Background

Breast cancer is the leading cause of cancer death in women worldwide. Despite advances in the understanding of cancer pathogenesis and improvement in diagnosis and treatment over the past few decades, biomarkers of clinical interest are not so numerous. Now it is well documented that endogenous estrogens known as an important regulator of development, growth and differentiation of the normal mammary gland play also a major role in the development and progression of breast cancer [1]. The mammary cell proliferation signals are mediated in part by the estrogen receptor alpha (ER). The expression of ER in breast tumors is frequently used to separate breast cancer patients in a clinical setting both as an important prognostic marker for prognosis and in predicting the likelihood of response to endocrine therapy. Although the majority of primary breast cancers are ER-positive (ER+) and respond well to antiestrogen therapy, up to one-third of patients with breast cancer lack ER (ER−) at the time of diagnosis, and a fraction of breast cancers that are initially ER+ lose ER expression during tumor progression [2]. These patients fail to respond to antiestrogen therapy and have higher tumor aggressiveness and poor prognosis. Previous studies have shown that ER absence is a result of hypermethylation of CpG islands in the 5’ region of ER coding gene (ESR1) in a fraction of breast cancer [2]. However, the molecular mechanism of the rest of the ER− breast cases and the molecule(s) involving ER hypermethylation remain largely unknown. Other mechanisms involved in altering ER expression have been identified, including mutations within the open reading frame of ESR1 [3] as well as ESR1 amplification increasing the ER protein expression [4]. Recently, ESR1 ligand-binding domain mutations were described in hormone-resistant breast cancers [5].

Since their first description in C. Elegans in 1993, increasing numbers of studies showing frequent deregulation of microRNAs (miRNAs) in human breast cancers and association of some of them with cancer metastasis and poor prognosis suggesting an important role of miRNAs in cancer development and progression [6, 7]. miRNAs are small non-coding RNA gene products able to regulate gene expression at the post-transcriptional level. Thus, today, miRNAs are increasingly seen as important regulators of gene expression in breast cancers, acting either as oncogenes (such as miR-21) or tumor suppressors (such as let-7), and affecting through different mechanisms many cellular processes that are routinely altered in cancer, such as differentiation, proliferation, apoptosis, metastasis and telomere maintenance [8–11]. MiRNAs are also thought of as biomarkers in cancer diagnosis and prognosis [12]. The diagnostic potential of circulating miRNAs is based mainly on their non-invasive detection in serum and plasma and on their high resistance under difficult environmental conditions, offering them therefore an emerging role in developing new follow-up markers and strategies for cancer treatment [13–15]. Moreover, studies suggested that expression profiles of miRNAs are informative for the classification of human breast cancers [16–18]. Numerous datas are available regarding the miRNA expression in ER+ and ER− breast cancer tissues and come mainly from studies using miRNA microarray techniques [16, 19, 20]. Results and conclusions from these old studies are generally not consistent and sometimes even conflicting. More recently, miRNA landscape in breast cancer was deciphering in a large cohort with matching detailed clinical annotation and long-term follow-up but not particularly taking into account ER+ and ER− contexts [17]. Taken together, these finding have prompted us to use the robust quantitative RT-PCR technology to identify miRNAs that are differentially expressed in ER+ and ER− in breast primary tumors with the aim to better understand the molecular basis for the phenotypic differences between these two sub-types of carcinomas and to find potential clinically relevant miRNAs.

Methods

Patients and samples

Breast tumor samples were obtained from 184 postmenopausal women with primary unilateral non metastatic breast adenocarcinoma who underwent biopsies or initial surgery at the Curie Institute/René Huguenin Hospital (Saint-Cloud, France) between 1984 and 2009. Each patient signed a written informed consent form and this study was approved by the Curie Institute/René Huguenin Hospital ethics committee. Immediately after biopsy or surgery, the tumor samples were stored in liquid nitrogen in −80 °C until RNA extraction. All samples analyzed contained more than 70 % of tumor cells. Tumor samples included 106 ER+ and 78 ER− tumors. ER status was determined at the protein level by using biochemical methods (Dextran-coated charcoal method until 1988 and enzyme immunoassay thereafter) and was confirmed at mRNA level by RT-PCR. Control samples consisted of twelve specimens of normal breast tissue obtained from women undergoing cosmetic breast surgery or adjacent normal breast tissue from breast cancer patients [21]. Thirty-one of breast tumor samples, comprising 21 ER+ and 10 ER−, as well as 8 normal breast samples, were used as a RT-PCR pan-miRNA screening set to identify and select miRNAs differentially expressed in ER+ compared to ER−. These selected miRNAs were then validated in the remaining 153 breast tumor samples comprising 85 ER+ and 68 ER− compared to eight normal breast samples. Clinicopathological characteristics of patients in relation to metastatic free survival in the screening and validation series are provided in Table 1. In the screening set, we voluntary included more SBR grade III tumors with the aim to facilitate identification of robust genes differentially expressed whereas the validation set is totally representative of breast cancers treated in the Curie institute/René huguenin hospital between 1984 and 2009.

Table 1 Pathological and clinical characteristics of patients in relation to metastasis free survival (MFS) in the screening and validation sets
Full size table

RNA extraction

Total RNA was extracted from breast tissue by using the acid-phenol guanidium method. Total RNA concentration was quantified using a NanoDrop™ spectrophotometer. RNA quality was determined by agarose gel electrophoresis and ethidium bromide staining. The 18S and 28S RNA bands were visualized under ultraviolet light.

miRNA expression profiling

MiRNA expression levels in samples were quantified by quantitative RT-PCR (RT-qPCR) using the SYBR Green Master Mix kit on the ABI Prism 7900 Sequence Detection System (Perkin-Elmer Applied Biosystems, Foster City, CA, USA). The Human miScript Primer Assays version 9.0 and 11.0 from Qiagen, designed to detect 804 human miRNA probes, were used according to the manufacturer’s guidelines. Small nucleolar RNA RNU44 (Qiagen) was used as endogenous control to normalize miRNA expression levels. The relative expression level of each miRNA, expressed as N-fold difference in target miRNA expression relative toRNU44, and termed "Ntarget", was calculated as follows: Ntarget = 2ΔCtsample. The value of the cycle threshold (ΔCt) of a given sample was determined by subtracting the Ct value of the target miRNA from the average Ct value of RNU44. The Ntarget values of samples were subsequently normalized such that the median Ntarget value of normal breast samples was one. To overcome limits of detection of RT-qPCR, and be sure in expression values of miRNAs, we have considered a miRNA as relevant when the Ct values were lower than 30 in at least 50 % of all samples analyzed. The relative expression of each miRNA was characterized by the median and the range, and a non-parametric Mann–Whitney test was used for statistical analysis of differences in miRNA expression between groups.

Gene expression profiling

In the validation series, mRNA expression levels of Dicer (NM_177438), Drosha (NM_013235), AGO2 (NM_012154), DGCR8 (NM_022720), four protein-coding genes required to the miRNA biogenesis, and six host genes CTDSPL(NM_005808.2),EVL(NM_016337.2), NFYC(NM_014223.4) OGFRL1(NM_024576.3),CTDSP1 (NM_021198.1), PTMA (NM_002823.4) containing the identified miRNAs were measured by RT-qPCR. Primers and PCR conditions are available on request, and the RT-qPCR protocol is described above. The mRNA expression level of each protein-coding gene is relative to the TBP gene (NM_003194).

Breast cancer cell lines

Expression levels of selected miRNAs were measured by RT-qPCR in a collection of RNAs from 30 human breast cancer cell lines commonly used including 19 ER− (BT-20, BT-549, HCC-38, HCC-70, HCC-202, HCC-1143, HCC-1187, HCC-1569, HCC-1599, HCC-1937, HCC-1954, Hs-578 T, MDA-MB-157, MDA-MB-231, MDA-MB-435 s, MDA-MB-436, MDA-MB-453, MDA-MB-468 and SK-BR-3) and 11 ER+ (BT-474, BT-483, CAMA1, HCC-1428, HCC-1500, MCF-7, MDA-MB-134VI, MDA-MB-361, MDA-MB-415, T-47D and ZR-75-1). These RNAs were provided by the transfer department of Curie Institute. For each miRNA and each cell line, mRNA levels were normalized such that the median value of the ER− breast cancer cell lines was one.

The effects of 17β-estradiol (E2) on the miRNA expression were studied on two ERα-positive breast cancer cell lines whose growth is known to be stimulated by E2 : MCF-7 cell line for all selected miRNAs and T-47D cell line for miR-190b. They were cultured in either minimum essential medium (MEM) or Dulbecco’s modified Eagle medium (DMEM) supplemented with 10 % fetal calf/bovine serum and antibiotics (penicillin 50 g/ml, streptomycin 50 g/ml and neomycin 100 g/ml) at 37 °C with 5 % CO2. For experiments using E2, MCF-7 and T-47D were grown in phenol red-free minimum essential medium (MEM) supplemented with 5 % charcoal-dextran-stripped fetal calf serum for at least 3 days before treatment. The cells were then treated with E2 (Sigma) diluted in ethanol (EtOH) at 1 nM for MCF-7 and 10 nM for T-47D, or with vehicle EtOH (control cells). RNAs were extracted from these cells after 6 h, 18 h and 4 days of the presence of E2 and the mRNA levels measured by RT-PCR were normalized such that the median value of control cells was of one. Three independent experiments were realized for each time and each condition. To verify the effects of E2 on growth of cells, mRNA expression of pS2/TFF1 (NM_003225), a well-known ERα-induced gene, was also measured by RT-qPCR on the treated cells.

The effects of miR-190b expression on cellular proliferation were studied on breast cancer cell lines ER+ MCF-7 and T-47D that were transfected with antagomir against miR-190b (sequence complementary to miR-190b which blocks its effect) and on breast cancer cell line ER− MD-MBA-231 that was transfected with a miR-190b mimic (double-stranded RNA which mimics mature endogenous miR-190b) using a 3-(4,5-dimethyl-2-thiazolyl)-2,5-diphenyl-2H-tetrazolium bromide (MTT) proliferation assay. In brief, after transient transfection of cells for 24 h with 40 nM of antagomir against miR-190b or mimic of miR-190b (synthetized by Qiagen), the cells were growth in normal medium for 48, 72 or 120 h to be then treated with 0.5 mg/ml of the MTT labeling reagent at 37 °C for 1 to 3 h and lysed in 150 μl of dimethyl sulfoxide at room temperature for 30 min. The cell viability was thus determined by reading the absorbance at 450 to 570 nm of signal generated by MTT reduction which is directly proportional to the cell number. For each cell line, the data were collected from three independent experiments and compared to the control group obtained by transfection of non-targeting siRNA as negative control in miRNA inhibition experiments or miRNA inhibitor as negative control in miRNA mimic experiment.

Survival analysis

Metastasis-free survival (MFS) was determined as the interval between initial diagnosis and detection of the first metastasis. Survival distributions were estimated by the Kaplan-Meier method, and the significance of differences between survival rates was ascertained with the log-rank test. The Cox proportional hazards regression model was used to assess prognostic significance, and the results are presented as hazard ratios and 95 % confidence intervals (CIs). Statistical analyses were performed using GraphPad Prism 5 software.

Results

Differential miRNA expression between ER+ and ER− breast tumors

To identify miRNA expression profiles in breast cancer according to ER status, expression levels of 804 miRNAs were measured by RT-qPCR technology in a well-defined series of 21 ER+ and 10 ER− breast tumors and in 8 normal breast tissues (Additional file 1: Table S1). MiRNAs with high Ct values in this screening set and miRNAs with very low expression levels (indicated by an asterisk after their name) were not more studied, resulting in a list of 333 informative miRNAs (Additional file 2: Table S2).

Among these 333 miRNAs a Mann–Whitney test identified 155 miRNAs that were significantly differently expressed in ER+ compared to ER− tumors with a p-value < 0.05 : 15 miRNAs were up-regulated and 140 miRNAs were down-regulated. We then selected miRNAs that were the most strongly deregulated and for which the specificity of RT-qPCR amplification was verified on the dissociation curve for RT-qPCR validation in a larger independent series of breast tumors. Thus, we focused our study on 11 miRNAs for which the expression level was increased by 2-fold in ER+ compared to ER− tumors and 7 miRNAs for which the expression level was decreased by 4-fold in ER+ compared to ER−tumors (Table 2).

Table 2 18 miRNAs significantly differentially expressed between ER+ and ER− breast tumors in the screening series
Full size table

miRNAs associated with ER status in an independent validation series

The expression levels of these 18 miRNAs selected in the screening series were then verified in a validation series including 153 breast tumors (85 ER+ and 68 ER−) and eight normal breast tissues (Table 3).

Table 3 Relative miRNA expression levels of the 30 selected miRNAs between ER+ and ER− breast tumors in the validation series
Full size table

In these validation series, we also measured the expression levels of 12 miRNAs reported by the literature to be particularly deregulated in ER+ breast tumors : let-7a and let-7b [22, 23], miR-18a and miR-18b [24], miR-21 [25], miR-22 [26], miR-155 [27, 28], miR-206 [29] and mir-221 and 222 [30] as well as miR-19a and miR-92a1, which, with miR-18a, belonged to the miR-17-92 cluster [31] (Table 3).

Among the 11 up-regulated miRNAs selected from the screening series, except miR-451, we validated the up-regulation of miR-190b, miR-101-1, miR-193b, miR-342-5p, miR-376c, miR-143, miR-30c2, miR-30e, miR-26a1 and miR-26b in ER+ compared to ER− (Table 3). Among the 12 miRNAs selected from the literature, we found 2 other miRNAs up-regulated in ER+ compared to ER−: let-7a1 and let-7b. However among these 12 up-regulated miRNAs, we identified 5 different expression profiles according to their expression in ER+/Normal and ER−/Normal. Eight miRNAs (miR-26a1, miR-101-1, let-7b, miR-30c2, miR-143, miR-26b, miR-376c and let-7a1) showed a significant decrease of their expression in both ER+ and in ER− compared to normal breast tissue (see miR-26a1 for example in Additional file 3: Figure S1A) but with a significantly greater decrease in ER−/Normal than in ER+/Normal (FC ranging from 2.1 to 11.1 and from 1.7 to 7.1, respectively) (Table 4). The down-regulation of miR-30e was specific to ER− (Additional file 3: Figure S1B) since its expression was not differentially expressed in ER+/Normal but significantly under-expressed in ER−/Normal. MiR-193b did not particularly retain our attention to the extent that this miRNA was deregulated neither in ER+/Normal nor in ER−/Normal (Additional file 3: Figure S1C). MiR-342-5p was significantly up-regulated in ER+/Normal but not differentially expressed in ER−/Normal (Additional file 3: Figure S1D), revealing a specific up-regulation of miR-342-5p in ER+. Finally, miR-190b retained our attention as it was the miRNA the most strongly over-expressed in ER+ compared to ER− with a FC upper to 23, much higher than FC of other up-regulated miRNAs (Table 3, Additional file 3: Figure S1E). Moreover the ER+ breast tumors showed a miR-190b up-regulation compared to normal breast tissue with a FC of 6.34 whereas the ER− breast tumors, a down-regulation with a FC of −3.70 (Table 4).

Table 4 Relative miRNA expression levels of the 30 selected miRNAs between ER+ or ER− breast tumors and normal breast tissues
Full size table

Among the seven down-regulated miRNAs selected from the screening series, under-expression of four miRNAs, miR-654-3p, miR-203, miR-146a and miR-1244, in ER+ compared to ER− was confirmed in the validation series (Table 3). Among the miRNAs selected from the literature, we found four other miRNAs down-regulated in ER+ compared to ER−: miR-18a, miR-18b, miR-92a1 and miR-155. Among these eight down-regulated miRNAs, we identified five different expression profiles according to their expression in ER+/Normal and RE−/Normal (Additional file 4 and Table 4). The first profile concerned miR-18a, miR-18b and miR-654-3p (see miR-654-3p for example in Additional file 4: Figure S2A) that were not differentially expressed in ER−/Normal but significantly under-expressed in ER+/Normal, revealing a specific down-regulation of miR-18a, miR-18b and miR-654-3p in ER+ (Table 4). We found two miRNAs that were not differentially expressed in breast cancer, miR-203 and miR-146a (see miR-146a for example in Additional file 4: Figure S2B) and one miRNA, miR-92a1, that was significantly down-regulated in breast cancer (Additional file 4: Figure S2C). The two last expression profiles concerned miR-155 (Additional file 4: Figure S2D) and miR-1244 (Additional file 4: Figure S2E). Although these two miRNAs were significantly up-regulated in ER− breast cancer compared to normal breast tissue, miR-155 showed also significant increase of its expression in ER+/Normal whereas miR-1244 was not differentially expressed in ER+/Normal, revealing a specific up-regulation of miR-1244 in ER− (Table 4).

Expression of genes required for miRNAs biogenesis in ER+ and ER− breast tumors

The majority of these 20 miRNAs deregulated in ER+ breast tumors, except miR-155 and miR-1244, were significantly down-regulated in breast cancers compared to normal breast tissue (Additional file 5: Table S3) so we explored if genes required for miRNAs biogenesis could be deregulated. In the validation series, we measured by RT-qPCR the expression levels of DICER1, DROSHA, AGO2 and DGCR8, four genes encoding proteins playing pivotal roles in the processing of mature miRNAs. We found that these genes were deregulated in ER+ compared to ER− breast tumors: DICER1, DROSHA and DGCR8 were significantly under-expressed in ER− while AGO2 was moderate over-expressed in ER−. We did not however observe significant expression changes in ER+/Normal for these four genes. On the other hand, we observed a significant under-expression of DICER1 in ER−/Normal (Table 5). These results revealed a deregulation of genes required for miRNA biogenesis in the absence of ER.

Table 5 Expression levels of 4 genes required for miRNA biogenesis in breast tissues
Full size table

Expression of host genes of miRNAs in ER+ and ER− breast tumors

Among the 20 miRNAs identified as deregulated in ER+ compared to ER− breast tumors, 6 miRNAs are located in intragenic regions: miR-26a1 in CTDSPL, miR-342-5p in EVL, miR-30e in NFYC, miR-30c2 in OGFRL1, miR-26b in CTDSP1 and miR-1244 in PTMA. The expression levels of these 6 host genes were then measured, by RT-qPCR, in the validation series (Table 6). These genes showed significant expression difference between ER+ and ER−similar to their miRNA (Table 4). Thus CTDSPL, EVL, NFYC, OGFRL1 and CTDSP1 are more expressed in ER+ than in ER− like miR-26a1, miR-342-5p, miR-30a, miR-30c2 and miR-26b respectively, and PTMAP2 is less expressed in ER+ than in ER− like miR-1244. Moreover Spearman’s rank correlation analysis revealed a significant and positive correlation between expression of all host genes and its resident miRNA in breast tumors : miR-26a1 and CDTSPL (r = 0.3157, p < 0.0001), miR-342-5p and EVL (r = 0.5931, p < 0.0001), miR-30e and NFYC (r = 0.3157, p < 0.0001), miR-30c2 and OGFRL1 (r = .02803, p = 0.0004), miR-26b and CDTSP1 (r = 0.2502, p = 0.0018) and miR-1244 and PTMA (r = 0.2258, p = 0.005), indicating a miRNA-host co-transcription. According to ER status, a significant correlation was observed for miR-342-5p/EVL in ER+ (r = 0.3817, p = 0.0004) and for miR-26b/CTDSP1 and miR-1244/PTMAP2 in ER− (r = 0.3584, p = 0.0029 and r = 0.2822, p = 0.0197, respectively) (data not shown).

Table 6 Expression levels of host genes containing miRNAs deregulated in ER+ compared to ER− breast tumors in the validation series
Full size table

miRNA expression in human breast cancer cell lines

We further evaluated the expression levels of 20 miRNAs identified as deregulated in ER+ compared to ER− breast tumors in 30 human breast cancer cell lines including 19 ER− and 11 ER+. The patterns of expression changes observed between ER+ and ER− breast tumors do not have been validated in breast cancer cell lines for all miRNAs (Table 7). We only confirmed the significant over-expression of miR-190b, miR-26a1, miR342-5p, miR-101-1, miR-26b and miR-193b in ER+ breast cancer cell lines. None down-regulation of miRNAs in ER+ compared to ER- tumors was validated in cell lines; miR-203 was even significantly up-regulated (p = 0.0226). It is worthy to note that miR-190b, the highest up-regulated miRNA in ER+ tumors, was also the highest miRNA expressed in ER+ breast cancer cell lines, with a FC of 43 compared to 8 for the second higher up-regulated miRNA, miR-342-5p (Table 7), and that this up-regulation was observed in most of ER+ breast cancer cell lines (Fig. 1) confirming thus that miR-190b may have an important role in ER-dependent tumorigenesis. This is why we decided to focus next experiments on the expression and function of miR-190b.

Table 7 Relative miRNA expression levels of 20 miRNAs in breast cancer cell lines. Results in breast cancer cell lines are expressed as the median (range) of miRNA level relative to ER- breast cancer cell lines
Full size table
Fig. 1
figure 1

Expression levels of miR-190b in breast cancer cell lines. Representation of miR-190b relative expression level in 30 breast cancer cell lines. For each cell line, the miRNA levels were normalized such that the median value of the ER− breast cancer cell lines was 1 (horizontal line)

Full size image

Confirmation of miR-190b up-regulation in ER+ compared to ER− breast tumors

The heightened increase of miR-190b expression in ER+ compared to ER− breast tumors identified previously by Qiagen quantitative RT-PCR was validated by another experimental technique provided by Applied System Biotechnologies. Indeed, on 20 breast tumor samples which showed a FC of 23 between ER+ and ER− with Qiagen technology, we found a similar increase of miR-190b expression levels with Applied technology (FC of 26) (data not shown) and a strong positive correlation of miR-190b expression between the two techniques (Spearman’s coefficient correlation of 0.8977 significant at p < 0.0001).

Prognostic value of miR-190b expression in breast cancer patients

Using a Kaplan-Meier analysis, we showed that high expression of miR-190b did not impact metastasis-free survival in ER+ and ER- separated subgroups (datas not shown). If we compared MFS according to the type of treatment, we observed no prognostic impact related on miR-190b expression level for patients who received hormone therapy alone (p = 0.40, datas not shown). All patients receiving other adjuvant treatment expressed miR-190b at low level. Interestingly high expression of miR-190b was associated with a prolonged metastasis-free survival independently to ER status and treatment (log rank test: p = 0.0173, HR = 1.869, 95 % CI = 1.12 to 3.13) (Fig. 2A), as well as a prolonged event-free survival (log rank test: p = 0.0046, HR = 2.048, 95 % CI = 1.248 to 3.360) (Fig. 2B). This result prompted us to explore functions of miR-190b in breast carcinogenesis.

Fig. 2
figure 2

Metastasis-free survival (a) and event-free survival (b) according to miR-190b expression level in breast tumors for the total cohort. Kaplan-Meier survival analysis stratified by the miR-190b expression level. The p value was determined using the log rank test

Full size image

Effect of estrogen on miR-190b expression

To identify if estrogen could explain the deregulation of miR-190b between ER+ and ER− breast tumors, we measured its expression levels on the ERα-positive MCF-7 and T-47D breast cancer cell lines treated with 17β-estradiol (E2). We did not observe an increase of miR-190b expression levels in MCF-7 or in T-47D treated by E2 whereas the expression of the well-known ERα-induced gene pS2 was highly increased in the two cell lines (Additional files 6: Figure S3A and S3B). Others 19 miRNAs did not respond to 17β-estradiol either (datas not shown). Obviously we neither observed effect of tamoxifen treatment on miR-190b expression in MCF-7 cell lines (datas not shown).

Role of miR-190b expression in tumor proliferation

The heightened increase of miR-190b in ER+ breast cancer prompted us to explore this possible biological significance in cell proliferation. As initial step, the capacity of proliferation induction was evaluated on breast cancer cell lines ER+ MCF-7 and T-47D that were transfected with an antagomir against miR-190b and on ER− MD-MBA-231 that was transfected with a miR-190b mimic. The efficacy of transfection was verified by quantifying miR-190b in RNA extracted from transfected cells by qRT-PCR (datas not shown). Antagomir did not affect proliferation of MCF-7 (Additional file 7: Figure S4A) and T-47D cell lines (Additional file 7: Figure S4B) as miR-190b mimic has no effect on MDA-MB-231 proliferation (Additional file 7: Figure S4C). Other experiments are therefore needed to decipher the role of miR-190b in mammary tumorigenesis.

Discussion

From the study of more than 800 miRNA in ER+ en ER− breast tumors and confirmation of our results in a large validation series, we identified a robust collection of 20 miRNAs significantly deregulated in ER+ compared to ER− breast cancers: 12 up-regulated and eight down-regulated miRNAs. Among these 20 miRNAs, we found ten miRNAs similarly deregulated in ER+ and ER−, independently to their ER status: let-7a, let-7b, miR-26a1, miR-101-1, miR-30c2, miR-143, miR-26b, miR-376c and miR-92a1 which were down-regulated in both ER+/Normal and in ER−/Normal and miR-155 which was up-regulated in both ER+/Normal and in ER−/Normal. Moreover we found six miRNAs only deregulated in ER+/Normal or in ER−/Normal (miR-30e, specifically down-regulated in ER−, miR-342-5p, specifically up-regulated in ER+, miR-18a, miR-18b and miR-654-3p, specifically down-regulated in ER+ and miR-1244, specifically up-regulated in ER−) and 3 miRNAs, not particularly attractive since not deregulated in ER+/Normal nor in ER−/Normal (miR-193b, miR-203 and miR-146a). In contrast we found a very interesting miRNA, miR-190b, which was not only the strongly up-regulated in ER+ compared to ER− (FC of 23.30) but also the only miRNA for which deregulation was different in breast cancer according to ER status.

Production and maturation of miRNAs require a set of proteins collectively known as the miRNA biogenesis machinery and it is now established that alterations in this machinery can generate changes in miRNA expression and contribute thus in the development and progression of cancer. We fully support the same view since we found that several key miRNA processing genes are differentially expressed between ER+ and ER− breast cancer, which may explain the different regulatory effects of miRNAs in these two breast cancer subtypes. Indeed DROSHA, DGCR8 and DICER1 were significantly down-regulated in ER− whereas AGO2 was moderate up-regulated. These results support previous observations [16, 20, 32–34] and strengthen the pertinence of alterations of the basic miRNA biogenesis machinery in breast cancer.

For all six miRNAs located in intragenic regions, we demonstrated a significant and positive correlation between expression of host gene and its resident miRNA, in particular the well-known co-regulation of the ER+ marker miR-342 and the ER-regulated gene EVL [16, 35, 36]. However, in the study of Dvinge, results suggested a limited miRNA-host co-transcription concerning only 49 out 227 same-strand intragenic miRNAs [17].

These 20 miRNAs have been already described in early studies identifying miRNAs implicated in breast cancers [7, 9, 15, 17] but miR-190b retained our attention because it was the only miRNA strongly up-regulated in ER+ compared to ER− breast tumors with a FC of 23 much higher than all other up-regulated miRNAs. This observation remained true also in breast cancer cell lines with a FC of 43 for miR-190b. Moreover miR-190b was strongly up-regulated in ER+ breast tumors compared to normal breast cancer and especially the only miRNA for which deregulation was different according to ER status. Indeed MiR-190b was significantly up-regulated in ER+/Normal and down-regulated in ER−/Normal suggesting a different deregulation of miR-190b depending on the ER status. We could note that we did not select miR-135b because of absence of expression in our screening series whereas this microRNA would be differentially expressed in ER+ and ER− breast tumors, and is described by Aakula et al. as a regulator of ER [37]

Few studies reported miR-190b implication in cancers. A recent next generation sequencing project identified mir-190b among seven others microRNAs as a biomarker for the diagnosis of Merkel cell carcinoma [38]. In lung cancer, miR-190b could be detected easily in serum of patients to facilitate diagnosis [39]. MicroRNA expression profile associated with response to neoadjuvant chemoradiotherapy in locally advanced rectal cancer patients included miR-190b [40]. However none of these studies explored mechanism of action of miR-190b and its targets did not have been well described contrary to miR-190 that could interfere with VEGF-mediated angiogenesis [41]. Morevover miR-190b has not been selected by previous microarray breast cancer studies [16, 17, 19, 42] so we tried to decipher its properties in breast cancer. By treating MCF7 and T47D cell lines with estradiol, we demonstrated that miR-190b is not directly regulated by this hormone whereas it seems particularly deregulated in ER+ breast tumors. We could speculate that expression of miR-190b is controlled by other mechanisms like ER-signaling pathways independent of estrogen but it remains to be demonstrated [43, 44]. Transfection experiments with anti-miR-190b in MCF7 and T47D cell lines, or with mimic of miR-190b in MDA-MB-231 cell line have shown that miR-190b has probably no effect on cell proliferation. Nevertheless, if the mechanisms of its expression regulation like its exact role in oncogenesis of ER+ breast cancers are elucidated, miR-190b could become a very interesting biomarker in ER+ breast cancers as it has the advantage to be highly expressed in this subtype and therefore easy to detect by RT-qPCR. We could speculate that miR-190b would be used as a circulating biomarker for minimal residual disease follow-up in hormone-dependent breast cancers to detect therapeutic resistance and early relapses.

Last but not least, a high expression of miR-190b was associated with a prolonged MFS and EFS in breast tumors, independently to ER status. To date miR-190b just appears in one study using global microRNA expression profiling to identify markers of recurrence in ER+ patients receiving tamoxifen [45]. Ten highly significant miRNAs including miR-190b could discriminate the patient samples according to outcome but this result was not confirmed in two validation cohorts. More interesting, miR-190b and its function have been explored in a very recent study in human hepatocellular carcinoma [46]. The authors have showed that up-regulation of miR-190b could play a role for decreased IGF-1 that induce insulin resistance in hepatocellular carcinoma. IGF-1 appears to be a direct target of miR-190b and another study have demonstrated that IGF-1 and ER expressions are raised in breast cancer cases which were likely to develop tamoxifen resistance [47]. Taking into account our present work and these two recent studies, we argued that the link between miR-190b and tamoxifen resistance could be very interesting to study in breast cancers.

Conclusion

This study identified miR-190b as the highest up-regulated miRNA in ER+ breast cancers compared to ER− tumors and to normal breast tissues. Surprisingly, expression of miR-190b is not directly regulated by estradiol. Using synthetic miRNA to mimic or to antagonize miR-190b, we demonstrated that miR-190b does not affect the proliferation of transfected breast cancer cell lines. However miR-190b affects MFS of breast cancer patients. Even if miR-190b exact role in breast carcinogenesis and regulation expression mechanisms remain to elucidate, this microRNA seems to be specifically expressed in ER+ breast cancers at higher level that all others miRNAs and could therefore represent a new biomarker of interest for the follow-up of this subtype of tumors.