Strategies to Modulate MicroRNA Functions for the Treatment of Cancer or Organ Injury

A. Biogenesis and Working Mechanism of MicroRNA

Approximately 70% of miRNA coding genes are associated with protein-coding host genes and are cotranscribed. Only 30% of miRNAs are transcribed from their own open reading frames located in intergenic areas (Bartel, 2009, 2018; Ha and Kim, 2014). In either case, during the expression process, miRNA coding sequences are first transcribed by RNA polymerase II into a long primary transcript of up to 10 kb, called primary-miRNA (pri-miRNA) (Lee et al., 2004) (Fig. 1). The pri-miRNA transcript is then processed by a nuclear RNase III called Drosha into a shorter length of transcript (about 70 bases), forming a hairpin-like structure, called precursor-miRNA (pre-miR) (Lee et al., 2003). The processed pre-miR is exported to the cytoplasm through nuclear membrane channel protein Exportin-5 (XPO5) (Yi et al., 2003). A recent study showed that XPO5 activity can be regulated by pre-miR phosphorylation. The phosphorylation by extracellular signal-regulated kinase suppresses the pre-miR export through XPO5 in cancers (Sun et al., 2016). Once translocated into the cytoplasm, the terminal hairpin loop of the pre-miR is further cleaved by Dicer, a cytoplasmic RNase III, to produce a 22-nucleotide–long mature miRNA (Hutvágner et al., 2001). Interestingly, it was suggested that intact RNAi machinery is critical to maintaining stem cell populations during early development. Studies in which Dicer was lost had a lethal effect, showing stem cell depletion in embryos during the early developmental stages of mice (Bernstein et al., 2003).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.

Biogenesis of miRNAs is a multistep process (Lee et al., 2003, 2004; Calin and Croce, 2006; Ha and Kim, 2014; Bartel, 2018). miRNA host genes are located in intragenic or intergenic regions and are primarily transcribed into a long, capped (Bartel, 2009), and polyadenylated transcript (pri-miRNAs) by RNA polymerase II, which is often longer than several kilobases (Lee et al., 2004). The pri-miRNAs are first processed in the nucleus into a shorter hairpin-structured transcript (pre-miRNAs) by the nuclear enzyme Drosha (Lee et al., 2003). The hairpin-loop pre-miRNAs are exported into the cytoplasm through a nuclear member channel protein, Exportin-5 (Yi et al., 2003). In the cytoplasm, pre-miRNAs are further processed by Dicer to a hairpin-free duplex form of miRNA called mature miRNAs (Hutvágner et al., 2001; Ketting et al., 2001). The duplex miRNAs bind with Argonaute 2 (Ago2) and transactivation-responsive RNA-binding protein (TRBP) to form RISC, and then a functional strand of the duplex miRNA remains in RISC until it binds to its target mRNA while the unfunctional strand is degraded. The eight-base-long seed sequences on the mature miRNA recognize and bind to their partial complementary sequences on the 3′ UTR of target gene mRNAs. The complex between miRNA and mRNA rapidly represses the translation of mRNA into proteins and eventually leads to mRNA degradation (Bartel, 2009, 2018; Eichhorn et al., 2014). DGCR8, Drosha and DiGeorge syndrome chromosome region; ORF, open reading frame.

The mature miRNA binds to Argonaute 2 and transactivation-responsive RNA-binding protein to form the RNA-induced silencing complex (RISC) (Redfern et al., 2013). Although associated with RISC, inactive strands, mostly 3′ end, are degraded, leaving the other active strand with the RISC until guided to the 3′ untranslated region (3′ UTR) of target mRNAs (Ha and Kim, 2014). siRNA is also processed, in a similar way to miRNAs, to a double-stranded RNA by Dicer and loaded onto RISC to bind to its target mRNA. The binding between the target mRNA and siRNA is fully complementary (Carthew and Sontheimer, 2009). Unlike the working mechanism of siRNA, miRNAs recognize their target mRNAs through partial complementary sequences between the second and the eighth nucleotide, called “seed sequences.” Once the seed sequence region of miRNA forms the partial base pairing with the 3′ UTR of target mRNA, the miRNA/mRNA complexes are unable to associate with active ribosomal complexes to complete protein synthesis (Bartel, 2009, 2018). This partial complementary base pairing between miRNAs and 3′ UTR of target mRNAs can vary depending on the target mRNA. Therefore, in theory, a single miRNA can bind to hundreds of different target mRNAs. There are several computer-based prediction algorithms to predict miRNA targeting. However, because of the complex nature of partial base pairing, it is still unsatisfactory to predict potent miRNA binding sites with reliable accuracy. In addition to the translational repression role of miRNAs (Olsen and Ambros, 1999), it has also been found that miRNAs can cause target mRNA degradation through the deadenylation of the target mRNA poly-A tail (Bagga et al., 2005). David Bartel’s laboratory proposed from their global steady-state measurements that the irreversible process of mRNA destabilization by miRNAs often overwhelms miRNA-based translation repression in a rapid, albeit relatively weak, manner (Eichhorn et al., 2014). Through these two major mechanisms of miRNAs, aberrantly regulated miRNAs can reprogram a significant amount of intracellular signal pathways over a threshold limit to cause disease development including human cancers (Bartel, 2018).

B. Regulation of MicroRNA Expression

Since mature miRNAs regulate the expression of multiple target genes, dysregulation of miRNA expression process can lead to abnormal gene expression profiles in cells, which in turn can provide a favorable environment for the development of organ injuries or even cancers (Ferrari et al., 2016; Neudecker et al., 2016, 2017c). Such deregulations of miRNA expression can take place at different stages of miRNA biogenesis. Transcription factors that regulate protein-encoding genes also regulate miRs, and some of these transcription factors are over- or underexpressed in diseases to modulate miR expression. For example, hypoxia-inducible factor (HIF) is stabilized during hypoxia and inflammation (Haeberle et al., 2008; Eltzschig and Carmeliet, 2011) and has been shown to be involved in the dysregulated expression of miRNAs in certain disease conditions (Eltzschig and Eckle, 2011; Choudhry et al., 2016). The center core region of tumor tissues becomes hypoxic because of the limited supply of oxygen via poor vasculatures and thus results in HIF-1α stabilization. In patients with breast cancer, miR-210 was induced by HIF-1α and showed an inverse correlation with disease-free and overall survival (Camps et al., 2008). Another hypoxia or ischemia-triggered transcription factor is specificity protein 1 (Sp1), which is stabilized during conditions of perioperative organ injury (Eltzschig et al., 2009; Hart et al., 2010). In cancers, Sp1 triggers a transcriptional activation of miRNAs, such as miR-205 in esophageal squamous cell carcinoma, to result in tumor progression and drug resistance (Pan et al., 2017). Besides, miRNAs also regulate Sp1 and other transcription factors and thereby serve as tumor suppressor–like molecules (Fulciniti et al., 2016; Hedrick et al., 2016; Li et al., 2018; Lv and Wang, 2018). Several studies demonstrated the cross talk between pro-oncogenic microRNAs and oncogene cellular myelocytomatosis (c-Myc) (Zheng et al., 2014; Chen et al., 2017; Yarushkin et al., 2017). The regulation of microRNAs by transcription factors will confer alternative strategies to target microRNAs as potential therapeutic interventions. Transcriptional inactivation of the miRNA-encoding gene through promoter hypermethylation has also been observed in many human diseases, including cancers. For example, the miR-127 expression can be silenced by promoter hypermethylation in bladder cancer cell lines and patients, and hypomethylating agents restored its expression level (Saito et al., 2006). miRNAs themselves can also modulate DNA methylation in cancers by interfering with the DNA methylation machinery (Holubekova et al., 2017). Single polymorphism (SNP) is another mechanism to produce variations on miRNA expressions and functions. Polymorphic genetic mutations of miRNA-binding sites in 3′ UTR of target mRNAs were proposed to be causative variants for human cancers (Chen and Rajewsky, 2006; Saunders et al., 2007). When the regions harboring two microRNA-binding sites in the proto-oncogene c-KIT gene from patients with papillary thyroid carcinoma were sequenced, a polymorphism of G>A SNP (rs17084733) was found in the proto-oncogene c-KIT 3′ UTR complementary sequences to the seed sequences of two miRNAs: miR-221/222 (He et al., 2005). In 5 out of 10 cases, this germline mutation increased the association between miR-221/222 and the target gene that was believed to contribute to cancer progression. Furthermore, when over 100 human cancer samples and cancer cell lines were examined by sequence variation comparison, it was found that the SNP (G>A mutation) at a 19th base pair of Lethal-7 (Let-7)e miRNA dramatically reduced the expression level of Let-7e (Wu et al., 2008b). On the other hand, SNP in miRNA sequences can also contribute to tumorigenesis. For instance, it was found that miR-16-1 carries germline mutations at a high frequency in chronic lymphocytic leukemia (CLL) (Calin et al., 2005). Overall, the frequency of SNPs in miRNA-binding sites or SNPs in miRNA sequences was predicted to be less than 1% because of the size of miRNA and its binding site (Saunders et al., 2007). However, successfully pinpointing such SNPs in miRNAs and its targets will provide a fine-tuned personalized miRNA signature for each cancer patient.

C. MicroRNA Detection

In the early days of miRNA research, a Northern blotting method using a specifically designed RNA probe, unlike Southern blotting, which uses DNA probes to hybridize to target DNA sequences, was widely used to detect miRNAs, similar to mRNA detection (Hunt et al., 2015). In Northern blotting, total RNAs were separated on urea-containing denatured PAGE and transferred to a positively charged nylon membrane. Then, a target RNA sequence is hybridized with a probe labeled with either radioactive isotopes (³²P) or chemiluminescent enzymes, such as alkaline phosphatase or horseradish peroxidase. To better detect miRNAs by Northern blotting, the probes are usually generated with locked nucleic acids (LNAs), which give at least 10-fold higher sensitivity than DNA probes (Valoczi et al., 2004). Because of the high specificity and sensitivity of LNA-based probes, fluorescently labeled LNA probes are also often used to visualize specific miRNAs directly from tissue or cell samples by fluorescence in situ hybridization (Urbanek et al., 2015). Although Northern blotting is a reliable RNA detection method in terms of sensitivity and selectivity, it is a time-consuming procedure and involves the potential use of radioactive isotope–labeled probes. Therefore, it was critical to develop new methods that are optimized for high-throughput detection with comparable sensitivity and selectivity. The laboratory of Thomas Schmittgen applied quantitative real-time polymerase chain reaction (qRT-PCR) using fluorescent SyBr-Green dye to detect 23 pre-miRs from six human cancer cell lines (Schmittgen et al., 2004), which has launched the quantitative comparison of miRNA expressions at high-throughput levels. As more miRNAs and their significant roles in human diseases were identified, global screening at high throughput levels was needed. A custom-built microarray platform using oligonucleotide probes against miRNA precursors and mature miRNAs was introduced in 2004 for screening the genome-wide miRNA expression profiling in human cancer specimens (Calin et al., 2004a). This platform has been used to identify several sets of miRNA signatures from various types of human cancer specimens. qRT-PCR has also been widely used to detect individual miRNA signatures from human cancer samples. For example, Megaplex reverse transcription format has been employed to use stem-loop–specific primers for qRT-PCR–based miRNA arrays after preamplified cDNA (Mestdagh et al., 2008), which significantly reduced the amount of input RNA to 10 ng. Unlike a one-way hybridization-based microarray, the qRT-PCR–based array was able to track the changes of miRNA expression on different stages of miRNA maturation. qRT-PCR–based arrays on a set of primary effusion lymphomas found that 68 primary effusion lymphoma–specific signature miRNAs are amplified at three levels of expression: gene alteration, transcription (pre-miRNA), and processing (mature miRNA) (O’Hara et al., 2008). Since the qRT-PCR–based array uses a set of primers for preselected miRNAs, it is relatively more flexible to customize the array for a specific range of pathways or disease conditions (Chugh et al., 2010). These advancements in the detection of miRNAs enabled high-throughput profiling on numerous human specimens obtained from patients, leading to the discovery of signature miRNAs in the disease types as useful biomarkers.

A. MicroRNAs as Tumor Suppressors

The loss of function of certain miRNAs due to genomic deletions, mutations, epigenetic silencing, and/or miRNA processing alterations can cause malignant transformation of normal cells. In this case, those miRNAs are considered to normally play a tumor-suppressive role during cancer development and progression. Two signature miRNAs in CLL, miR-15a and miR-16-1, were discovered from a 30-kb deleted region between exons 2 and 5 of the deleted in lymphocytic leukemia 2 (DLEU2) gene from the 13q14.2 region. Frequent hemizygous or homozygous deletions of this region occur in more than 50% of CLL cases (Calin et al., 2002, 2004a, 2005), and these miRNAs were found to target antiapoptotic protein B-cell lymphoma 2 (BCL-2). As a result of the epigenetic deletion of the miR-15a and miR-16-1, BCL-2 is widely overexpressed in a subset of patients with CLL. In myelodysplasia and therapy-related acute myeloid leukemia (AML), the miR-29 family, miR-29b-1/miR-29a, was found in chromosome 7q32, which is a frequently deleted region in AML (Garzon et al., 2008, 2009). It was also inversely correlated with upregulated oncogenic proteins BCL-2 and myeloid cell leukemia 1 (MCL-1) (Xu et al., 2014). In lung cancers, the miR-29 family is also frequently downregulated and targets DNA methyltransferase (DNMT) 3A and DNMT3B (de novo methyltransferases), known to be highly upregulated in lung cancers with poor prognosis (Fabbri et al., 2007). In sarcoma, it was suggested that miR-29 play as a decoy for human antigen R by preventing the binding of human antigen R to the 3′ UTR of tumor suppressor A20 transcript (Balkhi et al., 2013). MiR-34a has also been recognized as a tumor-suppressive miRNA since it is highly expressed in most normal tissues but downregulated in most types of cancer through methylation on its 5′-C-phosphate-G-3′ island (Lodygin et al., 2008). It is also directly activated by p53 (He et al., 2007). Ectopic overexpression of miR-34a in cancer cells causes cell cycle arrest by targeting tyrosine-protein kinase Met, c-Myc, Cyclin D1, and cyclin-dependent kinase 6 (CDK6) (Sun et al., 2008; Li et al., 2009; Yamamura et al., 2012) or apoptosis by targeting notch receptor 1 and Bcl-2 (Pang et al., 2010; Li et al., 2013a). Interestingly, miR-34a was shown to directly target sirtuin 1, a known inhibitor of p53, leading to activation of p53 as a positive-feedback loop (Yamakuchi et al., 2008). The let-7 family consists of 12 miRNAs and is frequently found to be correlated to the development of lung, breast, urothelial, and cervical cancers. Because of its location at a fragile site and frequent downregulation in cancer, it is thought that the Let-7 family member miRNAs function as tumor suppressors (Calin et al., 2004b). For example, loss of Let-7 expression in lung cancer leads to the upregulation of one of its targets, Rat sarcoma (Ras) (Johnson et al., 2005). Interestingly, many of the tumor suppressor miRNAs (miR-15-a/16-1, miR-29s, and let-7) have multiple genomic location that expresses identical mature miRNAs, which can be the conserved mechanism through evolution to preserve the function of important miRNAs. Global miRNA profiling on mouse embryonic stem cells has revealed stem cell–specific signature miRNAs, which function in the regulation of cell development and the maintenance of pluripotency (Houbaviy et al., 2003). In cancer cell populations, some of the miRNAs seemed to involve cancer stem cell (CSC) formation and maintenance. Expression of tumor-suppressive let-7 was found to be downregulated in breast cancer stem cells, leading to the overexpression of H-RAS and high-mobility group AT-hook 2, which is responsible for stem cell–like self-renewal and reduced differentiation properties (Yu et al., 2007). In breast cancers, miR-200b was found as a cancer-specific signature miRNA and was epigenetically lost during CSC formation (Iliopoulos et al., 2010). The loss of miR-200b released the expression of its direct target gene, suppressor of zeste 12 protein homolog, which epigenetically modulates polycomb-mediated repression of the E-cadherin gene. Since CSCs have been proposed to initiate tumor formation and induce cancer recurrences, miRNA signatures enriched in CSC populations may become attractive therapeutic targets or serve as diagnostic/prognostic markers.

B. MicroRNAs as Oncogenes

In contrast to the tumor-suppressive miRNAs, a certain number of miRNAs are significantly upregulated in human cancers. They are considered to possess oncogenic roles and are often called onco-miRs. For example, miR-155 was identified from a transcript of a noncoding RNA gene, B-cell integration cluster, located in chromosome 21q23 in human B-cell lymphomas (Eis et al., 2005). MiR-155 is highly expressed in many other human cancers, including pediatric Burkitt lymphoma, Hodgkin disease, primary mediastinal non-Hodgkin lymphoma, CLL, AML, lung cancer, and breast cancer (Jurkovicova et al., 2014). Mice with B-cell targeted miR-155 overexpression showed a polyclonal preleukemia pre-B cell proliferation phenotype accompanied by a high level of B cell malignancy (Costinean et al., 2006). A later study in mice identified Src homology 2 domain-containing inositol-5-phosphatase and CCAAT enhancer-binding protein β, both previously implicated in the interleukin-6 (IL-6) signaling pathway, as direct targets of MiR-155 (Costinean et al., 2009). Laboratory of Kalpana Ghoshal also found that nonalcoholic steatohepatitis is associated with a high level of miR-155 expression, leading to the development of hepatocellular carcinoma (HCC) (Wang et al., 2009). Single cistronic cluster miR-17-92 is located within 800 bp of the noncoding gene chromosome 13 in open reading frame 25 at 13q31.3. It is frequently amplified in follicular lymphoma and diffuse large B-cell lymphoma and produces six mature miRNAs, including miR-17, miR-18a, miR-19a, miR-20a, miR-19b-1, and miR-92-1 (Olive et al., 2013). The expression levels of the miR-17-92 cluster are significantly upregulated in hematologic malignancies such as lymphomas, and in various solid tumors such as breast, colon, lung, pancreas, prostate, and stomach tumors (Olive et al., 2013). In a mouse model, miR-17-92 promoted tumor proliferation and angiogenesis while inhibiting tumor cells by modulating E2F transcription factor 1 (E2F1) expression (O’Donnell et al., 2005). Both E2F1 and E2F transcription factor 3 (E2F3) activate the miR-17-92 cluster as a regulatory loop (Sylvestre et al., 2007). Additionally, miR-21 is the most frequently upregulated miRNA in human cancer, including both hematopoietic and solid organs (Pfeffer et al., 2015). It is one of the most extensively studied miRNAs, with several known targets, including phosphatase and tensin homolog (Meng et al., 2007), programmed cell death 4 (Gaur et al., 2011), and BTG antiproliferation factor 2 (Liu et al., 2009). Although miR-21 expression has been found at a high level in most human cancers, transgenic mice overexpressing miR-21 throughout the whole body did not develop tumor formation. However, providing a second hit by crossing the miR-21 transgenic mice to activated k-Ras mutant mice increased tumor incidents in lung cancer by negatively regulated multiple tumor suppressors, including Sprouty homolog 1, Sprouty homolog 2, BTG antiproliferation factor 2, and programmed cell death 4, through activation of the Ras/mitogen-activated protein kinase kinase/extracellular signal-regulated kinase pathway (Hatley et al., 2010). Interestingly, overexpression of miR-21 in the spleen has been shown to lead to lethal B-cell lymphoma in mice, and more importantly, the malignancy was related to the expression level of miR-21 (Medina et al., 2010). This addiction of tumorigenesis to a specific miRNA was also observed in a lymphoma study with miR-155 overexpressing mice (Cheng et al., 2015). These oncogenic miRNAs have been suggested as promising therapeutic targets since their downregulation or knockdown induces apoptosis, cell cycle arrest, and inhibition of invasion and metastasis of cancer cells. Further investigation into onco-miRs may benefit the development of cancer therapeutic strategies, since targeting onco-miRs may be sufficient to inhibit tumor initiation and progression.

C. Circulating MicroRNAs as Biomarkers of Human Cancers

Early detection is one of the key factors in disease diagnosis that leads to a positive outcome. This is especially true in cancer because of the higher chance of metastasis associated with late diagnosis. Global profiling of miRNAs is time- and cost-consuming; thus, it is essential to narrow down to a limited number of “disease signature” microRNAs. In this regard, defining accurate biomarkers and affordable detection methods for different types of cancer is critical for patients’ long-term outcomes. The various types of human cancer have allowed researchers to pinpoint the most critical miRNAs. For early diagnosis, the discovery of miRNA signatures can be useful biomarkers in the determination of disease diagnosis and prognosis. Profiling on circulating RNAs extracted from plasma or serum has been proposed as a diagnostic and prognostic marker since this method is relatively simple, noninvasive, low-cost, and rapid (Tsang and Lo, 2007). The first attempt to use a profile of circulating miRNAs as a diagnostic biomarker was in serum RNA samples from patients with diffuse large B-cell lymphoma (DLBCL). Three signature miRNAs, miR-155, miR-210, and miR-21, were highly elevated in DCBCL compared with normal sera (Lawrie et al., 2008). At that time, the miRNA profiles seemed to match those in their primary tumor cells. In colorectal cancer (CRC), onco-miR miR-17-92 cluster members were found as an miRNA signature in serum, and these RNAs were similar to their primary CRC cells (Ng et al., 2009). In HCC, a profiling result performed in HCC tissue miRNAs identified miR-21 as the most enriched miRNA (Meng et al., 2007), and miR-21 was also highly enriched in circulating miRNAs (Tomimaru et al., 2012). However, some other studies also reported that miRNA signatures are not always correlated between primary cancer cells and serum. For example, the serum miRNA signature from patients with breast cancer overlapped with only seven miRNAs from primary breast cancer tissues. More importantly, 16 of the top 20 miRNAs in the serum miRNA signature were not found in the tissue miRNA profile (Chan et al., 2013). The discrepancy may be caused by the heterogeneous nature of cancer origins and genetic backgrounds in each patient or may simply be due to the methodology of sampling procedures, microarray methods, and validation strategies (Pritchard et al., 2012; Moldovan et al., 2014). The accuracy of using the circulating miRNA profiles as diagnostic and prognostic markers is still debatable.

A. Acute Respiratory Distress Syndrome

ARDS is clinically defined by acute pulmonary onset after direct or indirect insult (such as pneumonia and sepsis), profound arterial hypoxemia, and bilateral radiographic opacities consistent with pulmonary edema (Ranieri et al., 2012). The pathogenesis of ARDS is characterized by exaggerated pulmonary inflammation, infiltration of immune cells, disrupted alveolar structure, and attenuated gas exchanges (Thompson et al., 2017). The incidence of ARDS is quite high in critically ill patients, as a recent study indicated that 23% of mechanically ventilated patients in the intensive care unit developed ARDS, with mortality up to 46% (Bellani et al., 2016). Patients who have recovered from ARDS also face long-term physiologic and psychologic challenges, with high risk for lifelong disability and decreased quality of life (Herridge et al., 2011). Previous studies demonstrated the importance of HIF-1A in the dampening of inflammation during acute lung injury (ALI) via transcriptional upregulation of the A2B adenosine receptor (Poth et al., 2013; Eckle et al., 2014). CD73 and adenosine receptors were also shown to be crucial in regulatory T cells during the resolution of ALI (Ehrentraut et al., 2012, 2013). Moreover, one of the mucin family members, Mucin 5AC, was identified as a proinflammatory mediator during ventilator-induced lung injury (VILI) (Koeppen et al., 2013). However, up to now, there is no pharmacologic therapy for the prevention and treatment of ARDS, whereas low tidal volume ventilation and conservative fluid management are the main lung-protective strategies in the management of patients with ARDS (Thompson et al., 2017). Thus, the discovery of a new molecular mechanism, such as the involvement of microRNAs, is essential for the development of novel therapeutic targets. Previous studies have identified several miRNAs that are involved in the resolution of inflammation through the regulation of resolvins, lipid mediators enzymatically generated to promote the crucial resolution process (Spite et al., 2009; Norling and Serhan, 2010). In particular, Dr. Charles Serhan’s laboratory found that resolvin D1 (RvD1) selectively upregulates miRNAs such as miR-21, miR-146b, and miR-219 and downregulates miR-208a, resulting in the reduction in the activity of nuclear factor kappa B and mothers against decapentaplegic family in macrophages (Recchiuti et al., 2011). They also found that the RvD1-triggered resolution of acute inflammation is mediated through two G protein–coupled receptors, formyl peptide receptor 2 and G protein–coupled receptor 32, which elevate the expression of miR-208a, resulting in interleukin-10 production (Krishnamoorthy et al., 2012). Interestingly, it was found that miR-4661 is overexpressed in neutrophils to initiate inflammation, whereas its overexpression in macrophages increases the production of specialized proresolving mediators, such as RvD1 and resolvin D5, to maintain homeostasis through accelerating the resolution process (Li et al., 2013b). Moreover, several studies have conducted screening experiments to identify differentially regulated pulmonary miRNAs during murine models of ARDS and in vitro mechanical stretch (Cai et al., 2012; Dong et al., 2012; Vaporidi et al., 2012; Yehya et al., 2012; Huang et al., 2014; Xiao et al., 2015; Mao et al., 2017). A recent study described the temporal changes in miRNA in peripheral blood mononuclear cells from patients with ARDS and identified 25 miRNAs differentially regulated in ARDS (Narute et al., 2017). However, because of the heterogeneity of human ARDS, a more comprehensive discovery study in patients with ARDS is still in dire need to identify potential biomarkers and therapeutic targets.

MiR-155 is one of the most profoundly upregulated microRNAs post–pulmonary infection and endotoxin exposure with a proinflammatory role to promote ALI (Rao et al., 2014; McAdams et al., 2015; Wang et al., 2016). Elevation of miR-155 is observed upon exposure to Staphylococcal enterotoxin B (SEB), a toxin known to cause food poisoning in humans (Rao et al., 2014). SEB exposure in mice significantly upregulates miR-155 levels in the lung-infiltrating mononuclear cells, and miR-155^−/− mice display attenuated pulmonary inflammation and histopathological injury. Suppressor of cytokine signaling (SOCS-1), a negative regulator of immune responses, was identified as a potential target of miR-155, as a gain of function of miR-155 inhibits its expression, whereas the loss of function of miR-155 enhances its expression. Besides SEB, intratracheal instillation of lipopolysaccharide (LPS) also induced the expression of miR-155, and its increased expression is mainly contributed to by alveolar macrophages (Wang et al., 2016). Similar to previously described SEB-induced ALI, miR-155–deficient animals are protected from LPS-induced lung injury marked by reduced lung permeability and dampened inflammatory responses. The same study identifies and confirms that miR-155 directly targets the 3′ UTR of SOCS-1 and that the inhibition of SOCS-1 by miR-155 allows the initiation of inflammatory responses in macrophages upon LPS stimulation. Additional study of fetal lung injury in nonhuman primates further suggests the proinflammatory role of miR-155 during Streptococcal infection (McAdams et al., 2015). The expression of miR-155 increases after group B Streptococcal–exposed fetal lung and in immortalized human fetal airway epithelial cells after exposure to IL-6 and tumor necrosis factor-α. In fetal airway epithelial cells, overexpression of miR-155 raises the expression level of chemokine (C-C motif) ligand 5, IL-6, and C-X-C motif chemokine ligand 10. However, besides its proinflammatory function, miR-155 also directly targets the 3′ UTR of fibroblast growth factor 9, implicating a potential role for miR-155 in fetal lung development. Summarized from the studies above, pharmacological inhibition of miR-155 could potentially benefit patients with ARDS to dampen pulmonary inflammation.

Besides proinflammatory roles, microRNAs also contribute to the protection against ALI. For example, miR-146a is upregulated in LPS-stimulated THP-1 cells (human monocytic), LPS-induced rat ALI, and acid-induced murine ALI (Nahid et al., 2009; Zeng et al., 2013; Vergadi et al., 2014). Nahid et al. (2009) were the first to report miR-146a as a toll-like receptor 4 responsive microRNA in human monocytic cell lines, which gradually increases after 4 hours of LPS challenge and then peaks at 24 hours post–LPS challenge. Interestingly, the elevated expression of miR-146a correlated with dampening of tumor necrosis factor-α production, and the following analysis suggests that miR-146a is the only microRNA upregulated in the tolerated state of LPS challenge after prolonged exposure. Functionally, THP-1 cells with overexpression of miR-146a are less responsive to LPS stimulation, representing a tolerance state, whereas inhibition of miR-146a dampens the LPS tolerance effect. After extensive studies of miR-146a during LPS stimulation in vitro, the upregulation of miR-146a is also demonstrated in rat LPS-induced ALI (Zeng et al., 2013). However, the anti-inflammatory role of miR-146a still needs to be further studied in LPS-induced ALI. In vivo function of miR-146a is further implicated by Vergadi et al. (2014) in murine hydrochloric acid aspiration–induced ALI. MiR-146a levels have been observed to be induced upon hydrochloric acid aspiration and to be significantly higher in AKT serine/threonine kinase 2 deficient (Akt2^−/−) mice, which are protected from acid-induced lung injury. MiR-146a mimic treatment significantly reduced acid-induced inducible NO synthase generation in alveolar macrophages, suggesting an anti-inflammatory function of miR-146a during aseptic ALI. In summary, the lung-protective role of miR-146a during ARDS/ALI has been suggested by several studies, and therapeutic enhancement of miR-146a could potentially improve the outcome of patients with ARDS.

Another microRNA and a new mechanism that is capable of dampening inflammation in ARDS is the shuttling of miR-223 from infiltrating neutrophils to alveolar epithelial cells (Neudecker et al., 2017a). Neudecker et al. (2017a) defined a novel lung-protective role of miR-223 shuttling in ARDS in a recent study (Fig. 3). The upregulation of miR-223 has been observed in both patients with ARDS and mice exposed to VILI. The function of miR-223 in ARDS was addressed by exposing mice deficient in miR-223 (miR-223^-/y mice) to VILI. Compared with wild-type counterparts, miR-223^-/y mice demonstrated exaggerated inflammation and tissue injury, characterized by increased bronchoalveolar lavage albumin levels to and elevated IL-6 and chemokine (C-X-C motif) ligand 1 levels in the lung. Further mechanistic studies identified the interaction between the shuttling of miR-223 from polymorphonuclear neutrophils (PMNs) to alveolar epithelial cells in vitro and in vivo during murine ARDS. Specifically, miR-223 is released from PMNs via microvesicles, which are transferred into alveolar epithelial cells and dampen inflammatory gene expression and lung inflammation. A similar shuttling mechanism for miR-21 and miR-29a between primary cancer cells and normal recipient cells via exosomes has been previously reported during prometastatic inflammatory responses (Fabbri et al., 2012). Besides the VILI model of ARDS, miR-223^-/y mice demonstrated exaggerated susceptibility to pulmonary bacterial infection, indicating a protective function of miR-223 in ALI. Moreover, poly (ADP-ribose) polymerase 1 (PARP-1) was identified as a target of miR-223 in this study, as the 3′ UTR of PARP-1 was directly targeted by miR-223. To confirm the functional relevance of the miR-223-PARP-1 interaction, pharmaceutical inhibition of PARP-1 in vivo attenuated the severity of VILI in miR-223^-/y mice. Subsequently, nanoparticle delivery of miR-223 before the onset of murine ALI dampened pulmonary inflammation and tissue injury. To conclude, the shuttling of miR-223 between neutrophils and alveolar epithelial cells during ALI established a unique mechanism for microRNA-mediated lung protection during ARDS, and direct targeting of alveolar epithelial cells to deliver miR-223 mimetics could potentially dampen pulmonary inflammation in patients with ARDS. Inhalation of nanoparticle-based therapy to achieve overexpression or inhibition of microRNA in the lung would lead to substantial advances in the treatment and prevention of ARDS.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.

Shuttling of miR-223 protects against acute lung injury. Infection or mechanical ventilation can result in acute lung injury, during which neutrophils will transmigrate through the vasculature to inflamed alveolae (Dengler et al., 2013). Recruited neutrophils release microvesicles containing miR-223, a microRNA highly expressed in myeloid lineage cells. MiR-223–containing microvesicles are shuttled to pulmonary alveolar type II cells, resulting in the transcriptional repression of its inflammatory target gene PARP-1. Inhibition of PARP-1 leads to the attenuation of pulmonary inflammation and tissue injury. Therapeutically, nanoparticle delivery of miR-223 to pulmonary epithelial cells could potentially dampen pulmonary inflammation to prevent and treat acute lung injury (Neudecker et al., 2017a,c).

B. Acute Myocardial Infarction

Acute myocardial infarction (AMI) is clinically diagnosed by abnormally high levels of plasma cardiac troponin concentration, ischemia, EKG abnormality, imaging of intracoronary thrombus, and/or wall motion abnormality (Thygesen et al., 2012; Anderson and Morrow, 2017). The incidence of MI ranges from two to four cases per 1000 person-years globally and with 750,000 episodes of MI each year in the United States (Yusuf et al., 2014; Mozaffarian et al., 2016). Multicenter international cohort studies by Devereaux and colleagues demonstrated that in perioperative settings, AMI results in about five-times-higher mortality rates, posing a significant impact on surgical outcome (Devereaux et al., 2005, 2011). Clinical management of AMI includes percutaneous coronary interventions and supplemental oxygen. Pharmaceutical interventions for MI include a combination of antiplatelet and anticoagulant agents, β-blockers, nitrates, and statins. Although the mortality of AMI has been reduced significantly by traditional management and pharmaceutical interventions, it is still crucial for the development of novel therapeutic targets to prevent perioperative AMI and to improve the long-term prognosis of AMI. Previous studies have indicated the important role of extracellular nucleotide and nucleoside signaling in vascular inflammation and ischemic preconditioning (Eltzschig et al., 2006, 2013; Hart et al., 2008, 2010; Riegel et al., 2011; Idzko et al., 2014). Furthermore, extensive studies have been conducted to understand the regulation of microRNAs during AMI in patients and murine models (Hullinger et al., 2012; Martinez et al., 2017). Here, we will summarize microRNAs identified to have a detrimental or protective role in AMI.

Several microRNAs and microRNA families have been identified to be detrimental in AMI. For example, the miR-15 family, including miR-15a, miR-15b, miR-16-1, miR-16-2, miR-195, and miR-497, are upregulated in the infarction area 24 hours after the onset of AMI (Hullinger et al., 2012). The same study showed that inhibition of miR-15 by anti-miRs reduced infarct size and improved cardiac function in the murine AMI model, suggesting the therapeutic potential of miR-15 inhibition. Another important microRNA, miR-34a, was first identified in a discovery screening of microRNA changes during cardiomyocyte aging and is involved in the recovery of AMI (Boon et al., 2013). The level of miR-34a increases as the heart ages, and its inhibition prevents age-associated cardiomyocyte death while facilitating the recovery of AMI. Furthermore, protein phosphatase 1 regulatory subunit 10 was identified as a direct target of miR-34a, and its main function includes reduction of aging damage and apoptosis. Additionally, the miR-17-92 cluster (miR-92a especially) is highly expressed in endothelial cells to modulate the growth of new vessels. In vivo silencing of miR-92a improved the functional recovery of murine AML via direct targeting of integrin subunit α5 (Bonauer et al., 2009). The therapeutic potential of miR-92a has been further evaluated in the large-animal AMI model (Hinkel et al., 2013). Indeed, only regional administration of miR-92a LNA reduced infarct size, whereas systemic delivery of miR-92a failed to render protection, thus implicating the necessity of precise delivery of miR-92a–based therapy. Moreover, a recent study by Martinez et al. (2017) identified a progressive response of miR-31 after the onset of rat AMI. Pharmaceutic miR-31 inhibition enhances the expression of its target genes, cardiac troponin-T, E2F transcription factor 6 (E2f6), nuclear receptor subfamily 3 group C member 2, and metalloproteinase inhibitor 4 mRNAs, and leads to improved recovery of cardiac function. To summarize, therapeutic inhibition of those abovementioned microRNAs could potentially improve the recovery of AMI.

In contrast to the miRNAs described in the previous paragraph, several microRNAs have been found to provide cardiac protection during AMI. For example, higher levels of miR-210 were detected in live cardiomyocytes when compared with apoptotic cells, and overexpression of miR-210 in vitro prevents apoptosis (Hu et al., 2010). Intramyocardial injections of nanoparticles carrying miR-210 precursor significantly improve cardiac function after recovery from murine AMI. The mechanistic study identified Eph-related receptor tyrosine kinase ligand 3 and protein-tyrosine phosphatase 1B as potential targets for miR-210. Interestingly, miR-210 was identified as one of the hypoxia-miRs, a group of microRNAs induced by hypoxia and dependent on HIF (Camps et al., 2008; Gou et al., 2012). Previous studies have identified a tissue-protective role of HIF1a and HIF2a in the setting of murine myocardial infarction (Eckle et al., 2008, 2012; Koeppen et al., 2018). Thus miR-210 could serve as an HIF responsive element and a hypoxia adaptive mechanism that facilitates HIF-elicited tissue protection during myocardial infarction. The tight link between HIF and miR-210 could also potentially serve as a mechanism to tie hypoxia and inflammation during ischemia/reperfusion injury, which was discussed in detail in our previous reviews (Eltzschig and Carmeliet, 2011; Eltzschig et al., 2014). Another microRNA with cardiac protective potential is miR-132, which is highly expressed and secreted by saphenous vein–derived pericyte progenitor cells (SVPs) (Katare et al., 2011). The authors demonstrate that SVP-conditioned media facilitates endothelial tube formation while inhibiting fibrosis. SVP transplantation improves cardiac function via inhibition of cardiomyocyte apoptosis and dampening of interstitial fibrosis, whereas miR-132 inhibition attenuates this effect. Following along the line of secreted microRNAs by progenitor cells, another study by Barile et al. (2014) identified miR-210, miR-132, and miR-146 as highly enriched microRNAs in extracellular vesicles (EVs) secreted by human cardiac progenitor cells (CPCs). In this study, most EVs are around 30–90 nm in diameter, which is similar to the size of exosomes, as measured by transmission electron microscopy and nanoparticle tracking analysis. Consistent with the other two reports, in vivo, functional studies suggest that injection of EV secreted from CPCs inhibits cardiomyocyte apoptosis and improves cardiac function. In conclusion, the delivery of EVs from CPCs and pharmaceutic enhancement of miR-210 and miR-132 could potentially attenuate cardiomyocyte apoptosis and aid the recovery from AMI. Notably, it is essential for the delivery of microRNA mimics or inhibitors in the heart, which could pave the way for important improvement in therapy for AMI.

C. Acute Gut Injury

The prevention of acute gut injury (AGI) in perioperative patients is pivotal to impede the development of multiorgan failure and sepsis (Chen et al., 2013; Yuan et al., 2018b). Several factors contribute to the development of AGI, including translocation of bacteria and/or their endotoxin products from the intestinal lumen into the systemic circulation or other tissues/organs across the intestinal mucosal barrier and activation of intestinal immunity and mesenteric ischemia resulting from the low-flow state commonly encountered in high-risk surgeries (Ohri and Velissaris, 2006; Deitch, 2010). AGI occurs in 0.3%–6.1% of patients undergoing high-risk surgery, such as lung transplant and cardiopulmonary bypass surgeries, and the mortality of AGI ranges from 18% to 58% (Rodriguez et al., 2010; Lahon et al., 2011). AGI involves disrupted intestinal barrier function and intensified intestinal inflammation, which can manifest into systemic inflammation in patients at risk. For example, A2B adenosine receptor signaling is beneficial in the protection against acute colitis and intestinal ischemia/reperfusion injury (Hart et al., 2009; Grenz et al., 2011; Aherne et al., 2015). Additionally, neuronal guidance molecule netrin-1 has also been shown to be important in the attenuation of acute experimental colitis (Aherne et al., 2012). Moreover, Charles Serhan’s laboratory found that the resolution of inflammation also plays a crucial role to reduce long-term tissue injury, and several microRNAs have been reported to function in this process (Fredman et al., 2012; Krishnamoorthy et al., 2012; Li et al., 2013b). Furthermore, in the attempt to understand the mechanism of AGI and the development of intestinal inflammation, several screening studies demonstrate differentially regulated microRNAs in intestinal inflammation (Wu et al., 2008a, 2010; Takagi et al., 2010; Neudecker et al., 2017b). Here, we will discuss several microRNAs that have been studied extensively in intestinal inflammation.

Several microRNAs have been identified to be detrimental in AGI. MiR-146a has been observed to be upregulated in human patients with inflammatory bowel disease (IBD) by several studies. MiR-146a is commonly known by its function in the regulation of T cells, especially T helper (Th) 1, Th17, regulatory T cells (Treg), and follicular helper T cells (Lu et al., 2010; Runtsch et al., 2015). MiR-146a^−/− mice have increased Th17 and Treg population in the lamina propria and increased follicular helper T cells in the germinal center (Runtsch et al., 2015). Increased T-cell and B-cell responses predict enhanced inflammation. Although surprisingly, miR-146a^−/− mice are protected in dextran sulfate sodium (DSS)-induced intestinal inflammation, suggesting a detrimental role of miR-146a in the intestinal barrier function. Additionally, the increased level of miR-214 has been observed in colon biopsies of patients with active colitis (Polytarchou et al., 2015). The upregulation of miR-214 is mediated by interleukin-6–induced signal transducer and activator of transcription 3 activation in colon tissues, and chemical inhibition of miR-214 attenuated the severity of DSS-induced murine intestinal inflammation. The above studies suggest that pharmacological inhibition of miR-146 and miR-214 could potentially benefit the outcome of patients with AGI by limiting intestinal inflammation and improving barrier functions.

In contrast to the miRNAs described in the previous paragraph, miR-223 is anti-inflammatory during intestinal inflammation (Zhou et al., 2015; Neudecker et al., 2017b). Increased levels of miR-223 have been observed in intestinal biopsies from patients with IBD and colon tissues in a murine model of intestinal inflammation (Neudecker et al., 2017b). In mouse models of colitis, Zhou et al. (2015) demonstrated that miR-223^-/y mice exhibit exaggerated intestinal inflammation. CCAAT enhancer-binding protein β was identified in this study as a proinflammatory target of miR-223. Further study also suggested the role of miR-223 in the protection of murine intestinal inflammation (Neudecker et al., 2017b). The function of miR-223 in intestinal inflammation was addressed by exposing miR-223^-/y mice to DSS-induced colitis. Compared with wild-type counterparts, miR-223^-/y mice demonstrated exacerbated intestinal inflammation marked by more weight loss and exaggerated tissue injury. Based on previously reported roles of miR-223 in myeloid cell compartments, bone marrow chimeric mice were generated to identify whether the myeloid or nonmyeloid compartment was crucial for the function of miR-223 during murine colitis. The importance of miR-223 in murine colitis lies within the myeloid compartment, as an miR-223 deficiency in the myeloid compartment showed a similar phenotype of miR-223^-/y mice. To identify the importance of different subtypes of myeloid cell in miR-223–mediated intestinal protection, neutrophils or monocytes in miR-223^-/y mice were deleted by antibody, and monocyte deletion disrupted the protective effect of miR-223. Furthermore, NLR family pyrin domain containing 3 (NLRP3) was identified as a functional target of miR-223 during murine DSS-induced colitis (Fig. 4). In support of the direct targeting in vivo, targeted deletion of the miR-223 binding site in the 3′ UTR region of NLRP3 in mice resulted in increased severity of DSS-induced colitis, which resembled that of miR-223^-/y mice. To investigate the therapeutic potential of miR-223 during intestinal inflammation, miR-223 mimic was delivered systemically in mice, and the delivery of miR-223 resulted in attenuation of DSS-induced colitis. To conclude, the protective role of miR-223 was suggested in experimental colitis, and this protection stems from innate immune responses controlled by the myeloid cell. Therapeutic enhancement of miR-223 in patients with IBD could serve as a potential novel treatment in the control of access intestinal inflammation. It is critical to achieving the overexpression of microRNA such as miR-223 in the intestinal tract; thus, the oral delivery of packaged miRNA mimic or inhibitor would be a feasible way to treat or prevent acute gut injury.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 4.

MiR-223 in inflammatory monocytes dampens intestinal inflammation. DSS-induced intestinal injury leads to the upregulation of MiR-223 in inflammatory monocytes. Increased miR-223 levels result in the inhibition of its target gene NLRP3, and dampened NLRP3 sequesters inflammasome activity and downstream production of inflammatory cytokines, such as IL-1β. Turning down IL-1β attenuates intestinal inflammation and histologic signs of intestinal injury. Pharmacological enhancement of miR-223 could potentially alleviate acute gut injury and intestinal inflammation in the perioperative period (Neudecker et al., 2016, 2017b,c; Yuan et al., 2018a).

A. Modification of Nucleotide Analogs to Increase Stability

RNA contains a 5-carbon sugar, ribose, with a hydroxyl group on its 2′ position (2′-OH), whereas DNA has a deoxyribose sugar (2′-H) without the hydroxyl group on the 2′ position. The 2′-OH is crucial for the degradation of RNA by self-hydrolysis and by endogenous RNases. Therefore, the removal and modification of the 2′-OH group from mimic or inhibitor RNAs will make them resistant to both self-hydrolysis or RNase-mediated enzymatic degradation. This approach includes direct modification of the sugar-phosphate backbone of RNA or using nucleic acid analogs that are chemically mimicking RNA molecules (Prakash and Bhat, 2007; Lennox and Behlke, 2011; Prakash, 2011). These approaches generate more chemically stable RNA analogs; therefore, they are suitable for designing anti-miRNA oligonucleotides (AMOs) to reduce a specific miRNA level. For example, phosphorothioate (PS) RNA is an RNA analog generated by replacing nonbridging oxygen with sulfur in the phosphodiester bond of the RNA strand, because of which the PS RNA is more stable (Lennox et al., 2006). Since PS RNAs lack phosphodiester bonds, which are attacked by RNases, PS AMOs are more resistant to the RNase-mediated enzymatic degradation. Replacement of the 2′-OH group on RNA ribose ring with a chemically inert group has also been studied by many different approaches to increase the AMO stability. The main idea is to deprive the hydrogen on the 2′-OH group to avoid the pH or enzyme-dependent deprotonation. For example, 2′-O-Methylation AMO duplexes were designed to specifically target mouse liver abundant miR-122 and were injected through mouse tail vein. When the miR-122 expression level was measured by Northern blotting 24 hours after the last injection, the hepatic miR-122 was completely degraded in the mouse group injected with 240 mg/kg per day (Krützfeldt et al., 2005). Davis et al. (2009) used 2′-fluoro/2′-methoxyethyl AMO to inhibit the endogenous miR-122 in mice. The 2′-O-MOE modification makes AMO form more stable binding with target RNAs than 2′-O-methyl by increasing 2°C of melting temperature (Tereshko et al., 1998), therefore increasing their tissue half-life up to 10-fold higher than PS RNA (Geary et al., 2001). 2′-Fluoro (2′-F) modification also increases the stability of AMO. However, combining the 2′-F modification with PS RNA ensures RNase resistance.

Besides the modification of 2′-OH, several other strategies have been employed to facilitate the delivery of RNA oligonucleotide in vivo. Peptide bond–based nucleic acid (PNA) mimics is another promising candidate for the targeting of microRNAs in vivo. The sugar-phosphodiester backbone of PNA is replaced with N-(2-aminoethyl)-glycine units, which present no surface charge, so that naked PNA molecules can directly penetrate cell membranes without the aid of transfection agent. When a PNA technology–based anti-miR against miR-155 was intraperitoneally given to mice, the endogenous miR-155 level in primary B cells was successfully inhibited, and the gene expression profiles from the treated mice were similar to that from miR-155–deficient mice (Fabani et al., 2010). Besides PNAs, locked nucleic acid (LNA) is another artificial RNA analog to facilitate the targeting of microRNAs in vivo. The 2′-oxygen and 4′-carbon on the ribose of LNA are chemically locked by a bridge. This closed conformation physically inhibits the binding of RNase, resulting in high stability in serum, but is still capable of pairing with complementary RNA strand. For this reason, LNAs are frequently used to either detect specific miRNAs for fluorescence staining or inhibit target miRNAs by complementarily binding to functional miRNA strands (Wahlestedt et al., 2000). Whereas unmodified oligonucleotides are degraded within 1.5 hours in serum, the half-life of LNA is up to 15 hours in serum, enabling longer intervals between administrations (Kurreck et al., 2002). Compared with PNA, LNA is superior in affinity to a target sequence than PNA. For example, the antisense efficiency of PNA against human telomerase mRNA was almost 200-fold lower than that of LNA (Elayadi et al., 2002). LNA is more cost-effective and has higher cellular uptake than PNA; therefore, LNA has been more frequently used as miRNA inhibitors than PNA.

B. Viral Delivery of MicroRNAs

Besides the modification of nucleotide to prevent degradation, viruses are also powerful genetic tools that have been studied extensively to use as a gene-editing vector system in mammalian cells. In general, two modifications were applied to those viruses, such as retroviruses, lentiviruses, adenoviruses, or adeno-associated viruses (AAVs), to control their cell type–specific expression of inquired genes and harmful side effects: 1) insertion of cell type–specific promoter sequences and 2) removal of pathologically harmful genes, except for the genes that are essential for viral entry and replication. Over the years, it has been shown that siRNAs or miRNAs can be expressed by these genetically engineered viruses to target disease cells (Davidson and Harper, 2005). Virus-mediated vector systems are especially useful to overexpress miRNAs that are found to be downregulated in diseased cells since the restoration of such downregulated miRNAs in the targeted cells is expected to reverse already ongoing pathologic phenomena. Viruses are suitable for continuous expression of miRNA by using a constitutive promoter. Thus, it enables sufficient overexpression of miRNA in the virus-infected cells with a single dose of administration. Because of the high specificity and transduction efficiency of viral vectors, high levels of miRNA expression can be achieved with no further chemical modification of miRNA sequences to protect the miRNAs from serum degradation.

Lentiviruses are useful to deliver the various sizes of genes since they are flexible for the cloning of a long size of DNA. Upon host cell infection, lentiviruses integrate into host genomes. The host gene integration enables the maximal and stable expression of the gene of interest by a single administration. However, the chromosomal integration of exogenous DNA sequences by lentivirus-based delivery may cause harmful genomic mutations when applied to long-term clinical trials (Milone and O’Doherty, 2018). AAVs, on the other hand, do not integrate into the host genomes, which reduces the chance of genomic mutations. AAVs are commonly used to deliver small noncoding RNAs, such as short hairpin RNAs or miRNAs, because their small packaging capacity is still large enough for small noncoding RNAs and because they are relatively safer than other viruses because of their lack of pathogenicity (Grimm, 2009). AAVs are physically stable in a wide range of temperatures and pH’s (Penaud-Budloo et al., 2018). Although the viral delivery of miRNA seems to be quite efficient for sustained expression of miRNAs, using such viral vector systems is often challenging because of their potential activation of oncogenes (depending on their integration location), systemic toxicity, or immunogenic response due to the nature of the viral infection. For these reasons, the Food and Drug Administration (FDA) recommends a level of residual cellular foreign DNA less than 10 ng per parenteral dose to limit the genotoxic risk of recombinant AAVs after systemic administration, which corresponds to 10¹¹ vector genomes (Penaud-Budloo et al., 2018). One of the most groundbreaking advancements in the AAV-mediated delivery for gene therapy is the FDA approval of Luxturna (voretigene neparvovec-rzyl) as a treatment of an inherited form of vision loss in children and adult patients. In clinical trials, Russell et al. (2017) have demonstrated that Luxturna (AAV2-hourRPE65v2) gene replacement significantly improved functional vision in retinal pigment epithelium-specific 65 KDa protein–mediated inherited retinal dystrophy. As the first directly administered gene therapy to be approved by the FDA, although Luxturna is approved for injection into the eye, a relatively immune-privileged organ, it opens many doors for the design of future gene therapy with AAV-mediated delivery of microRNA-based approaches. For example, Wang et al. (2019) have reported that AAV-mediated delivery of anti-miR-214 (AAV-anti-miR-214) successfully prevented the collapse of the femoral head in a rat model of osteonecrosis by rescuing the miR-214 targets activating transcription factor 4 and phosphatase and tensin homolog. Spark Therapeutics, the company that developed Luxturna, has recently announced that they evaluated the Luxturna-based miRNA delivery for the knockdown of Huntingtin gene to treat Huntington disease.

C. Nonviral Delivery of MicroRNAs

Using virus-based vector systems to target miRNAs is one of the most efficient methods because of their high level of expression controlled by strong promoters. However, some of the concerns and safety issues, as mentioned above, are still challenging to resolve. To overcome the limitation of viral delivery, delivering miRNA mimics or miRNA inhibitors by formulating with three major nonviral materials—polymers, lipids, and inorganic nanoparticles—seems to be a reasonable and promising approach. These nonviral delivery systems prevent serum degradation, endosomal capture, lysosomal degradation, and immunogenic response. Besides, many of the nonviral molecules are chemically flexible for further conjugation with various ligand molecules, which presents a great potential for cell-specific targeted delivery of miRNAs. All of these nonviral vector-based delivery approaches have shown adequate therapeutic capability and potential for clinical applications with a satisfactory safety profile compared to the viral vector systems. However, additional studies are needed to evaluate in vivo pharmacology of these nonviral particles to facilitate future clinical trials.

D. Small Molecule Drugs Targeting MicroRNAs

So far, we have outlined several approaches using oligonucleotides and other agents to target miRNAs for disease therapy. There is an important and novel strategy using small molecule drugs to directly target microRNAs or transcription factors that regulate miRNA expression (Wen et al., 2015; Safe et al., 2018). Small molecule drugs are capable of inhibiting or inducing microRNAs in disease settings. Lee et al. (2016) reported that Kenpaullone derivatives, small molecule inducers of miR-182, significantly improved heart function after ischemia/reperfusion injury in the rat. Another small molecule, enoxacin (Penetrex), is reported to promote the biogenesis of endogenous miRNAs (Shan et al., 2008). Moreover, neomycin B, an FDA-approved small molecule drug, can selectively inhibit the production of miRNA-525 and, in turn, inhibit the invasion of the HCC cell line (Childs-Disney and Disney, 2016). Small molecule drugs can also modulate the transcription factors to regulate and inhibit the constitutive expression of miRNAs. As mentioned above, c-Myc has been shown to activate the expression of several pro-oncogenic microRNAs. Thus, the downregulation of c-Myc by reactive oxygen species–inducing agents such as hydrogen peroxide, isothiocyanates, histone deacetylase inhibitors, penfluridol, piperlongumine, and many other natural products can target the critical reactive oxygen species-c-Myc-OncomiR interaction, resulting in downregulation of Sp transcription factors and pro-oncogenic Sp-regulated genes (Jutooru et al., 2010, 2014; Chintharlapalli et al., 2011; Chadalapaka et al., 2012; Kasiappan et al., 2016; Karki et al., 2017). To summarize, besides delivering microRNA mimics or inhibitors, small molecule compounds could serve as an opportunity to safely and effectively target microRNAs in cancer and perioperative organ injury.

A. MicroRNA Restoration

As described above, many miRNAs, such as tumor-suppressive miRNAs, are downregulated in cancer. In these cases, introducing high levels of such miRNAs will restore the capability of the diseased cells by inhibiting the expression of pathogenic or oncogenic target genes. Virus-mediated vector systems are especially useful to overexpress miRNAs that are found to be downregulated in diseased cells since the restoration of such downregulated miRNAs in the targeted cells is expected to reverse already ongoing pathologic phenomena. In addition, viruses are suitable for continuous expression of miRNA by using a constitutive promoter. It enables a single dose of administration to be enough to maintain a high level of miRNA expression in the virus-infected cells. Kota et al. (2009) were able to overexpress miR-26a in an Myc-induced HCC-bearing mouse model by using an engineered AAV as a delivery vector system. Another tumor-suppressive miRNA, let-7, was also delivered into breast cancer cells through lentiviruses and reduced proliferation, mammosphere formation, tumor formation, and metastasis in nonobese diabetic/severe combined immunodeficiency mice (Yu et al., 2007). Let-7–carrying lentiviruses were also shown to slow non–small-cell lung cancer (NSCLC) tumor growth in mice (Trang et al., 2010). Although the viral delivery vectors were proven to be efficient in overexpressing miRNAs in the infected cells, none of these preclinical studies has made it to clinical trials because of challenges including virus clearance, aberrant immune responses, and unwanted genomic mutations originated by genome integration. Cationic lipid-based nanoparticles were also used to deliver miRNA mimics since their positive charges make it convenient to carry negatively charged miRNA mimics. Wu et al. delivered tumor-suppressive miR-29b mimics into lung cancer cells via a systemic injection after formulation into 1,2-di-O-octadecenyl-3-trimethylammonium propane:cholesterol:TSGA lipoflex nanoparticles (Wu et al., 2013). They observed a 5-fold increase of miR-29b in the tumor region after systemic delivery and downregulation of target gene expressions including CDK6, DNMT3B, and MCL-1 by ∼57.4%, ∼40.5%, and ∼52.4%, respectively. As a result, the tumor growth was significantly inhibited by ∼60% compared with negative control miRNA carrying nanoparticles (Wu et al., 2013). Liposomes have replaced viral vector systems to restore downregulated miRNAs because of fewer safety concerns and have made it to clinical trials. In lung cancer cells, restoration of let-7 through an overexpressing plasmid DNA incorporated into liposome-sensitized EGFR inhibitor–resistant lung cancer cells by suppressing EGFR expression (Rai et al., 2011). The laboratory of Anil Sood generated chitosan-based nanoparticle to deliver miR-34a mimics into prostate cancer cells and observed the downregulation of tyrosine-protein kinase Met and tyrosine-protein kinase receptor UFO and c-Myc, which are responsible for the proliferation of prostate cancer cells (Gaur et al., 2015). Remlarsen/MRG-201, a synthetic mimic of microRNA-29 developed by miRagen for the potential treatment of patients with fibrotic diseases, is currently in phase 2 clinical trials for the treatment of keloid (NCT03601052). This study was built on phase 1 data which indicated the favorable safety profile of MRG-201 in healthy subjects (NCT02603224). Besides, small molecule compounds can also mediate the overexpression/restoration of microRNAs, and this therapeutic approach has been translated into clinical studies. According to Abivax, one of the small molecule compounds, ABX464, delivers its anti-inflammatory effect via potentiation of the splicing of a long, noncoding RNA, resulting in overexpression of miR-124. ABX464 is currently undergoing multiple phase 2 clinical trials to treat multiple diseases (Table 2). For instance, phase 2 trials have been completed to test the safety and efficacy of ABX464 in human immunodeficiency virus (HIV) infection and ulcerative colitis (NCT02735863; NCT03093259) (Rutsaert et al., 2019). The technology used in the new generation of microRNA mimics, as well as small molecule–mediated overexpression of microRNAs, is crucial for the therapeutic targeting of microRNAs by overexpression in cancers and perioperative organ injuries.

B. MicroRNA Inhibition

Upregulated miRNAs have been implicated in the pathogenesis of many human diseases, which makes aberrantly overexpressed miRNAs attractive targets for therapeutic interventions. Recently, many approaches have been employed to inhibit upregulated miRNAs. Elmen et al. (2008) systemically injected LNA-anti-miRs against miR-122 to nonhuman primates (African green monkeys) and observed successful uptake of the LNA-anti-miR-122 in hepatocytes and depletion of endogenous miR-122 through stable heteroduplexes formed between the LNA-anti-miR-122 and miR-122. Because of the stability in serum and profound ability to bind to target miRNAs, LNA-based specific miRNA inhibition has been extensively investigated in preclinical studies and clinical trials. For example, miR-122 inhibition by complementary LNAs suppressed hepatitis C virus (HCV) proliferation in chimpanzee models without serious cytotoxicity in hepatocytes (Lanford et al., 2010). LNA-based inhibition of onco-miRs can also successfully induce cancer cell suppression. For example, LNA targeting miR-221, known to be an onco-miR in many cancer types, inhibited the proliferation of multiple myeloma cells (Di Martino et al., 2014). In glioma models, LNA-based miR-21 inhibition inhibited cell proliferation and sensitivity to radiation (Griveau et al., 2013). In clinical applications, Santaris Pharma A/S, based in Denmark, has generated a pharmacological miR-122–inhibiting LNA named Miravirsen, which has completed phase 2 clinical trials in patients with HCV (NCT01200420) (Janssen et al., 2013). Additionally, miRagen Therapeutics has another LNA-based anti-miR drug candidate inhibiting miR-155 (Cobomarsen/MRG106), which is currently undergoing phase 2 clinical trial for patients with cutaneous T-cell lymphoma of the mycosis fungoides subtype who are given multiple dosages of MRG106 through intratumoral injection (NCT03713320). For the treatment of chronic ischemic disorders, miRagen is currently developing and testing the safety and efficacy of inhibiting miR-92a (MRG-110), a microRNA linked to the regulation of blood vessel growth (NCT03603431). It is speculated that the anti-miR-92a could potentially accelerate the revascularization to facilitate the healing process. To summarize, the recent implementation of multiple phase 2 clinical trials based on microRNA inhibition is an important step toward the future of microRNA-based therapy for cancers and perioperative organ injuries.

C. Challenges and Limitations

Recently, the focus of drug delivery has moved toward targeted therapy using specific strategies to recognize target cells. Without such specific targeting methods, miRNA delivery methods can cause unwanted side effects because of the delivery of therapeutic miRNAs into nondiseased cells. For example, the phase I clinical trial of MRX34 harboring miR-34a was terminated early because of the lack of a tissue-specific delivery method (NCT01829971). The liposome-based formula was likely delivered into many cells throughout the body, and there is little preclinical evidence regarding the effect of excess miR-34a when introduced to normal cells in different organs. This concern calls for further modification and optimization of current delivery methods to enable tissue-specific targeting. To achieve cell/tissue-specific targeting, the carrier molecules need to be modifiable to include targeting moieties. In the clinical trial of TargomiRs for patients with recurrent thoracic cancer, EGFR-targeting antibodies were included in the liposome formulations, which enabled the miR-16–containing drug to be specifically delivered to cancer cells that overexpress EGFR on their cell surface (Reid et al., 2016). Folate (FA) is another targeting moiety for the specific delivery of many miRNAs. For example, polymer nanoparticles made from PLGA/poly(L-lactide)-block-poly(ethylene glycol) loaded with miR-204 were conjugated with FA to target folate receptors on colon cancer cells. When injected into mice bearing xenograft tumor, the FA-conjugated miR-204–carrying nanoparticles caused the regression of the tumors by inducing apoptosis (Zheng et al., 2018). 3WJ RNA nanoparticles derived from the packaging RNA of the DNA packaging motor system in bacteriophage phi29 have also been chemically conjugated with FA to specifically target the brain tumor tissues that highly overexpress folate receptors (Shu et al., 2011). This cancer cell–specific FA ligand conjugation was especially beneficial for targeted delivery of miRNAs in brain tumor since already differentiated normal brain neuronal cells do not express folate receptors, whereas glioblastoma cells overexpress folate receptors (Lee et al., 2015a,b, 2017). Moreover, recently emerging RNA aptamer technologies could potentially replace antibodies to reduce the antibody-related host immune response. Many RNA aptamers have been developed for recognizing specific cell surface markers (Nimjee et al., 2017). For example, targeted nanoparticle delivery of miR-29b to NSCLC was mediated by MUC-1 aptamer conjugation, which enabled selective delivery of an miR-29b mimic in vivo to cancer cells (Perepelyuk et al., 2018). When 3WJ RNA nanoparticles were loaded with EGFR-targeting aptamer together with antisense inhibitor sequences against miR-21, systemic delivery of the EGFR aptamer and anti-miR-21–carrying RNA nanoparticles successfully inhibited the tumor growth of triple-negative breast cancer (TNBC) (Shu et al., 2015). Recently, surface modification technology has evolved to modify the exosome surface with cell-targeting specific moieties to enable targeted therapy. For example, systemic administration of siRNA to breast cancer–bearing mice with siRNA against Survivin-carrying exosomes that were chemically conjugated with epidermal growth factor receptor (EGFR)-targeting RNA aptamers specifically delivered the Survivin-targeting siRNAs into the breast cancer cells and reduced tumor growth (Pi et al., 2018). Using this method, the delivery of therapeutic miRNA-loaded exosomes specifically into cancer cells can be achieved.