- Email: [email protected]
Identification of cardiovascular microRNA targetomes
Journal of Molecular and Cellular Cardiology 51 (2011) 674–681
Contents lists available at SciVerse ScienceDirect
Journal of Molecular and Cellular Cardiology journal homepage: www.elsevier.com/locate/yjmcc
Review article
Identification of cardiovascular microRNA targetomes J. Fiedler a, S.K. Gupta a, T. Thum a, b,⁎ a b
Institute of Molecular and Translational Therapeutic Strategies (IMTTS), Germany Centre for Clinical and Basic Research, IRCCS San Raffaele, Rome, Italy
a r t i c l e
i n f o
Article history: Received 12 July 2011 Received in revised form 11 August 2011 Accepted 12 August 2011 Available online 24 August 2011 Keywords: Cardiac disease miRNA miRNA targetome RISCome RNA immunoprecipitation MRE Target prediction
a b s t r a c t MicroRNAs (miRNAs) are strong post-transcriptional regulators targeting multiple targets. Endogenously transcribed, miRNAs specifically bind to complementary sequences of mRNAs and repress their expression thus govern control of cellular signaling pathways. An altered miRNA expression is causally related to cardiovascular disease. Identification of miRNA-dependent pathways is therefore an important aim to develop new therapeutic approaches. To understand miRNA function in various cardiovascular cells, the identification of individual miRNA target genes is of utmost importance. Indeed, the biological function of a miRNA is dependent on the availability of potential targets in a cell. We here summarize and discuss current challenging approaches to identify miRNA targetomes which will help to understand miRNA function in cardiac homeostasis and disease. © 2011 Elsevier Ltd. All rights reserved.
Contents 1.
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.1. MicroRNAs (miRNAs)—biogenesis and biological function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.2. Cardiovascular disease is associated to miRNA deregulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2. Identification of miRNA target genes (the miRNA targetome or RISCome) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1. Bioinformatics, proteomics and other molecular biology-related approaches . . . . . . . . . . . . . . . . . . . . . . 2.1.1. Bioinformatics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2. Targetscan . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3. PicTar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.4. miRanda-mirSVR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.5. MicroCosm targets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.5.1. Transcriptomics, proteomics and other biochemical approaches . . . . . . . . . . . . . . . . . . . . . . . . . 2.6. Transcriptome and proteome approaches . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.7. Biochemical approaches . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.7.1. RIP (RNA immunoprecipitation)-Chip, RISCome (RNA-induced silencing complex) analysis . . . . . . . . . . . . 2.7.2. RNA-sequencing, HITS-CLIP (high-throughput sequencing of RNAs isolated by crosslinking immunoprecipitation), PAR-CLIP (photoactivatable-ribonucleoside-enhanced crosslinking and immunoprecipitation) . . . . . . . . . . . . . . . . . . 3. Conclusion and perspectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conflict of interest statement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
675 674 675 675 675 675 675 676 676 676 676 676 676 677 677
. . . .
. . . .
. . . .
. . . .
. . . .
678 679 680 680
⁎ Corresponding author at: Institute of Molecular and Translational Therapeutic Strategies (IMTTS), Hannover Medical School, Carl-Neuberg-Str. 1, 30625 Hannover, Germany. Tel.: + 49 511 532 5273; fax: + 49 511 532 5274. E-mail address: [email protected] (T. Thum). 0022-2828/$ – see front matter © 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.yjmcc.2011.08.017
J. Fiedler et al. / Journal of Molecular and Cellular Cardiology 51 (2011) 674–681
1. Introduction 1.1. MicroRNAs (miRNAs)—biogenesis and biological function Since discovered in the early nineties, microRNAs (miRNAs) have been shown to play major roles in different physiological settings by regulating gene expression post-transcriptionally [1–3]. Of note, miRNAs regulate 30–50% of the genome and thus are potent mediators of cellular signaling [4]. Several details about the regulation of miRNAs and their biogenesis especially in the context of cardiovascular biology have recently been reviewed [5] and therefore are not a main topic of this review article. Briefly, miRNAs are either transcribed from miRNA genes or spliced from host gene transcripts (overview on biogenesis reviewed in [6]). Once formed as so called primary (pri)-miRNAs in the nucleus, the Drosha complex cleaves the RNA to a stem– loop precursor structure (precursor (pre)-miRNA) of about 60–70 nucleotides in length. Cytosolic translocation mediated by exportin proteins triggers further processing at the Dicer complex. Here, the miRNA heteroduplex is unwinded to its biological active singlestranded format. The second strand (designated as miRNA*) is often not selected and subsequently degraded. RNA-binding proteins, e.g. proteins of the Ago family then recruit the mature miRNA to sites of RISC (RNA induced silencing complex). An antisense mechanism then mediates miRNA-dependent recognition in RISC to target mRNAs. Most commonly, miRNA responsive elements are found in 3′-UTR (untranslated region) of the mRNA and base pairing determines stringency of miRNA:mRNA interaction. In line, the strength of miRNA binding to the mRNA is crucially mediated by nucleotides 2 to 8 (the seed) and surrounding bases [7]. However, recently the central dogma of miRNA binding to 3′-UTR region has been challenged, as reports demonstrated additional miRNA binding to coding regions or even 5′-UTRs [8]. However, the main characteristics of a single miRNA or even miRNA families sharing the identical seed sequence, are their ability to simultaneously target many mRNAs which are accessible in a specific cellular context. It is crucial to decipher miRNA target genes in a single cell by different bioinformatic or experimental approaches to substantially identify intracellular signaling pathways especially in terms of cardiac disease. The still growing number of newly identified miRNAs emphasizes their cellular impact, calls for sequence annotation and a summary of characteristics in open-access databases (see below). 1.2. Cardiovascular disease is associated to miRNA deregulation MiRNAs are crucial in the control of cardiovascular signal transduction in the heart. Indeed, knockdown of the miRNA biogenesis related enzyme Dicer in cardiomyocytes led to cardiac remodeling, enhanced fibrosis and finally heart failure [9,10]. This key study of Costa Martins and colleagues emphasized the orchestrating role of cardiac miRNA expression to sustain and balance cardiac morphology and function. Cardiac disease animal models, such as experimental myocardial infarction lead to a differential pattern of miRNA expression [11]. Noteworthy, identified miRNAs deregulated in animal models of heart diseases such as miR-21 or miR-208a are also upregulated in human heart failure [11]. When comparing miRNA expression profiles in fetal and failing human hearts, many fetal miRNAs became re-expressed during heart failure demonstrating a potential role for miRNAs to be involved in general fetal gene reprogramming during heart failure [12]. In vitro studies further confirmed that modulating miRNA expression can trigger molecular and phenotypic changes in different cardiovascular cells. For instance, modulating miR-133 or miR-208a in cardiomyocytes had a strong impact on hypertrophic response of cardiomyocytes [13,14]. Modulation of miRNA biogenesis by Dicer or Drosha knockdown in endothelial cells interfered with the angiogenic response [15]. Manipulation of miR-21 or miR-29 expression in fibroblasts induced changes in
675
fibroblast proliferation or collagen expression implicating a potential therapeutic interest [16,17]. Using genetic animal techniques, miRNA gain- and loss-of function studies also revealed an essential role for many miRNAs in cardiac homeostasis and disease [18,19]. Taken together, these important findings have also been transferred to relevant therapeutic approaches based on antagomir or antimiR-mediated knockdown of miRNA expression in murine models of cardiac disease [17,20–22]. Of note, miRNA modulators were also translated into bigger animal studies such as in a primates' model of hepatitis C infection [23] and currently miRNA-based therapeutics also have entered the clinics based on ongoing phase I and II studies. 2. Identification of miRNA target genes (the miRNA targetome or RISCome) In the field of miRNA research, identifying miRNA target genes is one of the most important but unfortunately one of the toughest goals requiring different approaches (excellently reviewed in [24]). Detailed knowledge of downstream effectors of miRNAs is important to estimate their individual functional relevance in healthy conditions but especially during disease. This information is especially needed when miRNA expression is modulated by miRNA-therapeutics such as antagomirs or anti-miRs. Here, new tools were recently developed to investigate miRNA target genes on a more global level (miRNA targetome or the RISCome analysis) in cardiac cells [25]. Different technical approaches to uncover miRNA targets are now available and will be presented and discussed in the following sections (see Fig. 1). 2.1. Bioinformatics, proteomics and other molecular biology-related approaches 2.1.1. Bioinformatics Computational analysis of potential miRNA target genes relies mainly on open-access databases quite easily to use for the scientific community. Several working groups have developed software algorithms as target prediction tools which seem to be the first choice in miRNA target search. In common, the software tools screen mRNAs 3′-UTR for putative miRNA-responsive elements (MRE). As previously reported, MRE are more likely bound by miRNA when they are located either in the beginning or the end of the 3′-UTR of a gene [26]. Tandem repeats of MREs in a single 3′-UTR often enhance the possibility of miRNA binding. In contrast, secondary structure effects of RNA due to characteristic nucleotide sequence might impede miRNA binding in the center of a specific mRNA 3′-UTR. However, the underlying algorithms to predict targets vary and often the overlap of predicted
Fig. 1. Various approaches exist to delineate miRNA targets — the “targetome”. Bioinformatics and in silico predictions, RNA pulldown and genechip analysis, transcriptomics or proteomics and RNA sequencing methods can be applied to identify miRNA targets on a cellular level.
676
J. Fiedler et al. / Journal of Molecular and Cellular Cardiology 51 (2011) 674–681
target genes is small. In the proceeding paragraphs we will summarize the most important in silico features of several miRNA target prediction databases, such as Targetscan, PicTar, miRanda-mirSVR and MicroCosm, all of which are helpful bioinformatic tools to develop a general overview of potential miRNA targets. 2.2. Targetscan The Targetscan database is accessible via www.targetscan.org and is updated regularly by the Whitehead Institute for Biomedical Research, Cambridge, US. Recently, version 5.2 has been launched and offers the possibility for miRNA target searches in humans, mouse, worm and fly genomes. Primarily, the Targetscan-algorithm detects conserved seven to eight nucleotide (7-mer/8-mer) matches between the miRNA seed region and responsive mRNA 3′-UTRs [4]. Additionally, common target sites are listed, which may have a nucleotide mismatch in the seed region but can facilitate miRNA binding via supporting base pairing in the 3′-direction [27]. These potential binding events can also be applied for non-conserved target sites among the species. The final output list of target sites or target genes is ranked by a targetscan context score accordingly to Grimson et al. [26]. Here, four distinct features of miRNA binding are considered for the calculation: site-type, 3′ pairing, local sequence AU content and position contribution. The lowest score defines the most favorable chance of target regulation by direct miRNA:mRNA interaction. 2.3. PicTar PicTar has been developed by the Rajewsky lab at the Max Delbruck Centrum, Berlin and is online at www.pictar.mdc-berlin.de. It is a valuable source for potential miRNA target genes and is frequently used. The PicTar database also covers a wide range of species including vertebrates, Drosophila and nematode species [28–30]. This interface also highlights conserved miRNA binding sites and as a special feature, co-expressed miRNA and target mRNAs are listed if conservation is not the case [31]. The overall PicTar score is finally ranked in a consecutive list from most to less likely and can be applied for target identification. 2.4. miRanda-mirSVR The interconnected database miRanda-mirSVR is found at www. microrna.org and is provided by the Memorial Sloan Kettering Cancer Center, New York. Predictions are provided for several species and interestingly, various database links exist to support additional information e.g. on miRNA function and expression level in different tissues. The algorithm is based on sequence match, free energy calculation and evolutionary conservation [32]. These characteristics are summarized by miRanda [33], whereas the key score has been developed by mirSVR [34]. Taken together, target gene repression by miRNA binding is calculated and additional alignment views are given to depict miRNA binding. 2.5. MicroCosm targets This web interface is online at www.ebi.ac.uk/enright-srv/ microcosm/htdocs/targets/v5/ and was developed by the Enright Lab at the EMBL-EBI, Cambridge, UK. MicroCosm Targets also encompasses a comprehensive miRNA target prediction database and is regularly updated with new information. In accordance with miRandamirSVR, computational analysis is performed with the miRanda algorithm [33]. Perfect sequence complementarity within the seed region of the miRNA and target mRNA is of utmost importance for the calculation. Non-perfect matches are immediately negatively selected and thus are not processed. Thermodynamic stability of the mRNA– miRNA heteroduplex is acquired via the Vienna RNA folding routines
(www.tbi.univie.ac.at/~ivo/RNA/). Again, evolutionary conservation is considered for the MRE and must be present in at least two different species. Collectively, miRNA target prediction databases provide a first solid step into the identification of miRNA targets and offer additional valuable information including target gene annotation. It is obvious, that comparison and/or combination of target prediction algorithms should be used. This is a useful first step to identify target genes for a single miRNA or even miRNA families for further mechanistical or functional investigations. Here, we just introduce a small set of available databases and we think that other software tools might also be appropriate. These can be tools calculating free energy between two RNA sequences to hybridize, e.g. RNAhybrid provided by the Bielefeld University Bioinformatics Server. However as mentioned above, most predictions do not take into account if a certain target gene AND the miRNA are both expressed and/or available in the particular cell type of interest. Thus, cell-type specific expression of both miRNA and its targets needs to be assessed to narrow hits from the bioinformatic prediction. To summarize the current knowledge on previously reported miRNA/target gene expression, expression profiles in disease or after different treatments on miRNA expression specific databases such as miRGator (www.mirgator.kobic.re. kr:8080/MEXWebApp/index.jsp) or miRTarBase [35] have been developed and this is a great help in proper miRNA target identification. Next to these aforementioned conventional searches for miRNA binding sites in 3′-UTR, a preliminary database is available for screening whole genomes for miRNA binding (miRWalk, [36]) but this approach needs future evaluation to establish as a reliable source. Most importantly, computational predictions need to be validated, e.g. target gene expression should be detected via western blot when miRNA of interest is overexpressed in a particular cell type. In addition, novel technical approaches are now available as discussed below. 2.5.1. Transcriptomics, proteomics and other biochemical approaches Up on miRNA:mRNA base pairing targets are repressed at the mRNA and/or protein expression level. Thus deregulation of a miRNA leads to specific “foot-prints” in the targetome and/or proteome. Next to the aforementioned in silico target prediction analyses, several in vitro approaches have been developed to evaluate such direct interactions on a more global level. The ways to approach the changes in cellular signaling have in common that miRNA expression is correlated to target expression. Thereby, alteration of miRNA expression is accompanied by well-known “omic”-approaches describing high-throughput methods such as transcriptomics and proteomics. Next to these, other biochemical assays have been developed to support detection of miRNA targets. 2.6. Transcriptome and proteome approaches Analyzing conventional mRNA microarray-data sets after miRNA modulation of cells or organs identifies genes and gene network(s) being involved in the direct and indirect consequences of alteration of a miRNA. For instance, a certain miRNA can be modulated in cell cultures, whereas in controls only a scrambled miRNAs was used, followed by RNA isolation and transcriptome analysis by standard miRNA-microarrays. Several groups have used such an approach in the past. By way of example, hypoxia-regulated miR-210 was analyzed in such a way but using a combination of bioinformatics, transcriptomics and proteomics which will be discussed later on [37]. However, miRNA modulation itself does not only regulate direct miRNA targets but may have many secondary effects at the genome level. It is also difficult to identify the optimal time point for miRNA-mediated degradation of mRNAs using transcriptome approaches. To counteract such difficulties, bioinformatic-dependent statistics have been developed to support microarray data analysis
J. Fiedler et al. / Journal of Molecular and Cellular Cardiology 51 (2011) 674–681
[38,39]. Regulatory miRNAs are defined by an OR (odds-ratio)-statistic based on time-course expression of both miRNA and mRNAs as well as computational predictions. Time-course-dependent expression analysis on a microarray can help to estimate the correlation and significance of miRNA/mRNA sets. Applying the ORs on a set of cancer data the authors identified some miRNAs indeed having a regulatory potential towards transcriptional activity and thus cancer biology. Further improvement of the OR was gained by adding two clustering steps into the consideration based on the fact that single miRNAs target many mRNAs and vice versa. This clustering identified miRNA/mRNA (miRmR) modules that then can be studied for their biological impact. This approach can easily adapted to cardiovascular miRNA research for improved identification of cell-type specific miRNA targets. In vivo, investigating fetal and failing hearts linked alterations of target genes, the number of individual miRNA bindings sites in those targets and tissue-specific miRNA levels [12]. Interestingly, mRNAs with a high number of binding sites for miRNAs which were upregulated in failing human heart tissue were more likely repressed, whereas those with many binding sites for miRNAs, which were downregulated, tended to be more upregulated. Another paper reported the use of independent component analysis for miRNA profiling in a type 1 diabetes model [40]. The processing of microarray data was combined to computational calculations and also cross-checked against a prediction relying on the negative correlation between miRNA and target mRNA. However, in some cases the interaction of a miRNA with a mRNA target does not lead to mRNA degradation but to translational repression leading to alterations at the protein level. Thus proteome analyses may be of help to identify the targetome of individual miRNAs. Initially, two groups have applied miRNA modulation and amino acid labeling to investigate proteomic changes induced by a miRNA (see Table 1, [41,42]). SILAC (stable isotope labeling with amino acids in cell culture, [43] was combined with miRNA overexpression to measure the impact on protein output. Such approaches demonstrate a widespread change on cellular protein constitution upon miRNA modulation, even if the effects on protein level have been reported as moderate. Very recently, in the field of cancer, various quantitative proteomic approaches were performed and identified downstream targets for miR-143/373 [44,45]. Interestingly, observed changes at the protein level of miR targets did not coincide with decreased expression at the mRNA level [45]. Integrative analysis to dissect mammalian inner ear miRNA pathways also identified miRNA targets repressed in the proteomics approach only [46]. Different peptide labeling using iTRAQ (isobaric tag for relative quantitation) facilitated the identification of miR-21/193b targets emphasizing the general ability of proteomics to delineate miRNA targets [47,48]. Overexpression of miR-21 in endothelial cells followed by proteome analysis, using difference in-gel electrophoresis and subsequent mass spectrometric analysis of regulated proteins identified a number of deregulated proteins such as superoxide dismutase 2 [49]. This approach shows how rigorous utilization of
Table 1 Publications listed applying high-throughput screening to identify miRNA targets. High-throughput miRNA target identification miRNA(s)
Methodology
Reference
miR-133a, -499 miR-223, -1, -124, -181 miR-155, -1, -16, -30a and let-7b miR-21
RISC-Seq SILAC/Proteomics SILAC/Proteomics
[25] [41] [42]
2D-DIGE (differential in-gel electrophoresis)/Proteomics RIP-Chip, Microarray RIP-Chip, Microarray HITS-CLIP PAR-CLIP
[49]
miR-17/20/93/106 miR-103, -107,-16, -195 miR-124 miR-7, -124
[57] [62] [64] [66,67]
677
unbiased screening technologies can be used for mechanistic research [50]. 2.7. Biochemical approaches A bottleneck of the “omics”-based approaches such as microarrays or proteomics is their relatively high costs. Novel rapid assays have been now introduced to gain insight to miRNA target genes. A PCR technique designated “Hybrid-PCR” was used to screen for miRNA target genes in the genome of human cytomegalovirus lacking a complete mRNA annotation [51]. In HeLa and HepG2 cells miRNA targets were identified by determining the C/N (cytoplasmic to nucleic) ratio of mRNA after miRNA overexpression [52]. Here, a higher mRNA C/N ratio indicated a greater chance for a gene to be targeted by the miRNA. Biotin-tagged miRNAs have also been employed in a streptavidin–biotin pulldown system to demonstrate miRNA:mRNA interactions [53]. A recent study investigated miR-24 function in myocardial infarction (MI) and its outcome for endothelial cell biology [54]. Use of miR-24 overexpression in cultured endothelial cells and subsequent analysis of putative targets via western blot and luciferase reporter gene assays revealed an angiogenic signaling network regulating angiogenic response. Luciferase reporter gene assay is an appropriate tool to evaluate miRNA binding. The respective 3′-UTR of a target mRNA containing the MRE is cloned to luciferase gene and co-transfected with miRNA of interest and controls. Quite oftenly, the wild-type MRE is mutated at some nucleotides and experiments are performed according to wild-type conditions to indicate that perfect sequence complementary within the MRE is needed for miRNA target repression. Target mRNAs are also capable to function as a bait as described for cardiac transcription factor Hand2 [55]. A tagging strategy was applied within a reporter system to demonstrate binding of miR-133a to Hand2 3′-UTR. Of note, addition of biotinylated miR-1 revealed cooperative association of miR-1 and miR-133a with Hand2 3′-UTR. Affinity purifications may thus also be of interest when considering approaches to define miRNA target genes. 2.7.1. RIP (RNA immunoprecipitation)-Chip, RISCome (RNA-induced silencing complex) analysis A significant hurdle in the understanding of the biological function of an individual miRNA is to identify its target mRNAs precisely in particular cellular and event/disease contexts. Use of microarray and proteomics techniques as stated above result in first insights of what signaling networks may be affected by a single miRNA but such approaches do not necessarily identify direct targets as many secondary changes occur. In contrast to plants, where regularly a complete sequence complementarity between the miRNA and its target gene is observed, animals' miRNA/mRNA complementarity is significantly less making the identification of miRNA targets much more complicated. The RIP-Chip technique (RNA-binding protein immunoprecipitation followed by Genechip, Microarray) is a basic profiling approach to define miRNA targets in a high-throughput manner and combines genomic technologies with standard molecular biology tools to define miRNA target genes—the “targetome” [56] (see also Fig. 2). Briefly, members of the miRNA ribonucleoprotein complex such as argonaute [57], GW182 [58] and TNRC6 family members [59] or others are immunoprecipitated. Total RNA is isolated from the complexes and analyzed on a microarray which was also reported in a RNA immunoprecipitation-genechip protocol [57,60] to identify direct miRNA targets in cellular context. Briefly, RIP-Chip is the immunoprecipitation (IP) of RNA binding protein (RBP) complexes followed by RNA extraction and chip-based identification. In vitro, around 5–20 million cells are needed for the IP procedure along with antibody-coated sepharose beads. Finally, RNA is isolated from the IP-derived samples. The isolated RNA can then be processed for microarray (RIP-Chip) or sequencing analysis (RIP-Seq). Tan et al. have used RIP-Chip for identifying miRNA targets in Hodgkin
678
J. Fiedler et al. / Journal of Molecular and Cellular Cardiology 51 (2011) 674–681
Fig. 2. Target identification technique in miRNA research using RNA immunoprecipitation and genechip analysis (RIP-Chip). In this procedure cells are pre-treated either with miRNA precursor (pre-miR) or inhibitor (anti-miR) and scrambled miRNA (scr-miR). In case of pre-miR transfection, target mRNAs should be enriched with Ago2 whereas endogenous miRNA inhibition would lead to target mRNA depletion from Ago2. Finally, mRNA:miRNA-Ago2 complexes are immunoprecipitated and undergo RNA analysis. Recovered RNA is hybridized to a genechip and analyzed with bioinformatic tools.
Lymphoma cells [57]. Under basal conditions, wildtype argonaute 2 (Ago2) was used to immunoprecipitate RISC and the isolated RNA was subjected to gene array which showed enrichment for around 3000. Enriched probes were termed as miR-targetome and ~30–40% of the identified probes were indeed direct targets of the top 5% expressed miRNAs based on the presence of 8-mer seed sequences and Targetscan prediction analyses. These results show a successful enrichment of miRNA targets in the Ago2-IP fraction. To extend this approach, specific miRNA inhibition leading to mRNA depletion from Ago2 and RISC was used endogenously to decipher targets of the highly enriched miR-17 family. The depleted gene transcripts were categorized into three categories based on level of depletion. Gene transcripts with less than 2-fold depletion were included in non-miR-17 targets, probes with ≥2 depletion in miR-17 targets and a third group of transcripts with ≥4–7 fold depletion were termed as potent miR-17 targets. To summarize, about 50% (miR-17 targets) and 67% (potent miR-17 targets) of transcripts respectively, were depleted upon antisense inhibition and contained at least a 6mer or 8-mer seed sequence for the miR-17 family as demonstrated by bioinformatic analysis. This emphasizes a general suitability of this approach for the successful genome-wide identification of endogenous targets of selected miRNAs or miRNA families. A similar approach has been used by Zhang et al. who immunoprecipitated the GW182 proteins AIN-1 and AIN-2 of Caenorhabditis elegans to identify the respective miRNA targetome [58]. A comprehensive method has also been developed by Fasanaro et al. in order to identify targets of hypoxia-inducible miR-210 [37]. The authors combined a set of approaches (proteomics, transcriptomics, overexpression or inhibition of miRNAs) followed by RISC immunoprecipitation methods. The integration of different technical aspects revealed several new targets with MREs both in 3′-UTR but interestingly also in 5′-UTR and coding sequences. In order to reduce false-positive hits, transcripts which were reciprocally regulated (e.g. downregulated on overexpression as well as upregulated on inhibition of miR-210), were selected from transcriptomic profiling and 51 targets were found to comply these obligations representing potential targets for miR-210. Importantly, the targets that were downregulated in both the proteomic and transcriptome analyses showed no similarity raising a number of questions. One explanation could be that the time-dependent differences upon miRNA transfection in RNA and protein levels may differ at various time points. Secondly, proteomic techniques may not detect miRNA-induced changes accurately enough and other
approaches might be useful (reviewed in [61]). Fasanaro et al. additionally performed IP of Ago2–RNA complex upon miR-210 overexpression and controls. Some of the identified targets by proteomics and transcriptomics were found to be enriched in precipitates from miR-210 overexpressing cells. A very interesting finding was the enrichment of non-coding RNA Xist which was later on confirmed by qPCR highlighting downregulation of Xist RNA levels on miR-210 overexpression in HEK293 cells. The potential regulation of another non-coding RNA (Xist) by miR-210 points to a new mode of noncoding RNA–RNA regulation. A very recent study from Nelson et al. also used the RIP-Chip method to identify targets for miR-103, mir107, miR-16 and miR-195 [62]. In line with others, they also found that seed sequences for microRNAs are also present in the coding region along with 3′ UTR. Seed sequences for miR-107 along with other group members were found to be more enriched in the coding region (CDS) compared to 3′ and 5′ UTR in the immunoprecipitated fraction. Mechanistically, mutation of miR-107 in the 3′ region induced precipitate enrichment of the 3′ UTR sequence instead of CDS while mutations in the miRNA 5′ region (seed) did not. Despite having been mutated at the seed region, miRNA could still attach to a coding sequence although with an altered seed sequence complementarily comprising a new seed region. These observations evaluate the crucial role for the miRNA 3′ region in targeting a coding sequence of mRNAs. Another study termed TAP-Tar (tandem affinity purification of miRNA target mRNAs) also utilizes the concept of RIP-Chip in combination with use of biotin tagged miRNAs and Flag-tagged argonaute protein [63]. The authors first immunoprecipitate Ago–RISC complex followed by subsequent affinity purification with streptavidin beads. This method is applicable for miRNA-specific purification of target genes and importantly decreases the false positive rate by the two step purification procedure. 2.7.2. RNA-sequencing, HITS-CLIP (high-throughput sequencing of RNAs isolated by crosslinking immunoprecipitation), PAR-CLIP (photoactivatable-ribonucleoside-enhanced crosslinking and immunoprecipitation) Further technical advances of the aforementioned miRNA target identification strategies resulted in challenging protocols applying RNA:protein crosslinking and sequencing. Very recently, an elegant combination of both Ago2 IP (see above) and RNA-sequencing has been developed to identify the miRNA targetome of cardiac miR133a and miR-499 [25]. This approach is useful to identify miRNA
J. Fiedler et al. / Journal of Molecular and Cellular Cardiology 51 (2011) 674–681
targets more accurately with less false positive enrichment. Matkovich et al. directly immunoprecipitated the Ago2–RISC complex from murine ventricular tissue and compared bound mRNA fraction to the complete cardiac transcriptome. Related to the cardiac transcriptome, a significant enrichment of 15% was present in Ago2 immunoprecipitated RNA eluate, defining the group of cardiac expressed mRNAs which can be targets of cardiac miRNAs (RISCome). This observation demonstrates the involvement of many potential miRNA targets in heart tissue in vivo. From the enriched transcripts, 71% had MREs validated by deep sequencing methods and overlapped with bioinformatic predictions by Targetscan. In order to delineate the role of individual miRNAs in cardiac biology, a cardiomyocytespecific overexpression of miRNA-133a in vivo has been used. The cardiac RISCome of transgenic mice contained 47.4% of targets for miR-133a as predicted by Targetscan whereas the cardiac RISCome of wildtype mice displayed less (34.5%) miR-133a targets. This is a considerable enrichment of specific targets in the RISC complex depending on the overexpression of a certain miRNA. Next, RISCome comparisons of two different miRNAs, miR-133a and miR-499, showed little similarity confirming the specificity of the RISCsequencing method in the identification of specific targetomes of different miRNAs in vivo. A very similar method defined as HITS-CLIP (high-throughput sequencing of RNAs isolated by crosslinking immunoprecipitation) is also available [64]. Here, the RISC complex consisting of miRNA, mRNA and RISC protein is UV-crosslinked in advance to IP. This initial treatment had several advantages including a more stringent purification, reduced levels of false positive hits, removal of other small RNAs like tRNAs and rRNAs as well as an increased RNA amount. The Ago-mRNA HITS-CLIP experiments from mouse brain validated binding (Ago tags) to various sequences with a composition of 1% 5′ UTR tags, 40% 3′ UTR tags, 25% CDS tags, 12% intron tags and interestingly 4% non-coding RNA tags thus speculating about a novel relative function of Argonaute protein and miRNA action. A study done by Leung et al. applied the HITS-CLIP method along with silencing experiments of the miRNA biogenesis machinery [65]. Specifically, the miRNA targetome from wildtype and Dicer knockout mouse embryonic stem cells (ESC) were analyzed with this method. The RISCome from Dicer−/− ESC was still characterized by RNA cross-linked to Ago2 suggesting interaction of Ago2 with mRNAs to
679
be miRNA-independent. An updated and improved version of HITSCLIP is termed PAR-CLIP (photoactivatable-ribonucleoside-enhanced crosslinking and immunoprecipitation) [66,67]. This cell-based cross-linking approach determines transcriptome-wide binding sites of cellular RNA/miRNA binding proteins at high resolution. The technique facilitates crosslinking by incorporation of nucleoside-analogue 4-thiouridine in RNA, which results in higher yield of cross-linked RNA. The incorporation of 4-thiouridine itself led to base conversions (T to C) which in turn can be used to identify the miRNA/mRNA binding site to Ago2 very precisely. The crosslinking of 4-thiouridine to amino acid residues possibly changes their chemical structure which favors incorporation of dG rather than dA by reverse transcriptase. The change in chemical structure could be an alteration in stacking or hydrogen bond donor/acceptor properties. Specifically, 84% of the Ago-crosslinked center region from different Ago IPs belongs to exonic regions while 14% belongs to intronic sequences showing a possible involvement of Ago proteins in splicing or miRNA mediated alternative splicing. Briefly, 100–400 million cells (about 20–40 times of the RIP-Chip approach) are usually needed to perform the PAR-CLIP method (see Fig. 3). Prior to cell lysis, cells are tagged with photoactivatable ribonucleoside 4-thiouridine enabling crosslinks between RNA and RNA binding proteins by UV irradiation. The crosslinked RNA/RNA binding protein complex is then immunoprecipitated and subsequently radiolabeled. The complex is then separated on a SDS-PAGE gel, excised and subjected to electroelution followed by RNA isolation. The isolated RNA is proceeded through standard cDNA library preparation, deep sequencing and finally bioinformatic analysis. A more sensitive method could be potentially developed by combining some of the discussed approaches using biotinylated (tagged) miRNAs along with PAR-CLIP. This would add a two step purification process removing false positives and also due to pulldown of specific miRNA, the identified targetome may be more specific rather than a complete cellular miRNA targetome. 3. Conclusion and perspectives Different technical strategies have been developed to define and identify miRNA targets in a more or less comprehensive manner. Bioinformatics, transcriptomics, proteomics, RNA crosslink and sequencing
Fig. 3. Schematic representation of the PAR-CLIP technique. Cells are treated with 4-thiouridine prior to UV cross linking at 365 nm. Then, cells are lysed and Ago or TNRC complexes are immunoprecipitated. The RNA in immunoprecipitated complex is radiolabeled, separated on a SDS-PAGE gel and finally isolated from an immunoprecipitated complex. RNA is deep sequenced and sequence reads are analyzed using bioinformatics to gain insight into targets of miRNAs.
680
J. Fiedler et al. / Journal of Molecular and Cellular Cardiology 51 (2011) 674–681
methods can be combined to decipher miRNA target genes (summary in Table 1). The discussed approaches vary in their procedure and data analysis but are of great interest to be translated to cardiovascular research, which currently is clearly in its infancy. To truly validate a miRNA target it is still necessary to perform additional in vitro experiments to confirm direct miRNA binding to a specific sequence in a given cellular context. Conventional reporter gene assays using luciferase reporter enzyme activity are probably the best choice to prove the miRNA/gene interaction. New aspects of miRNA research beyond the classical cellular level just appear on the horizon. Recent reports suggested miRNAs as potential biomarkers in the circulation or even in organelles like mitochondria [68–70]. Whether circulating miRNAs or mitochondrial miRNAs are also able to bind to mRNA sequences is not entirely known. Since developmental effort is still ongoing, technical approaches in other scientific fields may be translated to miRNA target identification research in the future. Single-molecule pulldown, detection of uridylation modification at the miRNA level has been recently reported and could assist the exploration of new miRNA-regulatory pathways [71]. Besides the role of miRNAs for cellular signaling other regulatory RNAs like long non-coding RNAs (lncRNAs) might also impact specifically on target genes or protein machineries (reviewed in [72]). The knowledge of miRNA target identification strategies is therefore useful to understand the underlying molecular mechanisms and maladaptive responses in disease triggered by powerful non-coding RNAs. An understanding of the targetome of cardiac miRNAs is of outstanding importance as miRNAs play an important role in many cardiovascular diseases and recently emerged as powerful therapeutic targets.
Conflict of interest statement The authors disclose support from the IFB-Tx (BMBF 01EO0802; T.T.) and DFG TH 903/10-1 (T.T.). T.T. and J.F. have filed patents in the field of cardiovascular miRNA diagnostics and therapeutics.
References [1] Lee RC, Feinbaum RL, Ambros V. The C. elegans heterochronic gene lin-4 encodes small RNAs with antisense complementarity to lin-14. Cell 1993;75:843–54. [2] Wightman B, Ha I, Ruvkun G. Posttranscriptional regulation of the heterochronic gene lin-14 by lin-4 mediates temporal pattern formation in C. elegans. Cell 1993;75:855–62. [3] Pasquinelli AE, Reinhart BJ, Slack F, Martindale MQ, Kuroda MI, Maller B, et al. Conservation of the sequence and temporal expression of let-7 heterochronic regulatory RNA. Nature 2000;408:86–9. [4] Lewis BP, Burge CB, Bartel DP. Conserved seed pairing, often flanked by adenosines, indicates that thousands of human genes are microRNA targets. Cell 2005;120:15–20. [5] Bauersachs J, Thum T. Biogenesis and regulation of cardiovascular microRNAs. Circ Res 2011;109:334–47. [6] Esquela-Kerscher A, Slack FJ. Oncomirs — microRNAs with a role in cancer. Nat Rev Cancer 2006;6:259–69. [7] Lewis BP, Shih IH, Jones-Rhoades MW, Bartel DP, Burge CB. Prediction of mammalian microRNA targets. Cell 2003;115:787–98. [8] Fang Z, Rajewsky N. The impact of miRNA target sites in coding sequences and in 3′UTRs. PLoS One 2011;6:e18067. [9] da Costa Martins PA, Bourajjaj M, Gladka M, Kortland M, van Oort RJ, Pinto YM, et al. Conditional dicer gene deletion in the postnatal myocardium provokes spontaneous cardiac remodeling. Circulation 2008;118:1567–76. [10] Thum T. Cardiac dissonance without conductors: how dicer depletion provokes chaos in the heart. Circulation 2008;118:1524–7. [11] van Rooij E, Sutherland LB, Liu N, Williams AH, McAnally J, Gerard RD, et al. A signature pattern of stress-responsive microRNAs that can evoke cardiac hypertrophy and heart failure. Proc Natl Acad Sci USA 2006;103:18255–60. [12] Thum T, Galuppo P, Wolf C, Fiedler J, Kneitz S, van Laake LW, et al. MicroRNAs in the human heart: a clue to fetal gene reprogramming in heart failure. Circulation 2007;116:258–67. [13] Callis TE, Pandya K, Seok HY, Tang RH, Tatsuguchi M, Huang ZP, et al. MicroRNA208a is a regulator of cardiac hypertrophy and conduction in mice. J Clin Invest 2009;119:2772–86. [14] Care A, Catalucci D, Felicetti F, Bonci D, Addario A, Gallo P, et al. MicroRNA-133 controls cardiac hypertrophy. Nat Med 2007;13:613–8. [15] Kuehbacher A, Urbich C, Zeiher AM, Dimmeler S. Role of Dicer and Drosha for endothelial microRNA expression and angiogenesis. Circ Res 2007;101:59–68.
[16] van Rooij E, Sutherland LB, Thatcher JE, DiMaio JM, Naseem RH, Marshall WS, et al. Dysregulation of microRNAs after myocardial infarction reveals a role of miR-29 in cardiac fibrosis. Proc Natl Acad Sci USA 2008;105:13027–32. [17] Thum T, Gross C, Fiedler J, Fischer T, Kissler S, Bussen M, et al. MicroRNA-21 contributes to myocardial disease by stimulating MAP kinase signalling in fibroblasts. Nature 2008;456:980–4. [18] van Rooij E, Sutherland LB, Qi X, Richardson JA, Hill J, Olson EN. Control of stressdependent cardiac growth and gene expression by a microRNA. Science 2007;316: 575–9 (New York, N Y. [19] Small EM, Olson EN. Pervasive roles of microRNAs in cardiovascular biology. Nature 2011;469:336–42. [20] Bonauer A, Carmona G, Iwasaki M, Mione M, Koyanagi M, Fischer A, et al. MicroRNA-92a controls angiogenesis and functional recovery of ischemic tissues in mice. Science 2009;324:1710–3 New York, N Y. [21] Krutzfeldt J, Rajewsky N, Braich R, Rajeev KG, Tuschl T, Manoharan M, et al. Silencing of microRNAs in vivo with ‘antagomirs’. Nature 2005;438:685–9. [22] da Costa Martins PA, Salic K, Gladka MM, Armand AS, Leptidis S, el Azzouzi H, et al. MicroRNA-199b targets the nuclear kinase Dyrk1a in an auto-amplification loop promoting calcineurin/NFAT signalling. Nat Cell Biol 2010;12:1220–7. [23] Lanford RE, Hildebrandt-Eriksen ES, Petri A, Persson R, Lindow M, Munk ME, et al. Therapeutic silencing of microRNA-122 in primates with chronic hepatitis C virus infection. Science 2010;327:198–201 (New York, N Y. [24] Thomas M, Lieberman J, Lal A. Desperately seeking microRNA targets. Nat Struct Mol Biol 2010;17:1169–74. [25] Matkovich SJ, Van Booven DJ, Eschenbacher WH, Dorn II GW. RISC RNA sequencing for context-specific identification of in vivo microRNA targets. Circ Res 2011;108:18–26. [26] Grimson A, Farh KK, Johnston WK, Garrett-Engele P, Lim LP, Bartel DP. MicroRNA targeting specificity in mammals: determinants beyond seed pairing. Mol Cell 2007;27:91–105. [27] Friedman RC, Farh KK, Burge CB, Bartel DP. Most mammalian mRNAs are conserved targets of microRNAs. Genome Res 2009;19:92–105. [28] Krek A, Grun D, Poy MN, Wolf R, Rosenberg L, Epstein EJ, et al. Combinatorial microRNA target predictions. Nat Genet 2005;37:495–500. [29] Grun D, Wang YL, Langenberger D, Gunsalus KC, Rajewsky N. microRNA target predictions across seven Drosophila species and comparison to mammalian targets. PLoS Comput Biol 2005;1:e13. [30] Lall S, Grun D, Krek A, Chen K, Wang YL, Dewey CN, et al. A genome-wide map of conserved microRNA targets in C. elegans. Curr Biol 2006;16:460–71. [31] Chen K, Rajewsky N. Natural selection on human microRNA binding sites inferred from SNP data. Nat Genet 2006;38:1452–6. [32] Enright AJ, John B, Gaul U, Tuschl T, Sander C, Marks DS. MicroRNA targets in Drosophila. Genome Biol 2003;5:R1. [33] John B, Enright AJ, Aravin A, Tuschl T, Sander C, Marks DS. Human MicroRNA targets. PLoS Biol 2004;2:e363. [34] Betel D, Koppal A, Agius P, Sander C, Leslie C. Comprehensive modeling of microRNA targets predicts functional non-conserved and non-canonical sites. Genome Biol 2010;11:R90. [35] Hsu SD, Lin FM, Wu WY, Liang C, Huang WC, Chan WL, et al. miRTarBase: a database curates experimentally validated microRNA–target interactions. Nucleic Acids Res 2011;39:D163–9. [36] Dweep H, Sticht C, Pandey P, Gretz N. miRWalk — Database: prediction of possible miRNA binding sites by “walking” the genes of three genomes. J Biomed Inform 2011 [Electronic publication ahead of print]. [37] Fasanaro P, Greco S, Lorenzi M, Pescatori M, Brioschi M, Kulshreshtha R, et al. An integrated approach for experimental target identification of hypoxia-induced miR-210. J Biol Chem 2009;284:35134–43. [38] Jayaswal V, Lutherborrow M, Ma DD, Hwa Yang Y. Identification of microRNAs with regulatory potential using a matched microRNA–mRNA time–course data. Nucleic Acids Res 2009;37:e60. [39] Jayaswal V, Lutherborrow M, Ma DD, Yang YH. Identification of microRNA–mRNA modules using microarray data. BMC Genomics 2011;12:138. [40] Bang-Berthelsen CH, Pedersen L, Floyel T, Hagedorn PH, Gylvin T, Pociot F. Independent component and pathway-based analysis of miRNA-regulated gene expression in a model of type 1 diabetes. BMC Genomics 2011;12:97. [41] Baek D, Villen J, Shin C, Camargo FD, Gygi SP, Bartel DP. The impact of microRNAs on protein output. Nature 2008;455:64–71. [42] Selbach M, Schwanhausser B, Thierfelder N, Fang Z, Khanin R, Rajewsky N. Widespread changes in protein synthesis induced by microRNAs. Nature 2008;455:58–63. [43] Ong SE, Blagoev B, Kratchmarova I, Kristensen DB, Steen H, Pandey A, et al. Stable isotope labeling by amino acids in cell culture, SILAC, as a simple and accurate approach to expression proteomics. Mol Cell Proteomics 2002;1:376–86. [44] Yang Y, Chaerkady R, Kandasamy K, Huang TC, Selvan LD, Dwivedi SB, et al. Identifying targets of miR-143 using a SILAC-based proteomic approach. Mol Biosyst 2010;6:1873–82. [45] Yan GR, Xu SH, Tan ZL, Liu L, He QY. Global identification of miR-373-regulated genes in breast cancer by quantitative proteomics. Proteomics 2011;11:912–20. [46] Elkan-Miller T, Ulitsky I, Hertzano R, Rudnicki A, Dror AA, Lenz DR, et al. Integration of transcriptomics, proteomics, and microRNA analyses reveals novel microRNA regulation of targets in the mammalian inner ear. PLoS One 2011;6: e18195. [47] Yang Y, Chaerkady R, Beer MA, Mendell JT, Pandey A. Identification of miR-21 targets in breast cancer cells using a quantitative proteomic approach. Proteomics 2009;9:1374–84. [48] Leivonen SK, Rokka A, Ostling P, Kohonen P, Corthals GL, Kallioniemi O, et al. Identification of miR-193b targets in breast cancer cells and systems biological analysis of their functional impact. Mol Cell Proteomics 2011;10:M110.005322.
J. Fiedler et al. / Journal of Molecular and Cellular Cardiology 51 (2011) 674–681 [49] Fleissner F, Jazbutyte V, Fiedler J, Gupta SK, Yin X, Xu Q, et al. Short communication: asymmetric dimethylarginine impairs angiogenic progenitor cell function in patients with coronary artery disease through a microRNA-21-dependent mechanism. Circ Res 2010;107:138–43. [50] Brandes RP. A miR-thless perspective: how asymmetric dimethylarginine impairs the functions of angiogenic progenitor cells. Circ Res 2010;107:12–4. [51] Huang Y, Qi Y, Ruan Q, Ma Y, He R, Ji Y, et al. A rapid method to screen putative mRNA targets of any known microRNA. Virol J 2011;8:8. [52] Li J, Xia W, Huang B, Chen L, Su X, Li S, et al. A strategy to rapidly identify the functional targets of microRNAs by combining bioinformatics and mRNA cytoplasmic/ nucleic ratios in culture cells. FEBS Lett 2010;584:3198–202. [53] Orom UA, Lund AH. Isolation of microRNA targets using biotinylated synthetic microRNAs. Methods 2007;43:162–5. [54] Fiedler J, Jazbutyte V, Kirchmaier BC, Gupta SK, Lorenzen J, Hartmann D, et al. MicroRNA-24 regulates vascularity after myocardial infarction. Circulation 2011. [55] Vo NK, Dalton RP, Liu N, Olson EN, Goodman RH. Affinity purification of microRNA-133a with the cardiac transcription factor, Hand2. Proc Natl Acad Sci USA 2010;107:19231–6. [56] Jain R, Devine T, George AD, Chittur SV, Baroni TE, Penalva LO, et al. RIP-Chip analysis: RNA-binding protein immunoprecipitation-microarray (Chip) profiling. Methods Mol Biol 2011;703:247–63. [57] Tan LP, Seinen E, Duns G, de Jong D, Sibon OC, Poppema S, et al. A high throughput experimental approach to identify miRNA targets in human cells. Nucleic Acids Res 2009;37:e137. [58] Zhang L, Ding L, Cheung TH, Dong MQ, Chen J, Sewell AK, et al. Systematic identification of C. elegans miRISC proteins, miRNAs, and mRNA targets by their interactions with GW182 proteins AIN-1 and AIN-2. Mol Cell 2007;28:598–613. [59] Landthaler M, Gaidatzis D, Rothballer A, Chen PY, Soll SJ, Dinic L, et al. Molecular characterization of human Argonaute-containing ribonucleoprotein complexes and their bound target mRNAs. RNA 2008;14:2580–96. [60] Keene JD, Komisarow JM, Friedersdorf MB. RIP-Chip: the isolation and identification of mRNAs, microRNAs and protein components of ribonucleoprotein complexes from cell extracts. Nat Protoc 2006;1:302–7.
681
[61] Prokopi M, Mayr M. Proteomics: a reality-check for putative stem cells. Circ Res 2011;108:499–511. [62] Nelson PT, Wang WX, Mao G, Wilfred BR, Xie K, Jennings MH, et al. Specific sequence determinants of miR-15/107 microRNA gene group targets. Nucleic Acids Res 2011. [63] Nonne N, Ameyar-Zazoua M, Souidi M, Harel-Bellan A. Tandem affinity purification of miRNA target mRNAs (TAP-Tar). Nucleic Acids Res 2010;38:e20. [64] Chi SW, Zang JB, Mele A, Darnell RB. Argonaute HITS-CLIP decodes microRNA– mRNA interaction maps. Nature 2009;460:479–86. [65] Leung AK, Young AG, Bhutkar A, Zheng GX, Bosson AD, Nielsen CB, et al. Genome-wide identification of Ago2 binding sites from mouse embryonic stem cells with and without mature microRNAs. Nat Struct Mol Biol 2011;18: 237–44. [66] Hafner M, Landthaler M, Burger L, Khorshid M, Hausser J, Berninger P, et al. Transcriptome-wide identification of RNA-binding protein and microRNA target sites by PAR-CLIP. Cell 2010;141:129–41. [67] Hafner M, Landthaler M, Burger L, Khorshid M, Hausser J, Berninger P, et al. PARCliP—a method to identify transcriptome-wide the binding sites of RNA binding proteins. J Vis Exp 2010(41), doi:10.3791/2034 pii: 2034. [68] Fichtlscherer S, De Rosa S, Fox H, Schwietz T, Fischer A, Liebetrau C, et al. Circulating microRNAs in patients with coronary artery disease. Circ Res 2010;107: 677–84. [69] Zampetaki A, Kiechl S, Drozdov I, Willeit P, Mayr U, Prokopi M, et al. Plasma microRNA profiling reveals loss of endothelial miR-126 and other microRNAs in type 2 diabetes. Circ Res 2010;107:810–7. [70] Barrey E, Saint-Auret G, Bonnamy B, Damas D, Boyer O, Gidrol X. Pre-microRNA and mature microRNA in human mitochondria. PLoS One 2011;6:e20220. [71] Yeom KH, Heo I, Lee J, Hohng S, Kim VN, Joo C. Single-molecule approach to immunoprecipitated protein complexes: insights into miRNA uridylation. EMBO Rep 2011;12:690–6. [72] Ponting CP, Oliver PL, Reik W. Evolution and functions of long noncoding RNAs. Cell 2009;136:629–41.
Contents lists available at SciVerse ScienceDirect
Journal of Molecular and Cellular Cardiology journal homepage: www.elsevier.com/locate/yjmcc
Review article
Identification of cardiovascular microRNA targetomes J. Fiedler a, S.K. Gupta a, T. Thum a, b,⁎ a b
Institute of Molecular and Translational Therapeutic Strategies (IMTTS), Germany Centre for Clinical and Basic Research, IRCCS San Raffaele, Rome, Italy
a r t i c l e
i n f o
Article history: Received 12 July 2011 Received in revised form 11 August 2011 Accepted 12 August 2011 Available online 24 August 2011 Keywords: Cardiac disease miRNA miRNA targetome RISCome RNA immunoprecipitation MRE Target prediction
a b s t r a c t MicroRNAs (miRNAs) are strong post-transcriptional regulators targeting multiple targets. Endogenously transcribed, miRNAs specifically bind to complementary sequences of mRNAs and repress their expression thus govern control of cellular signaling pathways. An altered miRNA expression is causally related to cardiovascular disease. Identification of miRNA-dependent pathways is therefore an important aim to develop new therapeutic approaches. To understand miRNA function in various cardiovascular cells, the identification of individual miRNA target genes is of utmost importance. Indeed, the biological function of a miRNA is dependent on the availability of potential targets in a cell. We here summarize and discuss current challenging approaches to identify miRNA targetomes which will help to understand miRNA function in cardiac homeostasis and disease. © 2011 Elsevier Ltd. All rights reserved.
Contents 1.
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.1. MicroRNAs (miRNAs)—biogenesis and biological function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.2. Cardiovascular disease is associated to miRNA deregulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2. Identification of miRNA target genes (the miRNA targetome or RISCome) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1. Bioinformatics, proteomics and other molecular biology-related approaches . . . . . . . . . . . . . . . . . . . . . . 2.1.1. Bioinformatics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2. Targetscan . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3. PicTar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.4. miRanda-mirSVR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.5. MicroCosm targets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.5.1. Transcriptomics, proteomics and other biochemical approaches . . . . . . . . . . . . . . . . . . . . . . . . . 2.6. Transcriptome and proteome approaches . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.7. Biochemical approaches . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.7.1. RIP (RNA immunoprecipitation)-Chip, RISCome (RNA-induced silencing complex) analysis . . . . . . . . . . . . 2.7.2. RNA-sequencing, HITS-CLIP (high-throughput sequencing of RNAs isolated by crosslinking immunoprecipitation), PAR-CLIP (photoactivatable-ribonucleoside-enhanced crosslinking and immunoprecipitation) . . . . . . . . . . . . . . . . . . 3. Conclusion and perspectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conflict of interest statement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
675 674 675 675 675 675 675 676 676 676 676 676 676 677 677
. . . .
. . . .
. . . .
. . . .
. . . .
678 679 680 680
⁎ Corresponding author at: Institute of Molecular and Translational Therapeutic Strategies (IMTTS), Hannover Medical School, Carl-Neuberg-Str. 1, 30625 Hannover, Germany. Tel.: + 49 511 532 5273; fax: + 49 511 532 5274. E-mail address: [email protected] (T. Thum). 0022-2828/$ – see front matter © 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.yjmcc.2011.08.017
J. Fiedler et al. / Journal of Molecular and Cellular Cardiology 51 (2011) 674–681
1. Introduction 1.1. MicroRNAs (miRNAs)—biogenesis and biological function Since discovered in the early nineties, microRNAs (miRNAs) have been shown to play major roles in different physiological settings by regulating gene expression post-transcriptionally [1–3]. Of note, miRNAs regulate 30–50% of the genome and thus are potent mediators of cellular signaling [4]. Several details about the regulation of miRNAs and their biogenesis especially in the context of cardiovascular biology have recently been reviewed [5] and therefore are not a main topic of this review article. Briefly, miRNAs are either transcribed from miRNA genes or spliced from host gene transcripts (overview on biogenesis reviewed in [6]). Once formed as so called primary (pri)-miRNAs in the nucleus, the Drosha complex cleaves the RNA to a stem– loop precursor structure (precursor (pre)-miRNA) of about 60–70 nucleotides in length. Cytosolic translocation mediated by exportin proteins triggers further processing at the Dicer complex. Here, the miRNA heteroduplex is unwinded to its biological active singlestranded format. The second strand (designated as miRNA*) is often not selected and subsequently degraded. RNA-binding proteins, e.g. proteins of the Ago family then recruit the mature miRNA to sites of RISC (RNA induced silencing complex). An antisense mechanism then mediates miRNA-dependent recognition in RISC to target mRNAs. Most commonly, miRNA responsive elements are found in 3′-UTR (untranslated region) of the mRNA and base pairing determines stringency of miRNA:mRNA interaction. In line, the strength of miRNA binding to the mRNA is crucially mediated by nucleotides 2 to 8 (the seed) and surrounding bases [7]. However, recently the central dogma of miRNA binding to 3′-UTR region has been challenged, as reports demonstrated additional miRNA binding to coding regions or even 5′-UTRs [8]. However, the main characteristics of a single miRNA or even miRNA families sharing the identical seed sequence, are their ability to simultaneously target many mRNAs which are accessible in a specific cellular context. It is crucial to decipher miRNA target genes in a single cell by different bioinformatic or experimental approaches to substantially identify intracellular signaling pathways especially in terms of cardiac disease. The still growing number of newly identified miRNAs emphasizes their cellular impact, calls for sequence annotation and a summary of characteristics in open-access databases (see below). 1.2. Cardiovascular disease is associated to miRNA deregulation MiRNAs are crucial in the control of cardiovascular signal transduction in the heart. Indeed, knockdown of the miRNA biogenesis related enzyme Dicer in cardiomyocytes led to cardiac remodeling, enhanced fibrosis and finally heart failure [9,10]. This key study of Costa Martins and colleagues emphasized the orchestrating role of cardiac miRNA expression to sustain and balance cardiac morphology and function. Cardiac disease animal models, such as experimental myocardial infarction lead to a differential pattern of miRNA expression [11]. Noteworthy, identified miRNAs deregulated in animal models of heart diseases such as miR-21 or miR-208a are also upregulated in human heart failure [11]. When comparing miRNA expression profiles in fetal and failing human hearts, many fetal miRNAs became re-expressed during heart failure demonstrating a potential role for miRNAs to be involved in general fetal gene reprogramming during heart failure [12]. In vitro studies further confirmed that modulating miRNA expression can trigger molecular and phenotypic changes in different cardiovascular cells. For instance, modulating miR-133 or miR-208a in cardiomyocytes had a strong impact on hypertrophic response of cardiomyocytes [13,14]. Modulation of miRNA biogenesis by Dicer or Drosha knockdown in endothelial cells interfered with the angiogenic response [15]. Manipulation of miR-21 or miR-29 expression in fibroblasts induced changes in
675
fibroblast proliferation or collagen expression implicating a potential therapeutic interest [16,17]. Using genetic animal techniques, miRNA gain- and loss-of function studies also revealed an essential role for many miRNAs in cardiac homeostasis and disease [18,19]. Taken together, these important findings have also been transferred to relevant therapeutic approaches based on antagomir or antimiR-mediated knockdown of miRNA expression in murine models of cardiac disease [17,20–22]. Of note, miRNA modulators were also translated into bigger animal studies such as in a primates' model of hepatitis C infection [23] and currently miRNA-based therapeutics also have entered the clinics based on ongoing phase I and II studies. 2. Identification of miRNA target genes (the miRNA targetome or RISCome) In the field of miRNA research, identifying miRNA target genes is one of the most important but unfortunately one of the toughest goals requiring different approaches (excellently reviewed in [24]). Detailed knowledge of downstream effectors of miRNAs is important to estimate their individual functional relevance in healthy conditions but especially during disease. This information is especially needed when miRNA expression is modulated by miRNA-therapeutics such as antagomirs or anti-miRs. Here, new tools were recently developed to investigate miRNA target genes on a more global level (miRNA targetome or the RISCome analysis) in cardiac cells [25]. Different technical approaches to uncover miRNA targets are now available and will be presented and discussed in the following sections (see Fig. 1). 2.1. Bioinformatics, proteomics and other molecular biology-related approaches 2.1.1. Bioinformatics Computational analysis of potential miRNA target genes relies mainly on open-access databases quite easily to use for the scientific community. Several working groups have developed software algorithms as target prediction tools which seem to be the first choice in miRNA target search. In common, the software tools screen mRNAs 3′-UTR for putative miRNA-responsive elements (MRE). As previously reported, MRE are more likely bound by miRNA when they are located either in the beginning or the end of the 3′-UTR of a gene [26]. Tandem repeats of MREs in a single 3′-UTR often enhance the possibility of miRNA binding. In contrast, secondary structure effects of RNA due to characteristic nucleotide sequence might impede miRNA binding in the center of a specific mRNA 3′-UTR. However, the underlying algorithms to predict targets vary and often the overlap of predicted
Fig. 1. Various approaches exist to delineate miRNA targets — the “targetome”. Bioinformatics and in silico predictions, RNA pulldown and genechip analysis, transcriptomics or proteomics and RNA sequencing methods can be applied to identify miRNA targets on a cellular level.
676
J. Fiedler et al. / Journal of Molecular and Cellular Cardiology 51 (2011) 674–681
target genes is small. In the proceeding paragraphs we will summarize the most important in silico features of several miRNA target prediction databases, such as Targetscan, PicTar, miRanda-mirSVR and MicroCosm, all of which are helpful bioinformatic tools to develop a general overview of potential miRNA targets. 2.2. Targetscan The Targetscan database is accessible via www.targetscan.org and is updated regularly by the Whitehead Institute for Biomedical Research, Cambridge, US. Recently, version 5.2 has been launched and offers the possibility for miRNA target searches in humans, mouse, worm and fly genomes. Primarily, the Targetscan-algorithm detects conserved seven to eight nucleotide (7-mer/8-mer) matches between the miRNA seed region and responsive mRNA 3′-UTRs [4]. Additionally, common target sites are listed, which may have a nucleotide mismatch in the seed region but can facilitate miRNA binding via supporting base pairing in the 3′-direction [27]. These potential binding events can also be applied for non-conserved target sites among the species. The final output list of target sites or target genes is ranked by a targetscan context score accordingly to Grimson et al. [26]. Here, four distinct features of miRNA binding are considered for the calculation: site-type, 3′ pairing, local sequence AU content and position contribution. The lowest score defines the most favorable chance of target regulation by direct miRNA:mRNA interaction. 2.3. PicTar PicTar has been developed by the Rajewsky lab at the Max Delbruck Centrum, Berlin and is online at www.pictar.mdc-berlin.de. It is a valuable source for potential miRNA target genes and is frequently used. The PicTar database also covers a wide range of species including vertebrates, Drosophila and nematode species [28–30]. This interface also highlights conserved miRNA binding sites and as a special feature, co-expressed miRNA and target mRNAs are listed if conservation is not the case [31]. The overall PicTar score is finally ranked in a consecutive list from most to less likely and can be applied for target identification. 2.4. miRanda-mirSVR The interconnected database miRanda-mirSVR is found at www. microrna.org and is provided by the Memorial Sloan Kettering Cancer Center, New York. Predictions are provided for several species and interestingly, various database links exist to support additional information e.g. on miRNA function and expression level in different tissues. The algorithm is based on sequence match, free energy calculation and evolutionary conservation [32]. These characteristics are summarized by miRanda [33], whereas the key score has been developed by mirSVR [34]. Taken together, target gene repression by miRNA binding is calculated and additional alignment views are given to depict miRNA binding. 2.5. MicroCosm targets This web interface is online at www.ebi.ac.uk/enright-srv/ microcosm/htdocs/targets/v5/ and was developed by the Enright Lab at the EMBL-EBI, Cambridge, UK. MicroCosm Targets also encompasses a comprehensive miRNA target prediction database and is regularly updated with new information. In accordance with miRandamirSVR, computational analysis is performed with the miRanda algorithm [33]. Perfect sequence complementarity within the seed region of the miRNA and target mRNA is of utmost importance for the calculation. Non-perfect matches are immediately negatively selected and thus are not processed. Thermodynamic stability of the mRNA– miRNA heteroduplex is acquired via the Vienna RNA folding routines
(www.tbi.univie.ac.at/~ivo/RNA/). Again, evolutionary conservation is considered for the MRE and must be present in at least two different species. Collectively, miRNA target prediction databases provide a first solid step into the identification of miRNA targets and offer additional valuable information including target gene annotation. It is obvious, that comparison and/or combination of target prediction algorithms should be used. This is a useful first step to identify target genes for a single miRNA or even miRNA families for further mechanistical or functional investigations. Here, we just introduce a small set of available databases and we think that other software tools might also be appropriate. These can be tools calculating free energy between two RNA sequences to hybridize, e.g. RNAhybrid provided by the Bielefeld University Bioinformatics Server. However as mentioned above, most predictions do not take into account if a certain target gene AND the miRNA are both expressed and/or available in the particular cell type of interest. Thus, cell-type specific expression of both miRNA and its targets needs to be assessed to narrow hits from the bioinformatic prediction. To summarize the current knowledge on previously reported miRNA/target gene expression, expression profiles in disease or after different treatments on miRNA expression specific databases such as miRGator (www.mirgator.kobic.re. kr:8080/MEXWebApp/index.jsp) or miRTarBase [35] have been developed and this is a great help in proper miRNA target identification. Next to these aforementioned conventional searches for miRNA binding sites in 3′-UTR, a preliminary database is available for screening whole genomes for miRNA binding (miRWalk, [36]) but this approach needs future evaluation to establish as a reliable source. Most importantly, computational predictions need to be validated, e.g. target gene expression should be detected via western blot when miRNA of interest is overexpressed in a particular cell type. In addition, novel technical approaches are now available as discussed below. 2.5.1. Transcriptomics, proteomics and other biochemical approaches Up on miRNA:mRNA base pairing targets are repressed at the mRNA and/or protein expression level. Thus deregulation of a miRNA leads to specific “foot-prints” in the targetome and/or proteome. Next to the aforementioned in silico target prediction analyses, several in vitro approaches have been developed to evaluate such direct interactions on a more global level. The ways to approach the changes in cellular signaling have in common that miRNA expression is correlated to target expression. Thereby, alteration of miRNA expression is accompanied by well-known “omic”-approaches describing high-throughput methods such as transcriptomics and proteomics. Next to these, other biochemical assays have been developed to support detection of miRNA targets. 2.6. Transcriptome and proteome approaches Analyzing conventional mRNA microarray-data sets after miRNA modulation of cells or organs identifies genes and gene network(s) being involved in the direct and indirect consequences of alteration of a miRNA. For instance, a certain miRNA can be modulated in cell cultures, whereas in controls only a scrambled miRNAs was used, followed by RNA isolation and transcriptome analysis by standard miRNA-microarrays. Several groups have used such an approach in the past. By way of example, hypoxia-regulated miR-210 was analyzed in such a way but using a combination of bioinformatics, transcriptomics and proteomics which will be discussed later on [37]. However, miRNA modulation itself does not only regulate direct miRNA targets but may have many secondary effects at the genome level. It is also difficult to identify the optimal time point for miRNA-mediated degradation of mRNAs using transcriptome approaches. To counteract such difficulties, bioinformatic-dependent statistics have been developed to support microarray data analysis
J. Fiedler et al. / Journal of Molecular and Cellular Cardiology 51 (2011) 674–681
[38,39]. Regulatory miRNAs are defined by an OR (odds-ratio)-statistic based on time-course expression of both miRNA and mRNAs as well as computational predictions. Time-course-dependent expression analysis on a microarray can help to estimate the correlation and significance of miRNA/mRNA sets. Applying the ORs on a set of cancer data the authors identified some miRNAs indeed having a regulatory potential towards transcriptional activity and thus cancer biology. Further improvement of the OR was gained by adding two clustering steps into the consideration based on the fact that single miRNAs target many mRNAs and vice versa. This clustering identified miRNA/mRNA (miRmR) modules that then can be studied for their biological impact. This approach can easily adapted to cardiovascular miRNA research for improved identification of cell-type specific miRNA targets. In vivo, investigating fetal and failing hearts linked alterations of target genes, the number of individual miRNA bindings sites in those targets and tissue-specific miRNA levels [12]. Interestingly, mRNAs with a high number of binding sites for miRNAs which were upregulated in failing human heart tissue were more likely repressed, whereas those with many binding sites for miRNAs, which were downregulated, tended to be more upregulated. Another paper reported the use of independent component analysis for miRNA profiling in a type 1 diabetes model [40]. The processing of microarray data was combined to computational calculations and also cross-checked against a prediction relying on the negative correlation between miRNA and target mRNA. However, in some cases the interaction of a miRNA with a mRNA target does not lead to mRNA degradation but to translational repression leading to alterations at the protein level. Thus proteome analyses may be of help to identify the targetome of individual miRNAs. Initially, two groups have applied miRNA modulation and amino acid labeling to investigate proteomic changes induced by a miRNA (see Table 1, [41,42]). SILAC (stable isotope labeling with amino acids in cell culture, [43] was combined with miRNA overexpression to measure the impact on protein output. Such approaches demonstrate a widespread change on cellular protein constitution upon miRNA modulation, even if the effects on protein level have been reported as moderate. Very recently, in the field of cancer, various quantitative proteomic approaches were performed and identified downstream targets for miR-143/373 [44,45]. Interestingly, observed changes at the protein level of miR targets did not coincide with decreased expression at the mRNA level [45]. Integrative analysis to dissect mammalian inner ear miRNA pathways also identified miRNA targets repressed in the proteomics approach only [46]. Different peptide labeling using iTRAQ (isobaric tag for relative quantitation) facilitated the identification of miR-21/193b targets emphasizing the general ability of proteomics to delineate miRNA targets [47,48]. Overexpression of miR-21 in endothelial cells followed by proteome analysis, using difference in-gel electrophoresis and subsequent mass spectrometric analysis of regulated proteins identified a number of deregulated proteins such as superoxide dismutase 2 [49]. This approach shows how rigorous utilization of
Table 1 Publications listed applying high-throughput screening to identify miRNA targets. High-throughput miRNA target identification miRNA(s)
Methodology
Reference
miR-133a, -499 miR-223, -1, -124, -181 miR-155, -1, -16, -30a and let-7b miR-21
RISC-Seq SILAC/Proteomics SILAC/Proteomics
[25] [41] [42]
2D-DIGE (differential in-gel electrophoresis)/Proteomics RIP-Chip, Microarray RIP-Chip, Microarray HITS-CLIP PAR-CLIP
[49]
miR-17/20/93/106 miR-103, -107,-16, -195 miR-124 miR-7, -124
[57] [62] [64] [66,67]
677
unbiased screening technologies can be used for mechanistic research [50]. 2.7. Biochemical approaches A bottleneck of the “omics”-based approaches such as microarrays or proteomics is their relatively high costs. Novel rapid assays have been now introduced to gain insight to miRNA target genes. A PCR technique designated “Hybrid-PCR” was used to screen for miRNA target genes in the genome of human cytomegalovirus lacking a complete mRNA annotation [51]. In HeLa and HepG2 cells miRNA targets were identified by determining the C/N (cytoplasmic to nucleic) ratio of mRNA after miRNA overexpression [52]. Here, a higher mRNA C/N ratio indicated a greater chance for a gene to be targeted by the miRNA. Biotin-tagged miRNAs have also been employed in a streptavidin–biotin pulldown system to demonstrate miRNA:mRNA interactions [53]. A recent study investigated miR-24 function in myocardial infarction (MI) and its outcome for endothelial cell biology [54]. Use of miR-24 overexpression in cultured endothelial cells and subsequent analysis of putative targets via western blot and luciferase reporter gene assays revealed an angiogenic signaling network regulating angiogenic response. Luciferase reporter gene assay is an appropriate tool to evaluate miRNA binding. The respective 3′-UTR of a target mRNA containing the MRE is cloned to luciferase gene and co-transfected with miRNA of interest and controls. Quite oftenly, the wild-type MRE is mutated at some nucleotides and experiments are performed according to wild-type conditions to indicate that perfect sequence complementary within the MRE is needed for miRNA target repression. Target mRNAs are also capable to function as a bait as described for cardiac transcription factor Hand2 [55]. A tagging strategy was applied within a reporter system to demonstrate binding of miR-133a to Hand2 3′-UTR. Of note, addition of biotinylated miR-1 revealed cooperative association of miR-1 and miR-133a with Hand2 3′-UTR. Affinity purifications may thus also be of interest when considering approaches to define miRNA target genes. 2.7.1. RIP (RNA immunoprecipitation)-Chip, RISCome (RNA-induced silencing complex) analysis A significant hurdle in the understanding of the biological function of an individual miRNA is to identify its target mRNAs precisely in particular cellular and event/disease contexts. Use of microarray and proteomics techniques as stated above result in first insights of what signaling networks may be affected by a single miRNA but such approaches do not necessarily identify direct targets as many secondary changes occur. In contrast to plants, where regularly a complete sequence complementarity between the miRNA and its target gene is observed, animals' miRNA/mRNA complementarity is significantly less making the identification of miRNA targets much more complicated. The RIP-Chip technique (RNA-binding protein immunoprecipitation followed by Genechip, Microarray) is a basic profiling approach to define miRNA targets in a high-throughput manner and combines genomic technologies with standard molecular biology tools to define miRNA target genes—the “targetome” [56] (see also Fig. 2). Briefly, members of the miRNA ribonucleoprotein complex such as argonaute [57], GW182 [58] and TNRC6 family members [59] or others are immunoprecipitated. Total RNA is isolated from the complexes and analyzed on a microarray which was also reported in a RNA immunoprecipitation-genechip protocol [57,60] to identify direct miRNA targets in cellular context. Briefly, RIP-Chip is the immunoprecipitation (IP) of RNA binding protein (RBP) complexes followed by RNA extraction and chip-based identification. In vitro, around 5–20 million cells are needed for the IP procedure along with antibody-coated sepharose beads. Finally, RNA is isolated from the IP-derived samples. The isolated RNA can then be processed for microarray (RIP-Chip) or sequencing analysis (RIP-Seq). Tan et al. have used RIP-Chip for identifying miRNA targets in Hodgkin
678
J. Fiedler et al. / Journal of Molecular and Cellular Cardiology 51 (2011) 674–681
Fig. 2. Target identification technique in miRNA research using RNA immunoprecipitation and genechip analysis (RIP-Chip). In this procedure cells are pre-treated either with miRNA precursor (pre-miR) or inhibitor (anti-miR) and scrambled miRNA (scr-miR). In case of pre-miR transfection, target mRNAs should be enriched with Ago2 whereas endogenous miRNA inhibition would lead to target mRNA depletion from Ago2. Finally, mRNA:miRNA-Ago2 complexes are immunoprecipitated and undergo RNA analysis. Recovered RNA is hybridized to a genechip and analyzed with bioinformatic tools.
Lymphoma cells [57]. Under basal conditions, wildtype argonaute 2 (Ago2) was used to immunoprecipitate RISC and the isolated RNA was subjected to gene array which showed enrichment for around 3000. Enriched probes were termed as miR-targetome and ~30–40% of the identified probes were indeed direct targets of the top 5% expressed miRNAs based on the presence of 8-mer seed sequences and Targetscan prediction analyses. These results show a successful enrichment of miRNA targets in the Ago2-IP fraction. To extend this approach, specific miRNA inhibition leading to mRNA depletion from Ago2 and RISC was used endogenously to decipher targets of the highly enriched miR-17 family. The depleted gene transcripts were categorized into three categories based on level of depletion. Gene transcripts with less than 2-fold depletion were included in non-miR-17 targets, probes with ≥2 depletion in miR-17 targets and a third group of transcripts with ≥4–7 fold depletion were termed as potent miR-17 targets. To summarize, about 50% (miR-17 targets) and 67% (potent miR-17 targets) of transcripts respectively, were depleted upon antisense inhibition and contained at least a 6mer or 8-mer seed sequence for the miR-17 family as demonstrated by bioinformatic analysis. This emphasizes a general suitability of this approach for the successful genome-wide identification of endogenous targets of selected miRNAs or miRNA families. A similar approach has been used by Zhang et al. who immunoprecipitated the GW182 proteins AIN-1 and AIN-2 of Caenorhabditis elegans to identify the respective miRNA targetome [58]. A comprehensive method has also been developed by Fasanaro et al. in order to identify targets of hypoxia-inducible miR-210 [37]. The authors combined a set of approaches (proteomics, transcriptomics, overexpression or inhibition of miRNAs) followed by RISC immunoprecipitation methods. The integration of different technical aspects revealed several new targets with MREs both in 3′-UTR but interestingly also in 5′-UTR and coding sequences. In order to reduce false-positive hits, transcripts which were reciprocally regulated (e.g. downregulated on overexpression as well as upregulated on inhibition of miR-210), were selected from transcriptomic profiling and 51 targets were found to comply these obligations representing potential targets for miR-210. Importantly, the targets that were downregulated in both the proteomic and transcriptome analyses showed no similarity raising a number of questions. One explanation could be that the time-dependent differences upon miRNA transfection in RNA and protein levels may differ at various time points. Secondly, proteomic techniques may not detect miRNA-induced changes accurately enough and other
approaches might be useful (reviewed in [61]). Fasanaro et al. additionally performed IP of Ago2–RNA complex upon miR-210 overexpression and controls. Some of the identified targets by proteomics and transcriptomics were found to be enriched in precipitates from miR-210 overexpressing cells. A very interesting finding was the enrichment of non-coding RNA Xist which was later on confirmed by qPCR highlighting downregulation of Xist RNA levels on miR-210 overexpression in HEK293 cells. The potential regulation of another non-coding RNA (Xist) by miR-210 points to a new mode of noncoding RNA–RNA regulation. A very recent study from Nelson et al. also used the RIP-Chip method to identify targets for miR-103, mir107, miR-16 and miR-195 [62]. In line with others, they also found that seed sequences for microRNAs are also present in the coding region along with 3′ UTR. Seed sequences for miR-107 along with other group members were found to be more enriched in the coding region (CDS) compared to 3′ and 5′ UTR in the immunoprecipitated fraction. Mechanistically, mutation of miR-107 in the 3′ region induced precipitate enrichment of the 3′ UTR sequence instead of CDS while mutations in the miRNA 5′ region (seed) did not. Despite having been mutated at the seed region, miRNA could still attach to a coding sequence although with an altered seed sequence complementarily comprising a new seed region. These observations evaluate the crucial role for the miRNA 3′ region in targeting a coding sequence of mRNAs. Another study termed TAP-Tar (tandem affinity purification of miRNA target mRNAs) also utilizes the concept of RIP-Chip in combination with use of biotin tagged miRNAs and Flag-tagged argonaute protein [63]. The authors first immunoprecipitate Ago–RISC complex followed by subsequent affinity purification with streptavidin beads. This method is applicable for miRNA-specific purification of target genes and importantly decreases the false positive rate by the two step purification procedure. 2.7.2. RNA-sequencing, HITS-CLIP (high-throughput sequencing of RNAs isolated by crosslinking immunoprecipitation), PAR-CLIP (photoactivatable-ribonucleoside-enhanced crosslinking and immunoprecipitation) Further technical advances of the aforementioned miRNA target identification strategies resulted in challenging protocols applying RNA:protein crosslinking and sequencing. Very recently, an elegant combination of both Ago2 IP (see above) and RNA-sequencing has been developed to identify the miRNA targetome of cardiac miR133a and miR-499 [25]. This approach is useful to identify miRNA
J. Fiedler et al. / Journal of Molecular and Cellular Cardiology 51 (2011) 674–681
targets more accurately with less false positive enrichment. Matkovich et al. directly immunoprecipitated the Ago2–RISC complex from murine ventricular tissue and compared bound mRNA fraction to the complete cardiac transcriptome. Related to the cardiac transcriptome, a significant enrichment of 15% was present in Ago2 immunoprecipitated RNA eluate, defining the group of cardiac expressed mRNAs which can be targets of cardiac miRNAs (RISCome). This observation demonstrates the involvement of many potential miRNA targets in heart tissue in vivo. From the enriched transcripts, 71% had MREs validated by deep sequencing methods and overlapped with bioinformatic predictions by Targetscan. In order to delineate the role of individual miRNAs in cardiac biology, a cardiomyocytespecific overexpression of miRNA-133a in vivo has been used. The cardiac RISCome of transgenic mice contained 47.4% of targets for miR-133a as predicted by Targetscan whereas the cardiac RISCome of wildtype mice displayed less (34.5%) miR-133a targets. This is a considerable enrichment of specific targets in the RISC complex depending on the overexpression of a certain miRNA. Next, RISCome comparisons of two different miRNAs, miR-133a and miR-499, showed little similarity confirming the specificity of the RISCsequencing method in the identification of specific targetomes of different miRNAs in vivo. A very similar method defined as HITS-CLIP (high-throughput sequencing of RNAs isolated by crosslinking immunoprecipitation) is also available [64]. Here, the RISC complex consisting of miRNA, mRNA and RISC protein is UV-crosslinked in advance to IP. This initial treatment had several advantages including a more stringent purification, reduced levels of false positive hits, removal of other small RNAs like tRNAs and rRNAs as well as an increased RNA amount. The Ago-mRNA HITS-CLIP experiments from mouse brain validated binding (Ago tags) to various sequences with a composition of 1% 5′ UTR tags, 40% 3′ UTR tags, 25% CDS tags, 12% intron tags and interestingly 4% non-coding RNA tags thus speculating about a novel relative function of Argonaute protein and miRNA action. A study done by Leung et al. applied the HITS-CLIP method along with silencing experiments of the miRNA biogenesis machinery [65]. Specifically, the miRNA targetome from wildtype and Dicer knockout mouse embryonic stem cells (ESC) were analyzed with this method. The RISCome from Dicer−/− ESC was still characterized by RNA cross-linked to Ago2 suggesting interaction of Ago2 with mRNAs to
679
be miRNA-independent. An updated and improved version of HITSCLIP is termed PAR-CLIP (photoactivatable-ribonucleoside-enhanced crosslinking and immunoprecipitation) [66,67]. This cell-based cross-linking approach determines transcriptome-wide binding sites of cellular RNA/miRNA binding proteins at high resolution. The technique facilitates crosslinking by incorporation of nucleoside-analogue 4-thiouridine in RNA, which results in higher yield of cross-linked RNA. The incorporation of 4-thiouridine itself led to base conversions (T to C) which in turn can be used to identify the miRNA/mRNA binding site to Ago2 very precisely. The crosslinking of 4-thiouridine to amino acid residues possibly changes their chemical structure which favors incorporation of dG rather than dA by reverse transcriptase. The change in chemical structure could be an alteration in stacking or hydrogen bond donor/acceptor properties. Specifically, 84% of the Ago-crosslinked center region from different Ago IPs belongs to exonic regions while 14% belongs to intronic sequences showing a possible involvement of Ago proteins in splicing or miRNA mediated alternative splicing. Briefly, 100–400 million cells (about 20–40 times of the RIP-Chip approach) are usually needed to perform the PAR-CLIP method (see Fig. 3). Prior to cell lysis, cells are tagged with photoactivatable ribonucleoside 4-thiouridine enabling crosslinks between RNA and RNA binding proteins by UV irradiation. The crosslinked RNA/RNA binding protein complex is then immunoprecipitated and subsequently radiolabeled. The complex is then separated on a SDS-PAGE gel, excised and subjected to electroelution followed by RNA isolation. The isolated RNA is proceeded through standard cDNA library preparation, deep sequencing and finally bioinformatic analysis. A more sensitive method could be potentially developed by combining some of the discussed approaches using biotinylated (tagged) miRNAs along with PAR-CLIP. This would add a two step purification process removing false positives and also due to pulldown of specific miRNA, the identified targetome may be more specific rather than a complete cellular miRNA targetome. 3. Conclusion and perspectives Different technical strategies have been developed to define and identify miRNA targets in a more or less comprehensive manner. Bioinformatics, transcriptomics, proteomics, RNA crosslink and sequencing
Fig. 3. Schematic representation of the PAR-CLIP technique. Cells are treated with 4-thiouridine prior to UV cross linking at 365 nm. Then, cells are lysed and Ago or TNRC complexes are immunoprecipitated. The RNA in immunoprecipitated complex is radiolabeled, separated on a SDS-PAGE gel and finally isolated from an immunoprecipitated complex. RNA is deep sequenced and sequence reads are analyzed using bioinformatics to gain insight into targets of miRNAs.
680
J. Fiedler et al. / Journal of Molecular and Cellular Cardiology 51 (2011) 674–681
methods can be combined to decipher miRNA target genes (summary in Table 1). The discussed approaches vary in their procedure and data analysis but are of great interest to be translated to cardiovascular research, which currently is clearly in its infancy. To truly validate a miRNA target it is still necessary to perform additional in vitro experiments to confirm direct miRNA binding to a specific sequence in a given cellular context. Conventional reporter gene assays using luciferase reporter enzyme activity are probably the best choice to prove the miRNA/gene interaction. New aspects of miRNA research beyond the classical cellular level just appear on the horizon. Recent reports suggested miRNAs as potential biomarkers in the circulation or even in organelles like mitochondria [68–70]. Whether circulating miRNAs or mitochondrial miRNAs are also able to bind to mRNA sequences is not entirely known. Since developmental effort is still ongoing, technical approaches in other scientific fields may be translated to miRNA target identification research in the future. Single-molecule pulldown, detection of uridylation modification at the miRNA level has been recently reported and could assist the exploration of new miRNA-regulatory pathways [71]. Besides the role of miRNAs for cellular signaling other regulatory RNAs like long non-coding RNAs (lncRNAs) might also impact specifically on target genes or protein machineries (reviewed in [72]). The knowledge of miRNA target identification strategies is therefore useful to understand the underlying molecular mechanisms and maladaptive responses in disease triggered by powerful non-coding RNAs. An understanding of the targetome of cardiac miRNAs is of outstanding importance as miRNAs play an important role in many cardiovascular diseases and recently emerged as powerful therapeutic targets.
Conflict of interest statement The authors disclose support from the IFB-Tx (BMBF 01EO0802; T.T.) and DFG TH 903/10-1 (T.T.). T.T. and J.F. have filed patents in the field of cardiovascular miRNA diagnostics and therapeutics.
References [1] Lee RC, Feinbaum RL, Ambros V. The C. elegans heterochronic gene lin-4 encodes small RNAs with antisense complementarity to lin-14. Cell 1993;75:843–54. [2] Wightman B, Ha I, Ruvkun G. Posttranscriptional regulation of the heterochronic gene lin-14 by lin-4 mediates temporal pattern formation in C. elegans. Cell 1993;75:855–62. [3] Pasquinelli AE, Reinhart BJ, Slack F, Martindale MQ, Kuroda MI, Maller B, et al. Conservation of the sequence and temporal expression of let-7 heterochronic regulatory RNA. Nature 2000;408:86–9. [4] Lewis BP, Burge CB, Bartel DP. Conserved seed pairing, often flanked by adenosines, indicates that thousands of human genes are microRNA targets. Cell 2005;120:15–20. [5] Bauersachs J, Thum T. Biogenesis and regulation of cardiovascular microRNAs. Circ Res 2011;109:334–47. [6] Esquela-Kerscher A, Slack FJ. Oncomirs — microRNAs with a role in cancer. Nat Rev Cancer 2006;6:259–69. [7] Lewis BP, Shih IH, Jones-Rhoades MW, Bartel DP, Burge CB. Prediction of mammalian microRNA targets. Cell 2003;115:787–98. [8] Fang Z, Rajewsky N. The impact of miRNA target sites in coding sequences and in 3′UTRs. PLoS One 2011;6:e18067. [9] da Costa Martins PA, Bourajjaj M, Gladka M, Kortland M, van Oort RJ, Pinto YM, et al. Conditional dicer gene deletion in the postnatal myocardium provokes spontaneous cardiac remodeling. Circulation 2008;118:1567–76. [10] Thum T. Cardiac dissonance without conductors: how dicer depletion provokes chaos in the heart. Circulation 2008;118:1524–7. [11] van Rooij E, Sutherland LB, Liu N, Williams AH, McAnally J, Gerard RD, et al. A signature pattern of stress-responsive microRNAs that can evoke cardiac hypertrophy and heart failure. Proc Natl Acad Sci USA 2006;103:18255–60. [12] Thum T, Galuppo P, Wolf C, Fiedler J, Kneitz S, van Laake LW, et al. MicroRNAs in the human heart: a clue to fetal gene reprogramming in heart failure. Circulation 2007;116:258–67. [13] Callis TE, Pandya K, Seok HY, Tang RH, Tatsuguchi M, Huang ZP, et al. MicroRNA208a is a regulator of cardiac hypertrophy and conduction in mice. J Clin Invest 2009;119:2772–86. [14] Care A, Catalucci D, Felicetti F, Bonci D, Addario A, Gallo P, et al. MicroRNA-133 controls cardiac hypertrophy. Nat Med 2007;13:613–8. [15] Kuehbacher A, Urbich C, Zeiher AM, Dimmeler S. Role of Dicer and Drosha for endothelial microRNA expression and angiogenesis. Circ Res 2007;101:59–68.
[16] van Rooij E, Sutherland LB, Thatcher JE, DiMaio JM, Naseem RH, Marshall WS, et al. Dysregulation of microRNAs after myocardial infarction reveals a role of miR-29 in cardiac fibrosis. Proc Natl Acad Sci USA 2008;105:13027–32. [17] Thum T, Gross C, Fiedler J, Fischer T, Kissler S, Bussen M, et al. MicroRNA-21 contributes to myocardial disease by stimulating MAP kinase signalling in fibroblasts. Nature 2008;456:980–4. [18] van Rooij E, Sutherland LB, Qi X, Richardson JA, Hill J, Olson EN. Control of stressdependent cardiac growth and gene expression by a microRNA. Science 2007;316: 575–9 (New York, N Y. [19] Small EM, Olson EN. Pervasive roles of microRNAs in cardiovascular biology. Nature 2011;469:336–42. [20] Bonauer A, Carmona G, Iwasaki M, Mione M, Koyanagi M, Fischer A, et al. MicroRNA-92a controls angiogenesis and functional recovery of ischemic tissues in mice. Science 2009;324:1710–3 New York, N Y. [21] Krutzfeldt J, Rajewsky N, Braich R, Rajeev KG, Tuschl T, Manoharan M, et al. Silencing of microRNAs in vivo with ‘antagomirs’. Nature 2005;438:685–9. [22] da Costa Martins PA, Salic K, Gladka MM, Armand AS, Leptidis S, el Azzouzi H, et al. MicroRNA-199b targets the nuclear kinase Dyrk1a in an auto-amplification loop promoting calcineurin/NFAT signalling. Nat Cell Biol 2010;12:1220–7. [23] Lanford RE, Hildebrandt-Eriksen ES, Petri A, Persson R, Lindow M, Munk ME, et al. Therapeutic silencing of microRNA-122 in primates with chronic hepatitis C virus infection. Science 2010;327:198–201 (New York, N Y. [24] Thomas M, Lieberman J, Lal A. Desperately seeking microRNA targets. Nat Struct Mol Biol 2010;17:1169–74. [25] Matkovich SJ, Van Booven DJ, Eschenbacher WH, Dorn II GW. RISC RNA sequencing for context-specific identification of in vivo microRNA targets. Circ Res 2011;108:18–26. [26] Grimson A, Farh KK, Johnston WK, Garrett-Engele P, Lim LP, Bartel DP. MicroRNA targeting specificity in mammals: determinants beyond seed pairing. Mol Cell 2007;27:91–105. [27] Friedman RC, Farh KK, Burge CB, Bartel DP. Most mammalian mRNAs are conserved targets of microRNAs. Genome Res 2009;19:92–105. [28] Krek A, Grun D, Poy MN, Wolf R, Rosenberg L, Epstein EJ, et al. Combinatorial microRNA target predictions. Nat Genet 2005;37:495–500. [29] Grun D, Wang YL, Langenberger D, Gunsalus KC, Rajewsky N. microRNA target predictions across seven Drosophila species and comparison to mammalian targets. PLoS Comput Biol 2005;1:e13. [30] Lall S, Grun D, Krek A, Chen K, Wang YL, Dewey CN, et al. A genome-wide map of conserved microRNA targets in C. elegans. Curr Biol 2006;16:460–71. [31] Chen K, Rajewsky N. Natural selection on human microRNA binding sites inferred from SNP data. Nat Genet 2006;38:1452–6. [32] Enright AJ, John B, Gaul U, Tuschl T, Sander C, Marks DS. MicroRNA targets in Drosophila. Genome Biol 2003;5:R1. [33] John B, Enright AJ, Aravin A, Tuschl T, Sander C, Marks DS. Human MicroRNA targets. PLoS Biol 2004;2:e363. [34] Betel D, Koppal A, Agius P, Sander C, Leslie C. Comprehensive modeling of microRNA targets predicts functional non-conserved and non-canonical sites. Genome Biol 2010;11:R90. [35] Hsu SD, Lin FM, Wu WY, Liang C, Huang WC, Chan WL, et al. miRTarBase: a database curates experimentally validated microRNA–target interactions. Nucleic Acids Res 2011;39:D163–9. [36] Dweep H, Sticht C, Pandey P, Gretz N. miRWalk — Database: prediction of possible miRNA binding sites by “walking” the genes of three genomes. J Biomed Inform 2011 [Electronic publication ahead of print]. [37] Fasanaro P, Greco S, Lorenzi M, Pescatori M, Brioschi M, Kulshreshtha R, et al. An integrated approach for experimental target identification of hypoxia-induced miR-210. J Biol Chem 2009;284:35134–43. [38] Jayaswal V, Lutherborrow M, Ma DD, Hwa Yang Y. Identification of microRNAs with regulatory potential using a matched microRNA–mRNA time–course data. Nucleic Acids Res 2009;37:e60. [39] Jayaswal V, Lutherborrow M, Ma DD, Yang YH. Identification of microRNA–mRNA modules using microarray data. BMC Genomics 2011;12:138. [40] Bang-Berthelsen CH, Pedersen L, Floyel T, Hagedorn PH, Gylvin T, Pociot F. Independent component and pathway-based analysis of miRNA-regulated gene expression in a model of type 1 diabetes. BMC Genomics 2011;12:97. [41] Baek D, Villen J, Shin C, Camargo FD, Gygi SP, Bartel DP. The impact of microRNAs on protein output. Nature 2008;455:64–71. [42] Selbach M, Schwanhausser B, Thierfelder N, Fang Z, Khanin R, Rajewsky N. Widespread changes in protein synthesis induced by microRNAs. Nature 2008;455:58–63. [43] Ong SE, Blagoev B, Kratchmarova I, Kristensen DB, Steen H, Pandey A, et al. Stable isotope labeling by amino acids in cell culture, SILAC, as a simple and accurate approach to expression proteomics. Mol Cell Proteomics 2002;1:376–86. [44] Yang Y, Chaerkady R, Kandasamy K, Huang TC, Selvan LD, Dwivedi SB, et al. Identifying targets of miR-143 using a SILAC-based proteomic approach. Mol Biosyst 2010;6:1873–82. [45] Yan GR, Xu SH, Tan ZL, Liu L, He QY. Global identification of miR-373-regulated genes in breast cancer by quantitative proteomics. Proteomics 2011;11:912–20. [46] Elkan-Miller T, Ulitsky I, Hertzano R, Rudnicki A, Dror AA, Lenz DR, et al. Integration of transcriptomics, proteomics, and microRNA analyses reveals novel microRNA regulation of targets in the mammalian inner ear. PLoS One 2011;6: e18195. [47] Yang Y, Chaerkady R, Beer MA, Mendell JT, Pandey A. Identification of miR-21 targets in breast cancer cells using a quantitative proteomic approach. Proteomics 2009;9:1374–84. [48] Leivonen SK, Rokka A, Ostling P, Kohonen P, Corthals GL, Kallioniemi O, et al. Identification of miR-193b targets in breast cancer cells and systems biological analysis of their functional impact. Mol Cell Proteomics 2011;10:M110.005322.
J. Fiedler et al. / Journal of Molecular and Cellular Cardiology 51 (2011) 674–681 [49] Fleissner F, Jazbutyte V, Fiedler J, Gupta SK, Yin X, Xu Q, et al. Short communication: asymmetric dimethylarginine impairs angiogenic progenitor cell function in patients with coronary artery disease through a microRNA-21-dependent mechanism. Circ Res 2010;107:138–43. [50] Brandes RP. A miR-thless perspective: how asymmetric dimethylarginine impairs the functions of angiogenic progenitor cells. Circ Res 2010;107:12–4. [51] Huang Y, Qi Y, Ruan Q, Ma Y, He R, Ji Y, et al. A rapid method to screen putative mRNA targets of any known microRNA. Virol J 2011;8:8. [52] Li J, Xia W, Huang B, Chen L, Su X, Li S, et al. A strategy to rapidly identify the functional targets of microRNAs by combining bioinformatics and mRNA cytoplasmic/ nucleic ratios in culture cells. FEBS Lett 2010;584:3198–202. [53] Orom UA, Lund AH. Isolation of microRNA targets using biotinylated synthetic microRNAs. Methods 2007;43:162–5. [54] Fiedler J, Jazbutyte V, Kirchmaier BC, Gupta SK, Lorenzen J, Hartmann D, et al. MicroRNA-24 regulates vascularity after myocardial infarction. Circulation 2011. [55] Vo NK, Dalton RP, Liu N, Olson EN, Goodman RH. Affinity purification of microRNA-133a with the cardiac transcription factor, Hand2. Proc Natl Acad Sci USA 2010;107:19231–6. [56] Jain R, Devine T, George AD, Chittur SV, Baroni TE, Penalva LO, et al. RIP-Chip analysis: RNA-binding protein immunoprecipitation-microarray (Chip) profiling. Methods Mol Biol 2011;703:247–63. [57] Tan LP, Seinen E, Duns G, de Jong D, Sibon OC, Poppema S, et al. A high throughput experimental approach to identify miRNA targets in human cells. Nucleic Acids Res 2009;37:e137. [58] Zhang L, Ding L, Cheung TH, Dong MQ, Chen J, Sewell AK, et al. Systematic identification of C. elegans miRISC proteins, miRNAs, and mRNA targets by their interactions with GW182 proteins AIN-1 and AIN-2. Mol Cell 2007;28:598–613. [59] Landthaler M, Gaidatzis D, Rothballer A, Chen PY, Soll SJ, Dinic L, et al. Molecular characterization of human Argonaute-containing ribonucleoprotein complexes and their bound target mRNAs. RNA 2008;14:2580–96. [60] Keene JD, Komisarow JM, Friedersdorf MB. RIP-Chip: the isolation and identification of mRNAs, microRNAs and protein components of ribonucleoprotein complexes from cell extracts. Nat Protoc 2006;1:302–7.
681
[61] Prokopi M, Mayr M. Proteomics: a reality-check for putative stem cells. Circ Res 2011;108:499–511. [62] Nelson PT, Wang WX, Mao G, Wilfred BR, Xie K, Jennings MH, et al. Specific sequence determinants of miR-15/107 microRNA gene group targets. Nucleic Acids Res 2011. [63] Nonne N, Ameyar-Zazoua M, Souidi M, Harel-Bellan A. Tandem affinity purification of miRNA target mRNAs (TAP-Tar). Nucleic Acids Res 2010;38:e20. [64] Chi SW, Zang JB, Mele A, Darnell RB. Argonaute HITS-CLIP decodes microRNA– mRNA interaction maps. Nature 2009;460:479–86. [65] Leung AK, Young AG, Bhutkar A, Zheng GX, Bosson AD, Nielsen CB, et al. Genome-wide identification of Ago2 binding sites from mouse embryonic stem cells with and without mature microRNAs. Nat Struct Mol Biol 2011;18: 237–44. [66] Hafner M, Landthaler M, Burger L, Khorshid M, Hausser J, Berninger P, et al. Transcriptome-wide identification of RNA-binding protein and microRNA target sites by PAR-CLIP. Cell 2010;141:129–41. [67] Hafner M, Landthaler M, Burger L, Khorshid M, Hausser J, Berninger P, et al. PARCliP—a method to identify transcriptome-wide the binding sites of RNA binding proteins. J Vis Exp 2010(41), doi:10.3791/2034 pii: 2034. [68] Fichtlscherer S, De Rosa S, Fox H, Schwietz T, Fischer A, Liebetrau C, et al. Circulating microRNAs in patients with coronary artery disease. Circ Res 2010;107: 677–84. [69] Zampetaki A, Kiechl S, Drozdov I, Willeit P, Mayr U, Prokopi M, et al. Plasma microRNA profiling reveals loss of endothelial miR-126 and other microRNAs in type 2 diabetes. Circ Res 2010;107:810–7. [70] Barrey E, Saint-Auret G, Bonnamy B, Damas D, Boyer O, Gidrol X. Pre-microRNA and mature microRNA in human mitochondria. PLoS One 2011;6:e20220. [71] Yeom KH, Heo I, Lee J, Hohng S, Kim VN, Joo C. Single-molecule approach to immunoprecipitated protein complexes: insights into miRNA uridylation. EMBO Rep 2011;12:690–6. [72] Ponting CP, Oliver PL, Reik W. Evolution and functions of long noncoding RNAs. Cell 2009;136:629–41.