MicroRNAs and Cancer

Federica Calore, Muller Fabbri

Department of Molecular Virology, Immunology and Medical Genetics, The Ohio State University, Columbus, OH 43210, USA

Corresponding Author: Muller Fabbri, MD. 1092 Biomedical Research Tower,
460 West 12^th Avenue, Columbus, OH 43210- USA. Phone: +1-614-292-1019; Fax: +1-614-292-3558. E-mail: muller.fabbri@osumc.edu

August 2011

Keywords: microRNAs, non-coding RNAs, cancer, solid tumors, hematological malignancies, oncogene, tumor suppressor gene, angiogenesis, metastasis, therapy, biomarkers.

Abstract
MicroRNAs (miRNAs) are non-coding RNAs (ncRNAs) with gene expression regulatory functions, whose de-regulation has been documented in almost all types of human cancer (both solid and hematological malignancies), with respect to the non-tumoral tissue counterpart. After the initial discovery that the miRNome (defined as the full spectrum of miRNAs expressed in a specific genome) is de-regulated in cancer, contributes to human carcinogenesis, and to the mechanisms of angiogenesis and metastases (which are hallmarks of the malignant phenotype), new pieces of evidence have been provided that miRNAs can be detected in several human body fluids, and can also be successfully used as tumor biomarkers with diagnostic, prognostic and theranostic implications. These findings have cast a new 'translational light on the research in the miRNA field, providing the rationale for a miRNA-based cancer therapy.

Introduction
Tumor formation and progression is a complex multistep process characterized by several consecutive events: accumulation of genomic alterations, uncontrolled proliferation, angiogenesis, invasion and metastasis. Over the past few years an increasing number of studies have highlighted the key role that microRNAs have in the regulation of processes described above. 
MicroRNAs (miRNAs) are a family of single-stranded non-coding RNAs (ncRNAs) between 19-24 nucleotides in length that regulate the expression of target mRNAs both at transcriptional and translational level. In plants such regulation occurs by perfect base-pairing, usually in the 3' untranslated region (UTR) of the targeted mRNA, whereas in mammals the base-pairing is only partial (Lagos-Quintana et al., 2001; Lee and Ambros, 2001; Hu et al., 2010).
Evolutionarily conserved among distantly related organisms (Ambros, 2003), miRNA genes represent approximately 1% of the predicted genes in the genome of different species.  It has been demonstrated that each miRNA can have hundreds of different targets and that approximately 30% of the genes are regulated by at least one miRNA (Bartel, 2004). MiRNAs are known to be involved in several biological processes such as cell cycle regulation, proliferation, apoptosis, differentiation, development, metabolism, neuronal patterning and aging (Bartel, 2004; Bagga et al., 2005; Harfe, 2005; Boehm and Slack, 2006; Calin et al., 2006; Arisawa et al., 2007; Carleton et al., 2007).
The biogenesis of miRNAs starts in the nucleus (Figure 1), where for the most part an RNA polymerase II transcribes long primary precursors, up to several kilobases (pri-miRNAs) (Ambros and Lee, 2004). Such transcription occurs at the level of genomic regions located within the introns or exons of protein-coding genes (70%) or in intergenic areas (30%) (de Yebenes and Ramiro, 2010).
Long, capped and polyadenylated pri-miRNAs (Cai et al., 2004) are then processed by a ribonuclease III (Drosha) and by the double-stranded DNA binding protein DGCR8/Pasha, which enzymatically cut them into smaller fragments of 70-100 nucleotides (pre-miRNAs) (Ambros, 2004). Precursor molecules are then exported to the cytoplasm by Exportin 5 in a Ran-GTP-dependent manner (Allawi et al., 2004; Bohnsack et al., 2004) and through an additional step mediated by the RNAse III Dicer 22 nucleotides double-strand RNAs are generated (Bartel, 2004; Esquela-Kerscher et al., 2005). The duplex miR/miR* are finally incorporated into a large protein complex named RISC (RNA-induced silencing complex): the strand of the duplex which represents the mature miRNA remains stably associated with RISC and drives the complex to the target mRNA. If the base-pairing between miRNA and the 3' UTR of the target mRNA is perfect, the messenger is cleaved and degraded (as it occurs in plants), if the complementarity pairing is partial, translational silencing occurs without mRNA degradation (mechanism described in animals) (Achard et al., 2004; Gregory et al., 2006) (Figure 1).

The involvement of miRNAs in cancer arises from the observation that these small molecules are differentially expressed in neoplastic tissues in a tumor-specific manner when compared to normal tissues (Volinia et al., 2006), and in primary tumors when compared to metastatic tissues (Tavazoie et al., 2008).
Moreover the genomic localization of miRNAs often corresponds to tumor-associated regions, characterized by chromosomal translocations, genomic amplifications, fragile sites, breakpoint regions in proximity to oncogenes (OGs) or tumor suppressor genes (TSGs) (Calin et al., 2004). In 2002 Calin et al. showed that miR-15a and miR-16-1 genes are located at a chromosomic region (13q14) deleted in more than half of B cell chronic lymphocytic leukemias (B-CLL) and that both genes are deleted or down-regulated in the majority of CLL cases (68%) (Calin et al., 2002). Based on the miRNA profiling analysis the following studies aimed at investigating the functional role of these molecules in tumorigenesis by using various approaches, which have shed  light on a more complex role of miRNAs in cancer development: depending on the context they can act as OGs or TSGs, and some of them can even have a dual role of OG/TSG (Calin et al., 2007) (Table 1).

Table 1. The main de-regulated miRNAs in cancer.

Legend: CLL= chronic lymphocytic leukemia; AML= acute myeloid leukemia.

miRNAs as oncogenes
Profiling studies have revealed that several miRNAs show oncogenic properties. One of the first oncomiR identified was miR-155 (Metzler et al., 2004; Kluiver et al., 2005). It is located on chromosome 21 in a host non-coding RNA called the B cell integration cluster (BIC) and is highly expressed in pediatric Burkitt's lymphoma (Metzler et al., 2004), Hodgkin disease (Kluiver et al., 2005), primary mediastinal non-Hodgkin's lymphoma (Calin et al., 2005) , chronic lymphocytic leukemia (CLL) (Kluiver et al., 2005), acute myelogenous leukemia (AML) (Calin et al., 2008), lung, breast and pancreatic cancer (Volinia et al., 2006; Greither et al., 2010). A study conducted by Costinean et al. showed that transgenic mice with a B-cell targeted overexpression of miR-155 develop a lymphoproliferative disease (polyclonal pre-leukemic pre-B-cell proliferation followed by full-blown B-cell malignancy) resembling the human diseases, indicating that the deregulation mediated by miR-155 involves both the initiation and progression of the disease (Costinean et al., 2006).
Moreover the use of miR-155 knock out mouse model has revealed that miR-155 is strongly implicated into the induction of Th2 lymphocyte differentiation and altered cytokine production (de Yebenes and Ramiro, 2010).
Another miRNA which displays an oncogenic role is miR-21. Chan et al. demonstrated that knockdown of miR-21 in multiple glioblastoma cells induced caspase activation and apoptosis, indicating that miR-21 could function as an oncogene by blocking expression of critical apoptosis-related genes (Abdellatif, 2010). In fact miR-21 targets TSGs such as PTEN (phosphatase and tensin homolog) (Choong et al., 2007), PDCD4 (programmed cell death 4) (Dillhoff et al., 2008) and TPM1 (tropomyosin 1) (Beitzinger et al., 2007). Similarly to miR-155 it is expressed in a wide range of tumors such as glioblastoma (Ciafre et al., 2005), CLL (Calin et al., 2005), AML (Calin et al., 2008), prostate, pancreatic, gastric, colon, breast, lung (Costinean et al., 2006) and liver cancer (Choong et al., 2007).
The miR-17-92 cluster is characterized by six miRNAs (miR-17, miR-18a, miR-19a, miR-20a, miR-19b-1 and miR-92-1) highly expressed in breast, colon, lung, pancreatic, prostate and gastic cancer, lymphomas (Costinean et al., 2006; Nagel et al., 2007). It has been demonstrated that the miR-17-92 cluster induces B cell proliferation. Moreover, transgenic mice overexpressing miR-17-92 in lymphocytes developed lymphoproliferative disease and autoimmunity through the inhibition of tumor suppressor Pten and the pro-apoptotic protein Bim (de Yebenes and Ramiro, 2010).
Other miRNAs that have an oncogenic role are miR-372/373, which are involved in the development of human testicular germ cell tumors by neutralizing the TP53 pathway (Voorhoeve et al., 2006), miR-221/222 which induce proliferation of thyroid (Iorio et al., 2007), prostate (Galardi et al., 2007) and glioblastoma (Bai et al., 2007), miR-10b which promotes cell migration and invasion in breast cancer (Derby et al., 2007).

miRNAs as tumor suppressor genes
If several miRNAs are known for their pro-oncogenic role, then other miRNAs represent their counterpart by acting as a TSG. Their silencing due to mutations, chromosomal rearrangements or to promoter methylation (Calin et al., 2002; Calin et al., 2005; Ishii and Saito, 2006; Arisawa et al., 2007) contributes to the initiation and progression of cancer.
MiR-15a and miR-16-1 represent a typical example of TSG miRNA. Encoded as a cluster at the level of chromosome 13q14.3, a region frequently deleted in chronic lymphocytic leukemia (CLL) (Bullrich et al., 2001), miR-15a and -16 display expression levels inversely correlated to the BCL2 ones. These miRNAs in fact induce apoptosis in leukemic cells by directly targeting the anti-apoptotic gene (Calin et al., 2005). Moreover, it has been demonstrated that miR-15a and -16 exert a tumor-suppressor role also in prostate cancer by targeting BCL2, CCND1 (cyclin D1) and WNT3A (encoding a protein which promotes cell survival, proliferation and invasion) (Bonci et al., 2008). Taken together, these findings harbor therapeutic implications and bring new insights to the comprehension and treatment of cancer.
Chromosome 7q32 hosts the miR-29 family (comprising miR-29a, -29b and -29c), which is downregulated in lung cancer, CLL, AML, breast cancer and cholangiocarcinoma (Calin et al., 2005; Mott et al., 2007; Calin et al., 2008). It has been demonstrated that in lung cancer the expression of miR-29 family members is inversely correlated with DNMT3A and -3B (DNA methyltransferases 3A and 3B) and that these miRNAs directly target these enzymes, inducing global hypomethyation of tumoral cells (Calin et al., 2007) and reactivation of methylation-silenced TSGs such as WWOX, FHIT, MCL1 and TCL1 (Costinean et al., 2006; Mott et al., 2007).
Among the tumor suppressor miRNAs there is the let-7 family. Johnson et al. demonstrated an inverse correlation between the expression of the let-7 family members and the expression of the oncogene RAS in lung cancer tissue (Adai et al., 2005). Let-7 family targets as well other onco-genes such as C-MYC (Sampson et al., 2007), HMGA2 (high mobility group A2) (Barakat et al., 2007) and MYCN (Buechner et al., 2011). However, not all the members of this family display a tumor suppressor role since in lung adenocarcinoma let-7a-3 has an oncogenic function and promotes tumor cell proliferation (Brueckner et al., 2007). 
The miR-34 family (comprising miR-34a, -34b and -34c) is downregulated in lung cancer tumor cells with respect to normal tissue and their re-expression in pancreatic cancer cell lines inhibits cell growth and invasion, and induces apoptosis and cell cycle arrest in G1 and G2/M (Gallardo et al., 2009). Similarly to the tumor suppressor miRNAs described above, the miR-34 family exerts its function by targeting anti apoptotic mRNAs such as BLC2 and MYCN (Camps et al., 2008).
The list of miRNAs having a tumor suppressor function ends with the cluster miR-143 and -145. These miRNAs, downregulated in several tumors (Akao et al., 2007; Banaudha et al., 2011), have been found to target ERK5 (extracellular signal-regulated kinase 5) and c-MYC with consequent inhibition of tumor proliferation and increased apoptosis (Akao et al., 2007; Ibrahim et al., 2011).

miRNAs in solid tumors
Lung cancer
Lung cancer is the leading cause of cancer death around the world (Jemal et al., 2009). Gao et al. performed miRNA microarray expression profiling in order to compare miRNAs expression in primary squamous cell lung carcinoma with normal cells and determine miRNA potential relevance to clinicopathological factors and patient postoperative survival times. They found out that miR-21 was upregulated in nearly 75% of cancer specimens and that this modulation was significantly correlated with shortened survival time (Cheng et al., 2011).
Yanaihara and co-workers used the same approach and correlated miRNA expression profiles with survival of lung cancer, finding out that high miR-155 and low let-7a-2 expression were correlated with poor survival. Furthermore, they found a molecular signature for subset of lung cancer: they identified six miRNAs having a differential expression in adenocarcinoma and squamous cell cancer (mir-205, mir-99b, mir-203, mir-202, mir-102, and mir-204-Prec). Among these, the expression of miR-99b and miR-102 was found higher in adenocarcinoma (Volinia et al., 2006) .
Yu et al. found a five-microRNA signature (let-7a, miR-21, miR-137, miR-372, miR-182*) associated with survival and cancer relapse in NSCLC (non-small cell lung cancer) patients (Abdurakhmonov et al., 2008). Another specific marker for squamous cell lung carcinoma is miR-205, according to a microarray study performed by Lebanony et al., who found a strong association between the expression levels of miR-205 and squamous cell lung carcinoma histology (Barshack et al., 2010).
In addition to the already mentioned miRNAs, miR-31 is found to act as an oncogenic miRNA by targeting mRNAs encoding two anti-tumoral proteins, LATS2 (large tumor-suppressor 2) and PPP2R2A (PP2A regulatory subunit B alpha isoform) (Anand et al., 2010). Chou and co-workers discovered that miR-7 promotes EGFR-mediated tumorigenesis in lung cancer by targeting ERF (Ets transcriptional repressor) thus modulating cell growth (Choudhry and Catto, 2011). However, miR-7 seems to have a dual function of oncogene/tumor-suppressor miRNA. Xiong et al. indeed found that overexpression of miR-7 in NSCLC A549 cells inhibits cells proliferation and induces apoptosis by targeting anti-tumoral protein Bcl-2 (Shao et al., 2011).
Another miRNA that displays a tumor-suppressor role in lung cancer is miR-451. Wang et al. demonstrated not only that this miRNA is the most downregulated in NSCLC tissues, but also that it regulates survival of cells partially through the downregulation of the oncogene RAB14 (Ras-related protein 14) (Bian et al., 2011).

Breast cancer
Breast cancer is the second leading cause of cancer deaths in the developed world and the most commonly diagnosed cancer in women (Bonev et al., 2011). A miRNA expression profile study for breast cancer was conducted by Iorio et al. The authors found 13 miRNAs differentially expressed between tumor and normal tissues: among the upregulated ones there were oncogenic miR-21 and miR-155, while miR-10b, let-7 miR-125b, miR-145 and miR-205 were found downregulated (Calin et al., 2005). The latter directly targets HER3 receptor and blocks the activation of downstream Akt, inhibiting cell proliferation. Moreover, miR-205 sensitizes cells to Gefitinib and Lapatinib, two tyrosine-kinase inhibitors, promoting apoptosis (Iorio et al., 2009).
Shi et al. found that miR-301 has an oncogenic role in breast tumor by targeting FOXF2, BBC3, PTEN and COL2A1. Its upregulation promotes proliferation, migration, invasion and tumor formation. Moreover, by cooperating with its host gene SKA2, miR-301 promotes the aggressive breast cancer phenotype with nodal or distant relapses (Akao et al., 2011).
Heyn and co-workers identified miR-335 as a tumor-suppressor gene. It controls different factors of the upstream BRCA1 regulatory pathway (such as ERa, IGF1R, SP1), inducing an upregulation of the tumor suppressor gene BRCA1 (Heyn et al., 2011).

Colorectal cancer
In 2008 a study conducted by Schetter et al. the authors performed miRNA microarray expression profiling comparing 84 pairs of tumors (colon adenocarcinoma) and adjacent non-tumoral tissues (Schetter et al., 2008). They found 37 differentially expressed miRNAs; among them miR-20a, -21, -106, -181b and -203 levels were higher in tumor specimens. The overexpression of miR-21 and its role in tumor proliferation in several kind of cancers has already been described before. Also miR-20a belongs to the miR-17-92 cluster, whose overexpression promotes cell proliferation (Hayashita et al., 2005) and increased tumor size. 
One of the most recent tumor suppressor miRNAs found in colorectal cancer is miR-137. Balaguer et al. reported that this miRNA is constitutively expressed in the normal colonic epithelium but during the early events of colorectal carcinogenesis it is silenced through promoter hyper-methylation. Moreover, its re-expression in vitro inhibits cell proliferation in a cell specific manner. These findings suggest a prognostic role for miR-137 (Balaguer et al., 2010).
It has been recently demonstrated by Sarver et al. that miR-183 has an oncogenic role in colon cancer (but also in synovial sarcoma and rhabdomyosarcoma) through its regulation of the expression levels of 2 tumor suppressor genes, EGR1 and PTEN. The authors also provided evidence that knockdown of miR-183 affects cellular migration and they suggest that pharmaceutical intervention on tumor characterized by the upregulation of miR-183 may be useful as anti-cancer therapy (Chen et al., 2010).

Hepatocellular carcinoma
One of the most common malignant tumors is hepatocellular carcinoma. Murakami et al. analysed the miRNA expression profiles in 25 specimens of hepatocellular carcinoma compared with adjacent non-tumoral tissues and nine chronic hepatitis specimens (Murakami et al., 2006). miR-222, miR-17-92 and miR-106a exhibited higher expression in tumor tissues than in the normal ones and were found associated with the tumor differentiation status.
Pineau et al. performed profiling studies on 104 hepatocellular carcinoma tissue specimens, 90 cirrhotic, 21 normal and 35 hepatocellular carcinoma cell lines (Pineau et al., 2010). They found a 12 miRNA signature that characterizes tumor progression starting from normal liver, to cirrhosis to full blown tumor. Among them, miR-21, miR-221/222, miR-34a and miR-224 were found overexpressed in the progression signature. miR-224 overexpression is connected with the regulation of cell proliferation, cell migration and metastasis (Chemistry, 2010).
Su et al. reported that miR-101 is significantly downregulated in hepatocellular carcinoma and that its overexpression inhibits tumor development in nude mice, sensitizes tumor cell lines to serum starvation and chemotherapeutic treatment (Su et al., 2009).
Other tumor suppressor miRNAs are: miR-122, normally downregulated in hepatocellular carcinoma, whose overexpression induces apoptosis and cell cycle arrest through targeting of BCLW (Chemistry, 2010); miR-198, which inhibits migration and invasion in a c-MET dependent manner (Akao et al., 2011); miR-125b, which suppresses tumor cell growth in vitro and in vivo and induces cell cycle arrest at G1/S acting as a tumor suppressor gene through the suppression of LIN28B (Bates et al., 2010), a promoter of cell proliferation and metastasis through regulation of c-MYC and E-Cadherin (Ai et al., 2010).

miRNAs in hematological malignancies
Similarly to what has been reported in solid tumors, also in hematological malignancies the miRNome is frequently de-regulated with respect to the normal cell counterpart. Physiologic variations in miRNA expression occur during normal hematopoiesis, and affect differentiation and commitment of the multipotent hematologic progenitor (MPP). Hematologic tumors represent abnormal blocks in hematopoiesis. Interestingly, the aberrations of the miRNome occurring in these tumors can be explained, at least in some instances, as the result of the block of differentiation leading to the development of the malignancy. In other cases, the cause of the observed de-regulation has not been clarified, but the role of the de-regulated miRNAs in the acquisition of the malignant phenotype has been understood, based on the nature of the targeted genes.

miRNAs in leukemias
Chronic lymphocytic leukemia (CLL) is the most frequent leukemia of the adult in the Western world. Chromosomal aberrations recur in human CLL and harbor diagnostic and prognostic implications. Occurring in about 65% of cases, the 13q14 deletion is the most frequent chromosomal aberration observed in human CLL. Based on the analysis of a large number of CLL cases with monoallelic 13q14 deletion, a minimal deleted region (MDR) has been defined. This MDR includes a long ncRNA, called DLEU2 (deleted in leukemia 2), strongly conserved among vertebrates, and the first exon of the DLEU1 gene, another ncRNA (Migliazza et al., 2001; Chai et al., 2010). The miR-15a/16-1 cluster is located within intron 4 of DLEU2, and genetic alterations affecting DLEU2 mRNA expression would also affect miR-15a/16-1 cluster expression (Calin et al., 2002) . Therefore, the expression of miR-15a/16-1 is reduced in the majority of CLL patients carrying the 13q deletion (Calin et al., 2002). Interestingly, the same miRNA cluster is involved in cases of familial CLL, since a germ-line mutation in the sequence of pre-miR-16-1 (which leads to a reduced miR-16 expression both in vitro and in vivo), has been identified associated with the deletion of the normal allele in leukemic cells of two CLL patients, one of which with a family history of CLL and breast cancer (Calin et al., 2005). A similar point mutation, adjacent to the miR-16-1 locus has been described in the CLL prone New Zealand Black mouse strain model (Raveche et al., 2007). One of the most frequent molecular hallmarks of the malignant, mostly non-dividing B-cell of CLL, is the up-regulation of the antiapoptotic BCL2. It has been demonstrated that both miR-15a and miR-16 directly target BCL2 in CLL both in vitro and in vivo (Calin et al., 2005; Ambs et al., 2008), therefore suggesting that the miR-15a/16-1 cluster enacts a tumor suppressor function. Clinicians are aware that CLL is characterized by recurrent and common chromosomal aberrations, which harbor prognostic implications. Some of the most frequent of these abnormalities are the 13q deletion, the 17p deletion and the 11q deletion. While CLL patients with the 13q deletion experience the indolent form of the disease (characterized by IGV_H mutated and low levels of the prognostic surrogate marker ZAP70), those with the 17p or the 11q deletion (alone or in association with the 13q), experience an aggressive form of the disease (characterized by IGV_H unmutated and high levels of ZAP70) (Chiorazzi et al., 2005). Recently, a new molecular network explaining the role of these chromosomal aberrations and their prognostic implications for human CLL has been described. According to this model, the miR-15a/16-1 cluster (located at 13q), directly targets the pro-apoptotic TP53 (located at 17p), which in turn transactivates the miR-34b/34c cluster (located at 11q), directly targeting ZAP70 (Fabbri et al., 2011). Also, TP53 is able to transactivate the miR-15a/16-1 cluster, creating a feed-forward regulatory loop (Fabbri et al., 2011). These findings identify for the first time some of the molecular effectors connecting these three recurrent chromosomal aberrations in CLL and can explain both their prognostic implications and the observed levels of ZAP70 according to the degree of aggressiveness of the disease. Recently, Klein et al. (Danilov et al., 2010) generated two groups of transgenic mice models: one mimicking the MDR and the other containing a specific deletion of the miR-15a/16-1 cluster. Although the same spectrum of clonal lymphoproliferative disorders was observed in both animal models, the disease was more aggressive in the MDR group than in the miR-15a/16-1 group, suggesting that additional genetic elements in the 13q14 region may affect the severity of the disease. The oncogene TCL1 (T-cell leukemia/lymphoma 1A) is over-expressed in the aggressive CLL (Herling et al., 2006; Barlev et al., 2010), and is regulated by miR-29b and miR-181b (Costinean et al., 2006). Furthermore, miR-181a directly targets BCL2 (Ebert et al., 2007), suggesting a central role of miR-181 family and of the miR-15a/16-1 cluster in regulating BCL2 expression in CLL. Stamatopoulos et al. (Stamatopoulos et al., 2009) found that downregulation of miR-29c and miR-223 are predictive of treatment-free survival (TFS) and overall survival (OS). Low expression of miR-223, miR-29b, miR-29c, and miR-181 family are associated with disease progression in CLL cases harboring the 17p deletion, whereas patients carrying the trisomy 12 abnormality and high expression of miR-181a experience a more aggressive variant of CLL (De Martino et al., 2009). Interestingly, the miR-29 family has been demonstrated to control key epigenetic mechanisms (such as the expression of all three main DNA methyltranferases) both in solid tumors and in hematological malignancies (Calin et al., 2007; Garzon, 2009), therefore suggesting the involvement also of miRNA-mediated epigenetic factors in the pathogenesis and prognosis of human CLL.
Also miR-155 is up-regulated in CLL versus normal CD19+ B lymphocytes, suggesting that this miRNA might act as diagnostic biomarker of CLL (Marton et al., 2008).
The Philadelphia chromosome (reciprocal translocation t(9;22)) is the hallmark of the chronic myeloid leukemia (CML), generating the chimeric protein BCR-ABL1, which is able to activate the miR-17-92 cluster, together with the oncogene c-MYC, during the early chronic phase, but not in blast crisis CML CD34+ cells (Nagel et al., 2007). These findings suggest that the miR-17-92 cluster contributes to early phase CML pathogenesis, harboring CML diagnostic biomarker properties. ABL1 is also a direct target of miR-203, whose over-expression inhibits cancer cell proliferation in an ABL1-dependent manner (Bueno et al., 2008). Moreover, it has been shown that Philadelphia positive CMLs, often present a reduced expression of miR-203 because of its promoter hyper-methylation, while no methylation can be detected in other hematological malignancies that do not carry ABL1 alterations (Bueno et al., 2008). Finally, down-regulation of miR-10a has been observed in about 70% of CMLs, with an inverse correlation with the expression of the oncogene USF2 (upstream stimulatory factor 2) (Agirre et al., 2008). Overall, high levels of miR-17-92 cluster and low expression of miR-203 and miR-10a seem to be part of the diagnostic signature of human CML. More recently, miR-451 has emerged as another key player in CML. Indeed this miRNA can target BCR-ABL1, which in turn can inhibit miR-451 expression, creating a regulatory loop, whose disruption might have therapeutic implications in the disease (Lopotova et al., 2011). Another gene which inhibits cell growth and is frequently down-regulated in CML is CCN3 (also known as NOV or nephroblastoma overexpressed gene). A possible mechanism of its down-regulation in CML has been recently identified and is mediated by miR-130a and miR-130b, which are up-regulated by BCR-ABL1 in CML, and directly target CCN3, contributing to leukemic cell proliferation (Suresh et al., 2011).
Up-regulation of the miR-17-92 cluster has been described also in B- and T-cell acute lymphocytic leukemia (ALL) (Zanette et al., 2007; Nagel et al., 2009). Recently, the miR-17-92 cluster has been correlated with the development of mixed lineage leukemia (MLL)-rearranged acute leukemia (Chemistry, 2010). Up-regulation of this cluster was observed not only in MLL-associated AML, but also in ALL, and is possibly due to both DNA copy number amplification at 13q31 and to direct upregulation by MLL fusions (Chemistry, 2010). Interestingly, a specific miRNA signature of 4 miRNAs is able to distinguish the two forms of acute leukemias (ALL from AML (acute myeloid leukemia)) with an accuracy rate of 98%. Indeed, higher expression of miR-128a and miR-128b was found in ALL compared to AML, whereas down-regulation of let-7b, miR-223 indicates ALL vs AML (Science, 2007). At the moment, the leukemogenic mechanism of miR-128b is still poorly understood. Zhang et al., have identified a miRNA signature in children with ALL complicated by central nervous system (CNS) relapse (Ai et al., 2009). The high-risk-of-relapse signature is composed of over-expression of miR-7, miR-198, and miR-663, and down-regulation of miR-126, miR-345, miR-222, and miR-551a. MiR-16 has a prognostic significance in ALL. Indeed, Kaddar et al., found that low expression of miR-16 is associated with a better ALL outcome (Kaddar et al., 2009).
In AML with normal karyotype high levels of miR-10a, -10b, members of let-7 and miR-29 families, and down-regulation of miR-204, identify NPM1 (nucleophosmin-1) mutated versus unmutated cases (Calin et al., 2008). Recently, Ovcharenko et al., confirmed that miR-10a expression is highly characteristic for NPM1 mutated AML, and may contribute to the intermediate risk of this condition by interfering with the TP53 machinery, partly regulated by its target MDM4 (murine double minute 4) (Ovcharenko et al., 2011). Over-expression of miR-155 is associated with FLT3-ITD+ status, although there is evidence that this up-regulation is actually independent from FLT3 signaling (Calin et al., 2008). The fusion oncoprotein AML1/ETO (generated by the t(8;21) translocation), is the most frequent chromosomal abnormality in AML, and causes epigenetic silencing of miR-223, by recruiting chromatin remodeling enzymes at an AML1-binding site on the pre-miR-223 gene (Fazi et al., 2007). By silencing miR-223 expression, the oncoprotein inhibits the differentiation of myeloid precursors (promoted by high levels of miR-223), therefore actively contributing to the pathogenesis of this myeloproliferative disorder. A central role in the pathogenesis of AML is also played by miR-29b, a direct regulator of the expression of all three DNA methyltransferases (Calin et al., 2007; Garzon et al., 2009b). Re-expression of miR-29b induces de-methylation and re-expression of epigenetically silenced TSGs, such as ESR1 (estrogen-receptor alpha), and p15 (INK4b) (Garzon et al., 2009b). Moreover, restoration of miR-29b in AML cell lines and primary samples, suppresses the expression of OGs such as MCL1, CXXC6, and CDK6, which are direct targets of miR-29b (Garzon et al., 2009a). Abnormal activation of the proto-oncogene c-KIT contributes to leukemogenesis. Gao et al., found that miR-193a is silenced by promoter hyper-methylation in AML, and since this miRNA directly targets c-KIT, this epigenetic silencing is responsible, at least in part, for the aberrant up-regulation of the oncogene in AML (Cheng et al., 2011). Indeed, restoration of miR-193a expression by de-methylating agents, reduces the expression of c-KIT and induces cancer cell apoptosis and granulocytic differentiation (Cheng et al., 2011). Similarly, also miR-193b directly targets c-KIT in AML (Cheng et al., 2011). By using a novel approach based on the integration of miRNA and mRNA expression profiles, Havelange et al., found a strong positive correlation between miR-10 and miR-20a and HOX-related genes, a significant inverse correlation between genes involved in immunity and inflammation (such as IRF7 and TLR4) and a panel of 4 miRNAs (namely, miR-181a, -181b, -155, and -146), and a strong direct correlation between miR-23, -26a, -128a, and -145 and pro-apoptotic genes (such as BIM and PTEN) (Havelange et al., 2011). Also miR-100 has been described as an OG in AML, by targeting the TSG RBSP3 (CTD (carboxy-terminal domain, RNA polymerase II, polypeptide A) small phosphatase-like) (Cao et al., 2011). Also in AML, miR-17/20/93/106 have been shown to promote hematopoietic cell expansion by targeting sequestosome 1-regulated pathways in mice (Meenhuis et al., 2011). Down-regulation of miR-29a and miR-142-3p has been observed in AML with respect to controls (Bian et al., 2011), and miR-29a contributes to counteract leukemic proliferation by directly targeting the proto-oncogene SKI (Teichler et al., 2011).

miRNAs in lymphomas
De-regulation of miRNAs has been reported also in non Hodgkin lymphomas (NHL) and in Hodgkin's disease (HL). The first evidence of an involvement of miRNAs in lymphomagenesis was provided by Eis et al. who observed that the final part of the B-cell integration cluster (BIC) non-coding RNA (ncRNA), where miR-155 is located (Chen and Meister, 2005), was able to accelerate MYC-mediated lymphomagenesis in a chicken model (Bashirullah et al., 2003). Subsequently, high levels of BIC/miR-155 were described also in pediatric Burkitt's lymphoma (BL) (Metzler et al., 2004), in diffuse large B-cell lymphoma (DLBCL) (Lawrie, 2007; Hoefiget al., 2008), and in HL (Kluiver et al., 2005; Abdurakhmonov et al., 2008; Van Vlierberghe et al., 2009). In a B-cell specific miR-155 transgenic (TG) mouse model the onset of an acute lymphoblastic leukemia/high-grade lymphoma at approximately 9 months of age was observed (Costinean et al., 2006). In these TG mice, the B-cell precursors with the highest miR-155 expression were at the origin of the leukemias (Costinean et al., 2009). Indeed, miR-155 directly target SHIP (Src homology 2 domain-containing inositol-5-phosphatase), and C/EBPbeta (CCAAT enhancer-binding protein beta), two key regulators of the interleukin-6 signaling pathway, therefore triggering a chain of events that promotes the accumulation of large pre-B cells and acute lymphoblastic leukemia/high-grade lymphoma (Costinean et al., 2009). Also miR-155 knockout (KO) mice models have been generated, showing that the loss of miR-155 switches cytokine production toward T_H2 differentiation (de Yebenes and Ramiro, 2010), and also compromises the ability of dendritic cells (DC) to activate T cells, because of a defective antigen presentation or abnormal co-stimulatory functions (de Yebenes and Ramiro, 2010).
As observed in leukemias, also in NHLs, a specific signature of 4 de-regulated miRNAs (namely miR-330, -17-5p, -106a, and -210) can differentiate among reactive lymph nodes, follicular lymphomas (FL), and DLBCL (Hoefig et al., 2008). Noteworthy, miR-17-5p, and miR-106a belong to two paralogous clusters located on chromosome 13 and X, respectively, with a well established oncogenic role both in solid and hematological malignancies (Chang et al., 2008). The miR-17-92 cluster is located in a region frequently amplified in malignant B-cell lymphomas (Abbott et al., 2005), and is overexpressed in over 60% of B-cell lymphoma patients (Allawi et al., 2004). In murine pluripotent cells from MYC-transgenic mice, over-expression of this miRNA cluster accelerates lymphomagenesis (Allawi et al., 2004), whereas in miR-17-92 TG mice models a higher than expected rate of lymphoproliferative disorders and autoimmunity and premature death was observed (de Yebenes and Ramiro, 2010). These effects are at least in part due to the direct targeting of the PTEN and BIM, which controls B-lymphocyte apoptosis (de Yebenes and Ramiro, 2010). The miR-106a-363 polycistron is also overexpressed in 46% of acute and chronic human T-cell leukemias (Landais et al., 2007), claiming a role in leukemogenesis. Interestingly, both miR-106b-25 and miR-17-92 parologous clusters interfere with the transforming growth factor-beta (TGF-beta) signaling (Petrocca et al., 2008), which is inhibited in several tumors (Derynck et al., 2001). Moreover, Ventura et al. have shown that the miR-17-92 and miR-106b-25 double knockout mouse model has a more severe phenotype than the miR-17-92 single knockout mouse model (Ventura et al., 2008), suggesting that both clusters are implicated in the control of apoptosis in malignant lymphocytes. Interestingly, miR-17-5p and miR-20a (which belong to the miR-17-92 cluster) are induced by the proto-oncogene and transcription factor c-MYC (Nakamoto et al., 2005), and in turn the cluster directly targets E2F1, a c-MYC transactivated transcription factor promoting cell-cycle progression (Nakamoto et al., 2005). Therefore, the miR-17-92 cluster tightly regulates c-MYC-driven cell-cycle progression. From a more translational perspective, it has been also demonstrated that over-expression of the miR-17-92 cluster also significantly increases the resistance to radiotherapy in human mantle cell lymphoma cells (Ahn et al., 2010), revealing a role for this cluster as a theranostic biomarker. MiR-34a is negatively regulated by c-MYC (Abdurakhmonov et al., 2008). In c-MYC over-expressing B-lymphocytes miR-34a confers drug resistance by inhibiting TP53-dependent bortezomib-induced apoptosis (Sotillo et al., 2011). Finally, down-regulation of miR-143 and miR-145 has been described in B-cell lymphomas and leukemias (Akao et al., 2007), and re-expression of these miRNAs in a Burkitt lymphoma cell line demonstrated a dose-dependent growth inhibitory effect, mediated in part by miRNA-induced downregulation of the oncogene ERK5 (Akao et al., 2007).
In HL, Navarro et al. identified a distinctive signature of 25 miRNAs able to distinguish HL from reactive lymph nodes, and 36 miRNAs differentially expressed in the nodular sclerosis and mixed cellularity subtypes of HL (Navarro et al., 2007). Interestingly, 3 miRNAs (namely, miR-96, -128a, and -128b) are selectively downregulated in HL cells with Epstein-Barr virus (EBV) infection, but only one of these miRNAs is part of the signature of 25 de-regulated miRNAs in HL versus reactive lymph nodes, suggesting that EBV might not be relevant for HL pathogenesis (Navarro et al., 2007). Down-regulation of miR-150 and over-expression of miR-155 frequently occur in HL cell lines (Gibcus et al., 2009). Since HL develops in the lymph node germinal center, and high levels of miR-155 have been described in the germinal center also during normal lymphopoiesis, it can be postulated that the observed over-expression of miR-155 in HL might result from an abnormal block of lymphocyte differentiation at the germinal center level. Van Vlierberghe et al., have compared miRNA profiles of microdissected Reed-Sternberg cells and Hodgkin cell lines versus CD77+ B-cells (Van Vlierberghe et al., 2009). In this study a profile of 12 over and 3 under-expressed miRNAs was identified (Van Vlierberghe et al., 2009), showing only a partial overlap with Navarro's profile. This discrepancy might be due to the different procedure used to collect HL cells. Finally, also in HL miRNA expression profile can predict prognosis. Indeed, low levels of miR-135a are associated with a higher relapse risk and a shorter disease-free survival (Gallardo et al., 2009). A possible molecular explanation for this effect is that miR-135a directly targets the kinase JAK2 (Janus Kinase 2). Therefore, low levels of miR-135a are associated with higher expression of JAK2, which leads to up-regulation of the antiapoptotic BCL-XL, therefore leading to reduced apoptosis and increased cell proliferation (Gallardo et al., 2009).

miRNAs in body fluids as tumor biomarkers
MiRNAs have been successfully detected in blood and other human fluids. It has been shown that they circulate wrapped in circulating microvescicles called "exosomes" (Bar et al., 2008), and therefore are extremely stable and resistant to degradation (Aumiller and Forstemann, 2008; Kroh et al., 2010). In 2010, Weber et al. determined miRNA expression in 12 different types of body fluids (amniotic fluid, breast milk, bronchial lavage, cerebrospinal fluid (CSF), colostrum, peritoneal fluid, plasma, pleural fluid, saliva, seminal fluid, tears and urine) collected from healthy individuals, and showed that the highest concentrations of miRNAs were found in tears and the lowest in CSF, pleural fluid and urine (Black et al., 2010). The ability to detect miRNAs in body fluids has generated interest in their possible role as tumoral biomarkers. Several studies have demonstrated that miRNAs can indeed be successfully employed both as cancer diagnostic and prognostic biomarkers both in solid and in hematological malignancies. Table 2 summarizes some of these studies.

Table 2. MiRNAs detectable in body fluids and their diagnostic and prognostic significance for cancer patients.

Legend: The column "Biomarker property" should be read as each letter (or in parenthesis letters) referred to the miRNA reported in the column "miRNA", according to the sequence order in which these miRNAs are reported. D= Diagnostic biomarker; P= Prognostic biomarker; (D,P)= Diagnostic and Prognostic biomarker.OSCC= Oral Squamous Cell Carcinoma; HCC= Hepatocellular Carcinoma; DLBCL= Diffuse Large B-Cell Lymphoma.

miRNAs in body fluids as tumor biomarkers in solid tumors
Diagnostic biomarkers
The first evidence that circulating miRNAs can be effectively used to diagnose cancer was provided by Mitchell et al. in 2008 (Bar et al., 2008). They found that higher levels of miR-141 in the serum of 25 patients affected by prostate cancer, compared with 25 healthy control donors identify patients affected by cancer with a sensitivity of 60%, and a specificity of 100% (Bar et al., 2008). Subsequently, Taylor et al. showed that a signature of 8 circulating miRNAs (enclosed in tumor-derived exosomes of endocytic origin) can be used as diagnostic biomarker of ovarian cancer (Chang et al., 2008). Moreover, in a comparison of 152 patients affected by NSCLC versus 75 healthy donors, Chen et al., identified higher levels of miR-25, and miR-223 in the serum of cancer patients (Aumiller and Forstemann, 2008). Interestingly, these Authors also demonstrated that circulating miRNAs resist treatments with HCl, NaOH, and repeated freeze and thaw cycles, therefore acting as stable, reliable biomarkers (Aumiller and Forstemann, 2008). Patients affected by pancreatic cancer have higher concentrations of circulating miR-210 (Bar et al., 2008) , -200a, and -200b (Chemistry, 2010), suggesting that these miRNAs might be used to successfully screen for pancreatic cancer. High levels of circulating miR-29a, -92 and -17-3p have been found in patients affected by colorectal cancer (Anand et al., 2010). Interestingly, miR-92 is not elevated in the plasma of patients with irritable bowel disease, suggesting a role for this miRNA in the differential diagnosis between this benign condition and cancer. Moreover, the increased levels of circulating miR-29a and -92 occur already in presence of pre-cancerous conditions such as colon adenomas (Anand et al., 2010), revealing that the de-regulation of these two miRNAs is an early event in colon carcinogenesis and their increased plasma concentration might be helpful for the very early (even pre-cancerous) phase of colorectal tumorigenesis. In breast cancer, Asaga et al. showed that serum concentrations of miR-21 correlates with the presence and extent of breast cancer (Asaga et al., 2011), whereas Heneghan et al., showed that circulating miR-195 differentiates breast cancer from other malignancies and is a potential biomarker for the detection of non-invasive and early stage disease (Henegan et al., 2010). Finally, in oral squamous cell carcinoma (OSCC) high levels of circulating miR-31 differentiate patients from healthy controls and the concentration of this miRNA decreases after surgical resection of the tumor (Anand et al., 2010), suggesting that miR-31 might be helpful also for the early detection of OSCC recurrence.
In addition to blood and plasma, miRNAs can be detected also in other body fluids and have diagnostic biomarker properties. High levels of miR-31 (Anand et al., 2010), and lower levels of miR-200a and -125a (Addo-Quaye et al., 2009) have been identified in the saliva of OSCC patients. An increased expression of miR-126, -182, and -199a has been described in the urine of patients affected by bladder cancer with respect to healthy controls (Hanke et al., 2010), whereas the ratio miR-126/miR-152 and miR-182/miR-152 is higher in patients affected by bladder cancer versus carriers of urinary tract infections, with a sensitivity of 72% and 55%, respectively, and a specificity of 82% (Hanke et al., 2010). Similarly, in the blood of patients with hepatocellular carcinoma (HCC), Shigoka et al. found that the ratio of miR-92a/miR-638 is lower than healthy controls, suggesting a possible role of this non-coding RNA parameter in the diagnosis of HCC. Also in malignant pleural effusions of patients affected by lung cancer and gastric carcinoma, higher levels of miR-24, -26a, and -30d compared to controls were reported (Dai et al., 2010).

Prognostic biomarkers
In addition to their role as diagnostic biomarkers, miRNA can also act as prognostic and theranostic in several human solid tumors.
Low levels of circulating let-7a are associated with node positive breast cancer, compared to negative node disease (Henegan et al., 2010), whereas higher levels of miR-21 can be detected in patients with advanced breast cancer with respect to early stage disease (Asaga et al., 2011). Similarly, circulating miR-29a expression differs in early stage versus advanced colorectal cancer (Anand et al., 2010). In prostate cancer, higher serum levels of miR-375 and -141 are found in patients with advanced disease (Brase et al., 2011), whereas higher circulating miR-21 was found in hormone refractory prostate cancer, with respect to benign prostatic hyperplasia, localized prostate cancer and hormone dependent prostate cancer (Bo et al., 2011).

miRNAs in body fluids as tumor biomarkers in hematological malignancies
Diagnostic biomarkers
Higher levels of circulating miR-21, -155 and -210 have been described in patients affected by diffuse large B-cell lymphoma (DLBCL), compared to controls (Lawrie, 2008). Interestingly, the same group had previously shown that the expression of miR-155 in primary DLBCLs distinguishes between the activated B-cell phenotype (ABC) (higher expression of miR-155), than in the germinal center B-cell-like phenotype (GCB) (lower expression of miR-155) (Chen and Meister, 2005; Lawrie, 2007). Since, the 5-year survival rates of the ABC and the GCB subtypes of DLBCL are 30% and 59%, respectively (Kovanen et al., 2003), miR-155 expression in DLBCL has a prognostic value. A correlation between miR-155 and NFkB expression was found in DLBCL cell lines and patients (Abu-Elneel et al., 2008). In addition to miR-155, high levels of miR-21 and miR-221 are also associated with ABC-DLBCL and severe prognosis (de Yebenes and Ramiro, 2010). It would be interesting to investigate whether the expression of circulating miR-155 correlates with the expression of this miRNA in primary DLBCL, since it would indicate that miR-155 is a diagnostic biomarkers not only to put the diagnosis of DLBCL, but also of subtype of DLBCL.

Prognostic biomarkers
In DLBCL, increased serum levels of miR-21 are associated with a longer relapse-free survival (Lawrie, 2008), indicating that circulating miR-21 harbors prognostic implications in patients affected by DLBCL.
Overall, miRNAs can be detected in body fluids and increasing evidence shows that their expression in these fluids allows the diagnosis of cancer histotype and, in some cases histologic subtype. Finally, specific signatures of de-regulated miRNAs in body fluids harbor prognostic implications. These discoveries cast a new light on the translational implications of research in the miRNA field, by suggesting that these non-coding RNAs could be detected non-invasively and provide key diagnostic and prognostic clinical information.

miRNAs in invasion, angiogenesis and metastasis
In the last few years several studies have pointed out a critical role of miRNAs in tumor angiogenesis and metastasis. By regulating these processes miRNAs have emerged as crucial players, thus allowing primary tumor cells to invade adjacent tissues and reach through the systemic circulation distant sites in which they can finally proliferate as secondary tumors.
Depending on their role in the modulation of these processes, miRNAs can be subdivided into two groups: the anti-angiogenic and the pro-angiogenic ones.
Poliseno et al. demonstrated that the miR-221/miR-222 family has anti-angiogenic properties as it inhibits the angiogenic activity of stem cell factor SCF by targeting its receptor c-KIT in endothelial cells (Poliseno et al., 2006).
Since miR-21 plays a crucial role in cancer progression Sabatel et al. pondered whether it could also be involved in angiogenesis (Sabatel et al., 2011). Their in vitro and in vivo study revealed that mir-21 is a negative regulator of endothelial cell migration and tubulogenesis. Angiogenesis inhibition would occur through the targeting of RhoB, a  small GTPase which is responsible for the assembly of actin stress fibers (Aspenstrom et al., 2004). However, it seems that miR-21 has a dual role in the regulation of angiogenesis. Liu et al. in fact found that the overexpression of miR-21 in prostate cancer cell line increases the expression of HIF-1a and VEGF through the AKT and ERK pathway, thus acting as a pro-angiogenetic miRNA (Ayala de la Pena et al., 2011). Other miRNAs are known to be positive regulators for angiogenesis. For example, in vascular endothelial cells miR-130a downregulates the expression of the antiangiogenic homeobox genes HOXA5 and GAX in response to mitogens, proangiogenic and proinflammatory factors (Aumiller and Forstemann, 2008).
By using in vitro and in vivo studies Fang et al. found that miR-93 promotes angiogenesis and tumor growth by suppressing integrin-b8 expression and enhancing endothelial activity (Fang et al., 2011). Indeed this miRNA induces blood vessels formation, cell proliferation and migration by targeting the cell death-inducing antigen integrin-b8. The authors cannot exclude that miR-93 may also target other genes involved in tumorigenesis and angiogenesis.
Also, the miR-17-92 cluster promotes angiogenesis by inhibiting the expression of antiangiogenic protein thrompospondin-1 (TSP1) and connective tissue growth factor (CTGF) (Dews et al., 2006); miR-378 overexpression in glioblastoma cell line U87 enhanced angiogenesis and tumor growth through its targeting of tumor suppressor proteins SUFU and FUS-1 (Barakat et al., 2007); miR-296 is highly expressed in primary human brain microvascular endothelial cells and contributes to angiogenesis by directly targeting the hepatocyte growth factor-regulated tyrosine kinase substrate (HGS) mRNA, leading to decreased levels of HGS and thereby reducing HGS-mediated degradation of the growth factor receptors VEGFR2 and PDGFR-b (Gabriely et al., 2008).
Also in the regulation of the metastatic process miRNAs can be divided into two categories: pro-metastatic (such as miR-340, miR-92a, miR-10b, miR-373/520c) or anti-metastatic (such as miR-101, miR-34a, miR-126, miR-148a, miR-335) ones.
In breast cancer reduced miR-340 expression is associate with tumor cell migration, invasion and poor prognosis (Dong et al., 2011).
Of the six mature miRNAs produced by the miR-17-92a cluster, miR-92a is involved in the metastatization process. It has been reported that miR-92a is highly expressed in tumor tissue from ESCC (Esophageal Squamous Cell Carcinoma) patients (Cai et al., 2008). Chen et al. verified whether there is a correlation between the relative expression of miR-92a in tumor and normal tissues and lymph node metastasis in ESCC patients. Not only they found that miR-92a promotes ESCC cell migration and invasion through the inhibition (by direct targeting) of CDH1, which is known to mediate cell-to-cell adhesion, but also that ESCC patients with up-regulated miR-92a are prone to lymph node metastasis and poor prognosis (Bao et al., 2011).
In 2007, Ma et al. reported that miR-10b is highly expressed in metastatic breast cancer cells, when compared with non-metastatic cells. However, when overexpressed in the latter it promotes robust invasion and metastasis. Induced by the transcription factor Twist, miR-10 inhibits the translation of the messenger RNA encoding HOXD10 (homeobox D10), thus increasing the expression of the pro-metastatic gene RHOC and leading to tumor invasion and metastasis (Derby et al., 2007).
Through the transduction of a non-metastatic breast cancer cell line with a miRNA expression library Huang et al. studied which miRNAs could allow the cells to migrate. MiR-373 and miR-520c were found to promote cell invasion and metastasis both in vitro and in vivo through the inhibition of the expression of CD44, a protein involved in cell adhesion (Abdurakhmonov et al., 2008).
As previously reported, miRNAs are known also to have an anti-metastatic role. One of them is miR-101, whose expression decreases during prostate cancer progression, as depicted by Varambally et al. (Varambally et al., 2008). The authors showed that during this process there's a negative correlation between the expression of miR-101 and EZH2, a mammalian histone methyltransferase overexpressed in solid tumors (Varambally et al., 2002) and involved in the epigenetic silencing (Yu et al., 2007; Cao et al., 2008) of genes responsible for tumor invasion and metastasis.
By performing experiments based on computational analysis the authors showed also that miR-101 targets EZH2. Loss of miR-101, paralleled by increased levels of EZH2 in the tumor, leads to dysregulation of epigenetic pathways and cancer progression.
Another miRNA typically downregulated in tumors (colorectal cancer (Tazawa et al., 2007), pancreatic cancer (Chang et al., 2007), and neuroblastoma (Welch et al., 2007)) is miR-34a. Li et al. observed that in hepatocellular carcinoma miR-34a is also down-regulated (Li et al., 2009) and its expression is inversely correlated with that of the receptor for the hepatocyte growth factor c-MET (Leelawat et al., 2006), involved in cell invasion and metastasis. In their study the Authors demonstrated that miR-34a targets c-MET when ectopically expressed in Hep-G2 cells and observed reduced cell scattering, migration and invasion.
Crk (v-crk sarcoma virus CT 10 oncogene homolog) is a protein that regulates cell motility, differentiation and adhesion (Kobashigawa et al., 2007). High expression levels of this protein are found in several human tumors such as breast, ovarian, lung, brain, stomach and chondrosarcoma (Wang et al., 2007) and knock down of Crk decreases cell migration and invasion (Rodrigues et al., 2005; Wang et al., 2007). Crawford et al. showed that Crk is a functional target of miR-126 in NSCLC tumors and that overexpression of miR-126 induces a decrease in adhesion, migration and invasion (Crawford et al., 2008).
Finally, the list of anti-metastatic miRNAs includes miR-206 and miR-335. In a manuscript published in 2008 Tavazoie and coworkers took under consideration a set of miRNAs whose expression was lost in human breast cancer cells (Tavazoie et al., 2008). Among these they considered miR-206 and miR-335. By restoring their expression through retroviral transduction they found that the ability of these cells to migrate to the lung was lost. MiR-335 exerts its anti-metastatic role by targeting PTPRN2 (receptor-type tyrosine protein phosphatase) (Varadi et al., 2005), MERTK (the c-Mer tyrosine kinase) (Graham et al., 1995), SOX4 (SRY-box containing transcription factor), the progenitor cell transcription factor (van de Wetering et al., 1993; Hoser et al., 2007) and TNC (tenascin C) (Ilunga et al., 2004), which is an extracellular component of the matrix.

Therapeutic implications of miRNAs in oncology
The involvement of miRNAs in different aspects of human carcinogenesis, such as cell proliferation, apoptosis, differentiation, angiogenesis, motility and metastasis, has raised the question whether reverting these aberrations of the miRNome can be effectively used for therapeutic purposes. Preclinical data encourage this hypothesis and provide the biological rationale for clinical studies in this direction. Re-expression of miRNAs down-regulated in cancer (e.g. miR-15a and miR-16 in BCL2 positive CLL) and/or silencing of miRNAs up-regulated in the tumor (e.g. miR-155 in lung cancer) may lead to cancer cell apoptosis and exert a therapeutic effect. Before this becomes a reality in patients though, several issues need to be solved. First, there is a need to know the full spectrum of targets and effects that a given miRNA has on a given genome. It has been estimated that a single miRNA cluster (namely, the miR-15a/16-1 cluster) is able to affect, directly and indirectly, the expression of about 14% of the whole human genome (Calin et al., 2008). Also it is clear that each miRNA is able to target both OGs and TSGs, and that the phenotype induced by the external manipulation of a miRNA is the result of this combined targeting effect on several genes. Therefore, one of the goals of the preclinical research is to fully clarify this aspect before any clinical application can even be taken into consideration. Secondly, it needs to be established how can we reach a tumor-specific delivery of the miRNAs of interest? This question is more general, and involves the whole field of gene therapy, being not limited to the research on miRNAs. The advent of nanoparticles, able to target tumor-specific antigens hopefully will address this concern and allow tumor specificity. Another aspect of relevance consists in determining how the modulation of miRNA expression can integrate the existing anti-cancer therapies (chemo-, radio-, hormonotherapy)? Interestingly, some studies have been published showing that miRNAs can restore sensitivity to current therapeutic options to which the tumors became resistant, and this encourages a certain optimist on miRNA-inclusive association regimens. The other questions on what is the best formulation of miRNAs to be administered, and what are the pharmacokinetics and pharmacodynamics of these ncRNAs in humans will be answered (as always) by the established phases of the clinical studies.

Conclusion
The involvement of miRNAs in human cancer development and progression has been proven without any doubt by several studies. Other aspects of miRNA research are still under development, such as their role as molecular biomarkers (the published studies still suffer in most cases from a limited number of patients, which questions the statistical power of certain results), the identification of the full spectrum of targets of a given miRNA (in particular, there is a need to critically interpret the plethora of the identified targets in light of the specific genome in which the effect is observed, and in relation to the other identified and validated targets of that same miRNA), and their interaction with the existing treatments (the number of published studies on this regard is still relatively small to allow any safe conclusion). Nonetheless, despite there seems to be still a lot of work ahead, it is promising that in such a relatively small amont of time, from the discovery of their involvement in human cancer, till today so much has been discovered about miRNAs and cancer. The effort devoted by the scientific community in this research field is unprecedented, allowing a certain optimism for the years to come, in which the introduction of these ncRNAs in the clinical practice seems about to become a realistic option.