• Open Access

Systematic exploration of cancer-associated microRNA through functional screening assays


To whom correspondence should be addressed.
E-mail: hnakagam@ncc.go.jp


MicroRNA (miRNA), non-coding RNA of approximately 22 nucleotides, post-transcriptionally represses expression of its target genes. miRNA regulates a variety of biological processes such as cell proliferation, cell death, development, stemness and genomic stability, not only in physiological conditions but also in various pathological conditions such as cancers. More than 1000 mature miRNA have been experimentally identified in humans and mice, yet the functions of a vast majority of miRNA remain to be elucidated. Identification of novel cancer-associated miRNA seems promising considering their possible application in the development of novel cancer therapies and biomarkers. Currently, there are two major approaches to identify miRNA that are associated with cancer: expression profiling study and functional screening assay. The former approach is widely used, and a large number of studies have shown aberrant miRNA expression profiles in cancer tissues compared with their non-cancer counterparts. Although aberrantly expressed miRNA are potentially good biomarkers, in most cases a majority of them do not play causal roles in cancers when functional assays are performed. In contrast, the latter approach allows screening of ‘driver’ miRNA with cancer-associated phenotypes, such as cell proliferation and cell invasion. Thus, this approach might be suitable in finding crucial targets of novel cancer therapy. The combination of both types of approaches will contribute to further elucidation of the cancer pathophysiology and to the development of a novel class of cancer therapies and biomarkers. (Cancer Sci 2011; 102: 1615–1621)

MicroRNA (miRNA) belongs to a class of non-coding RNA of approximately 22 nucleotides, and post-transcriptionally represses expression of its target genes.(1) miRNA are sequentially processed from precursors, either primary transcripts transcribed from the genome or intronic sequences of protein coding genes (‘miRtrons’), by RNase III nuclease Drosha and Dicer. After their processing, miRNA are incorporated into the RNA-induced silencing complex (RISC), and the formed complex in turn represses the expression of the target genes, which have partially complimentary sequences with the miRNA in their 3′ untranslated regions (UTR), either by translational repression or cleavage of the target mRNA.(2) Although 3′ UTR is the main target of miRNA, 5′ UTR and open reading frames (ORF) were also reported as target sites of miRNA.(3) More than 60% of all protein coding genes have conserved miRNA binding sites in their 3′ UTR and are implicated to be the targets of miRNA.(4,5)

While miRNA was initially identified in Caenorhabditis elegans, it has been demonstrated that miRNA are evolutionarily conserved in many species, suggesting their universal roles in the regulation of gene expression.(6) The number of miRNA whose expression has been experimentally verified has grown rapidly over the last decade. This is partly because the use of sequencing-by-synthesis technology enabled the identification of novel miRNA with low-level expression.(7,8) Currently, 19 724 mature miRNA from 153 species are registered at the miRNA database (miRBase release 17, http://www.mirbase.org), including 1733 (1424 miRNA genes) in the human and 1111 (720 miRNA genes) in the mouse.(9) However, the functions of the vast majority of these miRNA remain to be elucidated. Because the roles of a miRNA depend on its target mRNA and the consequence of repressing multiple target mRNA under a specific cellular condition is difficult to predict, it is necessary to explore miRNA functions under experimental conditions of interest.

miRNA and cancer

It is estimated that more than 60% of all protein coding genes are the potential targets of miRNA.(4,5) Naturally the reported roles of miRNA are implicated in almost all aspects of cellular functions, including cell differentiation, cell death, cell cycle, developmental timing, inflammation, metabolism and stemness.(10–15) As expected from their involvement in normal physiological functions, dysregulation of miRNA expression has been shown to be involved in the pathogenesis of a wide variety of pathological conditions, such as heart disease, neurodegenerative disease and cancer. The dysregulation of miRNA is implicated in almost all aspects of cancer characteristics, including cell cycle, apoptosis, invasion/metastasis, angiogenesis and hypoxia-resistance.

The first evidence that miRNA is involved in the pathogenesis of cancer was obtained from the study of chronic lymphocytic leukemia (CLL), in which miR-15a and miR-16-1 were identified on a region of the genome that was frequently lost in CLL patients.(16) These miRNA target anti-apoptotic protein BCL2, and their downregulation promotes cancers. Hence, it was proposed that these miRNA have a ‘tumor-suppressive’ role in the pathogenesis of CLL.(17) Since then, a number of reports have demonstrated the involvement of miRNA in cancers (Table 1).

Table 1.   Cancer-associated miRNA
Cell cyclemiR-16Suppress CDK6, CARD10, CDC27; induce G0/G1 accumulation41
miR-17-92Suppress E2Fs42
miR-27aSuppress Myt-143
miR-34aSuppress Cdk4/6, cyclin E2, E2F320
miR-122aSuppress cyclinG144
miR-124aSuppress CDK645
miR-221/-222Suppress p27(Kip)46
let-7Suppress Cdk6, Cdc25a, Cyclin D2; induce G0/G1 accumulation41
ApoptosismiR-15a/16-1Suppress Bcl-2; induce apoptosis17
miR-22Suppress p21; induce apoptosis in p53 wild cells40
miR-29bSuppress Mcl-1; suppress apoptosis47
miR-34a/b/cSuppress Bcl-2; induce apoptosis20
Invasion/metastasismiR-10bSuppress HOXD10; promote invasion/metastasis48
miR-21Suppress PTEN, Pdcd4; promote motility/invasion49
miR-125a/bSuppress ERBB2/3; suppress motility/invasion50
AngiogenesismiR-27aSuppress Zbdb10, promote angiogenesis43
miR-17-92Suppress Tsp1, CTGF; promote angiogenesis42
miR-296Suppress HGS: promote angiogenesis51
miR-378Suppress SuFu, Fus-1; promote angiogenesis52
Oncogene-associated miRNAMycmiR-9, miR-17-9242
Tumor-suppressor-associated miRNAp53miR-26a, -34, -30c, -103, -107, -182, etc.18

Whereas miR-15a and miR-16-1 were identified through the study of aberrant chromosomes, most cancer-associated miRNA have been identified through expression analyses of miRNA in cancer tissues or by a ‘candidate approach’ in which potential cancer-associated miRNA that target known oncogenes or tumor-suppressor genes are evaluated as to whether they have cancer-related phenotypes. Interestingly, some miRNA in turn are directly regulated by cancer-associated transcriptional factors, including Myc, HIF, Stat3, p53 and Twist.(18) Thus, the emerging picture indicates that miRNA and transcriptional factors form an intertwined network during the development of cancers.

Oncogenic miRNA and tumor-suppressive miRNA.  miRNA associated with the pathogenesis of cancers can either be classified as oncogenic miRNA (oncomiR) or tumor-suppressive miRNA, although their roles are sometimes dependent on cellular context. As suggested by their name, oncogenic miRNA promote phenotypes associated with cancers, including cell proliferation, invasion and resistance to apoptosis. OncomiR are in many cases upregulated in cancers, and their elevated expression is indispensable for sustained growth of cancer cells.(19) Therefore, inhibition of these miRNA by anti-miRNA can be a new class of molecular-targeted therapy.

In contrast, tumor-suppressive miRNA are miRNA that have anti-tumor functions. Among them is the miR-34 family that represses E2F3, Cdk4, Bcl2 and MET in response to genotoxic stress.(20–22) Of note, miR-34 is a downstream effector of p53, and introduction of miR-34 in cancer cells induces either apoptosis or premature senescence. MiR-34a upregulates p53 by translational repression of SIRT1, thus miR-34a and p53 constitute a positive feedback loop.(23)

Interestingly, miRNA are globally downregulated in many cancer tissues compared with their non-cancerous counterparts.(24) Knockdown of miRNA processing components promote cellular transformation and tumor growth in vitro and in vivo.(25) Furthermore, reduced expression of Dicer in a subset of lung cancer has been shown to be associated with poor prognosis.(26) These observations suggest that there exists a subset of unidentified miRNA whose downregulation promotes cancer development. Identification of such tumor-suppressive miRNA will be a promising area of future cancer research.

Identification of cancer-associated miRNA for clinical application

Because oncogenic and tumor-suppressive miRNA confer cancer-promoting or cancer-suppressing characteristics to cancers, these miRNA are regarded as potential targets for novel cancer therapies. In addition, aberrantly expressed miRNA can be used for the diagnosis of cancers. Considering that the functions of a substantial proportion of miRNA are not known, systematic exploration of cancer-associated miRNA might be beneficial to detect such clinically relevant miRNA. There are currently two major approaches to explore cancer-associated miRNA: expression analysis and functional assay.

Elucidation of cancer-associated miRNA through expression profiling.  A body of evidence indicated there is a number of miRNA aberrantly expressed in cancer tissues in comparison with their non-cancerous counterparts (Table 1). Currently there are several techniques available for miRNA expression analysis including cloning, northern blotting, serial analysis of gene expression (SAGE), microarray, quantitative RT-PCR, in situ hybridization and sequencing-by-synthesis technology. Analyses of aberrant chromosomes and methylation status have also been performed to elucidate the underlying mechanisms of aberrant miRNA expression.(16,27) Among them, microarray is an experimental technique that is most widely used for genome-wide miRNA expression profiling. miRNA appear to be relatively stable in various storage conditions of cancer tissues, including fresh frozen tissues and formalin-fixed paraffin-embedded tissues. The relative stability of miRNA makes them good candidates for biomarkers, and expression profiling of miRNA by microarray has become an intense focus of current cancer research.(28) miRNA can also be detected in body fluids, especially blood, and they exhibit altered expression profiles in cancer patients, making them a new class of biomarker.(29–32)

Microarrays have been successfully used for genome-wide expression profiling of miRNA as well as protein-coding genes. However, there have been some technical drawbacks for microarray analysis of miRNA. Because mature miRNA are as short as ∼22 nucleotides and members of a miRNA family are highly similar, discrimination of mature miRNA and their precursors is necessary for precise expression profiling of mature miRNA. To overcome the problem, several novel technologies are beginning to emerge for miRNA profiling, including the use of hairpin-structured or locked-nucleic acid (LNA) probes, length-adjusted probes and microfluidics platforms.(33–35) Another drawback of microarray analyses of miRNA is its lack of a proper normalization method. Because of such drawback of the microarray, quantitative RT-PCR (qRT-PCR) is widely used for the quantification of individual miRNA as well as genome-wide miRNA profiling.

Alternative approach to detect cancer-associated miRNA: functional screening assay of miRNA

Another approach to elucidating miRNA that are associated with cancers is a functional screening assay (Fig. 1).(36–38) This is an assay that enables identification of miRNA that are causally linked to phenotypes of interest, irrespective of their levels of expression. Although there are several methodological variations among these assays, they are basically composed of the following two steps: (i) systematic introduction of miRNA into cells; and (ii) detection of exogenously introduced miRNA that confer cancer-associated phenotypes. Either the arrayed single-plex assay or the pooled multi-plex format with a virus-based expression vector is available for the functional screening assay of miRNA (Table 2).

Figure 1.

 Schematic view of the functional screening assays. In a single-plex format (A), miRNA undergo functional assays (e.g. cell proliferation assay, cell invasion assay, drug sensitivity assay, etc.) individually in separate wells of a multi-well plate. In a multi-plex format with viral-based expression vectors (B), a large number of cells are infected with a virus library expressing miRNA and then undergo phenotypic screening (e.g. cell proliferation). In the screening of proliferation-associated miRNA, cells transduced with each miRNA clone change their proportions in the whole cell population according to their effects on cell proliferation, which can be quantified using microarrays. For example, cells transduced with a proliferation-suppressive miRNA (miR-X1) or a proliferation-promotive miRNA (miR-X2) increase or decrease their proportion, respectively.

Table 2.   Representative cancer-associated phenotypes assayed in the functional screening assays of miRNA and genome-wide RNAi screening
Array based (single-plex)Cell proliferation/survival(54)Partners of KRAS(55)
Pool based (multi-plex)Cell proliferation/survival(36)Cell proliferation/survival(56–58)
Cellular transformation(38)Resistance to nutlin-3(59)
Cell migration and invasion(60)p53-dependent proliferation arrest(61)
Resistance to chemotherapy(62)
Suppressor of epithelial cell transformation(63,64)
Tumor suppressor in a mouse lymphoma model(65)

Single-plex format using multi-well plate.  In a single-plex assay, each miRNA is individually introduced to cells and their effects on cells are separately examined in a functional assay in multi-well plates (i.e. 96-well or 384-well format). Synthetic miRNA-mimics or vector-based miRNA can be used under this setting (Table 3). Using 319 synthetic miRNA in a 96-well plate format and chronometric cell viability assay, Nakano et al.(37) conducted single-plex gain-of-function analysis of miRNA associated with cell proliferation and identified a number of miRNA that increase or decrease cell viability in DLD-1 colon cancer cells. Among them are miR-362, -491 and -132, which do not exhibit aberrant expression in clinical colorectal cancer (CRC) samples.

Table 3.   Representative commercially available miRNA library
  1. FIV, feline immunodeficiency virus; H, human; HIV, human immunodeficiency virus; M, mouse; MSCV, murine stem cell virus; R, rat.

Synthetic miRNA-like molecules
miRNA mimic (gain-of-function analysis)
 Pre-miR miRNA precursor molecule (Ambion, Austin, TX, USA)H, MDouble stranded
 miRIDIAN microRNA Mimic (Thermo Fisher Scientific, Lafayette, CO, USA)H, M, RDouble stranded
 miScript miRNA Mimics (Qiagen, Hilden, Germany)H, M, RDouble stranded
 MISSION (Sigma-Aldrich, St. Louis, MO, USA)HDouble stranded
miRNA inhibitor (loss-of-function analysis)
 miRCURY LNA microRNA inhibitor (Exiqon, Vedbaek, Denmark)H, MSingle stranded
 Anti-miR miRNA inhibitors (Ambion)H, MSingle stranded
 miRIDIAN microRNA hairpin inhibitor (Thermo Fisher Scientific)H, M, RSingle stranded
 miScript miRNA inhibitors (Qiagen)H, M, RSingle stranded
 miArrest (GeneCopoeia, Rockville, MD, USA)H, M, RSingle stranded
Virus vector-based miRNA
miRNA (gain-of-function analysis)
 Virus vector (non-barcoded)
  Lenti-miR microRNA (System Biosciences,  Mountain View, CA, USA)HHIV based, expressing miRNA precursors
  miRNA library (miR-Lib) (NKI, Amsterdam,  The Netherlands)HMSCV based, expressing miRNA precursors
  miExpress (GeneCopoeia)H, M, RFIV based, expressing miRNA stem-loop
  miRIDIAN shMIMIC microRNA (Thermo Fisher Scientific)HHIV based, expressing mature miRNA incorporated into a universal scaffold
 Non-viral vector
  miExpress (GeneCopoeia)H, M, RNon-viral vector expressing miRNA precursor
miRNA inhibitor (loss-of-function analysis)
 miRZIP (System Biosciences)HHIV based
 miArrest (GeneCopoeia)H, M, RHIV based

Multi-plex format with viral-based expression vectors.  A pooled multi-plex format with a virus-based expression vector is more complex but can be used for a wide range of applications.(39) Vector-based miRNA, especially those expressing miRNA precursors by retrovirus- or lentivirus-based vectors, are generally used in multi-plex assays, and stable expression of introduced miRNA allows functional screening under various experimental settings (Table 3). In general, a pooled library of miRNA is transduced to a single large cell population and biological selection of the transduced cells is performed. After selecting a cell sub-population with a phenotype of interest, miRNA that cause phenotypic changes are identified by sequencing. Although the viral-based approach has been successfully used to detect oncogenes or OncomiR, in general detection of these factors relies on the growth advantage they confer, and it is difficult to detect anti-miR via such an approach.

An alternative approach to identify cancer-related miRNA from library-transduced cells was initially demonstrated by Voorhoeve et al.(38) Genetic screening of miRNA was performed using a retrovirus library of miRNA precursors (∼500 bp) and DNA barcode arrays. They successfully identified miR-372 and miR-373 as oncogenic miRNA that cooperate with oncogenic K-ras mutation in immortalized primary fibroblasts (BJ/ET cells). This is an example of the positive screening assay in which a cell population with phenotypes of interest (e.g. cell proliferation) increases during the selection process and the responsible clones are identified either by sequencing or microarrays. The combination of the miRNA-expressing virus library and the custom-made microarray can be used in the negative screening assay (‘drop-out’ screening) in which a cell population decreases in the selection process.

We have successfully identified miRNA that negatively regulate cell proliferation in pancreatic cancer cells using a lentivirus library of ∼450 miRNA precursors and custom-made microarray.(36) Changes in the relative abundance of a miRNA clone (e.g. miR-X1) in the whole cell population were quantified by the comparison of differently labeled miRNA clones recovered immediately after infection or after several passages (Fig. 1B). Five miRNA exhibited remarkable reduction in their abundance (log10 ratio <−1), indicating the proliferation-suppressive effect of these miRNA. Interestingly, one of these five miRNA was miR-34a, a representative tumor-suppressive miRNA that is transactivated by p53.(20,21) MiR-34a does not exhibit aberrant expression in pancreatic cancers or it exhibits mild upregulation in colorectal cancers. We have also reported that miR-222, which is upregulated in pancreatic cancers and has been shown to be tumor promotive by targeting p27, PUMA and PPP2R2A, has a tumor-suppressive effect in pancreatic cancer cells. These results suggest a methodological advantage of this functional screening assay to detect hidden cancer-related genes, and illustrates the need to explore miRNA functions in various tissues as the functions of miRNA are sometimes context-dependent, further showing the importance of experimental validation of cancer-associated miRNA using a functional screening assay. The flexibility of a multi-plex format assay also warrants exploration in more complicated settings, including in vivo settings. Moreover, use of lentivirus vectors broadens the possible application of the assay to primary or non-dividing cells, including neural and stem cells.

miR-22, a novel tumor-suppressor gene, identified by functional genetic and comprehensive genomic analyses.  The usefulness of a functional screening assay in the exploration of tumor suppressive miRNA can further be strengthened by combining the results of the assay with other data, especially those of clinical samples. We integrated the functional screening assay, expression profiling and chromosome analysis to identify miRNA species that function as tumor-suppressor genes in colon cancer(40). Tumor-suppressor miRNA were defined as miRNA that: (i) repress cell proliferation; (ii) is expressed in normal colon tissues; (iii) is located at a frequently lost region on the chromosome; and (iv) is downregulated in colon cancers. As indicated in Figure 2A, considerable numbers of miRNA clones dropped out during the culture of HCT 116 or SW480 cell lines. Expression analysis of proliferation-suppressive (‘drop-out’) clones in non-cancerous colon tissues (Fig. 2B) and copy number analysis of colon cancer tissues (Fig. 2C) led to a novel tumor-suppressor miRNA miR-22 that satisfies the aforementioned four criteria.

Figure 2.

 (A) Results of dropout screening. HCT116 and SW480 cells were transduced with a lentivirus pooled miRNA expression library at multiplicity of infection (MOI) of 3. Exogenous transduced miRNA precursor genes were PCR amplified from genomic DNA recovered from cells immediately after library infection (passage 1, P1) and after several passages (P9 in HCT 116 cells and P7 in SW480 cells). Amplified DNA were labeled with Cy3 (P1) and Cy5 (P9 and P7), and competitively hybridized onto a custom-made microarray. Graphs are a scattered plot of the log10 ratio of each array set. (B) Expression profile of miRNA in non-cancerous parts of four colon cancer specimens using microarrays. A heat map was made using expression levels of 73 dropout miRNA. (C) Copy number aberration of chromosome 17 in 24 human colon cancer patients. Green and red indicate loss and gain, respectively. The position of the miR-22 gene is shown by the red dot. (D) Cell proliferation assay. HCT 116 and SW480 cells were transfected with 5 nM of either miR-negative control (NC) or miR-22 (22) and incubated for 5 days. Cell viability was measured by MST (Promega, Fitchburg, WI, USA) assay. Error bars indicate standard deviation in triplicate cultures. (E) Possible roles of miR-22 as an intrinsic molecular switch. In the exposure to oncogenic stresses, activated p53 transcriptionally activates both p21 and miR-22. MiR-22 represses p21 directly through the inhibition of translation and enhancement of mRNA degradation. Repression of p21 might cause the changes of cellular state from cell cycle arrest to apoptosis.

The activity of repression for cell proliferation was again assessed by the introduction of miR-22 in both HCT 116 and SW480 cells (Fig. 2D). Interestingly, miR-22 induced apoptosis only in p53 wild-type cells, but it caused cell cycle arrest in p53 mutant cells.(40) Furthermore, we found that miR-22 is a transcriptional target of p53 and directly represses p21. Our findings define an intrinsic molecular switch that controls apoptosis by direct repression of p21 in response to strong stresses, in which cells should eliminate severely damaged cells to prevent malignant transformation (Fig. 2E).

In summary, functional screening of miRNA using either a single-plex or multi-plex format is a powerful genetic approach for the systematic elucidation of miRNA that cause cancer-associated phenotypes independent of their expression. Moreover, by combining the functional screening assay with expression profiling and genomic analysis, it should further facilitate the identification of novel cancer-associated miRNA that have a vital role in cancer pathophysiology, such as the miR-34 family and miR-22.


There is growing interest in the clinical application of miRNA. Aberrantly expressed miRNA in cancer tissues are good candidate biomarkers for the diagnosis of cancers and the prognosis of cancer patients, as has been shown by a large number of studies.(18) In addition, they are suitable as a good biomarker because of their ease of detection, high stability in clinical specimens and availability from the blood of cancer patients. Furthermore, miRNA themselves can be a novel class of molecular targets in cancer therapy. In theory, suppression of upregulated oncogenic miRNA by anti-miRNA or introduction of downregulated tumor-suppressive miRNA by synthetic or viral-vector based miRNA might be effective to cure cancer, although the development of effective drug delivery systems is another challenge. Identification of regulators of the mechanisms of cancer-associated miRNA will provide novel therapeutic targets. Despite many obstacles, exploration of cancer-associated miRNA will contribute to the elucidation of the pathogenesis of cancers and the development of novel cancer therapies and biomarkers.


This work is supported by the Program for Promotion of Fundamental Studies in Health Sciences of the National Institute of Biomedical Innovation (NIBIO), and a Grant-in-Aid for 3rd Term Comprehensive 10-Year Strategy for Cancer Control from the Ministry of Health, Labour and Welfare, Japan.

Disclosure Statement

The authors have no conflict of interest.