Non-coding RNAs: the cancer genome dark matter that matters!

Introduction

As stated in the central dogma of molecular biology, RNA is an essential component in the flow of genetic information from DNA to protein [1]. Because of the recognized importance of proteins in exerting biological functions, RNAs had been regarded for a long time as merely a mediator in passing genetic codes to final functional protein molecules. As a consequence, the functional activities of RNAs themselves are largely neglected [2]. Recent advances on the sequencing technologies revealed that vast majority of the human genome are transcribed, while the protein-coding genes only occupy 3% of the human genome [3]. The widespread transcription of the genome into non-protein-coding RNAs strongly suggest that RNAs process functions other than mere mediators between DNAs and proteins. Indeed, RNA have been suggested to be the earliest molecule of life and thought to possess both informational and catalytic function [4]. Emerging evidences from the last two decades have unambiguously proved the functional importance of noncoding RNAs in human biology and diseases [5].

Ribosome RNAs and RNAs are among the early-discovered non-coding RNA transcripts. However, because of their roles in facilitating protein translation, they are still considered part of the machinery translating genomic code into protein synthesis. The discovery of microRNAs in early 1990s [6], [7] opened a new chapter of gene regulation by non-coding RNAs and stimulated the scientific interest onto the many RNA molecules that do not code for proteins [8]. Numerous other types of small non-coding RNAs have also been subsequently identified, including endogenous small interfering RNAs (endo-siRNAs), PIWI-associated small RNAs (piRNAs), small nucleolar RNAs (snoRNAs), sno-derived RNAs (sdRNAs), transcription initiation RNAs (tiRNAs), miRNA-offset RNAs (moRNAs), and others [5]. Long non-coding RNAs, defined as non-coding transcripts longer than 200 nucleotides, also emerge as important regulator in a wide range of biological activities and human disease [9]. A systematic classification of long non-coding RNA is missing; currently, long non-coding RNAs are categorized as, according to their genomic localization or other features, natural antisense transcripts, long intergenic non-coding RNAs, transcribed ultraconserved regions (T-UCRs), circular RNAs, enhancer-associated RNAs, promoter-associated RNAs, and others [5].

In this review, we focus on the well-characterized microRNAs by providing a general overview on their production, function, regulatory mechanisms, and their clinical importance as cancer biomarkers and cancer therapeutics. Due to the space limit of this review, we only present the essential and relevant literatures and have to skip many excellent works.

What are microRNAs?

In 1993, Victor Ambros’s research group and Gray Ruvkun’s research team independently reported that a small RNA transcript around 20 nucleotides, known as lin-4, regulates developmental programming of Caenorhabditis elegans by posttranscriptional regulation of the heterochronic gene lin-14 via an antisense RNA-RNA interaction [6], [7]. These reports mark the first discovery of microRNAs, a new type of non-coding RNA with unique regulation mechanism. In 2000, the second microRNA let-7 was discovered from C. elegans [10]. Until now, the miRBase database (version 19) cataloged over 25,000 mature microRNA sequences in 193 species, including 2042 mature microRNAs in human beings.

Mature microRNAs are single-stranded non-coding RNAs with 19–24 nucleotides that are usually evolutionarily conserved [11]. The biogenesis of microRNA is a multi-step event [11] (Figure 1). First, microRNA gene is transcribed by RNA polymerase II into primary microRNA (pri-miRNA) of various lengths (usually 1000–3000 nucleotides). Second, the pri-miRNA is cleaved by the ribonuclease (RNase) III Drosha-DGCR8 nuclear complex into hairpin RNA of 60–100 nucleotides in length named precursor miRNA (pre-miRNA). Third, pre-miRNA is transported from the nucleus to cytoplasm by exportin-5, and further cleaved by the RNase III enzyme Dicer into imperfect microRNA duplex about 22 nucleotides in length. Finally, the microRNA duplex unwinds and the mature single-stranded microRNA incorporates into the RNA-induced silencing complex (RISC). Through Watson-Crick complementarities between the seed sequence of microRNA (positions 2–8) and the 3′ untranslated region (UTR) of its target messenger RNAs (mRNAs), microRNA recognizes and recruits mRNAs onto RISC. The resulting microRNA-mRNA interaction prevents mRNA translation or enhances RNA degradation in either case, altering protein output [1].

Figure 1:

microRNA biogenesis and action mechanisms.

microRNA genes are initially transcribed as primary microRNAs (pri-miRNAs), which are then processed by Drosha/DGCR8 complex into hairpin-structure microRNA precursors (pre-miRNAs). The precursors are transported from nucleus into cytosol, where Dicer further cleaves them into short microRNA duplex. The duplex unwinds, and the single-stranded mature microRNA loads onto the RNA-induced silencing complex to exert posttranscriptional regulation on target genes. The interaction of microRNA with 3′ untranslated region (UTR) of target messenger RNA based on sequence complementarity is considered as the canonical mechanism. Targeting other regions such as coding sequence or 5′UTR has also been reported.

Comparing with other types of novel non-coding RNAs, functioning mechanisms of microRNAs are well characterized. In the canonical way, microRNAs binds to 3′-untranslated region of the target mRNA based on sequence complementarity and regulates protein output via promoting mRNA degradation or interfering with the translation process [11]. The sequence complementarity does not need to be perfect; the seed sequence, in most cases situated at positions 2–7 from the 5′ end of microRNA, determines the target identification. As such, one microRNA could target tens or hundreds of mRNA candidates, and consequently affects various types of physiological activities. For example, miR-15a and miR-16 regulates B-cell survival; miR-181 regulates B-cell lineage fate; miR-125b and let-7 are involved in cell proliferation; miR-430 controls brain patterning; miR-375 regulates insulin secretion in pancreatic cells; and miR-143 regulates adipocyte development [12]. Of note, although the canonical mechanism applies to most cases of regulation by microRNAs, multiple non-canonical mechanisms have been reported. For instance, the fact that miR-29b is localized in nucleus challenged the canonical mechanism, which could only explains interaction occurring in the cytoplasm [13]; some microRNAs such as miR-10a also bind to 5′-untranslated region or gene body to modulate target gene expression [14]; miR-373 interacts with CDH1 promoter and up-regulates its transcription [15]; miR-369-3p interacts with AU-rich elements in TNF mRNA and recruits the protein complex composed of Ago2 and FXR1 to promote TNF translation during cell cycle arrest [16]; circulating miR-21 and miR-29a act as agonists of Toll-like receptors (TLRs) in the endosome of the recipient cells and modulate secretion of inflammatory cytokines [17].

microRNAs and human cancer

Despite the discovery of microRNAs lin-4 and let-7 in Caenorhabditis elegans in 1993 and 2000, respectively, the importance of microRNAs in human disease was not recognized. In 2002, George Calin and Carlo Croce reported the first microRNA-cancer link, discovering that a microRNA cluster, miR-15a/16-1 is frequently lost in chronic lymphocytic leukemia (CLL) [18]. This finding spurs immediate interest in the scientific society. Following this pioneer study, the role of miR-15a/16-1 as tumor suppressor was proved, and several other tumor suppressor microRNAs including let-7 family and miR-34 were demonstrated [8]. These tumor suppressor microRNAs perform their function by reducing levels of oncogenic protein-coding genes. For instance, the miR-15a/16-1 targets the BCL2, an oncogene associated with apoptotic cell death; let-7 reduces the levels of oncogenes KRAS and MYC; miR-34 mediates p53 signaling by targeting several important oncogenes CDK4, MYC and MET [12].

Studies also revealed several oncogenic microRNAs including the miR17-miR-92 cluster, miR-21, and miR-155 [19]. The miR-17/miR-92 cluster contains several well-studied oncogenic microRNAs and is frequently activated in lymphoma, lung cancer, breast cancer, gastric cancer, colon cancer, and pancreatic cancer [20]. These microRNAs are direct MYC targets and they work synergistically with the MYC oncogene in promoting B cell lymphoma in mice [21]. As a typical oncogenic microRNA, miR-21 expression is upregulated in many cancer types [8]. Overexpression of miR-21 induces pre-B cell lymphoma in mouse model, stimulates lung carcinogenesis by activating the MAPK signaling, and promotes metastasis of colorectal cancers by regulating PDCD4 [22]. Similarly, miR-155 alone was shown to induce lymphoblastic leukemia or high-grade lymphoma in transgenic mice [23]. These studies suggest that microRNAs could be the driving force of carcinogenesis.

Further studies demonstrated the functional role of microRNAs in cancer metastasis. For instance, a panel of microRNAs including miR-9, miR-10b, miR-103, miR373, and miR-520c has been shown to promote cancer metastasis in mouse models [24]. microRNAs including miR-31, miR-34a, miR-126, miR-200, miR-206, and miR-335 were able to block cancer metastasis [24]. In our own study, we identified via screening that miR-224 is a promoter in metastasis of colorectal cancer [25]. Forced expression of miR-224 dramatically increased cancer metastasis from colon to distant organs of lung and liver in orthotopic mouse models. We identified SMAD4 as a direct miR-224 target, as well as a mediator of miR-224′s pro-metastatic effect [25]. Among the metastasis-associated microRNAs, miR-126 is interesting because of its unique effect in impairing metastatic colonization [26]. This was achieved by suppressing recruitment of endothelial cells in the tumor microenvironment into the metastatic site through its targets IGFBP2, PITPNC1, and MERTK [26].

However, the concept of microRNAs as tumor suppressors and oncogenes should not be stereotyped. Depending on the cellular contexts, the same microRNA could play oncogenic role in one tumor type and act as a tumor suppressor in other(s). For instance, miR-200 is considered as a suppressor of tumor invasion because of its inhibitory effect on epithelial-mesenchymal transition (EMT) by targeting ZEB1 and ZEB2 [27], [28]. On the other hand, miR-200 was shown to promote colonization of circulating cancer cells in the distant metastatic organs [29]. Therefore, miR-200 exhibits oncogenic or tumor suppressor function, in different metastatic steps. A single microRNA could also have distinct function in different cancer types. For example, miR-221 targets the KIT oncogene to inhibit inhibits erythroleukemic cell proliferation [30]. In the scenario of hepatocellular carcinoma, miR-221 was shown to regulate DDIT4, a DNA damage-inducible transcript, to promote liver carcinogenesis [31]. The functions of a miRNA should thus be carefully examined and determined in specific biological conditions.

The plethora of reports argue that microRNAs are master regulator of cancer biology, and deregulated expression of microRNAs affects all hallmarks of cancer [32] (Figure 2): (1) sustaining proliferative signaling (miR-21, let-7) [33], [34]; (2) evading growth suppressors (miR-221 and miR-222) [35]; (3) resisting cell death (miR-34a, miR-15a/16-1) [36], [37]; (4) enabling replicative immortality (miR-29, miR-19b) [38], [39]; (5) inducing angiogenesis (miR-210, miR-17-92 cluster) [40], [41]; (6) activating invasion and metastasis (miR-10b, miR-224) [25], [42]; (7) reprogramming energy metabolism (miR-23, miR-103) [43], [44]; (8) evading immune destruction (miR-34a, miR-124a) [45], [46].

Figure 2:

microRNAs and cancer hallmarks.

microRNAs are functionally associated with all aspects of cancer hallmarks. Typical examples of oncogenic or tumor suppressor microRNAs for each hallmark are shown.

Aberrant microRNA expression in human cancer

Following the initial discovery of low miR-15a/16-1 expression in CLL, frequent deregulation of microRNA expression in many cancer types has been demonstrated. There are several mechanisms underlying the aberrant expression of microRNA in cancer. First, as demonstrated initially, microRNAs located in the frequently altered genomic regions could lost their expression due to chromosomal deletion or increase their expression as result of chromosomal gains. By comprehensive mapping of microRNAs in human genome, it had been discovered that microRNA genes are frequently located at cancer associated fragile sites in the genome, i.e. the chromosomal regions prone to deletions or amplifications [47]. For instance, the 7 mb genomic loci of 3p21.1 – 21.2, where let-7g and miR-135-1 genes are located, is frequently deleted in lung and breast cancer; miR-145 and miR-143 genes at 5q32 is recurrently lost in myelodysplastic syndrome; miR-123 gene at 9q33 is frequently deleted in non-small cell lung cancer; miR-34a-1 and miR-34a-2 at 11q23-q24 are lost in breast and long cancers [47]. On the contrary, many oncogenic microRNAs are genomically amplified in human cancers: the miR-17-92 cluster genes at 13q32-33 are commonly amplified in follicular lymphoma; the miR-21 gene at 17q23 is amplified in several cancer types including neuroblastoma [47].

The expression levels of microRNA are also controlled by other mechanisms. For instance, the miR-34a is transcriptionally activated by p53, and the miR-17-92 cluster is transcriptionally regulated by the MYC oncogene [21], [36]. Epigenetic mechanisms such as CpG island methylation also play important roles in regulating microRNA expression. With a genetic knockout model lacking DNA methyltransferase activity, Lujambio et al. identified that miR-124 is silenced by promoter CpG island hypermethylation in human cancer [48]. This group further reported a DNA methylation signature comprising miR-148a, miR-34b/c, and miR-9 that is associated with the risk of human cancer metastasis [49]. Similarly, the miR-200 family and miR-205 genes, which are key regulators of EMT by targeting ZEB1 and ZEB2, are tightly controlled by promoter CpG islands [50]. Aberrant promoter methylation of these microRNA genes and consequent low microRNA expression are associated with lung cancer initiation and progression [51]. Additionally, defects in microRNA-processing machineries could significantly alter the expression levels of mature microRNAs. Indeed, downregulation of or mutations in microRNA-processing genes including Dicer, TRBP, and Exportin have been observed in many cancer types, and affect a wide range of microRNA expression [52]. While one expects that the general effect of such defects lead to reduced microRNA processing, nucleolin, an element of the Drosha-DGCR8 complex, was shown to promote the processing of several pro-metastatic miRNAs, including miR-221/222, miR-21, and miR-103 [53].

In certain situations, multiple mechanisms work together to regulate a specific microRNA expression. In the case of miR-34a, it is located at 1p36, a frequently lost genomic region in neuroblastoma, transcriptionally regulated by p53 and MYC, and silenced via CpG methylation. The high levels of miR-17-92 cluster microRNAs in cancer are not only related to copy number gains, but also due to transcriptional activation by MYC.

Competitive endogenous RNAs (ceRNAs), which act as microRNA sponges through their microRNA binding sites, could de-repress target genes of the microRNA according to the ceRNA hypothesis [54]. This represents another layer of microRNA regulation in human cancer, with changes in the function but not expression of a microRNA. The debate on the physiological relevance of ceRNA concept is summarized in a recent excellent review [55].

microRNAs and growth factors signaling

As tumor initiation and progression are tightly controlled by growth factors and their intracellular signaling [56], [57], [58], it is not surprising that growth factors regulate microRNA expression while microRNAs orchestrate the expression and function of proteins within growth factor signaling pathways [59]. Over the last decade this bi-directional regulatory networks between microRNAs and signaling complexes coordinated by growth factors receptors have received particular attention revealing some conserved modules [60]. Among RTKs, the IGF and EGF system of ligands, receptors and binding proteins are major player in normal cellular growth and differentiation, as well as in cancer development [60], [61], [62], [63]. Both IGF and EGF systems are organized on three distinct levels: (i) the input layer of ligands, receptors and regulatory proteins of ligand-receptor interaction; (ii) the signal transduction layer, coordinated by adaptors and enzymes of the signaling cascades and the (iii) output layer of transcription factors, controlling the biological response. The receptors bind to their respective ligands with by far the greatest affinity, activating the signaling cascade by two-steps ligand induced dimerization and kinase activation (EGFR) or by inducing conformational changes of the preformed dimers (IGF-1R and IR) [60], [61]. Growth factor signaling pathways in general, including the ones activated by IGF-1 and EGFs are involved in microRNA processing [59], [60]. EGF stimulation of normal mammary epithelial cells dramatically modified microRNA expression profile in less than 1 h [64], [65]. Some of the identified microRNAs were demonstrated to control a defined malignant behavior, for instance miR-15b, upregulated by EGF stimulation, and miR-191, downregulated by EGF, were shown to positively regulates the metastatic phenotype (for extensive review see [60]). Similarly, in the case of IGF-1R stimulation, specific microRNA profiles were found to be associated with signaling activation as related to the malignant phenotype [66], [67]. In breast cancer cell lines, 18-h stimulation with IGF-1 lead to modified microRNA profile with increase of defined oncogenic microRNAs (103/107, 1826, 191, 93) and downregulation of tumor suppressive microRNAs (15b, 98, 195, 200b, let-7c, and let-7g) [67].

On the other hand, the microRNA network regulates the expression of IGF-1, their receptors as well as their regulatory proteins or effectors [68]. For instance miR-486, an IGF-1 inhibitor at transcriptional level was found to be decreased in NSCLC cancer cell lines [69]. In addition several other microRNAs including Let-7, miR-16, miR-122, or miR-196 were demonstrated to control the expression of the IGF-1R in different cancer types (for an extensive review see [68]). Not only the expression of the ligands and receptors are controlled by the microRNA network but also the downstream signaling proteins such as insulin-receptor substrate (IRSs) [70], β-arrestins [71], [72], the IGF-1R ubiquitination enzymes [73], [74], or the components of the P3K/Akt and MAPK pathways [68].

Potential use of microRNA as cancer biomarkers

microRNAs have several advantages as cancer biomarkers [75]. First, microRNAs are stable under harsh environments such as high temperatures, long-term storage, strong acidic or basic conditions, and frequent freeze-thaw. This feature makes possible the detection of microRNA expression from formalin fixed paraffin embedded (FFPE) samples, plasma, serum, urine, and other body fluids. Second, the detection of mature microRNA by either PCR-based method or hybridization-based method is straightforward. Third, the tissue-specific expression pattern of microRNAs makes them also ideal cancer biomarkers.

Cancer of unknown primary origin (CUP) represents around 3%–5% of all newly diagnosed cancers in the United States and remains an unmet medical challenge in cancer diagnosis [76]. The feature of tissue-specific of some microRNAs offers an opportunity to characterize the site and type of cancer origin. Early in 2005, The Golub lab reported that microRNA pattern is more accurate in establishing the right diagnosis than the mRNA criteria, in poorly differentiated metastatic cancer cases with CUP [77]. A large effort of microRNA profiling with 400 FFPE and fresh-frozen samples from Barshack group constructed a microRNA classifier composing 48 microRNAs that can accurately identify origin of cancer tissues [78]. The tissue-specific feature also allows for determination of histological subtypes. For instance, miR-205 has been reported as an accurate marker for squamous non-small cell lung cancer, with high sensitivity (96%) and specificity (90%) [79]. Because of the high performance of microRNA biomarkers in differentiating cancer subtypes, Biotech companies have started developing diagnostic products. Using 8 microRNA biomarkers, mi-LUNG (Rosetta Genomics Ltd.) accurately differentiates the four subtypes of lung cancer (small-cell lung cancer, carcinoid, squamous non-small cell lung cancer, and non-squamous NSCLC) with 94% sensitivity and 98% specificity [80].

Several advantages of microRNAs, including stability and easy detection, allow for potential use of microRNAs in body fluids as convenient and less invasive biomarkers for early cancer detection. Boeri et al. reported that microRNA signatures in plasma are early predictor of lung cancer occurrence, months before the diagnosis by computed tomography [81]. With a large set of lung cancer cohort, Montani et al. further validated that a curated serum microRNA signature, named miR-Test, accurately identify lung cancer cases in high-risk individuals, with 74.9% sensitivity and 74.8% specificity [82]. The diagnostic value of microRNA expression in other body fluids for cancer detection has also been explored. Using Sputum samples, Xing et al. showed that a panel of miR-21, miR-31, and miR-210 is able to differentiate malignant and benign solitary pulmonary nodules, with both high sensitivity and specificity (>80%), in all three cohorts of samples [83]. Hanke et al. reported that the ratio of miR-126 to miR-152, and the ratio of miR-182 to miR-152 are indicators of presence of bladder cancer [84].

Circulating microRNAs are also gold mines for biomarkers indicative of prognosis and therapeutic response. A four-microRNA signature comprising miR-486, miR-30d, miR-1 and miR-499 in the serum independently predicts survival of non-small cell lung cancer, showing potential of microRNAs as noninvasive prognosis predictor [85]. In a recent study, high expression levels of miR-155 were significantly associated with therapeutic resistance of patients with diffuse large B-cell lymphoma to the rituximab plus cyclophosphamide, doxorubicin, vincristine, and prednisone treatment [86]. These microRNA indicators for therapeutic response may help stratify patients for anticancer treatment to avoid unnecessary treatments.

Potential use of microRNA for cancer therapeutics

As cancer drivers, key promoters, or essential guardians, the therapeutic potential of microRNAs in cancer certainly attracts much attention [5]. microRNA therapeutic has the ability of targeting multiple protein-coding genes from the same pathways at different levels, thus probably preventing compensatory mechanisms such as mutation in the targeted oncogenes that leads to resistance. microRNAs are involved in all hallmarks of cancer, and targeting these microRNAs might offer therapeutic benefits. Identification of specific microRNAs associated with oncogenic IGF-1R signaling may also provide new tools for the molecular-designed treatments of cancer in clinical settings. The function of microRNAs with tumor suppressor activities can be restored with synthetic microRNA mimics, while the activities of oncogenic microRNAs can be blocked by oligonucleotides specifically interfering with the targeted microRNA [5]. Such strategies have been shown to be successful by many in vitro and in vivo studies. More impressively, MRX34, a liposome-based miR-34 mimetic developed by Mirna Therapeutics, has entered the phase of clinical trials for treating patients with liver cancer. The tumor suppressor miR-34a down regulates a number of key oncogenes, including MET, MEK1, CDK4/6, NOTCH, CD44, and immune evasion gene PDL1 [87]. The update on clinical trial with MRX34 American Society of Clinical Oncology (ASCO) Annual meeting 2016 reported that MRX34 has a manageable toxicity profile and evidence of activity in liver cancer, renal cell carcinoma, and melanoma during the phase I study (Hong et al. oral presentation at ASCO annual meeting 2016, abstract # 2508). In September 2016, Mirna Therapeutics voluntarily halted the phase I clinical trial because of multiple immune-related severe adverse events in patients treated with MRX34 (http://www.mirnarx.com/pipeline/mirna-pipeline.html). In another phase I trial in Australia, restoration of miR-16 led to a marked metabolic and radiological response on a patient with malignant pleural mesothelioma [88]. Whether it is an effective and safe therapy for patients with malignant pleural mesothelioma remains to be further evaluated.

Challenges and perspectives

Many challenges exist on the way of translating the preclinical findings into reliable cancer biomarkers and therapeutic agents. First, we still lack understanding on the basic mechanisms of how microRNAs determine their targets, besides their sequence complementarity, in a specific context. In addition, the frequent inconsistency between different reports on plasma microRNA biomarkers is a big concern on furthering such findings to clinical usages [89]. Sample collection time-point, extraction procedure, storage condition, variability in using internal controls, and different ways of data analysis are possible contributors to such discrepancies. microRNA therapeutics is even more challenging in achieving efficient and site-specific delivery of oligonucleotides, while avoiding side effects associated with oligonucleotides and delivery systems [5].

Despite these challenges, it is clear that the impact of microRNAs on cancer biology is tremendous. In two decades, microRNA concept has moved quickly from initial discovery to promising clinical candidates. With standardized, large-size, multiple-centered biomarker study design, it is possible to identify reliable cell-free cancer biomarkers in the near future. Following the initial launch of MRX34, other microRNA-based cancer therapeutics on the pipeline will soon enter the stage of clinical trials. The discovery of enrichment of microRNAs in extracellular vesicles such as exosomes provides further avenue for development of microRNA biomarker. Additionally, the feature of active packaging and release of exosome suggests that this might be an efficient microRNA delivery system for therapeutics [90]. We expect to see in the near future the use of microRNAs as cancer biomarkers, and the use of microRNA therapeutics, in combination with other therapeutic regimens, in cancer treatment.