Nội dung

REVIEW ARTICLE MicroRNAs – micro in size but macro in function Sunit K. Singh1,2, Manika Pal Bhadra3, Hermann J. Girschick2 and Utpal Bhadra4 1 2 3 4 Section of Section of Centre for Functional Infectious Diseases and Immunobiology, Centre for Cellular and Molecular Biology, Hyderabad, India Infectious Diseases, Immunology and Pediatric Rheumatology, Children’s Hospital, University of Wuerzburg, Germany Chemical Biology, Indian Institute of Chemical Technology, Hyderabad, India Genomics and Gene Silencing Group, Centre for Cellular and Molecular Biology, Hyderabad, India Keywords Dicer; microRNA; miRNA and cancer; miRNA and disease; miRNA and therapeutics; miRNA biogenesis; miRNA function; miRNA inhibitors; small RNA Correspondence S. K. Singh, Section of Infectious Diseases and Immunobiology, Centre for Cellular and Molecular Biology, Uppal Road, Hyderabad 500007, India Fax: +91 40 27160311 Tel: +91 40 27192523 E-mail: sunitsingh@ccmb.res.in MicroRNAs (miRNAs) are endogenous small RNAs that can regulate target mRNAs by binding to their 3¢-UTRs. A single miRNA can regulate many mRNA targets, and several miRNAs can regulate a single mRNA. These have been reported to be involved in a variety of functions, including developmental transitions, neuronal patterning, apoptosis, adipogenesis metabolism and hematopoiesis in different organisms. Many oncogenes and tumor suppressor genes are regulated by miRNAs. Studies conducted in the past few years have demonstrated the possible association between miRNAs and several human malignancies and infectious diseases. In this article, we have focused on the mechanism of miRNA biogenesis and the role of miRNAs in human health and disease. (Received 30 June 2008, revised 30 July 2008, accepted 1 August 2008) doi:10.1111/j.1742-4658.2008.06624.x Introduction Small RNAs exhibit a wide spectrum of biological functions. There are many classes of small RNAs, such as microRNAs (miRNAs), small interfering RNAs (siRNAs), repeat associated small interfering RNAs [1], small nuclear RNA, small nucleolar RNA, Piwiinteracting RNA [2] and transacting short interfering RNA [3]. miRNAs are single-stranded RNAs of 19–25 nucleotides in length originating from endogenous hairpin- shaped transcripts [4]. These miRNAs interact with their target mRNAs by base pairing, which could lead to translational repression; decapping, deadenylation and ⁄ or cleavage of target mRNA. The first known miRNA, lin-4, was discovered in 1993 by Ambros and coworkers in the nematode Caenorhabditis elegans [5,6]. The lin-4 gene plays a role in the developmental timing of stage-specific cell lineages in C. elegans. Later on, lin-4 was found to encode a 22-nucleotide noncoding RNA that negatively regulates the translation of lin-14. A few years later, another small RNA, Abbreviations AD, Alzheimer’s disease; AGO, argonaute; Ab, amyloid b-peptide; Dcp, decapping enzyme; DCR, Dicer; DGCR8, DiGeorge syndrome critical region gene 8; dsRBD, double-stranded RNA-binding domain; eIF, eukaryotic translation initiation factor; ES, embryonic stem; Exp-5, exportin-5; IRES, internal ribosome entry site; KSHV, Kaposi sarcoma herpes virus; LNA, lock nucleic acid; miRISC, microRNA-containing RNA-induced silencing complex; miRLC, microRNA-containing RNA-induced silencing complex loading complex; miRNA, microRNA; miRNP, microRNA ribonucleoprotein; P-body, processing body; Pol II, RNA polymerase II; Pol III, RNA polymerase III; pre-miRNA, precursor microRNA; pri-miRNA, primary microRNA; RISC, RNA-induced silencing complex; RLC, RNA-induced silencing complex loading complex; RNAi, RNA interference; siRISC, small interfering RNA-containing RNA-induced silencing complex; siRNA, small interfering RNA. FEBS Journal 275 (2008) 4929–4944 ª 2008 The Authors Journal compilation ª 2008 FEBS 4929 MicroRNAs and their roles S. K. Singh et al. let-7, was reported as an additional regulator of developmental timing in C. elegans [7]. Similar to lin-4, let-7 also functions by binding the 3¢-UTR of lin-41 and lin-57 to inhibit their translation. To date, 678 human miRNAs have been characterized in the Sanger miRBase sequence database [8], and many more are still to be identified. Approximately 50% of known human miRNAs are found in clusters [9,10]. The clustered miRNAs are often related to each other, but can also be unrelated. Clustered miRNAs may be functionally related in terms of targeting the same gene or different genes in the same biochemical pathway. Most mammalian miRNA genes have been reported to be located in defined transcription units Fig. 1. MicroRNA biogenesis and function. The miRNA gene is transcribed by Pol II into a pri-miRNA in the nucleus. The pri-miRNA is processed into pre-miRNA by the RNase III enzyme Drosha. The pre-miRNA is exported to the cytoplasm with the help of Ran-GTP cofactor and Exp-5. The miRNA duplex is cleaved from the pre-miRNA by the RNase III enzyme Dicer and TRBP. Helicase unwinds the mature miRNA duplex. Either each strand of the miRNA pair or only one strand of mature miRNA can be incorporated into miRISC. miRNAs bound to miRISC mediate the degradation or translational inhibition of their target mRNAs. 4930 FEBS Journal 275 (2008) 4929–4944 ª 2008 The Authors Journal compilation ª 2008 FEBS S. K. Singh et al. and intergenic regions [11]. Studies have revealed that miRNAs have key roles in diverse processes such as developmental control, hematopoietic cell differentiation, neural development, apoptosis, cell proliferation and organ development. In this review, we discuss the mechanism of miRNA biogenesis and the roles of miRNA during development and different pathological states. miRNA biogenesis miRNA biogenesis includes miRNA transcription in the nucleus, the export of miRNAs from the nucleus to the cytoplasm, and subsequent processing and maturation in the cytoplasm (Fig. 1). In most cases, the transcription of miRNA genes is mediated by RNA polymerase II (Pol II), resulting in long primary miRNA (pri-miRNA) transcripts with a fold-back structure comprising a stem loop along with flanking segments [12]. A few recent reports have shown the involvement of RNA polymerase III (Pol III) in miRNA transcription [13]. The sequence of the miRNA remains embedded in the arms of the stem loop. The pri-miRNA contains the 7-methylguanosine cap and a poly(A) tail, which is unique for Pol II transcripts, similar to mRNAs [12,14]. However, the cap and poly(A) tail are removed during miRNA processing. miRNA promoters have been identified in many studies [9,15,16], and reported to have typical Pol II elements such as a TATA box [17], although the recent report of Borchert et al. [13] suggests that members of the human chromosome 19 miRNA cluster (miR-515-1, miR-517a, miR-517c and miR-519a-1) are interspersed among Alu repeats and expressed through Pol III. The processing of pri-miRNAs into final mature miRNAs occurs in a stepwise fashion, which is discussed in detail in subsequent sections of this article. Enzymatic machinery involved in miRNA biogenesis and maturation Drosha In humans, the generation of precursor miRNA (pre-miRNA) from the pri-miRNA transcript takes place exclusively in the nucleus, through the action of the microprocessor complex, composed of the RNase III enzyme Drosha and the double-stranded RNA-binding domain (dsRBD) protein DiGeorge syndrome critical region gene 8 (DGCR8), into 70– 80 nucleotide pre-miRNAs [18,19]; this is followed by maturation of miRNA in the cytosol. The stem loop structure of pri-miRNAs is cleaved in the MicroRNAs and their roles nucleus by Drosha during the generation of pre-miRNA. This process is known as cropping. Drosha forms a large microprocessor complex of  650 kDa along with the dsRBD protein DGCR8 in humans [20] and a  500 kDa complex along with the dsRBD protein Pasha in flies (Drosophila melanogaster) [18,21,22]. In contrast to the siRNA pathway, the miRNA pathway is initiated in the nucleus [23]. The cleavage by Drosha generates a pre-miRNA hairpin bearing two nucleotide 3¢-overhangs. Precursor miRNAs are exported to the cytoplasm from the nucleus by exportin-5 (Exp-5) in the presence of Ran-GTP as a cofactor (Fig. 1) [22,24,25]. It is important to realize that many human miRNA genes remain located in intronic regions of coding genes, so their biogenesis remains coupled with mRNA splicing [12,26]. Drosha releases the pre-miRNA from the intron shortly before splicing, allowing the generation of both RNA species at the same time. The precision of Drosha–DGCR8 cleavage is very important for miRNA maturation. Any shift in the position of the Drosha cut, even by a single nucleotide on the primiRNA, will affect the position of Dicer cleavage. A shift in the Dicer cleavage site could result in different 5¢-ends and 3¢-ends in the mature miRNA. This type of nucleotide shift may invert the relative stability of the 5¢-end of the miRNA strand and of the other associated strand, which is opposite to the miRNA strand. Such a shift could result in the selection of the wrong strand as the mature miRNA. Even if the stability remains unchanged and the correct strand is loaded into the RNA-induced silencing complex (RISC), then the shift in the 5¢-end of the miRNA will change the position of the seed sequence (2–8 nucleotides of miRNA, which often match the target mRNA very closely), which could lead to a change in its target mRNA [27]. The RISC is a multiprotein complex that cleaves specific mRNAs, and that is targeted for degradation by homologous dsRNAs during the process of RNA interference. This complex plays a very important role in gene regulation by miRNAs and siRNAs. There is an interesting mechanism that determines the precision of cleavage by the Drosha–DGCR8 complex to generate pre-miRNA transcripts from pri-miRNA transcripts. Some structural features of the RNA have been shown to be involved in determination of the Drosha cleavage site [27]. The ssRNA segments flanking the base of the stem loop are crucial for Drosha cleavage [28]. The deletion of single-stranded regions or their conversion to dsRNA greatly impairs the conversion of pri-miRNA to pre-miRNA [28]. In a recent report, Davis et al. [29] have shown the role of SMAD protein in Drosha- FEBS Journal 275 (2008) 4929–4944 ª 2008 The Authors Journal compilation ª 2008 FEBS 4931 MicroRNAs and their roles S. K. Singh et al. mediated miRNA maturation. Transforming growth factor-b and bone morphogenetic protein signaling have been reported to promote the rapid increase in expression of mature miR-21 by promoting the processing of primary transcripts of miR-21 (pri-miR-21) into precursor miR-21 (pre-miR-21) by the Drosha (also known as RNASEN) complex [29]. Transforming growth factor-b-specific and bone morphogenetic protein-specific SMAD signal transducers are recruited to pri-miR-21 in a complex with RNA helicase p68 (also known as DDX5), a component of the Drosha microprocessor complex [29]. Thus, SMAD protein plays an important role in Drosha-mediated miRNA maturation [29]. Export and import of miRNAs between the nucleus and cytoplasm Exp-5 is a member of the karyopherin family of nucleocytoplasmic transport factors, and plays a role in the export of miRNAs from the nucleus to the cytoplasm [30]. The function of Exp-5 is dependent on the GTPbound form of Ran cofactor for specific binding to its export substrate in the cell nucleus. This process involves the hydrolysis of Ran-GTP to Ran-GDP by the cytoplasmic Ran GTPase-activating protein [31]. The role of Exp-5 in nucleocytoplasmic transport was verified by using RNA interference (RNAi). In the event of Exp-5 depletion by RNAi, the level of mature miRNAs goes down but pre-miRNA does not accumulate in the nucleus. The lack of accumulation could be due to instability of pre-miRNA. This suggests the possibility that interaction of pre-miRNA with Exp-5 is required for the stability of pre-miRNA [32]. Exp-5 has been reported to recognize the ‘minihelix motif’ of pre-miRNA, which consists of a > 14 bp stem and a short 3¢-overhang. Hwang et al. recently reported that a hexanucleotide element directs the process of nuclear import rather than export of miR-29b. In contrast to most of the animal miRNAs, miR-29b has been reported to be predominantly localized in the nucleus [33]. The special hexanucleotide terminal motif (AGUGUU) acts as a transferable nuclear localization element of miR-29b, and is responsible for the nuclear enrichment of miR-29b. These RNAs may prove to be useful tools for manipulation of gene expression in the nucleus. It is supposed that miR-29b could have a role in regulation of the transcription or splicing events of target transcripts. This role of miR-29b is quite unique and is different from the routine translational regulatory functions performed by other miRNAs [33]. 4932 Role of Dicer in miRNA maturation Dicer is an ATP-dependent multidomain enzyme of the RNase III family, and has been reported to be involved in cleavage of double-stranded siRNA and miRNA. Dicer was initially identified in Drosophila [34] and has been subsequently reported in humans, plants and fungi. The mechanism of recognition of the pre-miRNA by cytoplasmic Dicer is not known [35]. In the cytoplasm, the pre-miRNAs are processed into  22-nucleotide duplex miRNAs by the RNase III enzyme Dicer (Fig. 1). Some organisms have a single Dicer gene [36–39], whereas others have many [40,41]. In species with several Dicers, different homologs are required for different functions [40,42,43]. Two Dicer homologs (DCR1 and DCR2) have been reported in Drosophila. DCR1 processes pre-miRNA, whereas DCR2 processes long dsRNA in Drosophila [43–45]. The only Dicer gene in C. elegans, DCR1, is required for the processing of both the long dsRNA and pre-miRNAs. Dicer cleavage results in the release of a duplex with mature miRNA in one of the strands of the stem loop. Both arms of the pre-miRNA stem loop structures are imperfectly paired, containing G:U wobble pairs and single nucleotide insertions. These imperfections cause one strand of the duplex to be less stably paired at its 5¢-end [27]. The conversion from dsRNAs to ssRNAs is a complex process, involving several RNA–protein and protein–protein interactions. RISC loading complex (RLC) is an RNA–protein complex that initiates the formation of the RISC. The RLC puts a small RNA duplex in the correct orientation for subsequent RISC assembly [35]. The small RNAs (siRNAs and miRNAs) in the RLC remain ready to be unwound for functional RISC assembly. The siRISC loading complex (siRLC) of Drosophila contains a DCR2– R2D2 heterodimer and an siRNA duplex. R2D2 has been reported to be a DCR2 stabilizer as well as the asymmetric sensor for setting the siRNA orientation for RISC assembly [35]. Detailed information on miRISC loading complexes (miRLCs) is not available. In a recent report, MacRae et al. [46] have demonstrated the assembly of human RLC in vitro from purified components without any cofactors or chaperones. They demonstrated that reconstituted RLC maintains the endogenous RLC functional activities of dicing, slicing, guide-strand selection and argonaute (AGO)2 loading [46]. Dicer interacts with the dsRNA-binding protein partner, the TAR RNA-binding protein (TRBP), in humans [RDE4 in C. elegans and Loquacious (Loqs) in Drosophila], which probably bridges the initiation and effector FEBS Journal 275 (2008) 4929–4944 ª 2008 The Authors Journal compilation ª 2008 FEBS S. K. Singh et al. steps of miRNA action [47–49]. DCR binds with high affinity to the ends of dsRNAs bearing two-nucleotide 3¢-overhangs, which results in unwinding of duplexes. The thermodynamic properties of siRNA–miRNA duplexes play a critical role in determining molecular function and longevity [50]. The unwinding of the duplex strands starts at the ends with the lowest thermodynamic stability. The relative stabilities of the base pairs at the 5¢-ends of two strands determine the fate of the strand, which has to participate in the RNAi pathway [51]. Along with the thermodynamic stabilities, a role of proteins such as R2D2 has also been reported during the strand selection process. The orientation of the DCR2–R2D2 protein heterodimer on the siRNA duplex determines the siRNA strand, which has to associate with the core RISC protein AGO2 in Drosophila [52]. The exact mechanism by which R2D2 guides the asymmetric assembly of the RISC in Drosophila is not known. Dicer has an RNA helicase domain to cleave the dsRNA. In general, the miRNA strand, which has its 5¢-terminus at the lowest thermodynamic stability, acts as the mature miRNA (guide strand), and the other strand (passenger strand) is degraded. However, a recent report has shown that both strands could be coaccumulated as miRNA pairs in some tissues, and subjected to strand selection in other tissues [53]. Ro et al. [53] also reported that both strands of the miRNA pair can target equal numbers of genes, and were able to suppress the expression of their target genes. This study provided evidence for a novel mechanism involved in tissue-dependent miRNA biogenesis and miRNA target selection [53]. Mature miRNAs are incorporated into the effector complexes, known as miRNP (microRNA ribonucleoprotein), mirgonaute, or miRISC. The identification of the target by the RISC is based on the complementarity between mature miRNA and the mRNA. The degree of complementarity decides whether the complex has to undergo endonucleolytic cleavage of target mRNA or translational repression. In contrast to miRNAs, siRNAs are often synthesized in vitro or in vivo from viruses or repetitive sequences. siRNAs have been reported to be involved in antiviral defense, and also in protecting the genome against disruption by transposons. The presence of the selective AGO protein family is one of the several common features of siRISC and miRISC. AGO proteins in the RISC AGO proteins are well conserved in diverse organisms [54], and constitute a large family involved in develop- MicroRNAs and their roles mental regulation in eukaryotes. Several AGO homologs have been reported in eukaryotic organisms, such as eight in humans [55], five in Drosophila [54], 27 in C. elegans [56] and only one in fission yeast [36,56]. These homologs are characterized by the presence of two domains, PAZ (Piwi ⁄ Argonaute ⁄ Zwille) and PIWI. The PAZ domain of AGO proteins binds to the 3¢-end of the ssRNA, possibly by recognizing the 3¢-overhangs [57,58]. AGO proteins are the core components of the RISC in different organisms. Different AGO proteins specify distinct RISC functions. Cofractionation studies in Drosophila have shown that AGO2 cofractionates and remains functionally associated with DCR2, whereas AGO1 remains functionally associated with DCR1 [44,59]. These observations verify that DCR1 is involved in miRNA maturation, whereas DCR2 is involved in initiation of RNAi in Drosophila [43,44]. Although miRNAs and siRNAs have distinct biogenesis pathways in Drosophila, they have a common sorting pathway, which partitions them into AGO1containing or AGO2-containing effector complexes [60]. In contrast to Drosophila, humans and C. elegans contain only one Dicer, which initiates the formation of both siRISCs and miRISCs. In the case of humans, different AGO proteins (AGO1 to AGO4) have been reported to be involved during RISC assembly, but only AGO2-associated RISCs have been reported to be involved in the cleavage of target mRNA. Therefore, AGO2 is also called slicer argonaute [61,62]. Slicer activity has been reported in the PIWI domain of AGO proteins, on the basis of mutagenesis studies [61]. Specific amino acid residues of the PIWI domain of AGO2 are essential for slicer activity in AGO2 proteins of human and Drosophila [35]. Processing bodies (P-bodies) and their biological function It was thought that once mRNAs finish their job, enzymes in the cytoplasm simply break them down. Several groups reported that most of this degradation occurs in P-bodies (processing bodies) or glycine-tryptophan or decapping enzyme (Dcp) bodies. P-bodies are found as discrete cytoplasmic bodies in yeast and mammals. The conservation of P-bodies from yeast to mammals suggests their important role in the cytoplasmic function of eukaryotic mRNA. P-bodies include the Dcp1p ⁄ Dcp2p, activators of decapping, Dhh1p (referred to as RCK in mammals), Pat1p, Lsm1-7p, Edc3p and the 5¢–3¢-exonuclease Xrn1p [63–66]. P-bodies have been reported as the sites for decapping FEBS Journal 275 (2008) 4929–4944 ª 2008 The Authors Journal compilation ª 2008 FEBS 4933 MicroRNAs and their roles S. K. Singh et al. and degradation of mRNAs. In yeast, the major pathway of mRNA turnover is initiated with the shortening of a 3¢-poly(A) tail, by a process called deadenylation. Deadenylated transcript acts as a substrate for the Dcp1p ⁄ Dcp2p decapping complex, for removal of the 5¢-cap structure. The decapping process exposes the transcript to degradation by the 5¢ fi 3¢-exonuclease Xrn1p [67,68]. Alternatively, transcripts can be degraded in the 3¢ fi 5¢ direction following deadenylation in exosomes, by a conserved complex of 3¢-to-5¢exonucleases. Several observations suggest that the processes of mRNA decapping and translation are inversely related. Mutations in different translation initiation factors result in decreased rates of translation along with increases in the rate of mRNA decapping [69,70]. Recent reports have demonstrated that mRNA subjected to miRNA repression accumulates in Pbodies. P-bodies contain untranslated mRNAs and can serve as sites of mRNA degradation. This suggests that RISC proteins direct the mRNA to P-bodies, possibly for storage. So, the P-bodies do not just degrade mRNA, but also temporarily sequester them away from the translation machinery. Parker and coworkers have recently revealed the localization of AGO proteins in mammalian P-bodies. They found that mRNAs targeted for translational repression by miRNAs become concentrated in P-bodies in an miRNA-dependent manner [71]. This study provides a strong link between miRNAs and P-bodies, and suggests that translation repression by the RISC delivers mRNAs to P-bodies [71]. Other studies have also demonstrated that about 20% of let-7-repressed reporter mRNAs and 20% of fluorescently labeled microinjected let-7 miRNA colocalized with visible P-body structures [72,73]. The involvement of Pbodies in miRNA-based repression requires further investigation to determine the fraction of translationally repressed mRNAs, and the miRNAs localized in P-bodies. Ways to handle translational activities An miRISC represents an effector complex that mediates miRNA functions inside cells. The guide miRNAs are perfectly complementary to either the coding region or 3¢-UTR of target mRNA in plants [74]. In most cases, plant miRISCs can mediate mRNA degradation. The perfect complementarity between mature miRNA and target mRNA has not been reported in animals and humans, except for the HoxB8 gene in mice, which can be cleaved by miR-196 despite imperfect sequence identity [75]. Nucleotides 2–8 of miRNA, 4934 known as the ‘seed region’, do often match very closely to the target mRNA, and are considered to comprise the most critical region for selecting targets. The miRNAs sharing common seed sequences are grouped into miRNA families. These miRNAs possibly have overlapping targets and are considered to be redundant [33]. miRNAs handle the translational activities by mediating pretranslational, cotranslational or post-translational gene silencing. In eukaryotic translation, the step of initiation starts with the recognition of the 5¢-terminal cap structure (m7Gpp) of mRNA by the eIF4E subunit of the eukaryotic translation initiation factor (eIF), eIF4F and eIF4G [76]. The interaction of eIF4G with polyadenylate-binding protein 1 and eIF4E results in stimulation of translational initiation [76]. However, some cellular and viral mRNAs initiate translation without the involvement of the m7G cap and eIF4E. In such cases, the 40S ribosomes are recruited to mRNA through the internal ribosome entry site (IRES) [76,77]. Several reports have demonstrated that translation of m7G-capped mRNAs, but not of mRNAs containing the IRES, is repressed by miRNAs [72,76]. In such cases, AGO2 and related proteins might compete with eIF4E for m7G binding and thus prevent the translation of capped, but not IRES-containing, mRNAs [78]. However, other reports demonstrate the interaction of miRNA with the ORFs of genes whose translational activities are governed by IRES-mediated translational events [79,80]. MiRISCs can repress translational events at both initiation and postinitiation levels. MiRISCs are also known to increase cotranslational degradation of nascent proteins, reduce the elongation rate of translation, and increase the rate of mRNA deadenylation [72,73,80–82]. It is not well understood whether miRNAs always target the same or different steps of translational events under various physiological conditions [73]. In recent reports, it has been shown that the miRNA-mediated repression can be effectively reversed or prevented [83–85], and miRNPs can act as translational activators [86]. Cationic amino acid transporter-1 (CAT 1) mRNA has been reported to be translationally repressed by the liver-specific miRNA miR-122 in human hepatoma cells, and accumulates in cytoplasmic P-bodies. However, amino acid starvation results in the release of CAT 1 mRNA from P-bodies and its recruitment to polysomes [76]. APOBEC3G (apolipoprotein B mRNA editing enzyme catalytic polypeptide like 3G) has also been reported to interfere with miRNA action by altering the distribution of target messages between P-bodies and polysomes [87]. FEBS Journal 275 (2008) 4929–4944 ª 2008 The Authors Journal compilation ª 2008 FEBS S. K. Singh et al. miRNAs as developmental regulators miRNAs play important roles in the regulation of stem cells, organ differentiation and developmental timings [88]. Dicer mutational and knockout studies have shown a defect in miRNA biogenesis. Mutations in Dicer genes result in reduced levels of miRNAs. Knockout of mouse dcr1 results in depletion of embryonic stem (ES) cells. Dicer-deficient ES cells are viable, but do not form mature miRNAs, and fail to differentiate in vitro and in vivo [89,90]. Mouse and human ES cells express a specific set of miRNAs, which are downregulated upon differentiation into embryoid bodies [91]. Dicer mutant zebra fish embryos have been reported to develop normally for about 8 days postfertilization, but the process of development ceases after 8 days, when embryos run out of maternal Dicer. Giraldez et al. generated maternal zygotic Dicer mutants of zebra fish by a germ cell replacement technique to eliminate the maternal contribution of the Dicer gene. In maternal zygotic DCR mutants of Dicer knockout zebra fish, the pre-miRNAs are not processed into mature miRNAs [92], and show morphogenesis defects during gastrulation, brain formation and neural differentiation. Loss of Dicer leads to the defects in positioning as well as defasciculation of axons. These observations suggest that miRNAs are not only essential for cell fate determination and early patterning, but are also essential for subsequent later steps in early embryonic development in zebra fish [92]. The differential pattern of miRNA expression has also been reported during different stages of development. Most of the miRNAs show highly tissuespecific expression during the late stages of development [25,93]. Injection of miR-430 into maternal zygotic Dicer mutant zebra fish embryos rescues the brain morphogenesis defects and to some extent the other neuronal defects, indicating the importance of miR-430 in regulation of morphogenesis in the zebra fish [92]. The Dicer knockout studies have provided much strong evidence regarding the role of miRNAs in different species, but these results should be interpreted with caution, due to the role of Dicer in other functions, such as heterochromatin formation and chromosome segregation. miRNAs in health and disease miRNAs have already been implicated in a number of diseases, and both miRNA inhibition and activation show great promise in the treatment of various types of cancer, and viral and metabolic diseases. Aberrant MicroRNAs and their roles gene expression is the main reason for miRNA dysfunction in cancer, which results in abnormal mature ⁄ precursor miRNA expression in tumor samples [94]. MicroRNA germline and somatic mutations or polymorphisms in the protein-coding mRNAs targeted by miRNA also contribute to cancer predisposition, initiation or progression [94] The expression patterns of different miRNAs in various types of human tumor have been studied extensively [95]. Significant downregulation of most of the miRNAs has been reported in various tumors as compared to normal tissues [95]. Amplification, rearrangement and deletions have been reported among various miRNA location sites in cancer patients. This provides a clue about the association between miRNA and cancer pathogenesis [96]. The dysregulation of miRNA expression has been reported in many types of cancer, including Burkitt’s lymphoma [97], colorectal cancer [98], lung cancer [99], breast cancer [100] and glioblastoma [101]. MiR-143 and miR-145 miRNAs are downregulated in colon cancer tissue [98]. Let-7 miRNAs are downregulated in several lung cancers [99]. Overexpression of miRNAs with antiapoptotic activity has been reported in cancer cells. The miR-17 cluster (miR-17-5p, miR-17-18, miR-17-18a, miR-17-19b, miR-17-20 and miR-17-92) of miRNAs, located on human chromosome 13q31, has been shown to be associated with antiapoptotic activity. This region of the chromosome (chromosome 13q31) has often been associated with several types of lymphoma and solid tumor [95,102,103]. He et al. [104] also reported a higher level of expression of miR-17 cluster miRNAs in B-cell lymphoma samples. The lymphomas expressing the miRNAs of the miR-17 cluster show a high mitotic index without extensive apoptosis. The high mitotic index without apoptosis suggests that miR-17 cluster miRNAs suppress cell death [104]. It is worth mentioning here that the individual miRNAs of the miR-17 cluster could not accelerate tumor formation individually. This suggests that the oncogenic effect requires a cooperative interaction between the miRNAs in the cluster. The miRNA miR-21 with an antiapoptotic function was found to be overexpressed in breast cancer tissue [100], glioblastoma tumor tissues and cell lines [101]. Inhibition of miR-21 in a glioblastoma cell line resulted in caspase activation and enhanced apoptosis [95,101]. miRNAs with proapoptotic activity are likely to function as tumor suppressor genes, and have been reported to be underexpressed in cancer cells. The family of let-7 miRNAs falls into this category. RAS gene dysregulation has been reported among lung cancer patients. The let-7 miRNA has been demonstrated to FEBS Journal 275 (2008) 4929–4944 ª 2008 The Authors Journal compilation ª 2008 FEBS 4935 MicroRNAs and their roles S. K. Singh et al. regulate the RAS oncogenes. The level of RAS protein was inversely correlated with let-7 miRNA levels in lung cancer samples, without any change in RAS mRNA levels [95,105]. miR-15a and miR-16-1 are reported to have tumorsuppressing activity in B-cell chronic lymphocytic leukemia. MiR-15a and miR-16-1 are located on human chromosome 13 in a region that is frequently deleted in B-cell chronic lymphocytic leukemia. The conserved target site for miR-15a and miR-16-1 has been found in the 3¢-UTR of the antiapoptotic gene bcl2. Overexpression studies of miR-15a and miR-16-1 have shown reduced expression of Bcl2 protein in a leukemic cell line [95,100,106]. Ma et al. recently reported the association of miRNA with cancer cell invasiveness and the ability to metastasize. Investigators named the miR-10b prometastatic miRNA, due to its ability to promote tumor cell invasion [107]. Fragile X syndrome is one of the commonly inherited mental retardation syndromes. The gene responsible for fragile X syndrome, FMR1 (fragile X retardation 1), is located on human chromosome 10. This syndrome is caused by loss of an RNA-binding protein called familial mental retardation protein, which has been reported to be regulated by miRNAs [108]. Tourette’s syndrome is another neuropsychiatric disorder among humans in which the role of miRNAs has been reported [109]. The 3¢-UTR of the SLITRK1 gene contains the binding site of miR-189, which is mutated in some Tourette’s syndrome patients [110]. In situ hybridization of SLITRK1 mRNA and miR-189 revealed coexpression in the neuroanatomical circuits most commonly implicated in Tourette’s syndrome. This demonstrates how an miRNA can be involved in the establishment of a disease phenotype [110]. miRNA expression profiles have been reported to be altered in sporadic Alzheimer’s disease (AD). Small, soluble oligomers of amyloid b-peptide (Ab) have been reported to have a role in the molecular basis for memory failure in AD. Ab oligomeric ligands (also known as ADDLs) are known to be potent inhibitors of hippocampal long-term potentiation. In a recent study, Hebert et al. [111] reported the interaction of miRNAs with BRACE-1 ⁄ b secretase genes. BRACE1 ⁄ b secretase is a rate-limiting step for Ab production, and its increased expression has been reported among AD patients. Hebert et al. [111] reported that miR29a, miR-29b-1 and miR-9 can regulate BRACE-1 expression in vitro. They found that expression of the miR-29a ⁄ b-1 cluster is significantly decreased in AD patients, which results in abnormal accumulation of high BRACE-1 protein and Ab levels among AD 4936 patients [111]. The altered expression of miRNAs has also been reported in postmortem samples of cerebellar cortex from autism patients [112]. The roles of miRNAs have also been reported in various viral infections. Some viruses perturb miRNA expression of the host cells, for their survival, and others encode their own miRNAs, which target various host genes [113]. Viruses have been reported to encode miRNAs [114], but the functions of most of them are not known. Herpes viruses such as Epstein–Barr virus or Kaposi sarcoma herpes viruses (KSHVs) have been reported to express miRNAs. Epstein–Barr virus induces cellular miRNAs such as miR-21, miR-155 or miR-146a during its infection cycle. Out of these, miR146 has been reported to be upregulated in various tumors [115]. In a recent report, 12 miRNA genes were identified within the genome of KSHV, and these miRNAs affect the expression of large number of cellular genes during KSHV infection [116]. The miR-28, miR125b, miR-150, miR-223 and miR-382 cluster of cellular miRNAs have been reported to contribute to the maintenance of HIV-1 latency in resting primary CD4+ T-lymphocytes [117]. Hepatitis B virus also encodes viral miRNA as a means of regulating its own gene expression [118]. Hepatitis C virus utilizes the liver-specific host miRNA miR-122 as a positive regulator of its own replication [119,120]. It is now time to study the function of virus-encoded miRNAs by utilizing bioinformatics and molecular biology tools. Irrespective of diseases, miRNAs are also involved in many other physiological functions. The expression of miR-375 takes place in murine pancreatic islets cells and plays an important role in regulation of the myotrophin gene and thereby glucose-stimulated insulin exocytosis [121]. Higher expression levels of miR-375 have been reported in pituitary glands of zebra fish embryos [25], which indicates its possible involvement in neuroendocrine activities [90]. MiR-122 and miR-1 play roles in mammalian liver development and cardiomyocyte differentiation, respectively [90,122]. The role of miRNAs is well known in ES cell differentiation, lineage specification and organogenesis, especially neurogenesis and cardiogenesis [123]. The miR-1 gene has been reported to be a direct transcriptional target of muscle differentiation regulators, including serum response factors, myogenic differentiation factor D, and myocyte-enhancing factor 2 [124]. The higher level of miR-1 results in a reduction in the number of proliferating ventricular cardiomyocytes in the developing heart. This suggests that miR-1 modulates the effects of critical cardiac regulatory proteins to control the balance between differentiation and proliferation during cardiogenesis [125]. miR-1, miR-133 and miR-206 have FEBS Journal 275 (2008) 4929–4944 ª 2008 The Authors Journal compilation ª 2008 FEBS S. K. Singh et al. MicroRNAs and their roles Fig. 2. Interference in the miRNA pathway by modified antisense synthetic oligonucleotides. Inhibition of miRNA can be achieved by introducing antisense synthetic oligonucleotides against miRNAs in the cytoplasm (shown as continuous lines). The possible targets of antisense synthetic oligonucleotides against miRNAs in the nucleus are pri-miRNA and pre-miRNA (shown as dotted lines). been reported to be involved in proliferation, differentiation and regeneration of skeletal muscles [126]. Recently, Chim et al. [127] have reported the existence of placental miRNA in maternal plasma, which opens up new possibilities of using the miRNAs as molecular markers for pregnancy monitoring. Four placental miRNAs (miR-141, miR-149, miR-299-5p and miR-135b) were found to be present at higher levels in maternal plasma during the predelivery period than after delivery. The measurement of miRNA in maternal plasma for prenatal monitoring and diagnosis would be an interesting future research direction [127]. Clinical implications of miRNA research The widespread role of miRNAs in the biological system makes them valuable targets for therapeutic intervention. The base pair interaction between miRNAs and their target mRNAs is key for miRNA function. Modified synthetic antisense oligonucleotides act as potential inhibitors of the miRNAs. Antisense oligonucleotides against miRNA pair with the miRNAs, occupying their binding sites and leaving their target mRNA in the unbound state [128]. Antisense oligonucleotides are useful tools for the inhibition of specific miRNAs. These have the potential to develop into a new class of therapeutic agents [129]. An abnormal phenotype might appear through aberrant suppression of any specific mRNA, due to the induction of its corresponding miRNA. In such cases, antisense oligonucleotides complementary to either the mature miRNA or its precursors can be designed (Fig. 2) to release the suppressive effect on mRNA [129,130]. Boutla et al. [131] demonstrated the inhibition of miRNA in Drosophila embryos by using antisense modified oligonucleotides against miRNAs through microinjection. Modified oligonucleotides have previously been shown to be effective inhibitors of both coding and noncoding RNAs in vitro and in vivo, and some of them, such as a 20-mer FEBS Journal 275 (2008) 4929–4944 ª 2008 The Authors Journal compilation ª 2008 FEBS 4937 MicroRNAs and their roles S. K. Singh et al. phosphorodiamidate morpholino oligomer targeting cMyc, are currently under investigation in human clinical trials [129,132]. The most important property of such oligonucleotides is specificity and high binding affinity for RNA. Two independent groups have transiently transfected 2¢-O-methyl-modified antisense RNAs directly in cultured cells, and have shown miRNA-specific inhibition [133,134]. These in vitro studies look promising, but the real challenge is in vivo use of modified miRNA inhibitors. In-depth studies are required to develop precise strategies for the pharmacological delivery of small RNAs into their target cells, for utilization of the potential of miRNAs as therapeutics. Several studies have been conducted to determine the role of RNAi in the suppression of disease-associated molecular pathologies in various animal models of disease [135]. The role of miR-122 in regulating cholesterol biosynthesis and in maintaining the adult liver phenotype, its association with hepatocarcinogenesis and its role in hepatitis C virus replication make it an invaluable target with which to expand our knowledge of the pathophysiology of diverse liver diseases [136]. Recently, it has been shown that inhibition of miR-122, a liver-enriched miRNA, has therapeutic potential in mice. Krutzfeldt et al. synthesized a 23-nucleotide RNA molecule (antagomir) complementary to miR-122 in such a way as to stabilize the RNA and protect it from degradation. They conjugated this small nucleotide with cholesterol molecules for their easy delivery into liver cells. This group successfully demonstrated inhibition of endogenous miR-122 in mice after injecting this small nucleotide complex through the tail vein [137]. The silencing of miR-122 by antagomir-122 resulted in a 44% decrease in plasma cholesterol levels in mice. Investigators expected that miR-122 might downregulate any repressor of the genes associated with the cholesterol biosynthetic pathway. Antagomir-122 may enhance the level of expression of the possible repressor, after binding with miR-122, which in turn results in inhibition of the transcription of cholesterol-synthesizing enzymes. Approximately 11 genes involved in cholesterol biosynthesis were reported to be downregulated by antagomir-122 [135,137]. Although there are many reports demonstrating the silencing of specific miRNAs by the use of miRNA inhibitors in mice, Elmen et al. [138] have recently demonstrated the silencing of miR-122 by a lock nucleic acid (LNA)-based miRNA inhibitor (LNA-antimiR). They demonstrated that delivery of NaCl ⁄ Pi-formulated LNA-antimiR inhibited the expression of miR-122 in the liver of nonhuman primates [138]. 4938 Krutzfeldt et al. [137] have developed new methods for the effective delivery of antisense oligonucleotide against miRNAs. These include modification in the RNA backbone, at each nucleotide, by an O-methyl moiety at the 2¢-ribose position. The terminal nucleotides at both ends are also modified with a phosphorothioate linkage, in contrast to the standard phosphodiester linkage in RNA and DNA. Unmodified oligonucleotides were used to inhibit the expression of let-7 miRNA in C. elegans, but this strategy was not effective, due to the unstable nature of unmodified oligonucleotides in vivo [133]. Therefore, modifications of synthetic antisense oligonucleotides against miRNAs are required to make them thermostable and nuclease-resistant, which protects them once they are exposed to serum and cellular nucleases. The third modification is cholesterol functionality at the 3¢-end of the nucleic acid. This improves pharmacokinetic properties by increasing binding to serum proteins, and improving stability and half-life in serum and cellular uptake [128,139]. Pharmaceutical companies such as Regulus and Santaris have focused their drug discovery research on the development of miRNA-based therapeutics for viral infectious diseases and metabolic disorders. Problems in therapeutic application Although there have been successful attempts at delivery of antisense RNAs to cells and tissues, successful delivery of small RNAs to the brain is one of the major challenges in the development of small RNAbased neurotherapeutics. Appropriate access of plasmid–lipid complexes or viral vectors to the desired tissues and cells of neural origin is a critical issue. Several modifications and improvements in delivery methods are ongoing, but they still need precision. The blood–brain barrier is a big hurdle in the treatment of neurological diseases, because it inhibits the passive entry of therapeutic molecules from the peripheral circulation into the brain. The study conducted by Kumar et al. [140] suggests that the short peptide derived from rabies virus glycoprotein potentiates the transvascular entry of siRNAs into the brain. They demonstrated that rabies virus glycoprotein-9R-bound antiviral siRNA provided effective protection against viral encephalitis in mice, without any induction of inflammatory cytokines and antibodies against peptides [140]. The intracellular concentrations of the target RNA and the small RNA-based drugs will determine the extent and duration of suppression. Therefore, there is a need to conduct studies on dose optimization and modes of delivery of miRNAs, in FEBS Journal 275 (2008) 4929–4944 ª 2008 The Authors Journal compilation ª 2008 FEBS