Abstract

Introduction

MicroRNAs (miRNAs) are short noncoding RNAs ~22 nucleotides in length that were first discovered in Caenorhabditis elegans to control developmental timing (Lee et al., 1993; Reinhart et al., 2000). They are generated from large RNA precursors (termed pri-miRNAs) that are processed in the nucleus by the RNase III enzyme Drosha and the protein Pasha/DGCR8 into ~70 nucleotides pre-miRNAs, which fold into imperfect stem-loop structures (Lee et al., 2003). The pre-miRNA undergoes an additional processing step within the cytoplasm, and a small double-stranded RNA structure approximately 22 nucleotides in length is excised from the pre-miRNA hairpin by another RNase III enzyme, Dicer (Hutvagner et al., 2001; Ketting et al., 2001). The mature single-stranded miRNA is then loaded into the miRNA-associated, multiprotein RNA-induced silencing complex (miRISC), where it guides the RISC to target sequences through the binding of imperfect complementary sites within the 3′ untranslated regions (UTR) of messenger RNA (mRNA) transcripts and regulates gene expression by either translational repression or degradation of the mRNA (Bartel, 2004).

These novel classes of cellular regulators have been shown to be involved in regulating various processes, such as cell cycle, proliferation, apoptosis, differentiation and development (Esquela-Kerscher and Slack, 2006; Johnson et al., 2007; Hermeking, 2009; Wang and Olson, 2009). Recent studies have also provided evidence that aberrant expression of specific miRNAs is implicated in a broad spectrum of human diseases, including diabetes, cardiovascular disease and cancer (Esquela-Kerscher and Slack, 2006; Pandey et al., 2009; Zorio et al., 2009).

One of the first studies to document the link between miRNAs and cancer was carried out by Calin et al. (2002). In this work, the authors observed that mir-15 and mir-16 located on chromosome 13q14 were preferentially deleted and downregulated in about 68% of patients with B-cell chronic lymphocytic leukemia, a common form of adult leukemia. The same group of investigators also found that miRNA genes are frequently located at cancer-associated chromosomal fragile sites, such as minimal regions of loss of heterozygosity, minimal regions of amplification and common breakpoint regions (Calin et al., 2004). Frequent reduced expression or deletion of these miRNAs in cancer strongly implicated them as tumor suppressors.

Other well-studied potential tumor suppressor miRNAs are the members of the let-7 family (Roush and Slack, 2008). Work from the our lab previously found that let-7 is consistently downregulated in lung cancers relative to normal adjacent tissue (NAT). We also found that in lung cancer tissue, where let-7 levels are low, RAS protein levels are elevated relative to the NAT. In support of the significance of our findings, reduced expression of let-7 is correlated with poor prognosis, as lung cancer patients with the least let-7 expression had shorter time of survival after surgery (Takamizawa et al., 2004; Yanaihara et al., 2006). We and others subsequently showed that let-7 directly regulates multiple oncogenes, including RAS, MYC and HMGA2 (Johnson et al., 2005; Lee and Dutta, 2007; Mayr et al., 2007; Sampson et al., 2007), and promoters of cell-cycle progression, such as CDC25A, CDK6 and Cyclin D2 (Johnson et al., 2007). Moreover, administration of let-7 blocks the growth of cultured lung cancer cells (Johnson et al., 2007).

Conversely, miRNAs have also been identified as potential oncogenes. A potentially oncogenic cluster of miRNAs, the polycistron, miR-17–92, containing seven miRNAs: miR-17–5p, miR-17–3p, miR18a, miR-19a, miR-19b-1, miR-20, and miR-92–1 was found to be overexpressed in many kinds of lymphoma samples compared with normal tissues (He et al., 2005b). Using a mouse B-cell lymphoma model, He et al. (2005b) found that miR-17–92 in conjunction with MYC accelerates tumor development. At the same time, O’Donnell et al. (2005) independently identified the miR-17–92 cluster as a group of potentially cancer-related genes.

Various studies have reported that miRNAs are also aberrantly expressed in brain tumors. During the analysis of a global expression level of 245 miRNAs in glioblastoma multiforme, Ciafre et al. (2005) observed that miR-221 was highly upregulated. In a recent work, Gillies and Lorimer (2007) have reported that miR-221 and miR-222 directly target p27^Kip1, a key negative regulator of the cell cycle, in glioblastoma. In another study, Chan et al. (2005) also observed that miR-21 was strongly overexpressed (5- to 100-fold) in highly malignant glioblastoma tumor tissues. The investigators also found that the knockdown of miR-21 in cultured glioblastoma cells activated caspases and resulted in more cell death by an apoptotic pathway. In a comprehensive analysis of miRNA expression signature of human solid tumors, Volinia et al. (2006) found that the over-expression of miR-21 is shared by all the six solid tumors studied, breast, colon, lung, pancreas, prostate and stomach. This finding suggests that miR-21 has an important regulatory role in a pathway shared by all these sites. Despite the evidence that miRNAs cause cancer, we still understand little about the mechanism of these miRNAs and how they affect cancer progression.

MiRNA profiling in cancers: diagnosis, classification prognosis and risk factors

miRNA expression profiling studies using microarrays and other methods can differentiate normal from cancer tissues through a unique miRNA expression signature. These differences can classify different tumor types and tumor grades. Certain miRNA signatures are correlated with prognosis and can potentially be used to determine the specific course of treatment (Takamizawa et al., 2004; He et al., 2005a; Lu et al., 2005; Yanaihara et al., 2006; Shell et al., 2007; Visone et al., 2007).

Using a novel bead-based miRNA microarray assay, Lu et al. (2005) conducted the expression analysis of 217 miRNAs in a panel of 334 samples that included primary tumors, tumor-derived cell lines and normal tissues. The investigators found that miRNA profiles can distinguish between normal and cancer tissues, separate different cancer types, stratify the cancer differentiation state and cluster sample groups according to their embryonic lineage. Moreover, miRNA profiles were more accurate than mRNA profiles in classifying tumor types and were able to categorize 17 poorly differentiated tumors where histological appearance was not diagnostic into their specific tissue lineages.

Studying colorectal cancer, Michael et al. (2003) were the first to recognize aberrant miRNA expression in solid tumors as the investigators identified 28 different miRNAs in colonic adenocarcinoma and normal mucosa, and found that miR-143 and miR-145 were consistently downregulated in the cancer. In breast cancer, the first miRNA profiling study showed that a set of 15 miRNA signatures correctly predicted normal versus cancer tissues with 100% accuracy (Iorio et al., 2005). Papillary thyroid carcinoma (PTC) is the most common malignancy in thyroid tissue. Profiling of PTC by He et al. (2005a) yielded a signature of five miRNAs (which included miR-221, miR-222 and miR-146) that can separate PTC from normal thyroid. Bloomston et al. (2007) showed that a signature of 21 upregulated and 4 downregulated miRNAs correctly differentiated pancreatic cancer from benign pancreatic tissue in 90% of samples. In the profile, two commonly malignancy-associated miRNAs, miR-21 and miR-155, were uniquely overexpressed in pancreatic cancer compared to normal pancreas. Of significant importance is a study performed by Rosenfeld et al. (2008), which needed only 48 miRNA markers to identify 22 tissue origins with an accuracy of 90%. This work holds promise for the 3–5% of patients from all new cancer cases diagnosed with metastatic cancer of unknown primary origin (Pimiento et al., 2007). Enabling oncologists to identify the tissue of origin will markedly improve treatment decision and prognosis.

Recently, a very new and exciting venue of potential cancer diagnosis has emerged through the detection of miRNAs in serum. In one of the first reports of this method, Lawrie et al. (2008) showed that sera levels of miR-21 were associated with relapse-free survival in patients with diffuse large B-cell lymphoma. In another study, Mitchell et al. (2008) showed that miRNAs can enter the circulation even when they originate from an epithelial cancer cell type. The investigators found that serum levels of miR-141 can distinguish patients with prostate cancer from healthy subjects. Another two groups were able to identify functional miRNAs in circulating tumor exosomes with specific miRNA signature-discerning patients with ovarian and lung cancers from healthy controls (Taylor and Gercel-Taylor, 2008; Rabinowits et al., 2009).

One continuing concern for oncologists is giving patients the right prognosis and being able to predict the outcome for certain types of cancer. Several studies profiling the different levels of Dicer and Drosha and miRNAs between normal and tumor tissues were able to shed some light on this problem. In a recent study, showed that levels of the miRNA-processing enzymes Drosha and Dicer were able to predict survival in ovarian cancer, with low level of Dicer associated with advanced tumor stage, reduced response to chemotherapy and poor clinical outcome. Low levels of Drosha are associated with suboptimal surgical cyto-reduction, a poor prognostic factor. High levels of Dicer and Drosha are associated with increased median survival (Merritt et al., 2008). In a lung cancer outcome study, expression levels of DICER and DROSHA in 67 non-small-cell lung cancer (NSCLC) samples were examined in the study and reduced expressions of DICER were found to correlate with shortened postoperative survival (Karube et al., 2005).

Analyzing the miRNA expression in 104 pairs of primary lung cancers and corresponding noncancerous tissues, Yanaihara et al. (2006) found that among the miRNA signatures in a unique set of 43 miRNAs, high miR-155 and low let-7a-2 expression independently correlated with poor survival for adenocarcinoma patients. In a comparable study, Takamizawa et al. found that let-7 miRNA expression can classify 143 postoperative lung cancer patients into two prognosticator groups. Those with low levels of let-7a were associated with significantly shorter survival after surgery (Takamizawa et al., 2004). Recently, a five-miRNA signature (let-7a, miR-221, miR-137, miR-372 and miR-182*) has been identified to predict survival and cancer relapse in NSCLC patients after surgical resection (Yu et al., 2008).

Single nucleotide polymorphisms (SNPs) within the miRNA coding genes or within miRNA target genes are likely to be deleterious and can affect an individual’s risk to develop diseases such as cancers. In one of the first studies to illustrate this, He et al. (2005a) observed that in 5 of 10 thyroid cancer cases analyzed where the upregulation of miR-221, -222 and -146 was highest, there was a corresponding loss of KIT transcript and Kit protein in addition to germ-line single nucleotide changes in the two recognition sequences in KIT for these miRNAs in its 3′ UTR. Owing to the high familial incidence, these single nucleotide polymorphic changes are hypothesized to predispose this group of carriers to thyroid cancer. In another example, when comparing the dbSNP database against human cancer specimens, Yu et al. (2007b) found that 12 miRNA-binding SNPs show an aberrant allele frequency in human cancers.

Recently, our lab identified an SNP in let-7 complementary site 6 (LCS6) in the KRAS 3′ UTR that is associated with smoking-induced lung cancer risk. This variant allele is found in 20% of the 74 NSCLC patients in the study. We also found that o6% of DNA from 2433 healthy individuals tested had this allele. In a case–control lung cancer study, the presence of the aberrant allele predicts for an increased risk of NSCLC with an odds ratio of 2.3 in patients with a <41 pack-year smoking history. Luciferase reporter assay suggested that this variant in LCS6 leads to decreased let-7 binding to the KRAS 3′ UTR and thus to increased KRAS expression (Chin et al., 2008). In this study, we have identified the first binding-site SNP that alone can predict a significant increase in NSCLC risk in people with a moderate smoking history. In a follow-up study, Christensen et al. (2009) determined that in head and neck squamous cell carcinoma, the KRAS-LCS6 SNP is not associated with overall disease risk, but cases with the NSP variant have significantly reduced survival.

miRNAs as therapeutics

The revealing role of miRNAs functioning as potential oncogenes and tumor suppressors in tumorigenesis has generated great interest in using them as targets for cancer therapies. General therapeutic strategies involving antisense-mediated inhibition of oncogenic miRNAs and miRNA replacements with miRNA mimetics or viral vector-encoded miRNAs will be discussed. Synthetic anti-miRNA oligonucleotides (AMOs) with 2′-O-methyl modification have been shown to be effective inhibitors of endogenous miRNAs in cell culture and xenograft mice models. Application of 2′-O-methyl AMOs targeting the onco-miR-21 potently inhibited glioblastoma and breast cancer cell growth in vitro and tumor growth in an MCF-7 breast cancer xenograft mice model (Chan et al., 2005; Si et al., 2007).

As one of the first studies illustrating the utility of AMOs in vivo, modified AMOs conjugated with cholesterol termed ‘antagomirs’ were systematically delivered through intravenous injection to target the liver-specific miR-122 (Krutzfeldt et al., 2005). Impressively, a single injection of 240 mg/kg body weight conferred specific miR-122 silencing for up to 23 days. Moreover, as miR-122 was predicted to affect cholesterol biosynthesis (Krutzfeldt et al., 2005; Esau et al., 2006), the serum cholesterol level was seen to decrease by 44% in antagomir-treated mice. As an alternative to 2′-hydroxyl-modified AMOs, lock nucleic acid (LNA)-based oligonucleotides (LNA-antimiR) have been shown to be more stable and less toxic in inhibiting endogenous miRNAs in vivo (Vester and Wengel, 2004; Elmen et al., 2008). In this study, systemic administration by intravenous injections at a dose of 3 or 10 mg/kg LNA-antimiR to African green monkeys with three intravenous infusions over 5 days resulted in the depletion of mature miR-122 and a dose-dependent and long-lasting lowering of plasma cholesterol for up to 3 months (Elmen et al., 2008).

In an alternative strategy to single anti-miRs, Ebert et al. (2007) recently developed miRNA inhibitors called ‘miRNA sponges’, which can be transiently expressed in cultured cells. These molecules are transcripts expressed from strong promoters, containing multiple tandem binding sites to specific miRNAs, and competitively inhibit them. Similarly, Gentner et al. (2009) recently illustrated that the stable knockdown of miRNA in vivo can be achieved through the construction of lentiviral vectors carrying multiple complementary binding sites for the targeted miRNA acting as an anti-miRNA decoy. In this study, the authors retransplanted hematopoietic stem cells carrying lentiviruses expressing the anti-miR-223 decoy into lethally irradiated mice, which resulted in the functional knockdown of miR-223 that phenocopied important aspects of the mouse miR-223 knockout (Johnnidis et al., 2008). One advantage of these systems is that one could construct sponges and anti-miRNA decoys with a combination of miRNA binding sites to potentially inhibit an entire miRNA family or an miRNA cluster.

With anti-miRNA strategies showing great therapeutic promise, the reverse approach of miRNA replacement may be equally attractive for anticancer therapy. The reasoning follows the observation that miRNA expression profiling studies found that although some specific miRNAs are upregulated in cancer, most miRNAs have reduced expression in tumor tissues compared with normal tissues (Lu et al., 2005; Gaur et al., 2007). Kumar et al. (2007) also found that the inhibition of miRNA biogenesis through DICER knockdown promoted cellular transformation and tumor development in both in vitro and in vivo lung cancer models. Furthermore, Chang et al. (2008) reported that the widespread repression of miRNAs occurred through the hyperactivation of c-Myc, a common event in cancer progression. These studies suggest that the reintroduction of specific miRNAs underexpressed in cancer cells could have a therapeutic benefit in reversing tumorigenesis.

To examine the therapeutic potential of let-7 miRNA replacement in NSCLC, our lab previously examined two murine models of human lung cancer, a xenograft model using human lung cancer cells and a well-characterized autochthonous NSCLC mouse model, LSL-Kras-G12D. In this mouse model, tumorigenesis is initiated by the activation of a gain-of-function Kras G12D gene through the inhalation of adenovirus-expressing Cre recombinase (Jackson et al., 2001). Using these models, our group showed that let-7 suppresses lung tumor initiation, including a 66% reduction of the tumor burden when comparing Kras G12D mice treated with adenovirus expressing let-7 miRNA with those that were treated with a control miRNA, respectively (Esquela-Kerscher et al., 2008). Similar findings were made by Jack’s lab (Kumar et al., 2008). Since in both studies let-7 miRNA was delivered at the same time as tumorigenesis was initiated, these two studies showed that let-7 can be used as a preventive therapy against lung cancer in the LSL-Kras G12D mice. In a similar experiment using let-7 treatment, Yu et al. (2007a) showed that in a xenograft model of breast cancer, tumor-initiating cells, SK-3rd cells infected with lentivirus expressing let-7, developed significantly fewer tumors than cells given the control virus.

In a recent report, Bonci et al. showed a link between the deletion of the mir-15a and mir-16 tumor suppressors and prostate cancer. The group analyzed miR-15a and miR-16 in the primary cells of 20 patients with stage 2 and 3 prostate cancers and found consistent downregulation of both miRNAs in about 80% of the tumor samples (Bonci et al., 2008). The investigators showed that the reconstitution of both miRNAs in LNCaP prostate cancer cells and primary tumor cells resulted in growth arrest and apoptosis. Restoration of miR-15a and miR-16 expression in vivo through the intratumoral injection of lentivirus expressing both miRNAs to LNCaP-derived tumor xenograft showed that the tumors underwent growth arrest within 1 week and considerable volume reduction thereafter, whereas the empty vector-treated tumors continued to progress.

In a very recent work, Mendell and colleagues showed that the systemic delivery of a single miRNA can cause tumors from a mice model of liver cancer to regress (Kota et al., 2009). In this study, Kota et al. delivered adeno-associated virus 8 (AAV8)-expressing miR-26a intravenously through the tail veins of MYC-induced mice harboring preformed liver tumors. The authors showed that after 3 weeks after AAV8 infection, they observed a significant regression of tumors in mice receiving the miR-26a treatment compared with mice given the control treatment. The result from this study is exciting, as the ability to treat existing tumors closely reflects clinical scenarios. Furthermore, it illustrated that a single miRNA is powerful enough to have a dramatic suppression of tumor progression.

The use of miRNA therapy to complement traditional anticancer treatments appears to have great potential as it has been reported by several groups that miRNAs can enhance the response and suppress resistance to anticancer cytotoxic therapies. Meng et al. (2006) reported the use of anti-miR-21 and anti-miR-200b AMOs to increase the susceptibility of colangiocarcinoma cells to the chemotherapy drug gemcitabine. MiR-200c levels were reported to be high in well-differentiated endometrial, breast and ovarian cancer cell lines, but extremely low in poorly differentiated cancer cells and restoration of miR-200c in these cells increases their sensitivity to microtubule-targeting agents by 85% (Cochrane et al., 2009).

Work from our own lab showed that let-7 family miRNAs can radiosensitize A549 lung cancer cells and that a C. elegans in vivo model of radiation induced cell death, whereas decreasing the let-7 level causes radio-resistance (Weidhaas et al., 2007). We have also shown recently that the exogenous addition of miR-34 protects breast cancer cells from radiation-induced cell death and that lowering the level of miR-34a with anti-miR radiosensitized the cells (Kato et al., 2009). Significantly, works by Meng et al. (2006) and Weidhaas et al. (2007) also showed that chemotherapy and radiation treatment alter miRNA expression, perhaps as part of the cellular damage repair pathway. As the development of chemotherapy- and radiotherapy-resistant cancer cells accounts for the majority of treatment failure and recurrence, it would be interesting to profile the miRNA differences between the resistant and sensitive populations to identify miRNAs that are involved in the resistant mechanism or pathways, which can then be used to resensitize tumors to available treatments.

Conclusions

The roles of miRNAs in cancer have been very well established over the last few years. Although there is still much to be learned concerning the mechanism of miRNAs in tumorigenesis, scientists have been able to apply their knowledge to use miRNAs for cancer diagnosis and prognosis and identification of cancer risks. Within the last few years, many studies involving either anti-miR knockdown or miRNA replacement therapy have moved into animal models with highly encouraging results for cancer therapeutics.

Footnotes

References