZEB1, TGFB2 and TGFBR1are potential target genesof miRNA-148a and miRNA-625-3p, supporting the roles of these miRNAs as regulators of EMT. In addition, our results are consistent with published data, regarding the role of miRNA-148a and miRNA-625-3p in EMT regulation.

Conclusions
The presented study has identified a significant relationship between low miRNA-148a and miRNA-625-3p expression and tumor budding, which is thought to represent EMT. The results show that the studied miR-NAs may be associated with a more aggressive phenotype and could be the potential prognostic and predictive biomarkers in CRC. Further investigation is needed to confirm the involvement these miRNAs in EMT, and their prognostic and predictive value.

Additional files
Additional file 1: Table S1. Full list of genes associated with EMT which could be potentially regulated by miR-148a or miR-625-3p miRNAs. (DOCX 11 kb) Additional file 2: Table S2. Funding National Cancer Institute covered the cost of study organization and article publication.

Availability of data and materials
Clinical data of study participants obtained in this study except the personal data is available on request addressed to corresponding author.
Authors' contributions EB prepared study design, performed all the linical part of the study, assisted in performing pathology and miRNA examination, collected and analyzed the data and drafted the manuscript. DS performed miRNA analysis, discussed the data of experiments and drafted the manuscript. VS performed bioinformatics analysis of potential miRNAs functions in cancer and potential miRNAs target genes and drafed the manuscript. UM permormed pathology analysis. TZ performed statistical analysis. EA contributed to the conception of the idea, helped to prepare study design, discussed the data of experiments and critically reviewed the final manuscript part. JB discussed the data of experiments and drafted the manuscript. KS contributed to the conception of the idea, miRNA analysis, discussed the data of experiments and critically reviewed the final manuscript part. All authors contributed to experimental work and preparation of the manuscript, including the final one. All authors read and approved the final manuscript.

Ethics approval and consent to participate
The study (No. 158200-06-347-88) has been approved by Vilnius Regional Biomedical Research Ethics Committee (Vilnius). All participants of the study have signed the informed consent to participate before study specific procedures have been started.

Consent for publication
Not applicable.

Competing interests
The authors declare that they have no competing interests.

Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.