Downregulation of microRNAs 145-3p and 145-5p Is a Long-term Predictor of Postmenopausal Breast Cancer Risk: The ORDET Prospective Study

The identification of molecular biomarkers associated with cancer initiation and progression represents a fundamental step for the risk assessment and development of new prevention strategies. A recently identified class of noncoding small RNAs, miRNA, may provide new insights into cancer prevention and early detection methodology. MiRNAs are small, stable noncoding RNAs whose function is to bind the messenger RNAs (mRNA) of expressed genes and target them for degradation or inhibition of translation, resulting in reduced expressed protein levels (1). It is estimated that miRNAs may actually regulate up to two thirds of the human genome (2). MiRNAs have been shown to be directly involved in many human cancers, including breast, lung, brain, liver, colon, prostate, ovarian cancers, and leukemia (3). Some miRNAs function as tumor suppressors by negatively inhibiting oncogenes that control cell differentiation and apoptosis, whereas others act as oncogenes (oncomirs; refs. 4, 5). There is a growing consensus that miRNA downregulation has a profound impact on the genesis of tumors (6–8). There is evidence that an extensive downregulation of miRNAs is one of the first biologic response of the deregulation in a signaling cascade downstream of specific growth factor receptors implicated in human cancers, including breast cancer. For example, EGF signaling rapidly and simultaneously induces a massive downregulation of multiple miRNAs, reflecting coordinated regulation at the level of miRNA synthesis, processing, or degradation (9).

In the present study, we aimed to test the hypothesis that miRNAs (as a single entity or as a signature) may represent early indicators of future breast cancer incidence. Previous evidence indicated that miRNAs are deregulated, and, in particular, mainly downregulated in response to environmental/metabolic risk factors for cancer (10–12). To test the working hypothesis, we compared leukocyte miRNA profiles of healthy women who subsequently became affected with breast cancer with women who remained healthy. This was performed using a case–control study design nested in the ORDET [hORmones and Diet in the ETiology of Breast Cancer) prospective cohort study over a follow-up period of 20 years. The prospective study design also allowed corroborating the evidence that the downregulation of miRNAs represents one of the very early molecular alterations during the development of the disease. Subsequently, we evaluated whether miRNAs differently expressed in the ORDET women candidate to become breast cancer cases versus control subjects were also modulated in breast cancer tissues and had prognostic value using the well-characterized METABRIC cohort of 1,359 breast cancer cases (13). As the final phase of the study, we investigated the functional activity of the identified miRNAs in breast cancer cell lines.

The study has been conducted in the context of the ORDET prospective cohort study; the analysis included 133 incident postmenopausal breast cancer cases and 133 matched control subjects.

The ORDET cohort was established in northern Italy between June 1987 and June 1992, in which 10,786 healthy women ages 35 to 69 years were enrolled (14). They were all residents of the Varese province, an area covered by the population-based Lombardy Cancer Registry (15). They had heard about the study through the media, at public meetings, and volunteered to participate. At recruitment, we measured anthropometric variables and collected demographic information and blood samples. Because the study's focus was on endogenous hormones in relation to breast cancer risk, we also applied stringent inclusion criteria and highly standardized conditions on the collection of biologic samples.

Information on cancer outcomes available from the Lombardy Cancer Registry has been linked to the ORDET cohort to identify incident breast cancer cases up to December 31, 2006 (16).

Case subjects were women who developed breast cancer after their recruitment into the ORDET cohort and before the end of the follow-up. We randomly chose one control for each case, from appropriate risk sets consisting of all cohort members who satisfied the matching criteria and were alive and free of cancer at the time of diagnosis of the index case. Matching characteristics were age (±3 years) at enrollment and date of recruitment (±180 days). We applied an incidence density sampling protocol for control selection (17).

After exclusion of women with a history of cancer and women who, immediately after baseline, were lost to follow-up (observed time = 0), 10,633 participants remained to form the base population of ORDET.

In the ORDET study, as well as in other cohort studies, we found that risk factors differ either in their phenotypic expression or in their distribution by menopausal status (18). Thus, the present report focused on the postmenopausal group of the cohort, defined as those cohort members who had the last menstrual period at least 12 months before their enrollment in the study. In summary, within the postmenopausal members of the cohort and because of the selection criteria, we identified and included in the study 133 incident breast cancer cases and 133 matched control subjects.

We evaluated the miRNA expression profile of leukocytes derived from buffy coats collected at recruitment.

Blood samples were drawn after overnight fasting between 7:30 am and 9:00 am from each woman and stored at −80°C.

Samples from each case and related control were handled identically and assayed together in the same laboratory session. Laboratory personnel were blinded to case–control status.

Leucocytes were lysed in 1 mL of TRI Reagent, a lysis reagent from Ambion, according to the manufacturer's instructions. The concentration and purity of total RNA were assessed using a NanoDrop 1000 Spectrophotometer (NanoDrop Technologies). Total RNA (100 ng) was labeled, hybridized to Human microRNA Microarray V2 (Agilent Technologies), and scanned with Agilent DNA Microarray Scanner (P/N G2565BA) according to the manufacturer's instructions. Feature Extraction Software (Version 10.5) was used for data extraction from raw microarray image files using the microRNA_105_Dec08 FE protocol. This miRNA Agilent expression profile was submitted to the Gene Expression Omnibus (GEO) with the accession number GSE54470. Minimum Information About a Microarray Gene Experiment (MIAME) guidelines were followed as instructions. Furthermore, representative RNA preparations were evaluated for integrity using the 2100 Bioanalyzer RNA 6000 Nano Kit (Agilent Technologies; data not shown).

We also assessed the expression of miR223 in randomly selected samples by Northern blot analysis (data not shown). miR223 is highly specific for hematopoietic cells and constitutes a regulator of myelopoiesis (19). The blot was hybridized with a [³²P]γATP-radiolabeled LNA oligonucleotide complementary to miR223 sequence. The specificity and strength of hybridization of ORDET RNA samples was as good as that of human promyelocytic HL60 cells treated with retinoic acid (10⁻⁶ mol/L), a known inducer of miR223.

Data were verified and extracted by the Agilent Extraction 10.7.3.1 software and analyzed using an in-house built routines by Matlab (The MathWorks Inc.). Background-subtracted signal of 851 human miRNA assays was used in the study. All arrays were quantile normalized, assuming that all samples were measured and analyzed under the same condition, enforcing all the arrays to assume the same mean distribution. The Pearson coefficient was calculated to assess the correlation between technical replicates of some randomly chosen samples.

We fitted a linear model to the expression values for each miRNA, to assess the significance of differential expression between case and control. In addition, we used empirical Bayes methods implemented in the LIMMA package to construct moderated t statistics and incorporated the statistical tools to adjust for the multiplicity of the tests. The Benjamini and Hochberg method (1995) was used to control for false discovery.

We considered the linear model including the matched case–control study design, the case–control status, and the error term.

Data preprocessing and differential expression analysis were done using the Bioconductor AgiMicroRna package (20). The Total Gene Signal (TGS) provided by the Agilent Feature Extraction image analysis software was used as the quantitative measure of miRNA expression. We set all negative TGS values to 0.5 before log transformation, so that the log ratios are shrunk toward zero at lower intensities. The miRNA expression data (i.e., TGS) were quantile normalized before determining differential expression. The data were analyzed using the R software package. For differential expression analysis, the AgiMicroRna package incorporates the linear model with matched pair features from the Bioconductor LIMMA package (21). The LIMMA approach fits a linear model to the expression value for each miRNA to assess the significance of differential expression between different experimental conditions. In addition, the method uses empirical Bayes methods (22) to construct moderated t statistics and incorporates statistical tools to adjust for multiple testing. The Benjamini and Hochberg method (23) was used to control for false discovery rate (FDR), and we ranked the miRNAs according to FDR. We considered the top-ranked 20 miRNAs and investigated the upregulated and downregulated miRNAs identified from postmenopausal samples. We computed agglomerative hierarchical clustering of the dataset. At first, each object is assigned to its own cluster and then the algorithm proceeds iteratively. At each stage, the two most similar clusters are combined to form a larger cluster, continuing until there is just a single cluster. At each stage, distances between clusters are computed by the Lance–Williams dissimilarity update formula. Details about the clustering algorithm are given in the book by Kaufman and Rousseeuw (24). This clustering method partitions the dataset into clusters, in which similar miRNA expression patterns are assigned to the same cluster.

We identified predictive pathways using pathway analysis. Pathway analysis was performed by DIANA miRPath v2.0. The software calculated the union of targeted genes by the selected miRNAs (UNION_SET, all genes targeted by at least one selected miRNA). The UNION_SET set was used for the statistical analysis. This enrichment analysis identified the pathways significantly enriched with genes belonging to the UNION_SET (25).

To assess the miRNAs' prognostic value, we conducted a survival analysis in the Metabric cohort of breast cancer cases. This analysis was not possible on the ORDET cohort study database for the limited sample size of the breast cancer–specific mortality events. The Metabric cohort is a very well-characterized breast cancer database provided with matching detailed clinical annotation, long-term follow-up, and genomic and miRNA expression data (13). The database includes 1,302 breast tumors, which included a subgroup of 81 breast cancer cases, where each case was provided with samples derived from both the tumoral lesion and the related normal breast tissue. As an initial step of the survival analysis, we tested the consistency of the disregulation (down- versus upregulation) between the observed 20 top-ranked miRNAs in the ORDET cohort and in the tumor tissue versus the normal tissue in the subgroup of 81 breast cancer cases described in the Metabric study. For instance, the downregulated miRNAs in ORDET breast cancer cases versus control subjects were expected to have lower expression levels in breast tumors versus normal tissue. In the subsequent survival analysis, higher expression levels of these miRNAs were expected to be associated with better survival. We expected the opposite effect for the ORDET upregulated miRNAs. For testing the prognostic value, for each miRNA, we computed the P value for its differential expression between tumor tissue and normal tissue (in the subgroup of patients provided with samples of both tumor and normal tissue). Subsequently, for each miRNA, we performed survival analysis and generated Kaplan–Meier plots.

Human breast cancer cell lines MDA-MB-231 and MDA-MB-468 were obtained from the American Type Culture Collection (ATCC; www.atcc.org). ATCC uses morphology, karyotyping, and PCR-based approaches to confirm the identity of human cell lines and to rule out both intra- and interspecies contamination (www.atcc.org). After purchasing, (5 months ago) the cells were routinely tested with PCR approaches. For mature miR145-5p and miR145-3p expression, we used Pre-microRNA Precursor-Negative Control (Ambion), Pre–miR145-5p (Ambion), and Pre–miR145-3p (Ambion) at a final concentration of 5 nmol/L.

The expression levels of miR145-3p and miR145-5p were evaluated by PCR (Supplementary Fig. S1)

MDA-MB-231 cells were seeded into 6-well dishes and transfected in triplicates as indicated. Cells (6 × 10⁴) were seeded for this assay.

Cells were collected and manually counted at 0, 24, 48, 72 hours after transfection.

Migration assay was performed using a 24-well Boyden chamber with a noncoated 8-mm pore size filter in the insert chamber (BD Falcon). Cells (mimic 145-5p and control-transfected MDA-MB-231 and MDA-MB-468; 5 × 10⁴) were suspended in 0.5 mL DMEM without FBS and seeded into the insert chamber. Cells were allowed to migrate for 48 hours into the bottom chamber containing 0.7 mL of DMEM containing 5% FBS in a humidified incubator at 37°C in 5% CO₂. Migrated cells that had attached to the outside of the filter were visualized by staining with DAPI and counted.

MDA-MB-231 cells were grown to 70% confluence and transfected as indicated. Colony staining and counting were performed as described by Biagioni and colleagues (7).

Most of the baseline characteristics did not differ between the 133 breast cancer cases and 133 controls (Table 1), in particular for age, reproductive, hormonal, and life-style risk factors that could have represented confounders of the studied association.

When we conducted class comparisons to identify differentially expressed miRNAs, we first performed a moderated t test (22) for each miRNA on all 266 postmenopausal women. Most of the difference in miRNA expression between cases and controls was toward the downregulation in women who were later affected with breast cancer.

Table 2 reports all the top-ranked 20 miRNAs (ranked according to FDR values). Of these 20 miRNAs, 15 (75%) were downregulated. Among the upregulated miRNAs, miRNA892b was characterized by the lowest FDR (close to 10%), resulting in the most statistically significant differentially expressed miRNA.

An examination of miRNA expression revealed that the 20 top-ranked miRNAs were grouped in three different clusters. Figure 1 shows the heatmap of the distinct miRNA profiles following an unsupervised hierarchical clustering analysis. None of the reported 20 top-ranked miRNAs correlate to age (in either cases or controls), body mass index (BMI), serum fasting glucose, and serum fasting insulin. When we looked at the hormone receptor status of the ORDET incident breast cancer cases (e.g., estrogen and progesterone receptor status in breast cancer cases), in the subgroup of progesterone receptor–positive cases, we found that low expression levels of miR99b were statistically associated with a higher probability to develop progesterone receptor–positive breast cancer.

In Table 3, we describe the most significant cancer predicted pathways targeted by the three clusters. It is worth noting that most of the pathways identified by the clusters of differentially expressed miRNAs were related to both breast cancer and more general cancer development (e.g., ErbB, mTOR, TGFβ, Hedgehog, and Wnt pathways; refs. 26–30). We then observed that the three miRNA clusters targeted pathways also related to metabolic and endocrine systems, such as cholesterol synthesis, steroid biosynthesis, and the insulin pathway recognized in the ORDET cohort, as well as in other studies as pathways involved in breast cancer development (27, 31, 32). Finally, the mitogen-activated protein kinase (MAPK) pathway was targeted at a very high level of statistical significance by all the three clusters.

We also noticed that among the top 20 differentially regulated miRNAs, a pair of downregulated miRNAs share the same precursor: the miR145-3p and miR145-5p located on chromosome 5.

To further investigate the consistency of the observed downregulation of miR145-3p and miR145-5p in breast cancer, we analyzed three different publicly available databases of miRNA expression in breast cancer [GSE28884 (33) and GSE19536 (34) and The Cancer Genome Atlas Network (35)] and Biagioni and colleagues article database (7). As shown in Tables 4 and 5, miR145-3p and miR145-5p were consistently downregulated across all databases in tumor tissue versus peritumoral or normal tissues.

In Table 6, we also included, as the fifth database, the comparison between the tumor tissue and the normal tissue observed in the Metabric cohort study: Again, both miRNAs were downregulated in the tumor tissue versus normal tissue at a very high level of statistical significance. We conducted the same analysis for all the remaining 18 miRNAs; however, we did not notice a similar level of consistency for any of them (Table 6 and Supplementary Tables S1–S18). For instance, the downregulated miRNA-199a-5p allocated in the first cluster was equally downregulated in both tumor tissue and normal tissue in the Metabric study (13) and upregulated in the Biagioni and colleagues study (ref. 7; Supplementary Table S1).

To evaluate whether miR145-3p and miR145-5p played a role in the progression of breast cancer, we analyzed the disease-specific survival rate as a function of their low and high expression. In the Metabric breast cancer cohort, we saw that the high expression of both miR145-3p and miR145-5p was associated with longer survival: For miR145-3p, the difference with its low expression reached statistical significance (Fig. 2).

As the final step of our study, we investigated the functional activity of miR145-3p and miR145-5p in MDA-MB-231 and MDA-MB-468 breast cancer cell lines. We used these breast cancer cell lines, which are both representative of a basal-like phenotype, because they were previously used by Sachdeva and Mo, to demonstrate the involvement of miR145-5p in breast cancer invasion and metastatization (36). Because this breast cancer histotype is known to be the most aggressive one, it is also the best model to perform a migration assay. As reported in Fig. 3A–E, we observed that the two miRNAs had different and complementary activities in controlling breast cancer cells. Ectopic expression of miR145-5p inhibited cell migration but did not affect proliferation of the analyzed breast cancer cell lines (Fig. 3B). On the contrary, overexpression of miR145-3p impinged on cell proliferation and colony formation, but had no effect on cell migration (Fig. 3C and D); we did not observe any effect of miR145-3p on the migration of breast cancer cells (Fig. 3E).

In this study, we provide evidence that miRNAs might serve as early indicators of breast cancer occurrence. We observed that in leukocytes collected from healthy women who later became affected with breast cancer, differences were found in miRNA expression profiles, in comparison with leukocytes of women who did not develop the disease during the same follow-up period. The observation that among the microRNAs listed in the top 20 most differentially expressed microRNAs, 15 were down-regulated, supports our working hypothesis that downregulation of regulatory miRNAs might herald breast cancer initiation. This concept has also been experimentally validated by the progressive downregulation of miRNAs observed when passing from healthy breast tissue to breast cancer with high cell-proliferation rates (37).

The high blood stability of miRNAs, their resistance to RNA degradation, and their reproducible detection make miRNAs suitable biomarker candidates (38). A number of lifestyle factors and conditions, often related to inflammation status, have been reflected into specific deregulation of miRNA in peripheral leukocytes (39). Lifestyle and dietary factors are both related to inflammation status and breast cancer risk. Macronutrient intake, characteristic of the Western diet, and obesity may activate inflammatory signaling pathways (40). Elevated levels of the proinflammatory cytokines promote angiogenesis, tumor progression, and metastasis (41).

The role of miRNAs isolated from leukocytes as biomarkers of cancer occurrence has been underlined in this study by the predicted pathway analysis conducted on the basis of the three miRNA clusters. All clusters seem to target, with the highest level of statistical significance, pathways related to both breast and other cancer development. At the same time, the clusters targeted genes involved in metabolic and hormone risk factors for breast cancer.

When considering miR145-3p and miR145-5p downregulation, we found that it was consistent across five unrelated databases, with their high expression associated with better survival. This association was statistically significant for miR145-3p. This was further corroborated by our findings that miR145-3p apparently controls cell proliferation, whereas miR145-5p's activity seems to modulate cell migration and invasion. Thus, we propose that biologic functions of these two miRNAs might differ according to the different phases of cancer development. The two miRNAs derive from each arm of their pre-miR hairpin. In a previous publication of our group, Biagioni and colleagues (7) observed a similar opposite and complementary behavior of two miRNAs that, similar to miR145-3p and miR145-5p, shared the same precursor: miR10b-5p and miR10b-3p. The overexpression of miR10b-5p was associated with tumor invasion and metastasis (42), whereas the overexpression of miR10b-3p was related to cell-cycle progression and proliferation (7). Following this model, we may consider miR145-3p downregulation as a very early signal of breast cancer development for its specific control of the first step of carcinogenesis: cell proliferation. miR145-5p has been shown to act as a tumor suppressor in different tumor types (36). Its downregulation was associated with neoplastic cell growth, invasion, and metastasis and is paired with increased expression of different oncogenic mRNA targets expression, such as EGFR, MYC, MUC1, OCT4, and RTKN (8, 36, 43).

There is very scarce evidence on the involvement of miR145-3p in cancer progression. Indeed, Camps and colleagues (44) observed downregulation of miR145-3p expression in MCF-7 cells exposed to hypoxia.

The evidence showing that only two miRNAs out of 20 were corroborated in all different databases was expected. Differences in phase of the natural history of the disease at sample collection (ORDET was based on samples derived from women still healthy, whereas all the other samples were derived from breast cancer lesions) and differences in study design reflected in differences in study populations by age and/or menopausal status may explain the results. In particular, a comparison of miRNA profiling between leukocyte samples (ORDET) and tissue samples (Metabric and the other database) represents a limitation for our direct study inference because miRNA profiling was performed in different tissues. However, we hypothesized that leukocyte miRNA profile (ORDET samples) did reflect, in still healthy women, exposure to nutritional and metabolic determinants of breast cancer. Then, differences in miRNA profiling between women who later developed breast cancer and women who remained healthy had potential etiologic meaning. Subsequently, in the Metabric study, as well as in other databases, we investigated whether tissue from breast cancer cases, as a later event, reflected similar miRNA profiling, effect of exposure to risk factors, we observed in leukocytes collected from those ORDET women candidate to develop breast cancer.

The prospective nature of the observation that specific miRNAs were deregulated in leukocytes collected from healthy women to develop breast cancer 20 years before disease onset makes our study unique when compared with previous studies. The prospective design of our study assigns to both miR145-3p and miR145-5p a role as long-term breast cancer predictors and it also opens a new avenue for prevention.

No potential conflicts of interest were disclosed.

Conception and design: P. Muti, F. Berrino, S. Strano, G. Blandino

Development of methodology: P. Muti, J. Beyene

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): P. Muti, S. Donzelli, F. Ganci, S. Sieri, V. Krogh, F. Berrino

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): P. Muti, A. Sacconi, A. Hossain, N.B.B. Moshe, V. Krogh, J. Beyene

Writing, review, and/or revision of the manuscript: P. Muti, A. Hossain, S. Sieri, V. Krogh, F. Biagioni, S. Strano, J. Beyene, G. Blandino

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): P. Muti, Y. Yarden

Study supervision: P. Muti, G. Blandino

Other (provided microarray data): F. Biagioni

The authors thank the 10,786 ORDET participants. They also thank Dr. Paolo Contiero and the staff of the Lombardy Cancer Registry for technical assistance.

This work was supported by the Department of Defense grant W81 XWH 04 1 0195 and by the Veronesi Foundation.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.