The base-pair interaction between miRNAs and their target mRNAs is essential for the function of miRNAs. Therefore, to silence miRNAs, nucleic acids that were antisense to the target miRNA have been developed. These antisense oligodeoxyribonucleotides (AMOs) specifically and stoichiometrically bind to their targets and silence them efficiently and irreversibly [50]. The AMO approach has recently been used successfully for miRNA therapy of human diseases [65,66]. By modifying this approach and by engineering various antisense miRNAs into a one single unit, Lu et al. showed that multiple target miRNAs can be silenced simultaneously in cell culture studies with the breast cancer cell line MCF-7 [50]. Despite this success, dose-response actions and in vivo stability of AMOs is largely unknown. Further studies are needed to further evaluate their potential in cancer therapy.

One of the first modifications to be carried out to enhance the stability of small RNAs was the substitution of 2'-OH in the ribose ring with 2'-O-methyl or 2'-fluorin. In in vivo experiments with rodents, antisense miRNA inhibitors carrying this substitution displayed enhanced stability and had a higher binding affinity to the miRNA target [67,68]. The use of phosphorothioate bond, that replaces the oxygen atom in the phosphodiester linkage with a sulfur atom, also improved stability of these small RNA to nuclease degradation [69]. AntagomirRs against miR-17-5p and miR-221/miR-222 containing the 2-O-methyl modification and terminal phosphothioates, together with a 3'-conjugated cholesterol group, significantly reduced tumor growth when administered subcutaneously in mouse models of neuroblastoma and prostate cancer, respectively [52,53]. This approach has since faced setbacks because of the requirement for high doses of antagomiRs [70].

The most commonly used miRNA inhibitors in cancer studies are the locked nucleic acid (LNA) antisense oligonucleotides [71]. LNAs are conformational RNA analogs that bind complementary RNA with extremely high affinity making them ideal candidates for cancer detection and analysis of cancer diagnostics. LNA-antimiRs have locked ribose rings that connect 2'-O and 4'-C atoms through methylene bridges, thus conferring unprecedented specificity, due to a significant increase in the RNA:RNA melting temperature [71]. The advantage of LNA-antimiRs is that they can be used in low doses, as they are relatively more resistant to nuclease digestion. Successful use of the LNA-based approach was also shown by high affinity inhibition of miR-91a and miR-21 in animal models of breast cancer [72,48].

Double-stranded RNA molecules containing the same sequence as the depleted, naturally occurring miRNA are commonly used as “miRNA mimics” for replacement therapy. They are expected to target the same set of mRNAs that is also regulated by the natural miRNAs [46]. To protect these dsRNAs from nuclease digestion and to avert immune responses, they were modified with the addition of inverted bases and alkyl or biotin groups [70,73]. While designing these dsRNAs, it is important to preserve target specificity and to ensure miRISC loading [69]. In a mouse model of human colorectal cancer, miR-143 duplex with the modified passenger strand and 3'-overhangs displayed greater activity and stability to nucleases in a mouse xenograft model of human colorectal cancer [54]. This chemically modified synthetic miR-143 also showed a significant tumor suppressive effect than the endogenous miR-143 [54].

More recently, a novel class of small molecule inhibitors of miRNAs have been developed [74]. By screening over 1,000 compounds at the 10 μM concentration, Gumireddy et al. identified two diazobenzene compounds that can inhibit miR-21 in three tumor cell lines [74]. It would be interesting to see if any of these could inhibit miR-21 in vivo.

Liposomes are commonly used to deliver DNA and RNA into the cells as they have higher half-lives in systemic circulation than naked oligonucleotides. Moreover, they protect oligonucleotides from degradation and nuclease digestion. In one study, neutral liposomes carrying synthetic miR-34a and let-7 mimics were able to reach the lung tissues when administered systemically in mice and inhibit tumor growth [60,63]. Dosage issues and in vivo stabilities of liposomes have resulted in the search for suitable alternatives. Polyethanolimine (PEI)-based polymers were developed as an alternative to liposomes. They are biodegradable and are capable of delivering miRNA mimics to the target tissues [68]. When administered in mouse models of colon cancer [61], glioblastoma [55] and adenocarcinoma of the lung [56], PEI-polymers carrying miR-34 and miR-145 mimics strongly inhibited tumor formation.

Nanoparticles are popular delivery vehicles for human studies both in vivo and in vitro. Even though the nanoparticles made with type I collagen were frequently used to deliver siRNAs into cells, N- and C-linked telopeptides of collagen were found to elicit immune responses, thereby hindering the delivery of RNAs into cells [68]. To overcome these adverse effects, nanoparticles carrying miRNA inhibitors were made with atelocollagen, a type I biocompatible collagen [75]. These atelocollagen nanoparticles carrying miR-34a and LNA-miR-135b were shown to effectively inhibit colon cancer and lymphoma in mice, respectively [62,49]. More recently, nanoparticles carrying miR-143b were shown to inhibit tumors induced by human osteosarcoma cells [76] and orthotopic inoculation of nanoparticles with miR-516a-3p inhibited peritoneal dissemination of gastric cancer in mice [77]. These findings demonstrate that atelocollagen-mediated delivery of miRNAs would be an effective strategy for cancer treatment.

Another approach for the delivery of miRNA inhibitors is the use RNA ligands, termed as “RNA aptamers”. They target specific cell surface receptors to facilitate the entry of the miRNA modulators that are coupled to them. One such chimera Chi-29b composed of a mucin 1 (MUC1) aptamer (that targets tumor cell surface MUC1 protein) coupled to miR-29b was recently shown to inhibit the growth of epithelial ovarian carcinoma cells in vitro by inducing apoptosis in a dose-dependent manner [57]. In a similar study, MUC1 aptamer coupled to let-7i miRNA inhibited the proliferation of these cells and sensitized them to paclitaxel treatment [78]. These findings suggest that RNA aptamers are useful vehicles for the delivery of miRNAs into tumor cells. However, their therapeutic utility in vivo is yet to be realized.

Fluorescent DNA probes termed “molecular beacons” are being used to detect mature miRNAs in real-time imaging [79]. Kim et al. developed a cancer-targeting theranostics probe in a single system using the AS1411 aptamer and miRNA-221 molecular beacon (miR-221 MB)-conjugated magnetic fluorescence (MF) nanoparticle (MFAS miR-221 MB) to simultaneously target to cancer tissue, image intracellularly expressed miRNA-221 and to inhibit miR-221 in various brain cancer cell lines [58]. This novel method was sensitive enough to detect the miR-221 biogenesis in real-time and to produce antitumor therapeutic effects by inhibiting the miRNA function. Additional studies are needed to extend this simultaneous miRNA detection and inhibition approach to other cancers.

A more recent strategy to target cell surface receptors for miRNA delivery is based on antibodies. In this approach, single-chain variable fragment (scFv) antibodies with variable regions of immunoglobulin heavy and light chains were coupled to a short linker peptide of 10–25 amino acids and encapsulated into nanoparticles [68,80]. It has been established in melanoma studies that systemic delivery of scFv antibodies coupled to miR-34a mimics were able to reduce tumors without causing any adverse side-effects [59], thus demonstrating their potential use in cancer treatment. Synthetic RNA-based regulatory systems that integrate sensing and gene-regulatory functions, where the former are encoded in RNA aptamer sequences that recognize small molecule ligands, have also been developed [57]. Such integrated ligand-responsive RNA-based control systems offer several advantages over more traditional protein-based regulatory systems in avoiding the potential immunogenicity of heterologous protein components and providing a more tunable, compact control system [57].

Viral vectors commonly used in human gene therapy are modified, noninfectious adenoviruses, adeno-associated viruses (AAV), herpes simplex viruses (HSV) and lentiviruses. Each of these can encode miRNA mimics and express them in infected cells. Since viral vectors can be engineered to alter their tropism for transduction into any cell type, they are the most preferred vectors for gene therapy. Recombinant adenoviral (Ad) and AAV vectors were commonly used in such studies, because they do not integrate into the host genome, but replicate as episomes in the nucleus and stably express miRNAs. Both Ad and AAV vectors are ideal for transient expression of transgenes in tumor cells or in cells that divide slowly [81]. Moreover, these vectors are of greater relevance, because random integration of lentiviral vectors may result in the activation of oncogenes and induce tumorigenesis. It was shown recently that AAV vectors transduce hepatocytes efficiently in vivo when injected through the intra-portal or intra-hepatic artery of healthy and cirrhotic rat livers, suggesting that they are convenient for transgene expression in the liver [82]. A major drawback of viral vectors, in general, is that they elicit strong immune responses in transduced cells, which reduces gene transfer and limits re-administration. Therefore, chemical or physical modification of Ad vectors with polymers has been used as a strategy for cancer therapy. Lately, to endow targeting moiety to polymer-coated Ad vectors, a diversity of ligands, such as tumor-homing peptides, growth factors or antibodies, have also been introduced to avoid unwanted transduction and to enhance therapeutic efficacy [83]. In addition, efforts are being made to encapsulate Ad vectors into poly (lactide-co-glycolide) (PLG) microspheres prior to administration [84]. When compared to free Ad vectors, PLG-encapsulated Ad vectors are known to have a significantly higher transduction efficiency rate in both E1 complementing and non-complementing cells [84]. When Ad was enveloped by cationic lipids, significantly high levels of viral uptake was observed in cultured HepG2 cells [85]. When administered in mice, these artificial envelopes can significantly alter the interactions with blood components and divert viral particles from their natural liver tropism, resulting in reduced hepatotoxicity [85]. Another added advantage of encapsulated viral vectors is that their level of transduction can be controlled by varying the quantity of encapsulated viral particles in the microspheres and the amount of microspheres administered.

AAV vectors, on the other hand, are single stranded adenoviruses devoid of critical genes involved in viral infection, making them relatively safe to administer in gene therapy trials. However, because of the difficulties in the construction of AAV vectors, self-complementary AAVs (scAAVs) that contain inverted repeats of its genome were developed. scAAVs can fold into dsDNA without the need for additional DNA synthesis or base-pairing between multiple vector sequences [68]. Successful therapeutic delivery of miR-26a-expressing AAV vector has recently been reported in a murine liver cancer model [64].

Efforts to knock-down miRNAs by using the classical RNA interference (RNAi) techniques were tested for inhibiting miRNAs. However, these approaches have largely been unsuccessful. Therefore, to induce the knockdown of miRNAs, genetic “sponges” were developed. These “miRNA sponges” are competitive inhibitors expressed from vectors containing multiple, tandem binding sites to a miRNA of interest [86]. When transiently transfected into cultured cells, miRNAs expressed from a strong promoter on these sponges will compete with endogenous miRNAs and sequester mature miRNAs away from their natural targets. This approach has been successfully used recently to knockdown miR-223 using a lentiviral vectors [87].

All these diverse methods of miRNA delivery and encapsulation would allow us to perform both loss-of-function and gain-of-function assays and to modulate the expression of miRNAs in vivo, thereby facilitating the application of miRNAs for cancer therapy.

OncomiRs promote tumor growth and proliferation by blocking the activities of cellular tumor suppressors, cell cycle regulators and pro-apoptotic genes. For instance, miR-224 increases tumor cell progression by inhibiting the expression of apoptosis inhibitor-5 [18,44], and miR-221 promotes cell proliferation by controlling the cell cycle inhibitors (CDKI) CDKN1C/p57 and CDKN1B/p27 [42]. miR-519d, on the other hand, contributes to hepatocarcinogenesis by targeting CDKNA/p21, PTEN, AKT3 and TIMP2 [88]. Intratumoral administration of miR-143 showed that high levels of miR-143 can significantly promote HCC metastasis in an athymic nude mouse model by repressing the expression of fibronectin type III domain containing 3B (FNDC3B), which regulates cell motility [89]. miR-222, a frequently overexpressed miRNA in HCC, also regulates cell motility by enhancing Akt signaling [90], whereas miR-423 directly binds to 3'-UTR of p21Cip1/Waf1 to suppress its expression and to promote cell cycle progression and tumorigenesis [91].

While miR-146a increases the resistance of HCC cells from the cytotoxic effects of interferon-α by inhibiting the expression of Smad4 [92], upregulation of hepatic transforming growth factor (TGF)-β and its downstream mediators, Smads 2, 3 and 4, was found to be correlated with an increased expression of miR-181 in the livers of mice fed with a choline-deficient L-amino acid-defined (CDAA) diet [93]. Depletion of miR-181b inhibited tumor growth in nude mice, whereas its expressed enhanced resistance of HCC cells to the anticancer drug, doxorubicin [93]. miR-17-5p upregulates the migration and proliferation of HCC cells by activating the p38 mitogen-activated protein kinase MAPK pathway and increasing the phosphorylation of heat shock protein 27 [94], suggesting its potential application in HCC therapy. The miR-34 family members are direct transcriptional targets of tumor suppressor, p53, and loss of miR-34 function can impair p53-mediated cell cycle arrest and apoptosis [19].

miR-210, which is often upregulated in HCC, promotes hypoxia-mediated tumor cell metastasis by targeting the vacuole membrane protein 1 [95]. Therefore, inhibition of miR-210 expression could lead to reduction in HCC metastasis and should facilitate the development of novel therapeutic strategy against hypoxic tumor cells. Similarly, miR-30d, a miRNA associated with intrahepatic metastasis of HCC, promotes tumor cell migration and invasion in vitro and intrahepatic and distal pulmonary metastasis in vivo by targeting the Galphai2 (GNAI2) protein [96]. The expression of miR-148a is elevated in HepG2 and Hep3B cells, where it promotes cell proliferation, cell cycle progression and cell migration. When anti-miR-148a was overexpressed from a lentivirus, it inhibited the activity of miR-148a, in addition to blocking the Akt signaling pathway, in these cells, suggesting that it could serve as an early diagnostic marker and/or therapeutic target [97]. Additionally, miR-1 and miR-499 inhibited invasion and migration of HepG2 cells by targeting the ets1 proto-oncogene, which causes degradation of extracellular matrix [98]. Another antagomiR, miR-219-5p, induced tumor suppressive effects in HCC cell lines by targeting glypican-3 and by causing cell cycle arrest at the G1 to S transition [99].

While successful strategies to inhibit oncomiRs for HCC therapy have not been reported so far, each of these miRNAs represent a strong candidate for therapeutic intervention towards achieving a gain of function with the use of miRNA antagonists, RNAi and small molecule inhibitors. A recent study has showed the effects of a modified AMO against miR-221 in reducing the number and size of HCC tumors in a miR-221 transgenic mouse model [51]. While this study is promising, further in-depth studies on candidate miRNAs will allow us to ascertain their potential as therapeutic targets for HCC treatment.

The first successful demonstration of miRNA replacement to restore the expression levels of a downregulated miRNA by delivery to the tumor was reported using a miR-26a-encoding AAV vector in a mouse model of HCC [64]. miR-26a is normally downregulated in HCC, and its overexpression in mouse livers results in the inhibition of cancer cell proliferation and induction of tumor-specific apoptosis [64]. More recently, another miRNA downregulated in HCC, miR-34a, has been successfully tested using this approach and by delivering it with NOV340 liposomes in an orthotopic model of HCC [100]. In both cases, significant tumor reduction, dramatic protection from disease progression without toxicity and prolonged survival of animals has been reported. While these reports were aimed to restore a loss of function, similar success has not yet been achieved with approaches to inhibit miRNA expression levels. All these studies are built on the regulation of multiple cellular pathways associated with human disease, which now appears to be a requirement for successful cancer therapy [101].

Along these lines, another downregulated miRNA in HCC, miR-375, was overexpressed in liver cancer cells. It decreased cell proliferation, clonogenicity, migration/invasion, induced G1 arrest and apoptosis [102] in vitro, indicating that it could be a good candidate for in vivo studies. Indeed, when cholesterol-conjugated 2'-O-methyl-modified miR-375 mimics (Chol-miR-375) were administered in nude mice, they were able to significantly suppress the growth of hepatoma xenografts [103].

In other promising studies, overexpression of miR-376a suppressed cell proliferation in HuH7 cells [101] and miR-138 induced cell cycle arrest by targeting cyclin D3 [103]. In an in vitro experiment with HCC cell lines transfected with miR-637 mimics or transduced with Lv-miR637 vectors, miR-637 suppressed tumor growth by negatively regulating the phosphorylation of STAT3 protein [104].

Downregulation of miR-29b correlates with rapid recurrence and poor survival of individuals with HCC. miR-29b dramatically suppressed the ability of HCC cells to promote capillary tube formation of endothelial cells and to invade extracellular matrix gel in vitro and in mouse xenografts where it regulated the expression of matrix metalloproteinase 2 [105]. Restoration of miR-29b expression resulted in the deregulation contributes to angiogenesis, invasion and metastasis of HCC [105], suggesting it can be used as a novel therapeutic target in anti-HCC therapy.

Intratumoral injection of cholesterol-conjugated miR-99a mimics was recently demonstrated to significantly inhibit tumor growth and to reduce α-fetoprotein levels in HCC-bearing nude mice. In this study, miR-99a expression inversely correlated with protein levels of insulin-like growth factor 1 receptor (IGF-1R) and mammalian target of rapamycin (mTOR), inducing cell cycle arrest at the G1 phase [106]. This finding suggested a potential tumor suppressor role for miR-99a. Another miRNA often downregulated in HCC is miR-199a-3p. By transfecting pre-miR-199a-3p and anti-miR-199a-3p oligonucleotides, it was shown to target mammalian target of rapamycin (mTOR) and c-Met in HCC cells [107]. Restoring attenuated levels of miR-199a-3p in these cells led to cell cycle arrest at the G1 phase, reduced invasive capability, enhanced susceptibility to hypoxia and increased sensitivity to doxorubicin-induced apoptosis, suggesting that enhancement of miR-199a-3p levels may have a therapeutic benefits in HCC [107].

miR-122 binds to the 3'-UTR of Bcl-2 family member, Bcl-w, an anti-apoptotic protein, to induce cell death in HepG2 and Hep3B cells [108]. When expressed from an Ad vector, miR-122 sensitized HCC cells to adriamycin and vincristine treatment by causing cell cycle arrest at the G2/M phase and by modulating the expression of multidrug resistant protein, MDR-1 [109]. Using several 3' cholesterol-conjugated, 2'-O-Me oligonucleotides and an unconjugated 2'-O-methoxyethyl-phosphorothioate-modified oligonucleotide, miR-122 was inhibited in mouse livers to demonstrate its function in the lipid metabolism [110,111]. Elmen et al. also subsequently reported the inhibition of miR-122 in the mouse liver after the systemic administration of a 16-mer LNA-modified antimiR oligonucleotide [112]. These studies provide strong evidence that miR-122-based HCV therapy can be developed with antagomiRs. To pursue this lead, Lanford et al. administered a LNA-modified phosphorothioate oligonucleotide (SPC3649) complementary to the 5'-end of miR-122 in chimpanzees with chronic HCV infection and showed that SPC3649 suppression of miR-122 had long-lasting suppression of HCV viraemia with no evidence for viral resistance or side effects in the treated animals [113]. In addition to demonstrating the feasibility and safety of prolonged administration of a LNA oligonucleotide drug in vivo, this study showed that miR-122 is essential for HCV accumulation and that the miR-122 seed sites were conserved in all HCV genotypes and subtypes, suggesting the genotype independent nature of this therapy. More recently, Janssen et al., further extended this approach and reported successful therapy in HCV patients with an oligonucleotide drug, MiraVirsen, that targets miR-122 [114]. This was the first miRNA-based therapeutic drug developed to treat a liver disease and is currently being tested by Santaris Pharma (Hørsholm, Denmark) in phase 2 clinical trials [28].

These are some of the promising candidates for miRNA replacement therapy for HCC. It is important to exercise caution while designing clinical trials with miRNA mimics for targeting multiple genes relevant to human disease, because of the concerns about potential toxicity in normal tissues, especially under conditions where the therapeutic delivery of miRNA mimics will also lead to an accumulation of exogenous miRNA in normal cells. These toxic effects might be the result of overloading RISC with the exogenous miRNA, thereby competing with endogenous miRNAs necessary for normal cellular welfare and/or hyperactivating cellular pathways that will also reduce the viability of normal cells [46].

An important aspect to consider for miRNA-based therapy is the assessment of off-target effects of miRNA modulators. When using these oligonucleotides, there is a potential risk of affecting cellular RNAs other than the target miRNA [115]. Therefore, proper understanding on the effects of unwanted interactions between antimiR molecules and endogenous RNAs and appropriate designing of antimiRs to minimize the off-target effects is critical. As discussed earlier, the incorporation of chemical modifications, such as LNAs, into antimiRs have been shown to improve mismatch discrimination between a given antimiR and its target miRNA. However, these exogenous miRNAs compete with endogenous ones for miRISC, causing the saturation of miRISC complexes, resulting in the loss of endogenous miRNA regulation and fatality, due to liver toxicity, as demonstrated in animal studies [116,117].