MicroRNAs (miRNAs) are a class of small noncoding RNAs that control gene expression by targeting mRNAs and triggering either translation repression or RNA degradation. Their aberrant expression may be involved in human diseases, including cancer. Indeed, miRNA aberrant expression has been previously found in human chronic lymphocytic leukemias, where miRNA signatures were associated with specific clinicobiological features. Here, we show that, compared with normal breast tissue, miRNAs are also aberrantly expressed in human breast cancer. The overall miRNA expression could clearly separate normal versus cancer tissues, with the most significantly deregulated miRNAs being mir-125b, mir-145, mir-21, and mir-155. Results were confirmed by microarray and Northern blot analyses. We could identify miRNAs whose expression was correlated with specific breast cancer biopathologic features, such as estrogen and progesterone receptor expression, tumor stage, vascular invasion, or proliferation index.

MicroRNAs (miRNAs) represent a class of naturally occurring small noncoding RNA molecules, distinct from but related to small interfering RNAs. Mature miRNAs are 19- to 25-nucleotide-long molecules cleaved from 70- to 100-nucleotide hairpin pre-miRNA precursors (1). The precursor is cleaved by cytoplasmic RNase III Dicer into ∼22-nucleotide miRNA duplex: one strand (miRNA*) of the short-lived duplex is degraded, whereas the other strand serves as mature miRNA. In animals, single-stranded miRNA binds through partial sequence homology to the 3′ untranslated region (3′ UTR) of target mRNAs, and causes either block of translation or, less frequently, mRNA degradation. The discovery of this class of genes has identified a new layer of gene regulation mechanisms, which play an important role in development and in various cellular processes, such as differentiation, cell growth, and cell death (2). Deviations from normal pattern of expression may play a role in diseases, such as in neurologic disorders (3).

Among human diseases, it has been shown that miRNAs are aberrantly expressed or mutated in cancer, suggesting that they may play a role as a novel class of oncogenes or tumor suppressor genes. The first evidence of involvement of miRNAs in human cancer came from molecular studies characterizing the 13q14 deletion in human chronic lymphocytic leukemia (CLL), which revealed that two miRNAs, mir-15a and mir-16-1, were the only genes within the smallest common region of deletion. The same two genes were affected by a chromosomal translocation in a CLL patient. mir-16-1 and/or mir-15a were then found down-regulated in 50% to 60% of human CLL (4). Following this initial finding, miRNA expression deregulation in human cancer has been proven in other instances. For example, miR143 and miR145 are down-regulated in colon carcinomas (5). Let-7 is down-regulated in human lung carcinomas and restoration of its expression induces cell growth inhibition in lung cancer A549 cells (6). The BIC gene, which contains the miR155, is strongly up-regulated in some Burkitt's lymphoma and several other types of lymphomas (7, 8). The findings that miRNAs have a role in human cancer is further supported by the fact that >50% of miRNA genes are located at chromosomal regions, such as fragile sites, and regions of deletion or amplification that are genetically altered in human cancer (9), suggesting that the relevance of miRNAs in human cancer may be presently underestimated.

Only recently, the possibility of analyzing the entire miRNAome has become possible by the development of microarrays containing all known human miRNAs (1015). The use of miRNA microarrays made possible to confirm miR-16 deregulation in human CLL, but also recognize miRNA expression signatures associated with well-defined clinicopathologic features of human CLL (16). Recognition of miRNAs that are differentially expressed between normal and tumor samples may help to identify those that are involved in human cancer and establish the basis to unravel their pathogenic role. Here, we present results of a genome-wide miRNA expression profiling in a large set of normal and tumor breast tissues demonstrating the existence of a breast cancer–specific miRNA signature.

Breast cancer samples and cell lines. RNAs from primary tumors were from 76 samples collected at the University of Ferrara (Italy), Istituto Nazionale dei Tumori, Milano (Italy), and Thomas Jefferson University (Philadelphia, PA). Clinicopathologic information was available for 58 tumor samples. RNAs from normal samples consisted of six pools of five normal breast tissues each and four additional single breast tissues. RNAs of human breast cell lines were from Hs578-T, MCF7, T47D, BT20, SK-BR-3, HBL100, HCC2218, MDA-MB-175, MDA-MB-231, MDA-MB-361, MDA-MB-435, MDA-MB-436, MDA-MB-453, and MDA-MB-468.

Immunohistochemical analysis of breast cancer samples. Hormonal receptors were evaluated with 6F11 antibody for estrogen receptor α and PGR-1A6 for progesterone receptor (Ventana, Tucson, AZ). The proliferation index was assessed with MIB1 antibody (DAKO, Copenhagen, Denmark). ERBB2 was detected with CB11 (Ventana) and p53 protein expression was examined with DO7 (Ventana). Staining procedures were done as described (17). Only tumor cells with distinct nuclear immunostaining for estrogen receptor, progesterone receptor, Mib1, and p53 were recorded as positive. Tumor cells were considered positive for ERBB2 when they showed distinct membrane immunoreactivity. To perform a quantitative evaluation of biological markers, the Eureka Menarini computerized image analysis system was used. For each tumor section, at least 20 microscopic fields of invasive carcinoma (40× objective) were measured. The following cutoff values were used: 10% of positive nuclear area for estrogen receptor, progesterone receptor, c-erbB2, and p53; 13% of nuclei expressing Mib1 was introduced to discriminate cases with high and low proliferative activity.

MicroRNA microarray. Total RNA isolation was done with Trizol (Invitrogen, Carlsbad, CA) according to the instructions of the manufacturer. RNA labeling and hybridization on miRNA microarray chips was done as previously described (10). Briefly, 5 μg of RNA from each sample was biotin-labeled during reverse transcription using random examers. Hybridization was carried out on miRNA microarray chip (KCI version 1.0; ref. 10), which contains 368 probes, including 245 human and mouse miRNA genes, in triplicate. Hybridization signals were detected by biotin binding of a Streptavidin–Alexa 647 conjugate using a Perkin-Elmer ScanArray XL5K. Scanner images were quantified by the Quantarray software (Perkin-Elmer, Wellesley, MA).

Statistical and bioinformatic analysis of microarray data. Raw data were normalized and analyzed using the GeneSpring software version 7.2 (Silicon Genetics, Redwood City, CA). Expression data were median centered. Statistical comparisons were done by ANOVA, using the Benjamini and Hochberg correction for false-positive reductions. Prognostic miRNAs for tumor versus normal class prediction were determined by using both the Prediction Analysis of Microarrays software (PAM; ref. 18)7

and the Support Vector Machine (19) tool. Both algorithms were used for cross-validation and test-set prediction. All data were submitted using MIAMExpress to the Array Express database (accession numbers to be received upon revision).

Northern blotting. Northern blot analysis was done as previously described (4). RNA samples (10 mg each) were electrophoresed on 15% acrylamide, 7 mol/L urea Criterion precasted gels (Bio-Rad, Hercules, CA) and transferred onto Hybond-N+ membrane (Amersham Biosciences, Piscataway, NJ). Hybridization was done at 37°C in 7% SDS/0.2 mol/L Na2PO4 (pH 7.0) for 16 hours. Membranes were washed at 42°C, twice with 2× standard saline phosphate [0.18 mol/L NaCl/10 mmol/L phosphate (pH 7.4)], 1 mmol/L EDTA (saline-sodium phosphate-EDTA, SSPE), and 0.1% SDS and twice with 0.5× SSPE/0.1% SDS. The oligonucleotides used as probes are the complementary sequences of the mature miRNA (miR Registry):8

miR21 5′-TCAACATCAGTCTGATAAGCTA-3′; miR125b1: 5′-TCACAAGTTAGGGTCTCAGGGA-3′; miR145: 5′-AAGGGATTCCTGGGAAAACTGGAC-3′. An oligonucleotide complementary to the U6 RNA (5′-GCAGGGGCCATGCTAATCTTCTCTGTATCG-3′) was used to normalize expression levels. Two hundred nanograms of each probe was end labeled with 100 mCi [γ-32P]ATP using the polynucleotide kinase (Roche, Basel, Switzerland). Blots were stripped in boiling 0.1% SDS for 10 minutes before rehybridization.

A microRNA expression signature discriminates between normal and cancer breast tissues. We used a miRNA microarray (10) to evaluate miRNA expression profiles of 10 normal and 76 neoplastic breast tissues. Each tumor sample was derived from a single specimen; 6 of the 10 normal samples consisted of pools made of five different normal breast tissue RNAs; hence, 34 normal breast samples were actually examined in the study.

To identify miRNA whose expression was significantly different between normal and tumor samples and could identify the different nature of these breast tissues, we made use of ANOVA and class prediction statistical tools.

To identify differentially expressed miRNAs among all the human miRNAs spotted on the chip, the ANOVA analysis on normalized data generated a list of differentially expressed miRNAs (at P < 0.05) between normal breasts and breast cancers (Table 1). Cluster analysis, based on differentially expressed miRNA, generated a tree with clear distinction between normal and cancer tissues (Fig. 1A).

Table 1.

miRNAs differentially expressed between breast carcinoma and normal breast tissue

PBreast cancer
Normal breast
Median
Range
Median
Range
NormalizedMinMaxNormalizedMinMax
let-7a-2 1.94E−02 1.67 0.96 6.21 2.30 1.34 5.00 
let-7a-3 4.19E−02 1.26 0.81 3.79 1.58 1.02 2.91 
let-7d (=7d-v14.61E−03 0.90 0.59 1.54 1.01 0.83 1.25 
let-7f-2 6.57E−03 0.84 0.51 1.58 0.92 0.76 1.03 
let-7i (= let-7d-v23.38E−02 2.05 1.02 7.49 1.53 1.01 3.47 
mir-009-1 (mir-131-19.12E−03 1.36 0.69 4.16 1.01 0.61 2.44 
mir-010b 4.49E−02 1.11 0.69 4.79 1.70 0.96 6.32 
mir-021 4.67E−03 1.67 0.66 26.43 1.08 0.80 2.31 
mir-034 (=mir-1701.06E−02 1.67 0.70 6.40 1.09 0.65 3.17 
mir-101-1 4.15E−03 0.83 0.52 1.26 0.90 0.77 1.05 
mir-122a 3.43E−03 2.21 0.93 8.08 1.48 1.06 3.67 
mir-125a 3.28E−03 1.20 0.69 2.36 1.73 1.21 3.34 
mir-125b-1 2.65E−02 1.30 0.55 8.85 2.87 1.45 18.38 
mir-125b-2 2.33E−02 1.26 0.69 6.29 2.63 1.40 16.78 
mir-128b 1.60E−02 1.12 0.68 7.34 1.02 0.89 1.27 
mir-136 2.42E−03 1.32 0.74 10.26 1.06 0.76 1.47 
mir-143 7.11E−03 0.87 0.68 1.33 0.96 0.81 1.17 
mir-145 4.02E−03 1.52 0.92 8.46 3.61 1.65 14.45 
mir-149 2.75E−02 1.11 0.53 1.73 1.03 0.83 1.22 
mir-155 (BIC1.24E−03 1.75 0.95 11.45 1.37 1.11 1.88 
mir-191 4.26E−02 5.17 1.03 37.81 3.12 1.45 14.56 
mir-196-1 1.07E−02 1.20 0.57 3.95 0.95 0.66 1.75 
mir-196-2 1.16E−03 1.46 0.57 5.55 1.04 0.79 1.80 
mir-202 1.25E−02 1.05 0.71 2.03 0.89 0.65 1.20 
mir-203 4.06E−07 1.12 0.50 5.69 0.86 0.71 1.04 
mir-204 2.15E−03 0.78 0.48 1.04 0.89 0.72 1.08 
mir-206 1.42E−02 2.55 1.22 6.42 1.95 1.34 3.22 
mir-210 6.40E−13 1.60 0.98 12.13 1.12 0.97 1.29 
mir-213 1.08E−02 3.72 1.42 40.83 2.47 1.35 5.91 
PBreast cancer
Normal breast
Median
Range
Median
Range
NormalizedMinMaxNormalizedMinMax
let-7a-2 1.94E−02 1.67 0.96 6.21 2.30 1.34 5.00 
let-7a-3 4.19E−02 1.26 0.81 3.79 1.58 1.02 2.91 
let-7d (=7d-v14.61E−03 0.90 0.59 1.54 1.01 0.83 1.25 
let-7f-2 6.57E−03 0.84 0.51 1.58 0.92 0.76 1.03 
let-7i (= let-7d-v23.38E−02 2.05 1.02 7.49 1.53 1.01 3.47 
mir-009-1 (mir-131-19.12E−03 1.36 0.69 4.16 1.01 0.61 2.44 
mir-010b 4.49E−02 1.11 0.69 4.79 1.70 0.96 6.32 
mir-021 4.67E−03 1.67 0.66 26.43 1.08 0.80 2.31 
mir-034 (=mir-1701.06E−02 1.67 0.70 6.40 1.09 0.65 3.17 
mir-101-1 4.15E−03 0.83 0.52 1.26 0.90 0.77 1.05 
mir-122a 3.43E−03 2.21 0.93 8.08 1.48 1.06 3.67 
mir-125a 3.28E−03 1.20 0.69 2.36 1.73 1.21 3.34 
mir-125b-1 2.65E−02 1.30 0.55 8.85 2.87 1.45 18.38 
mir-125b-2 2.33E−02 1.26 0.69 6.29 2.63 1.40 16.78 
mir-128b 1.60E−02 1.12 0.68 7.34 1.02 0.89 1.27 
mir-136 2.42E−03 1.32 0.74 10.26 1.06 0.76 1.47 
mir-143 7.11E−03 0.87 0.68 1.33 0.96 0.81 1.17 
mir-145 4.02E−03 1.52 0.92 8.46 3.61 1.65 14.45 
mir-149 2.75E−02 1.11 0.53 1.73 1.03 0.83 1.22 
mir-155 (BIC1.24E−03 1.75 0.95 11.45 1.37 1.11 1.88 
mir-191 4.26E−02 5.17 1.03 37.81 3.12 1.45 14.56 
mir-196-1 1.07E−02 1.20 0.57 3.95 0.95 0.66 1.75 
mir-196-2 1.16E−03 1.46 0.57 5.55 1.04 0.79 1.80 
mir-202 1.25E−02 1.05 0.71 2.03 0.89 0.65 1.20 
mir-203 4.06E−07 1.12 0.50 5.69 0.86 0.71 1.04 
mir-204 2.15E−03 0.78 0.48 1.04 0.89 0.72 1.08 
mir-206 1.42E−02 2.55 1.22 6.42 1.95 1.34 3.22 
mir-210 6.40E−13 1.60 0.98 12.13 1.12 0.97 1.29 
mir-213 1.08E−02 3.72 1.42 40.83 2.47 1.35 5.91 
Figure 1.

Cluster analysis and PAM prediction in breast cancer and normal breast tissues. A, tree generated by a cluster analysis showing the separation of breast cancer from normal tissues on the basis of miRNA differentially expressed (P < 0.05) between breast cancer and normal tissue (see Supplementary Table S1). The bar at the bottom indicates the group of cancer samples (red) or the group of normal breast tissues (yellow). B, PAM analysis displaying the graphical representation of the probabilities (0.0-1.0) of each sample for being a cancer or a normal tissue. All breast cancer and normal tissues were correctly predicted by the miR signature shown in Table 1. C, Northern blot analysis of human breast carcinomas and breast cancer cell lines with probes mir-125b, mir-145, and mir-21. The U6 probe was used for normalization of expression levels in the different lanes.

Figure 1.

Cluster analysis and PAM prediction in breast cancer and normal breast tissues. A, tree generated by a cluster analysis showing the separation of breast cancer from normal tissues on the basis of miRNA differentially expressed (P < 0.05) between breast cancer and normal tissue (see Supplementary Table S1). The bar at the bottom indicates the group of cancer samples (red) or the group of normal breast tissues (yellow). B, PAM analysis displaying the graphical representation of the probabilities (0.0-1.0) of each sample for being a cancer or a normal tissue. All breast cancer and normal tissues were correctly predicted by the miR signature shown in Table 1. C, Northern blot analysis of human breast carcinomas and breast cancer cell lines with probes mir-125b, mir-145, and mir-21. The U6 probe was used for normalization of expression levels in the different lanes.

Close modal

To identify the smallest set of predictive miRNAs differentiating normal versus cancer tissues, we used the Support Vector Machine (GeneSpring software; ref. 19) and PAM (18).9

Results from the two types of class prediction analysis were largely overlapping (Table 2; Fig. 1B). Among miRNAs listed in Table 2, 11 of 15 have an ANOVA P value of <0.05.

Table 2.

Normal and tumor breast tissue class predictor miRNAs

miRNA nameMedian expression
ANOVA* PSVM prediction strengthPAM score
Chromosome map
CancerNormalCancerNormal
mir-009-1 1.36 1.01 0.0091 8.05 0.011 −0.102 1q22 
mir-010b 1.11 1.70 0.0449 8.70 −0.032 0.299 2q31 
mir-021 1.67 1.08 0.0047 10.20 0.025 −0.235 17q23.2 
mir-034 1.67 1.09 0.0106 8.05 0.011 −0.106 1p36.22 
mir-102 (mir-29b1.36 1.14 >0.10 8.92 0.000 −0.004 1q32.2-32.3 
mir-123 (mir-1260.92 1.13 0.0940 9.13 −0.015 0.138 9q34 
mir-125a 1.20 1.73 0.0033 8.99 −0.040 0.381 19q13.4 
mir-125b-1 1.30 2.87 0.0265 14.78 −0.096 0.915 11q24.1 
mir-125b-2 1.26 2.63 0.0233 17.62 −0.106 1.006 21q11.2 
mir-140-as 0.93 1.10 0.0695 11.01 −0.005 0.050 16q22.1 
mir-145 1.52 3.61 0.0040 12.93 −0.158 1.502 5q32-33 
mir-155 (BIC1.75 1.37 0.0012 10.92 0.003 −0.030 21q21 
mir-194 0.96 1.09 >0.10 11.12 −0.025 0.234 1q41 
mir-204 0.78 0.89 0.0022 8.10 −0.015 0.144 9q21.1 
mir-213 3.72 2.47 0.0108 9.44 0.023 −0.220 1q31.3-q32.1 
miRNA nameMedian expression
ANOVA* PSVM prediction strengthPAM score
Chromosome map
CancerNormalCancerNormal
mir-009-1 1.36 1.01 0.0091 8.05 0.011 −0.102 1q22 
mir-010b 1.11 1.70 0.0449 8.70 −0.032 0.299 2q31 
mir-021 1.67 1.08 0.0047 10.20 0.025 −0.235 17q23.2 
mir-034 1.67 1.09 0.0106 8.05 0.011 −0.106 1p36.22 
mir-102 (mir-29b1.36 1.14 >0.10 8.92 0.000 −0.004 1q32.2-32.3 
mir-123 (mir-1260.92 1.13 0.0940 9.13 −0.015 0.138 9q34 
mir-125a 1.20 1.73 0.0033 8.99 −0.040 0.381 19q13.4 
mir-125b-1 1.30 2.87 0.0265 14.78 −0.096 0.915 11q24.1 
mir-125b-2 1.26 2.63 0.0233 17.62 −0.106 1.006 21q11.2 
mir-140-as 0.93 1.10 0.0695 11.01 −0.005 0.050 16q22.1 
mir-145 1.52 3.61 0.0040 12.93 −0.158 1.502 5q32-33 
mir-155 (BIC1.75 1.37 0.0012 10.92 0.003 −0.030 21q21 
mir-194 0.96 1.09 >0.10 11.12 −0.025 0.234 1q41 
mir-204 0.78 0.89 0.0022 8.10 −0.015 0.144 9q21.1 
mir-213 3.72 2.47 0.0108 9.44 0.023 −0.220 1q31.3-q32.1 
*

ANOVA (Welch t test in the Genespring software package) as calculated in Table 1.

Support Vector Machine prediction analysis tool (from Genespring 7.2 software package). Prediction strengths are calculated as negative natural log of the probability to predict the observed number of samples, in one of the two classes, by chance. The higher is the score, the best is the prediction strength.

Centroid scores for the two classes of the PAM (18).

To confirm results obtained by microarray analysis, we carried out Northern blot analysis on some of the differentially expressed miRNAs. We analyzed the expression of mir-125b, mir-145, and mir-21 in human breast cancers and in breast cancer cell lines. All Northern blots confirmed results obtained by microarray analysis, and in many cases differences seemed even stronger than that anticipated from microarray studies (Fig. 1C).

Given that biological significance of miRNA deregulation relies on their protein-coding gene targets, we analyzed the predicted targets of the most significantly down-regulated and up-regulated miRNAs: miR-10b, miR125b, miR-145, miR-21, and miR-155. The analysis was done using the three algorithms, miRanda, TargetScan, and PicTar, commonly used to predict human miRNA gene targets (2022). Because any of the three approaches generates an unpredictable number of false positives, results were intersected to identify the genes commonly predicted by at least two of the methods. Results are shown in Supplementary Table S1.

Biopathologic features and microRNA expression. We analyzed results from miRNA expression profiles in breast cancer to evaluate whether a correlation existed with various biopathologic features associated with tumor specimens. We analyzed lobular versus ductal histotypes, breast cancers with differential estrogen receptor α or progesterone receptor expression, lymph nodes metastasis, vascular invasion, proliferation index, expression of ERBB2, and immunohistochemical detection of p53. Lobular versus ductal and ERBB2 expression classes did not reveal any differentially expressed miRNA, whereas all other comparisons revealed a small number of differentially expressed miRNAs (P < 0.05). Tumor grade was not analyzed because the only two grade 1 samples were a size too small to be compared with a large number of grade 2 or 3 samples. Complete results are shown in Table 3.

Table 3.

Differentially expressed miRNAs associated with invasive breast cancer biopathologic features

Median expression
No. samples
20
13
P
FeatureER+ER−
mir-26a 2.473 1.483 0.0273 
mir-26b 3.751 1.932 0.0273 
mir-29b 1.280 0.935 0.0188 
mir-30a-5p 1.779 1.202 0.0191 
mir-30b 1.810 1.184 0.0250 
mir-30c 1.587 1.040 0.0191 
mir-30d 2.986 1.736 0.0273 
mir-185 1.568 2.296 0.0399 
mir-191 6.354 2.908 0.0273 
mir-206 1.811 2.373 0.0273 
mir-212 2.811 3.905 0.0403 
    
No. samples
 
18
 
14
 
P
 
Feature
 
PR+
 
PR−
 

 
let-7c 1.445 1.129 0.0130 
mir-26a 2.451 1.673 0.0474 
mir-29b 1.283 0.997 0.0194 
mir-30a-5p 1.879 1.219 0.0012 
mir-30b 1.898 1.220 0.0044 
mir-30c 1.643 1.089 0.0047 
mir-30d 3.211 1.777 0.0055 
    
No. samples
 
9
 
22
 
P
 
Feature
 
pT1
 
pT2-3
 

 
mir-9-2 0.894 0.840 0.0078 
mir-15a 0.905 0.830 0.0024 
mir-21 1.080 1.348 0.0040 
mir-30a-s 0.944 0.875 0.0065 
mir-133a-1 0.928 0.843 0.0025 
mir-137 0.894 0.818 0.0100 
mir-153-2 0.896 0.833 0.0096 
mir-154 0.924 0.852 0.0062 
mir-181a 1.024 1.225 0.0045 
mir-203 0.905 1.102 0.0011 
mir-213 1.915 3.197 0.0003 
    
No. samples
 
16
 
6
 
P
 
Feature
 
pN0
 
pN10+
 

 
let-7f-1 1.195 1.053 0.0378 
let-7a-3 1.191 1.039 0.0303 
let-7a-2 1.470 1.213 0.0300 
mir-9-3 1.634 1.344 0.0152 
    
No. samples
 
21
 
11
 
P
 
Feature
 
Vascular invasion absent
 
Vascular invasion present
 

 
mir-9-3 1.059 0.988 0.0451 
mir-10b 1.048 0.972 0.0210 
mir-27a 1.104 0.992 0.0317 
mir-29a 1.101 0.970 0.0346 
mir-123 1.125 0.852 0.0161 
mir-205 1.299 0.762 0.0451 
    
No. samples
 
26
 
23
 
P
 
Feature
 
Low PI
 
High PI
 

 
let-7c 1.817 1.361 0.0071 
let-7d 1.594 1.310 0.0073 
mir-26a 2.602 1.928 0.0492 
mir-26b 4.039 2.695 0.0297 
mir-30a-5p 1.783 1.394 0.0257 
mir-102 1.389 1.037 0.0017 
mir-145 1.557 1.281 0.0136 
    
No. samples
 
39
 
14
 
P
 
Feature
 
p53+
 
p53−
 

 
mir-16a 0.895 1.030 0.0026 
mir-128b 0.964 1.059 0.0096 
Median expression
No. samples
20
13
P
FeatureER+ER−
mir-26a 2.473 1.483 0.0273 
mir-26b 3.751 1.932 0.0273 
mir-29b 1.280 0.935 0.0188 
mir-30a-5p 1.779 1.202 0.0191 
mir-30b 1.810 1.184 0.0250 
mir-30c 1.587 1.040 0.0191 
mir-30d 2.986 1.736 0.0273 
mir-185 1.568 2.296 0.0399 
mir-191 6.354 2.908 0.0273 
mir-206 1.811 2.373 0.0273 
mir-212 2.811 3.905 0.0403 
    
No. samples
 
18
 
14
 
P
 
Feature
 
PR+
 
PR−
 

 
let-7c 1.445 1.129 0.0130 
mir-26a 2.451 1.673 0.0474 
mir-29b 1.283 0.997 0.0194 
mir-30a-5p 1.879 1.219 0.0012 
mir-30b 1.898 1.220 0.0044 
mir-30c 1.643 1.089 0.0047 
mir-30d 3.211 1.777 0.0055 
    
No. samples
 
9
 
22
 
P
 
Feature
 
pT1
 
pT2-3
 

 
mir-9-2 0.894 0.840 0.0078 
mir-15a 0.905 0.830 0.0024 
mir-21 1.080 1.348 0.0040 
mir-30a-s 0.944 0.875 0.0065 
mir-133a-1 0.928 0.843 0.0025 
mir-137 0.894 0.818 0.0100 
mir-153-2 0.896 0.833 0.0096 
mir-154 0.924 0.852 0.0062 
mir-181a 1.024 1.225 0.0045 
mir-203 0.905 1.102 0.0011 
mir-213 1.915 3.197 0.0003 
    
No. samples
 
16
 
6
 
P
 
Feature
 
pN0
 
pN10+
 

 
let-7f-1 1.195 1.053 0.0378 
let-7a-3 1.191 1.039 0.0303 
let-7a-2 1.470 1.213 0.0300 
mir-9-3 1.634 1.344 0.0152 
    
No. samples
 
21
 
11
 
P
 
Feature
 
Vascular invasion absent
 
Vascular invasion present
 

 
mir-9-3 1.059 0.988 0.0451 
mir-10b 1.048 0.972 0.0210 
mir-27a 1.104 0.992 0.0317 
mir-29a 1.101 0.970 0.0346 
mir-123 1.125 0.852 0.0161 
mir-205 1.299 0.762 0.0451 
    
No. samples
 
26
 
23
 
P
 
Feature
 
Low PI
 
High PI
 

 
let-7c 1.817 1.361 0.0071 
let-7d 1.594 1.310 0.0073 
mir-26a 2.602 1.928 0.0492 
mir-26b 4.039 2.695 0.0297 
mir-30a-5p 1.783 1.394 0.0257 
mir-102 1.389 1.037 0.0017 
mir-145 1.557 1.281 0.0136 
    
No. samples
 
39
 
14
 
P
 
Feature
 
p53+
 
p53−
 

 
mir-16a 0.895 1.030 0.0026 
mir-128b 0.964 1.059 0.0096 

Abbreviations: ER, estrogen receptor; PR, progesterone receptor; pT, tumor stage; pN, positive lymph nodes; low PI, low proliferation index, MIB-1 < 20; high PI, high proliferation index, MIB-1 > 30.

We have analyzed 76 breast cancer and 10 normal breast samples to identify miRNAs whose expression is significantly deregulated in cancer versus normal breast tissues. We have indeed identified 29 miRNAs whose expression is significantly deregulated (at P < 0.05) and a smaller set of 15 miRNAs that were able to correctly predict the nature of the sample analyzed (i.e., tumor or normal breast tissue) with 100% accuracy. These results leave few doubts that aberrant expression of miRNA is indeed involved in human breast cancer.

Among the differentially expressed miRNAs, miR-10b, miR-125b, miR145, miR-21, and miR-155 emerged as the most consistently deregulated in breast cancer. Three of them, miR-10b, miR-125b, and miR-145, were down-regulated and the remaining two, miR-21 and miR-155, were up-regulated, suggesting that they may potentially act as tumor suppressor genes or oncogenes, respectively.

It has been reported that the miR-125b, a putative homologue of lin-4 in Caenorhabditis elegans, and the let-7 miRNAs are induced during in vitro retinoic acid–induced differentiation of Tera-2 or embryonic stem cells. Furthermore, high expression of human miR-125b seems to be present in differentiated cells or tissues (23). Here, we show that breast cancer primary tumors and cell lines show evidence of a decreased level of miR-125b expression, suggesting that lack of miR-125 may impair differentiation capabilities of cancer cells.

At present, the lack of knowledge about bona fide miRNA gene targets hampers a full understanding on the biological functions deregulated by miRNA aberrant expression. To partially overcome this limitation, we made use of presently available computational approaches to predict gene targets (21, 22, 24). Supplementary Table S1 shows targets that were predicted by at least two of the methods, and shows that various cancer-associated genes are potentially regulated by miRNAs aberrantly expressed in breast cancer.

It may be expected that targets of down-regulated miRNAs include oncogenes or genes encoding proteins with potential oncogenic functions. Indeed, among putative targets, several genes with potential oncogenic functions could be found, such as FLT1 and the v-crk homologue, the growth factor BDNF, and the transducing factor SHC1 predicted as miR-10b targets. Among putative targets of miR-125b, potential oncogenic functions included the oncogenes YES, ETS1, TEL, and AKT3; the growth factor receptor FGFR2; or members of the mitogen-activated signal transduction pathway VTS58635, MAP3K10, MAP3K11, and MAPK14. The oncogenes MYCN, FOS, YES, and FLI1; integration site of Friend leukemia virus; cell cycle promoters such as cyclins D2 and L1; and MAPK transduction proteins such as MAP3K3 and MAP4K4 were predicted targets for miR-145. Interestingly, the proto-oncogene YES and the core-binding transcription factor CBFB were potential targets of both miR-125 and miR-145.

For the up-regulated miRNAs miR-21 and miR-155, it may be expected that gene targets belong to the class of tumor suppressor genes. For miR-21, the TGFB gene was predicted as target of miR-21 by all three methods. For miR-155, potential targets included the tumor suppressor genes SOCS1 and APC, and the kinase WEE1, which blocks the activity of Cdc2 and prevents entry into mitosis. The hypoxia-inducible factor HIF1A was also a predicted target. Interestingly, among predicted genes, the tripartite motif-containing protein TRIM2, the proto-oncogene SKI, and the RAS homologues RAB6A and RAB6C were found as potential targets of both miR-21 and miR-155.

miRNAs were found differentially expressed in various biopathologic features distinctive of human breast cancer. Some of these findings are worth noticing. For example, mir-30s are all down-regulated in both estrogen receptor– and progesterone receptor–negative tumors, suggesting that expression of these miRNAs is regulated by these hormones. Another interesting observation is the finding that the expression of various let-7 miRNAs was down-regulated in breast cancer samples with either lymph node metastasis or higher proliferation index, suggesting that a reduced let-7 expression could be associated with a poor prognosis. An association between let-7 down-regulation and poor prognosis was previously reported in human lung cancer (6). The finding that the let-7 family of miRNAs regulates the expression of the RAS oncogene family provides a potential explanation for the role of the let-7 miRNAs in human cancer (25). Two miRNA, miR-145 and miR-21, whose expression could differentiate cancer versus normal tissues, were also differentially expressed in cancers with different proliferation indexes or different tumor stage. In particular, miR-145 is progressively down-regulated from normal breast to cancer with high proliferation index. Similarly, but in opposite direction, miR-21 is progressively up-regulated from normal breast to cancers with high tumor stage. These findings suggest that deregulation of these two miRNAs may affect critical molecular events involved in tumor progression. Another miRNA potentially involved in cancer progression is miR-9-3. miR-9-3 was down-regulated in breast cancers with either high vascular invasion or presence of lymph node metastasis, suggesting that its down-regulation was acquired in the course of tumor progression and, in particular, during the acquisition of cancer metastatic potential.

It has been reported that miRNA genes are frequently located in chromosomal regions characterized by nonrandom aberrations in human cancer, suggesting that resident miRNA expression might be affected by these genetic abnormalities (9). miR-125b, which is down-modulated in breast cancer, is located at chromosome 11q23-24, one of the regions most frequently deleted in breast, ovarian, and lung tumors (26, 27). The recognition of a bona fide tumor suppressor gene located at 11q23-24 involved in the pathogenesis of human breast cancer is still lacking. The miR-125b gene establishes itself as an important candidate for this role.

Results reported here increase our understanding of the molecular basis of human breast cancer and suggest that aberrant expression of miRNA genes may be important for the pathogenesis of this human neoplasm.

Note: Supplementary data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org/).

M.V. Iorio and M. Ferracine contributed equally to this work. R. Spizzo is a recipient of an Associazione Italiana per la Ricerca sul Cancro fellowship.

Grant support: Associazione Italiana per la Ricerca sul Cancro; Ministero dell'Istruzione, dell'Università e della Ricerca Programma Post-genoma (FIRB no. RBNE0157EH); Ministero della Salute Italiano; Progetto CAN2005—Comitato dei Sostenitori (M. Negrini); Program Project grants P01CA76259, P01CA81534, and CA083698 (C.M. Croce) from the National Cancer Institute; and a Kimmel Scholar award (G.A. Calin).

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.

1
Bartel DP. MicroRNAs: genomics, biogenesis, mechanism, and function.
Cell
2004
;
116
:
281
–97.
2
Cheng AM, Byrom MW, Shelton J, Ford LP. Antisense inhibition of human miRNAs and indications for an involvement of miRNA in cell growth and apoptosis.
Nucleic Acids Res
2005
;
33
:
1290
–7.
3
Ishizuka A, Siomi MC, Siomi H. A Drosophila fragile X protein interacts with components of RNAi and ribosomal proteins.
Genes Dev
2002
;
16
:
2497
–508.
4
Calin GA, Dumitru CD, Shimizu M, et al. Frequent deletions and down-regulation of micro-RNA genes miR15 and miR16 at 13q14 in chronic lymphocytic leukemia.
Proc Natl Acad Sci U S A
2002
;
99
:
15524
–9.
5
Michael MZ, O' Connor SM, van Holst Pellekaan NG, Young GP, James RJ. Reduced accumulation of specific microRNAs in colorectal neoplasia.
Mol Cancer Res
2003
;
1
:
882
–91.
6
Takamizawa J, Konishi H, Yanagisawa K, et al. Reduced expression of the let-7 microRNAs in human lung cancers in association with shortened postoperative survival.
Cancer Res
2004
;
64
:
3753
–6.
7
Metzler M, Wilda M, Busch K, Viehmann S, Borkhardt A. High expression of precursor microRNA-155/BIC RNA in children with Burkitt lymphoma.
Genes Chromosomes Cancer
2004
;
39
:
167
–9.
8
Eis PS, Tam W, Sun L, et al. Accumulation of miR-155 and BIC RNA in human B cell lymphomas.
Proc Natl Acad Sci U S A
2005
;
102
:
3627
–32.
9
Calin GA, Sevignani C, Dumitru CD, et al. Human microRNA genes are frequently located at fragile sites and genomic regions involved in cancers.
Proc Natl Acad Sci U S A
2004
;
101
:
2999
–3004.
10
Liu CG, Calin GA, Meloon B, et al. An oligonucleotide microchip for genome-wide microRNA profiling in human and mouse tissues.
Proc Natl Acad Sci U S A
2004
;
101
:
9740
–4.
11
Babak T, Zhang W, Morris Q, Blencowe BJ, Hughes TR. Probing microRNAs with microarrays: tissue specificity and functional inference.
RNA
2004
;
10
:
1813
–9.
12
Nelson PT, Baldwin DA, Scearce LM, et al. Microarray-based, high-throughput gene expression profiling of microRNAs.
Nat Methods
2004
;
1
:
155
–61.
13
Thomson JM, Parker J, Perou CM, Hammond SM. A custom microarray platform for analysis of microRNA gene expression.
Nat Methods
2004
;
1
:
47
–53.
14
Liang RQ, Li W, Li Y, et al. An oligonucleotide microarray for microRNA expression analysis based on labeling RNA with quantum dot and nanogold probe.
Nucleic Acids Res
2005
;
33
:
e17
.
15
Barad O, Meiri E, Avniel A, et al. MicroRNA expression detected by oligonucleotide microarrays: system establishment and expression profiling in human tissues.
Genome Res
2004
;
14
:
2486
–94.
16
Calin GA, Liu CG, Sevignani C, et al. MicroRNA profiling reveals distinct signatures in B cell chronic lymphocytic leukemias.
Proc Natl Acad Sci U S A
2004
;
101
:
11755
–60.
17
Querzoli P, Albonico G, Ferretti S, et al. Quantitative immunoprofiles of breast cancer performed by image analysis.
Anal Quant Cytol Histol
1999
;
21
:
151
–60.
18
Tibshirani R, Hastie T, Narasimhan B, Chu G. Diagnosis of multiple cancer types by shrunken centroids of gene expression.
Proc Natl Acad Sci U S A
2002
;
99
:
6567
–72.
19
Furey TS, Cristianini N, Duffy N, et al. Support vector machine classification and validation of cancer tissue samples using microarray expression data.
Bioinformatics
2000
;
16
:
906
–14.
20
Enright AJ, John B, Gaul U, et al. MicroRNA targets in Drosophila.
Genome Biol
2003
;
5
:
R1
.
21
Lewis BP, Shih IH, Jones-Rhoades MW, Bartel DP, Burge CB. Prediction of mammalian microRNA targets.
Cell
2003
;
115
:
787
–98.
22
Krek A, Grun D, Poy MN, et al. Combinatorial microRNA target predictions.
Nat Genet
2005
;
37
:
495
–500.
23
Lee YS, Kim HK, Chung S, Kim KS, Dutta A. Depletion of human micro-RNA miR-125b reveals that it is critical for the proliferation of differentiated cells but not for the down-regulation of putative targets during differentiation.
J Biol Chem
2005
;
280
:
16635
–41.
24
John B, Enright AJ, Aravin A, et al. Human microRNA targets.
PLoS Biol
2004
;
2
:
e363
.
25
Johnson SM, Grosshans H, Shingara J, et al. RAS is regulated by the let-7 microRNA family.
Cell
2005
;
120
:
635
–47.
26
Negrini M, Rasio D, Hampton GM, et al. Definition and refinement of chromosome 11 regions of LOH in breast cancer: identification of a new region at 11q23-q24.
Cancer Res
1995
;
55
:
3003
–7.
27
Rasio D, Negrini M, Manenti G, Dragani T, Croce CM. Loss of heterozygosity at chromosome 11q in lung adenocarcinoma: identification of three independent regions.
Cancer Res
1995
;
55
:
3988
–91.

Supplementary data