Identification of novel microRNA targets based on microRNA signatures in bladder cancer



MicroRNAs (miRNAs) are small noncoding RNAs that negatively regulate protein-coding genes. To identify miRNAs that have a tumor suppressive function in bladder cancer (BC), 156 miRNAs were screened in 14 BCs, 5 normal bladder epithelium (NBE) samples and 3 BC cell lines. We identified a subset of 7 miRNAs (miR-145, miR-30a-3p, miR-133a, miR-133b, miR-195, miR-125b and miR-199a*) that were significantly downregulated in BCs. To confirm these results, 104 BCs and 31 NBEs were subjected to real-time RT-PCR-based experiments, and the expression levels of each miRNA were significantly downregulated in BCs (p < 0.0001 in all). Receiver-operating characteristic curve analysis revealed that the expression levels of these miRNAs had good sensitivity (>70%) and specificity (>75%) to distinguish BC from NBE. Our target search algorithm and gene-expression profiling in BCs (Kawakami et al., Oncol Rep 2006;16:521–31) revealed that Keratin7 (KRT7) mRNA was a common target of the downregulated miRNAs, and the mRNA expression levels of KRT7 were significantly higher in BCs than in NBEs (p = 0.0004). Spearman rank correlation analysis revealed significant inverse correlations between KRT7 mRNA expression and each downregulated miRNA (p < 0.0001 in all). Gain-of-function analysis revealed that KRT7 mRNA was significantly reduced by transfection of 3 miRNAs (miR-30-3p, miR-133a and miR-199a*) in the BC cell line (KK47). In addition, significant decreases in cell growth were observed after transfection of 3 miRNAs and si-KRT7 in KK47, suggesting that miR-30-3p, miR-133a and miR-199a* may have a tumor suppressive function through the mechanism underlying transcriptional repression of KRT7. © 2009 UICC

Bladder cancer (BC) is among the 5 most common malignancies worldwide, and it is the 2nd most common tumor of the genitourinary tract and the 2nd most common cause of death in patients with genitourinary tract malignancies.1, 2 Gene-expression profiling by microarray analysis is an excellent tool for screening candidate genes that have a tumor suppressive or oncogenic function in BC.3, 4 We previously identified that SKP2 and CKS1 contribute the progression and prognosis in BC by our microarray analysis.5, 6 However, the hundreds of candidate genes derived from microarray analysis may embarrass investigators to decide which gene should be studied. DNA hypermethylation is aberrant epigenetic event that negatively regulates the expression levels of tumor suppressor genes or oncogenes.7 Hence, it seems to be a good strategy to find cancer-related genes by sorting methylated genes. Another major epigenetic event is caused by microRNA (miRNA). miRNAs are small noncoding RNA gene products about 22-nucleotide (nt) long, which are cleaved from 70- to 100-nt hairpin-shaped precursors (pre-miRNA).8 Although their precise biology is not fully understood, miRNAs are found in diverse organisms and epigenetically function as negative regulators of gene expression through posttranslational mRNA degradation.8, 9 A more recent link between miRNA function and cancer pathogenesis is supported by studies examining the signatures of miRNA in clinical samples. miRNAs are aberrantly expressed in human cancer, indicating that they may have a novel oncogenic or tumor suppressive function.10 The first evidence of involvement of miRNAs in human cancer came from molecular studies characterizing the 13q14 deletion in human chronic lymphocytic leukemia, which revealed 2 miRNAs, miR-15a and miR-16-1.11 Subsequently, altered miRNA expression has been reported in lung cancer,12 breast cancer,13 glioblastoma,14 hepatocellular carcinoma,15 papillary thyroid carcinoma,16 colorectal cancer,17 pancreatic tumors18 and, most recently, bladder and kidney cancer.19 However, the target genes by these miRNAs were not fully elucidated especially in BC.

In our study, we determined the miRNA expression signatures specific to BC by evaluating 156 mature miRNAs expressions in 14 BCs, 5 normal bladder epithelium (NBE) specimens, and 3 BC cell lines. We identified a subset of 7 downregulated miRNAs that have a tumor suppressive function based on miRNA expression signatures of BC. Subsequently, we validated the expression levels of the miRNAs by using more than 100 clinical specimens. Our target search algorithm based on the gene-expression profiling in BCs5 indicated that Keratin7 (KRT7) transcript was a common target of the downregulated miRNAs. We transfected the downregulated miRNAs in a BC cell line (KK47) to examine their transcriptional repression of KRT7. Furthermore, we also examined the cell growth effect on KK47 cells transfected with the miRNAs to examine the functional roles of these miRNAs and KRT7 in BC development.

Material and methods

Clinical BC specimens and BC cells culture

Tissue specimens were from 104 BC patients who had undergone cystectomy, or transurethral resection of BC at Kagoshima University Hospital and 3 affiliated hospitals between 2003 and 2007. The patients' backgrounds and clinicopathological characteristics are summarized in Table I. Our study was approved by the Bioethics Committee of Kagoshima University; written prior informed consent and approval were given by these patients. These samples were staged according to the American Joint Committee on Cancer-Union Internationale Contre le Cancer tumor-node-metastasis classification and histologically graded.20 There was a significant difference in median age between BC and non-BC patients (76 vs. 70 years, respectively, p < 0.001). Of the 104 BCs, 14 BCs were randomly selected for the screening test of 156 miRNAs that were available in November 2006 (Table II). We could not use the samples with G1 grade for the screening test, because these samples were obtained from transurethral resection, and they were too small to extract enough total RNA for screening 156 miRNAs.

Table I. Patients' Characteristics
Bladder cancer (BC)
 Total number104
 Median age (range)76 (33–100) years
  Superficial (pTa)71
  Invasive (≥pT1)33
 Follow-up period (range)476 (5–1440) days
Normal bladder epithelium (NBE)
 Total number31
 Median age (range)70 (30–78)
Table II. Patients' Characteristics for miRNA Screening Test

The human BC cell lines (T24, KK47 and BOY21) were maintained in the recommended medium mixed with 10% fetal bovine serum, 50 μg/ml streptomycin and 50 U/ml penicillin in a humidified atmosphere of 95% air/5% CO2 at 37°C. Routine tests for mycoplasma infection were negative.

RNA isolation

Tissues and cells, freshly harvested and immediately frozen in liquid nitrogen and stored at −80°C until processing, were dissolved in ISOGEN (Nippon Gene, Tokyo, Japan); we followed the manufacturer's protocol for total RNA extraction. The concentrations of RNA were determined spectrophotometrically; integrity was checked by gel electrophoresis. The RNA quality was confirmed in an Agilent 2100 bioanalyzer (Agilent Technologies, Santa Clara, CA).

miRNA expression signature and data normalization

Each pooled sample in the 156 panel assay was analyzed in duplicate. Analysis of relative miRNA expression data was performed using GeneSpring GX version 7.3.1 software (Agilent Technologies) according to the manufacturer's instruction. We employed 2 different approaches for normalizing the miRNA expression data, global normalization and endogenous gene (ACTB) normalization. Briefly, the Ct values were transformed by the following formula: expression score = 2○ (40 − Ct), and the calculated data were uploaded onto GeneSpring software. We chose the following methods for global normalization: (i) Per Chip, normalize to 50th percentile and (ii) Per Gene, normalize to median; and for ACTB normalization: (i) Per Gene, normalize to median and (ii) Per Gene, normalize to positive control (ACTB) genes. To prepare a low-level filtered gene list, before analysis, we excluded miRNAs with Ct values of 35 cycles. The low-level filtered gene list was further processed to select only those genes with a fold change >2.0 or <0.5 in comparison with normal samples; these were considered as differential expressed genes. Furthermore, a Welch analysis of variance test (parametric test, with variances not assumed equal, cutoff p value of 0.001), using the Benjamini and Hochberg method to control the false discovery rate, was performed on the list of differentially expressed genes to finally generate a list of statistically differential expressed miRNAs. TaqMan® probe and primers for ACTB were assay-on-demand gene expression products (P/N: Hs99999903_m1, Applied).

The unsupervised clustering analysis was performed on the statistically differential expressed miRNAs using the condition tree, gene tree options and the Pearson correlation equation (GeneSpring software, Agilent Technologies).

miRNAs signatures in BC determined by stem-loop RT-PCR

Stem-loop RT-PCR (TaqMan® MicroRNA Assays, Applied Biosystems, Foster City, CA) was used to quantitate miRNAs according to the previously published conditions.22 To prepare cDNA specific to the miRNAs, each 15 μl RT reaction contained 10 ng of purified total RNA, 50 nM stem-loop RT primer, 1× RT buffer, 0.25 mM each of dNTPs, 3.33 U/μl MultiScribe™ reverse transcriptase and 0.25 U/μl RNase inhibitor. The reactions were incubated in a 96-well plate for 30 min at 16°C, 30 min at 42°C, followed by 5 min at 85°C and then held at 4°C. Each stem-loop RT-PCR for each miRNA assay was carried out in triplicate, and each 20 μl reaction mixture included 1.33 μl of diluted RT product, 10 μl of 2×TaqMan® Universal PCR Master Mix and 1 μl of 20×TaqMan® MicroRNA Assay Mix. The reaction was incubated in a 7900HT Fast Real-Time PCR System in 384-well plates at 95°C for 10 min, followed by 40 cycles at 95°C for 15 sec and 60°C for 1 min. RNA U6B small nuclear (RNU6B) (42 bp, a small RNA) served as an endogenous control according to the manufacturer's primary data across several human tissues and cell lines (P/N: 4373381, Applied).

Real-time quantitative RT-PCR

We synthesized first-strand cDNA with 1 μg of total RNA using random primers of the reverse transcription (RT) system (Promega, Tokyo, Japan). The initial PCR step was a 10-min hold at 95°C; the cycles (n = 40) consisted of a 15-sec denaturation step at 95°C followed by 1-min annealing/extension at 63°C. The TaqMan® probe and primers for KRT7 (P/N: Hs00818825_m1) were assay-on-demand gene expression products (Applied). All reactions were performed in triplicate, and a negative control lacking cDNA was included.

Algorithm for target genes

Our previous study has demonstrated the original method for target.23 Because many miRNAs cause degradation of their target mRNAs, expression patterns of miRNAs and their targets can be expected to indicate reverse relationships. Hence, our method identifies mRNAs as targets that not only contain complementary sequences but also show inverse expression patterns to the miRNA. To predict target mRNAs for the downregulated miRNAs, we identified the upregulated mRNAs that showed a more than 2.0-fold increase in expression level in BCs relative to NBEs in our corresponding expression data.5 Subsequently, the targets including the complementary sites to the miRNAs in the 3′-UTRs were selected from the upregulated mRNAs using the miRanda target prediction tool.24

Mature miRNA transfection

Mature miRNA molecules, Pre-miR™ (Applied), were incubated with Lipofectamine™ RNAiMAX (Invitrogen, Tokyo, Japan) before plating. Subsequently, the complex was just added to suspended 4 × 104 cells per well at plating (cotransfecting) in 24-well plates. We first tested the transfection efficacy of miRNA precursors in BC cell lines based on the downregulation of PTK9 mRNA by overexpression of miR-1 (This method was recommended by the manufacturer).

Small interfering RNA treatment and XTT assay

After cotransfecting small interfering RNA (siRNA)-KRT7 (si-KRT7) (L-019277-00; Dharmacon, Lafayette, CO) or nonsilencing siRNA (si-control) (D-001810-10; Dharmacon), KK47 cells were seeded in 96-well plates at 3 × 103 cells per well. After 48 hr, cell viability was determined by an XTT assay following the manufacturer's protocol (Roche Applied Sciences, Tokyo, Japan).


After 48-hr cotransfecting, the cells were harvested, and total protein lysate was prepared with common detergent lysis buffer in the presence of a protease inhibitor. The 20 μg protein lysate was separated by NuPAGE® on 4–12% bis-tris gel (Invitrogen) and transferred to a polyvinylidene difluoride membrane. Immunoblot was done with diluted (1:300) monoclonal cytokeratin7 antibody (clone OV-TL 12/30; No. 7018, Dako, Tokyo, Japan) and GAPDH antibody (MAB374; Chemicon, Temecula, CA). The membrane was washed and then incubated with goat anti-mouse IgG horseradish peroxidase conjugate (Bio-Rad, Hercules, CA). Specific complexes were visualized with an echochemiluminescence detection system (GE Healthcare, Buckinghamshire, UK).

Statistical analysis

The relationship between the 2 variables and numerical values obtained by real-time PCR-based experiments was analyzed with the Mann-Whitney U-test. The relationship between downregulated miRNA expression and KRT7 mRNA expression was analyzed by the Spearman rank correlation. All statistical values were age-adjusted because the mean ages of the BCs and NBEs were statistically different. The analysis software was Expert StatView (version 4, SAS Institute, Cary, NC); differences of p < 0.05 were considered statistically significant. The cutoff scores for the analysis were obtained by receiver-operating characteristic (ROC) curve analysis using MedCalc software (MedCalc Software, Mariakerke, Belgium).


Identification of differentially expressed miRNAs between BC, NBE and BC cell lines

We evaluated 156 mature miRNA expression levels of 14 BCs, 5 NBE specimens and 3 BC cell lines. After global normalization of the raw data, we identified 27 differentially expressed miRNAs between BCs and NBEs when we employed a cutoff p value of <0.01 to narrow down the candidates (Table III). Among these miRNAs, 8 were up- and 19 were downregulated in BCs relative to NBEs. When we applied a Bon-ferroni correction (p < 0.00032), 9 miRNAs were still differentially expressed between BCs and NBEs with a statistical significance (Table III). After global normalization, hierarchical clustering analysis based on differentially expressed miRNAs generated a tree with clear distinction between BCs, BC cells and NBEs (Fig. 1).

Figure 1.

Hierarchical clustering of BCs, NBEs and BC cells. 14 BCs, 5 NBEs and 3 BC cells were clustered according to the expression signature of 27 miRNAs differentially expressed between BCs and NBEs.

Table III. Twenty-Seven Differentially Expressed miRNAs in Bladder Cancer after Global Normalization
miRNAsp-valueCancer scoreNormal scoreFold change (cancer/normal)Status in BC
  1. Underlined miRNAs are commonly detected with global- and beta-actin normalization.


When we employed another normalization using the internal reference, ACTB, 10 miRNAs still had a p value of <0.05, and all were downregulated in BCs. Seven miRNAs were commonly detected with the 2 different approaches; miR-145, miR-30a-3p, miR-133a, miR-133b, miR-195, miR-125b and miR-199a* (Table III, underlined).

Verification of downregulated miRNAs expression in clinical specimens

To validate the results of our screening test, we subjected 6 miRNAs (miR-145, miR-30a-3p, miR-133a, miR-195, miR-125 and miR-199a*), which were selected as downregulated miRNAs by both different normalization methods to stem-loop RT-PCR on BCs (n =104) and NBEs (n = 31). We did not examine miR-133b because it was considered to function similar to miR-133a, and they have very similar sequences (miR-133a: UUGGUCCCCUUCAACCAGCUGU, miR-133b: UUGGUCCCCUUCAACCAGCUA). The 6 miRNAs shown by our signature to be downregulated did indeed have a lower expression in BCs than in NBEs (miR-145: 0.079 ± 0.019 (BC) vs. 0.897 ± 0.197 (NBE), p < 0.0001; miR-30a-3p: 0.145 ± 0.036 (BC) vs. 0.747 ± 0.095 (NBE), p < 0.0001; miR-133a: 0.048 ± 0.016 (BC) vs. 0.960 ± 0.196 (NBE), p < 0.0001; miR-195: 0.124 ± 0.012 (BC) vs. 0.688 ± 0.088 (NBE), p < 0.0001; miR-125b: 0.069 ± 0.013 (BC) vs. 0.939 ± 0.171 (NBE), p < 0.0001; miR-199a*: 0.099 ± 0.015 (BC) vs. 1.043 ± 0.194 (NBE), p < 0.0001) (Fig. 2a). Our cohort showed no significant relationship between the miRNA expression and clinicopathological parameters.

Figure 2.

(a) The 6 miRNAs that were commonly downregulated with 2 different normalizations in the screening test were significantly downregulated in BCs relative to NBEs (p < 0.0001). (b) ROC curve analysis showed each miRNA had good sensitivity and specificity to distinguish BC from NBE.

ROC curve analysis demonstrated that each miRNA had good sensitivity and specificity with optimal cutoff values as follows: miR-145: 90.5% (sensitivity), 77.4% (specificity), 0.173 (cutoff); miR-30a-3p: 94.3% (sensitivity), 76.7% (specificity), 0.310 (cutoff); miR-133a: 93.3% (sensitivity), 77.4% (specificity), 0.140 (cutoff); miR-195: 91.4% (sensitivity), 80.0% (specificity), 0.273 (cutoff); miR-125b: 89.4% (sensitivity), 77.4% (specificity), 0.206 (cutoff); miR-199a*: 72.1% (sensitivity), 90.3% (specificity), 0.087 (cutoff). Each miRNA had a good area under curve (AUC) of >0.7 (Fig. 2b).

Identification of target genes of the downregulated miRNAs in our algorithm

We devised an original algorithm based on our BC mRNA profile to predict target genes for the miRNAs differentially expressed in BCs and NBEs. Table IV shows the predicted target genes for the downregulated miRNAs, which were sorted by number of predicted target sites. Interestingly, 12 genes were common targets among the 39 predicted target genes for the 6 miRNAs as follows: KRT7 for all, ALK for miR-145 and miR-30a-3p, LOC400451 for miR-145 and miR-133a, PSMA7 for miR-145 and miR-133a, S100A14 for miR-145 and miR-125b, WWOX for miR-145 and miR-125b, FLNB for miR-145, miR-30a-3p, miR-133a and miR-195, CNIH4 for miR-30a-3p and miR-199a*, LEPRE1 for miR-30a-3p and miR-125b, CSF3 for miR-30a-3p and miR-133a, HSPA14 for miR-30a-3p and miR-199a* and KRT8 for miR-133a and miR-195 (Table IV and Supporting Information Table SI).

Table IV. Algorithm for Predicted Target Genes for the Downregulated miRNAs in Bladder Cancer
Unigene IDGene symbolTarget sitesUnigene IDGene symbolTarget sitesUnigene IDGene symbolTarget sites
Unigene IDGene symbolTarget sitesUnigene IDGene symbolTarget sitesUnigene IDGene symbolTarget sites
Hs.476448FLNB1   Hs.159918PHF142
Hs.443831PDCD51   Hs.80545RPL371
Hs.411501KRT71   Hs.411501KRT71

KRT7 mRNA expression and its correlation with each miRNA expression

We focused on KRT7 because it was the most frequent target of all miRNAs and was listed at the top in 3 of 6 miRNAs in our algorithm (Table IV, Supporting Information Table SI). Real-time RT-PCR revealed that the mRNA expression levels of KRT7 were significantly higher in BCs than in NBEs (8.220 ± 0.973 (BC) vs. 3.119 ± 0.762 (NBE), p = 0.0004) (Fig. 3a). Spearman's rank correlation test indicated that there were significant inverse correlations between KRT7 mRNA expression and the downregulated miRNAs (miR-145, miR-30a-3p, miR-133a, miR-195, miR-125 and miR-199a*) (p < 0.0001 for all miRNAs examined) (Fig. 3b).

Figure 3.

(a) KRT7 mRNA expression was significantly higher in BCs than in NBEs. (b) There were significant inverse correlations between KRT7 mRNA expression and the downregulated miRNAs.

Effects of miRNA transfection on KRT7 expression in BC cell lines

To determine whether KRT7 expression is actually downregulated by miRNAs from our algorithm, we transfected the 6 miRNAs (mature miRNAs; Pre-miR™) to KK47 cells, which overwhelmingly expressed KRT7 mRNA compared to T24 or BOY (Fig. 4a). After 24- or 48-hr transfection, KRT7 mRNA expression was significantly repressed (p < 0.05) by miR-30a-3p, miR-133a and miR-199a* (Fig. 4b). Particularly, miR-199a* transfection caused KRT7 mRNA expression to decrease by 35% of the control (Fig. 4b). Immunoblot also showed that the protein expression of various KRT7 isotypes was markedly decreased in miR-199a* transfection and slightly decreased in miR-145- and miR-30a-3p transfection (Fig. 4c).

Figure 4.

(a) KRT7 mRNA expression in 3 BC cells (BOY, T24 and KK47) by real-time RT-PCR. (b) KRT7 mRNA expression after 24- and 48-hr transfection of miRNAs (miR-145, miR-30a-3p, miR-133a, miR-195, miR-125 and miR-199a*) (right). *p < 0.05. (c) Immunoblot revealed that various KRT7 isotypes were markedly decreased in miR-199a* transfectant and slightly decreased in miR-145- and miR-30a-3p transfectant.

Effects of si-KRT7 and miRNA transfection on the cell growth

We then determined whether KRT7 and the 6 miRNAs contribute to cell viability. After 48-hr transfection, an XTT assay demonstrated that cell growth was significantly decreased in si-KRT7-transfected KK47 cells in comparison with the si-control-transfectant or moc (lipofectamine only) (80.3% of the moc, p < 0.001) (Fig. 5a). In addition, significant decreases in cell growth (p < 0.05) occurred in miR-30-3p-, miR-133a-, miR-195- and miR-199a* transfectant (88, 86, 82 and 88% of the si-control-transfectant, respectively) (Fig. 5b).

Figure 5.

(a) Effects of si-KRT7 transfection on the cell growth determined by XTT assay. si-KRT7-transfected KK47 cells exhibited a significant decrease in cell growth in comparison with si-control-transfectant or moc (lipofectamine only). (b) The cell growth was also significantly reduced in comparison with the control by transfection of the downregulated miRNAs (miR-30a-3p, miR-133a, miR-195 and miR-199a*). *p < 0.05, **p < 0.001.


The screening test identified 27 miRNAs that were up- or downregulated in BCs differently from those in NBEs. Among them, 7 miRNAs (miR-145, miR-30a-3p, miR-133a, miR-133b, miR-195, miR-125b and miR-199a*) were persistently remained as the downregulated miRNAs in BC after 2 different normalizations were applied. Stem-loop RT-PCR confirmed that these miRNAs were actually downregulated in 104 BCs when compared to 31 NBEs with a significant p value. Recent investigations detected several up- or downregulated miRNAs in breast cancer (miR-145 and miR-125b), hepatocellular cancer (miR-199a, miR-195 and miR-224) and colorectal cancer (miR-145, miR-133a, miR-133b, miR-195, miR-125b, miR-199a, miR-96 and miR-215), which were identical to our results.13, 15, 17 These miRNAs might be part of the underlying mechanism of carcinogenesis in various cancers. Recently, Gottardo et al. identified 10 miRNAs that were significantly upregulated in 25 BCs in comparison with only 2 NBEs, by using their custom oligonucleotide microchip for miRNA.19 However, our study and theirs had no miRNAs in common, and they did not detect any miRNAs that were downregulated in BCs.19 We found that both up- and downregulated miRNAs in BCs in the stem-loop RT-PCR experiment, which carries the advantages of accuracy, specificity and reproducibility for miRNA-detection.22 ROC curve analysis demonstrated that the miRNAs showed a good AUC of >0.7 with good sensitivity and specificity (>70%) to distinguish BC from NBE when an appropriate cutoff value was employed. These results suggest that miRNA can be used as an outstanding diagnostic biomarker for BC if it could be detected in serum or urine samples. These ideas are still speculative, and studies of miRNAs as diagnostic biomarkers are now underway in our laboratory.

We devised an algorithm based on our BC mRNA profile5 to predict target genes for the miRNAs differentially expressed in BCs and NBEs. Twelve genes were the common targets among the 39 predicted target genes, implying that one target gene is not regulated by one miRNA, and one miRNA regulates several target genes. Namely, there are one-to-many and many-to-one relationships between miRNAs and the target genes. To verify our algorithm's results, we focused on KRT7, which was the most frequent target of all miRNAs and was at the top of most miRNA lists. We found significant inverse correlations between KRT7 mRNA expression and each miRNA expression in clinical BCs. Moreover, significant decreases in KRT7 mRNA expression were observed in some miRNA transfectants (miR-30a-3p, miR-133a and miR-199a*). Unexpectedly, miR-145 transfection did not decrease the KRT7 mRNA expression, regardless of the fact that miR-145 has the most predicted targeting sites17 on KRT7, while miR-199a* has only one target site in our algorithm (Table IV, Supporting Information Table SI). Our data suggest that, for mRNA cleavage of the target genes, the specific region targeted by the corresponding miRNA may have an advantage over the number of the targeting sites. This hypothesis is consistent with a previous report demonstrating that an arrangement of miRNA recognition site through interaction with other miRNAs as well as the number of the targeting sites can influence the degree and specificity of miRNA-mediated gene repression.25 Considering this phenomenon from another aspect, immunoblot revealed that miR-145 transfection decreased small KRT7 isoformes (30–40 kDa). TaqMan® probes and primers for KRT7 transcript (assay-on-demand, Applied) were designed to cover the last exon-junction (around 4,000 bp from 5′ region) but unfortunately, uncovered small transcript variants (for example, accession No: BC107082.1 or BC107083.1), suggesting that miR-145 may target these small transcript variants of KRT7. Thus, our algorithm is incomplete, and some targets might be regulated by secondary effects of the miRNAs. However, the algorithm could provide novel candidates of miRNA targets based on miRNA signatures in human BC. In our study, miR-195 transfectant showed significant cell growth inhibition, even though it had no effect on KRT7 expression in the BC cell line. Other targets of miR-195 may be involved in cell growth inhibition. Further investigations are necessary to validate the interactions between the miRNAs and predicting miRNA-targeted genes.

KRT7 is a member of the cytokeratin gene family and encodes type II cytokeratin, which is a useful biomarker for detecting BC or making a differential diagnosis of the origin of a tumor.26 However, the functional role of KRT7 in cancer development has not been elucidated. In our study, cell proliferation assays showed that a significant decrease in cell growth occurred in si-KRT7-transfectants in comparison with the controls. This result suggests that the KRT7 gene may have an oncogenic function in human BC and that these miRNAs may play a role of tumor suppressive function. In our cohort, there was no significant relationship between the miRNA expressions and clinicopathological parameters including tumor stage, tumor grade or patient's prognosis. Altered expression of the miRNAs is a frequent event, but it may function in the early step of BC carcinogenesis.

Ours is the first study demonstrating that miR-30-3p, miR-133a and miR-199a* have a tumor suppressive function in human BC. These miRNAs from our algorithm lead us to a new strategy to find novel oncogenes, such as KRT7, and we believe that they are promising candidates for biomarkers and gene therapy of human BC.


The authors thank Ms. M. Miyazaki for her excellent laboratory assistance.