The microRNA-205-5p is correlated to metastatic potential of 21T series: A breast cancer progression model

MicroRNA is a class of noncoding RNAs able to base pair with complementary messenger RNA sequences, inhibiting their expression. These regulatory molecules play important roles in key cellular processes including cell proliferation, differentiation and response to DNA damage; changes in miRNA expression are a common feature of human cancers. To gain insights into the mechanisms involved in breast cancer progression we conducted a microRNA global expression analysis on a 21T series of cell lines obtained from the same patient during different stages of breast cancer progression. These stages are represented by cell lines derived from normal epithelial (H16N2), atypical ductal hyperplasia (21PT), primary in situ ductal carcinoma (21NT) and pleural effusion of a lung metastasis (21MT-1 and 21MT-2). In a global microRNA expression analysis, miR-205-5p was the only miRNA to display an important downregulation in the metastatic cell lines (21MT-1; 21MT-2) when compared to the non-invasive cells (21PT and 21NT). The lower amounts of miR-205-5p found also correlated with high histological grades biopsies and with higher invasion rates in a Boyden chamber assay. This work pinpoints miR-205-5p as a potential player in breast tumor invasiveness.


Introduction
Breast cancer is the most frequent carcinoma in women and a highly heterogeneous disease. Diagnostic and treatment decision are mainly based on classical biological variables including morphology, tumor grade, presence of lymph-node metastasis and molecular markers [1]. Cancer progression is a multistep process, progressing from normal epithelia to an atypical ductal hyperplasia then to ductal carcinoma in situ (DCIS), invasive ductal carcinoma (IDC) culminating in metastasis [2]. The knowledge on the DCIS transition to IDC is still incomplete and there are many questions concerning breast cancer progression. It is not clear if the enrichment of a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 miRNA microarray The miRNA global expression analysis was performed using the SurePrint Human miRNA Microarrays assays-Human miRNA Microarray Release 16.0, 8x60K. Based on miRBase version 16.0. (Agilent Technologies, Santa Clara, CA, USA).
Briefly, total RNA samples (100ng each) were dephosphorylated and labeled with the fluorophore Cyanine 3-pCp at the 3'end of the RNA molecule. After labeling, samples were purified with Micro Bio-spin 6 columns (Bio-Rad) and hybridized on array slides at 55˚C, 20 rpm for 20h. The label and hybridization reactions were performed using miRNA Complete Labeling and Hyb Kit (Agilent Technologies, Santa Clara, CA, USA). The microarray slides were scanned on Agilent microarray scanner version C.

Microarray data analysis
To perform microarray data analysis we used R language and the Limma package on Bioconductor software essentially as described in [10]. Data were background corrected using AgiMicroRNA algorithm and normalized between arrays by the quantile method in order to compensate for systematic technical differences between probe affinity and arrays. After background correction, differential expression was analyzed using the Limma package. A 2.0 times fold expression cutoff was applied to minimize the effect of probe background signal.

Quantitative RT-PCR
For miRNAs quantification, cDNA synthesis was carried out with total RNA obtained from the 21T cell lines and tissue biopsies, using NCode VILO miRNA cDNA Synthesis Kit (Thermo Fisher Scientific, Waltham, MA, USA). For mRNAs quantification, cDNA synthesis was achieved, using High Capacity cDNA Reverse Transcription Kit (Thermo Fisher Scientific, Waltham, MA, USA). Real-time PCR was performed using SYBR PCR Master Mix (Thermo Fisher Scientific, Waltham, MA, USA). The primers sequences employed in qRT-PCR reactions were: b-actin (F) 5´CATCGAGCACGGCATCGT 3´; b-actin (R) 5´GCCTGGATAGCAA CGTACAT 3´, miR-205-5p (F) 5´CTTCATTCCACCGGAGTCTG 3´with the universal reverse primer provided on miRNA cDNA synthesis kit. All experiments were performed in triplicate. β-actin was chosen as reference gene since its expression levels were constant in our experimental conditions. Expression results were analyzed using the ΔΔCt method [11] and statistical analysis was done using unpaired T-test.

Invasion assay
For cell invasion assay, 0.5 x 10 4 cells were seeded on the upper container of 8μm pore Boyden chamber (Merck Millipore, Darmstadt, Germany) coated with 10% Matrigel (BD Biosciences Franklin Lakes, NJ, USA), diluted in the respective culture media. Cells were incubated in a starvation media, without fetal calf serum for 22h at 37˚C, 5% CO 2 and the recipient dish underneath the cells was filled with a nutrient media containing 10% fetal calf serum. The non-migrating cells were removed from the upper part of the chamber with a cotton swab. Cells able to transmigrate to the lower part of the chamber were counted after being fixed with 4% paraformaldehyde and having the nucleus stained by Hoechst 3342. Results are displayed as a percentage ratio between tumoral and the normal cell line H16N2 and between the respective non-treated control in the functional assays. miR-205-5p contributes for metastasis in 21Tcells miR-205-5p target prediction To increase the accuracy in miR-205-5p in silico target prediction we used as approach a combined miRNA and mRNA transcriptome profiling [10]. First, the mRNAs that had the expression increased more than two fold in 21T cells during the transition in situ to metastatic tumor (21MT1 vs 21NT) were selected from a published database [8]. After this first selection, we performed a search for miR-205-5p binding sites among the overexpressed mRNA, using the software RNA22 [12]. The most likely interaction sites (p 0.05) were considered as possible miR-205-5p target.

Results
Differential expression of miRNAs in 21T series reveals miR-205-5p downregulation during breast cancer progression The transition from normal tissue to hyperplasia (H16N2-21PT) caused a significant increase in the expression levels (! 2 times fold up-regulated) of 29 miRNAs (Fig 1, Fig A in S1 File). The miR-141, miR-29 family, miR-96 and miR-106b presented the highest increase in the expression levels (! 5 times fold up-regulated). Those early stage up-regulation persisted at all stages of tumor progression (Fig 1, Figs A-D in S1 File).
Among the down-regulated miRNAs, the miR-4299 ( 6 times fold down-regulated) and miR-3934 ( 2.6 down-regulated) were altered throughout all stages of tumor progression (Fig 1, Figs A-D in S1 File).
The validation of microarray data was done by qPCR confirming that miR-205-5p is downregulated through breast cancer progression to metastasis ( Fig 2B).

miR-205-5p downregulation is observed in high-grade breast cancer biopsies
miR-205-5p expression can be related to tumor aggressiveness in vivo. miR-205-5p expression was analyzed by qRT-PCR in paraffin embedded formalin fixed biopsies of breast tissue samples (S5 Table with patient description and tumor characteristics). Breast cancer cases diagnosed as invasive ductal carcinoma were distributed according Elston's classification in I, II and III [13]. Elston grade III representing the most proliferative and aggressive stage. miR-205-5p fold change in breast cancer tissue samples were calculated in comparison to miR-205-5p mean expression of 6 biopsies obtained from normal breast tissues (Fig A in S2 File) for the non-normalized distribution of the miR-205-5p expression). As shown in Fig 3, we could demonstrate a significant downregulation of miR-205-5p displayed in the most aggressive Elston grade III breast tumors in comparison to normal breast tissues. miR-205-5p contributes for metastasis in 21Tcells miR-205-5p low expression correlates to 21T cells invasive phenotype The 21T invasive phenotype was measured by the capacity of the 21T cells to transmigrate through 8μm pored membrane coated with 10% Matrigel matrix. The cells were seeded on top of an 8μm pored membrane coated with 10% Matrigel matrix. The percentage of 21T cells able to transmigrate to the lower part of the membrane is shown in Fig 4. These results confirm our assumption that miR-205-5p expression levels are inversely related to the invasion capacity. To verify if the exclusive change in miR-205-5p expression levels is able to alter the 21T cells invasiveness, we transfected 21MT1 and 21MT2 cells with the miR-205-5p precursor and H16N2, 21PT and 21NT cells with a miR-205-5p silencer. miR-205-5p silencing in the noninvasive cell lines (H16N2, 21PT and 21NT) partially increased migration rates while miR-205-5p overexpression in metastatic cells (21MT1 and 21MT2) reduced their migration capacity (S1 Fig). Although not statistically significant, these observations suggest that miR-205-5p plays an important role reverting breast cancer cells to a less aggressive behavior.

miR-205-5p has potential cancer-related targets
A simultaneous correlation between mRNA overexpression and miR-205-5p downregulation during in situ to metastatic transition in 21T cells was used as a filter for a subsequent  identification of miR-205-5p binding sites in the selected mRNAs. Eight miR-205-5p potential target mRNAs were selected (Table 1). A total of eleven miR-205-5p binding sites were identified using this approach (S1 Table). The predicted miR-205-5p targets comprises genes with  functions related to cell invasion such as SOCS3, PTPRN2, and MMP3 and on epithelial to mesenchymal transition such as TGFB1 (Table 1).

Discussion
Breast cancer is one of the most common cancer type concerning women worldwide [14].
Tumor progression for a metastatic phenotype is the major complication causing patient death. The high heterogeneity among breast tumors is a challenge for the development of more specific therapies and for the identification of consistent prognostic biomarkers. In this sense, we conducted a study to detect differences in microRNA expression during breast cancer progression using the 21T series of cell lines [15,16]. The 21T cells were isolated from the same patient at different tumor stages. This allowed us to map the changes in miRNAs expression during the tumor progression, excluding inter-individual genetic differences. The 21T series is composed of five cell lines: H16N2, representative of adjacent non-tumoral breast cells; 21PT, representative of Atypical Ductal Hyperplasia (ADH) cells; 21NT of Ductal Carcinoma in Situ (DCIS) and 21MT1 and 21MT2, of Invasive Metastatic Carcinoma (IMC) [8].
Previous research have demonstrated that this series of cell lines is a very interesting model to investigate breast cancer progression [8,16] and epigenetic alterations [9]. We have recently demonstrated a global hypomethylation pattern as well as chromatin changes, such as the accumulation of pericentric heterochromatin in metastatic 21MT1 cells [9]. We have first analyzed miRNA expression patterns in the five cell lines using microarray expression assay.
Among the altered miRNAs, miR-4299 and miR-3934 presented a strong downregulation in all tested cell lines, during all stages of tumor progression. In silico target prediction combined with gene ontology analysis reveals that miR-4299 can potentially regulate ErbB signaling pathway in different points by targeting BRAF, EIF4EBP1, PIK3R3 and BTC [17]. miR-3934 predicted targets include components of the oncogenic Wnt signaling pathway such as WNT7, WNT9A, MAPK10, APC and ROCK2 [17]. The up-regulated miRNAs in all 21T cell lines, suggesting an early event in tumorigenesis include the oncomiRs miR-29a/b/c [18,19], miR-141 [20,21], miR106b [22] and miR-96 [23,24]. In this study, we focus our attention at the miRNA profile during the progression from an in situ tumor to a metastatic phenotype. The downregulation of miR-205-5p observed between 21PT/21NT and 21MT1/21MT2 cells, representing the transition hyperplasia/in situ to metastasis suggest a specific role of miR-205-5p in cell invasion. miR-205-5p down-regulation was later confirmed by qRT-PCR ( Fig 2C). Indeed, miR-205-5p expression decreased with the tumor stage from Elston histological grade I through III (Fig 3), where grade III indicates tumor aggressiveness and patient poor survival [13]. Similar results were also observed in other clinical breast cancer cohorts where miR-205-5p expression levels were measured by qRT-PCR [25,26] and by in situ hybridization [27]. In addition to the miR-205-5p downregulation in high grade tissue biopsies, the presence of blood circulating miR-205-5p, indicating an overall high expression correlated with tumor chemo sensitivity and higher rates of disease free survival [28].
In all cases, miR-205-5p downregulation was linked to a worse prognosis. These findings correlate to our both observations in 21T cell lines and breast tumors highlighting the need of including miR-205-5p in more comprehensive clinical studies.
miR-205-5p has an established role in regulating epithelial to mesenchymal transition (EMT) and tumorigenesis [29][30][31]. When normally expressed, miR-205-5p can inhibit its targets such as ZEB1 and ZEB2 and consequently result in the expression of E-cadherin, keeping cell polarity and cell-cell junction integrity [32]. miR-205-5p validated targets, besides ZEB1 and ZEB2, include E2F1, HER3, VEGF-A, PTEN and NOTCH2 [6,29,[33][34][35]. Our data miR-205-5p contributes for metastasis in 21Tcells support the notion that miR-205-5p downregulation contributes to breast cancer progression to metastasis. Derived from the same patient, 21T cells have a decreasing expression of miR-205-5p from non-tumoral H16N2 to metastatic 21MT1/2 consequently promoting higher proliferation behavior and tumor aggressiveness. In addition to the already known miR-205-5p targets we could by crossing mRNA and miRNA expression levels predict other genes involved in tumor progression that can contributing to the transition to a mesenchymal and more proliferative phenotype.
miRNAs can bind to their targets with a partial base pair complementarity. The level of expression of miRNA targets in a system is relevant for increasing the success rate in the target prediction [10]. By selecting just the mRNAs that were overexpressed in the transition in situ to metastatic cells, we take into consideration the specific cellular context allowing a more precise prediction.
Among our predicted targets, TGFB has already been experimentally validated in osteosarcoma cells [36]. However, other predicted targets such as SOCS3, PTPRN2, MMP3, PRG1 and BASP1 have not yet been experimentally validated but could be of particular interest on cancer research. SOCS3 draws special attention by having three miR-205-5p binding sites and for being able to promoting growth, metastasis and EMT in breast cancer [37,38].
Our results show that cell migratory capacity inversely correlates with miR-205-5p expression, (Fig 4). We observed a lower migration rate in Boyden chamber after miR-205-5p transfection in 21MT1 and 21MT2 (S3). The opposite trend was also observed: higher migration rates were observed after miR-205-5p was silenced in non-invasive cell lines. Although not significant, it indicates that miR-205-5p contributes but it´s not the only factor responsible to 21T series invasiveness.
Altogether, our results reveal a shift on the miRNA expression during tumorigenesis in 21T cells and highlight miR-205-5p as a major player on breast tumor evolution from carcinoma in situ to metastasis.

Conclusions
We analyzed differences in microRNA expression during breast cancer progression using the 21T series of cell lines. miR-205-5p appeared as the only deregulated microRNA in the transition carcinoma in situ to metastatic, showing a significant 5.4 times downregulation. This observation correlates to the cells invasive potential in Boyden chamber assay and was strengthened, with the in vivo downregulation of miR-205-5p in aggressive breast tumors. Our results show miR-205-5p as a tumor suppressor miRNA and contribute to understanding breast cancer progression to an invasive phenotype. Further studies will help to validate experimentally miR-205-5p target genes involved in this process.  Table. miR-205-5p binding sites and estimated folding energy for the predicted targets. (DOCX)