- Open Access
Advancing the functional utility of PAR-CLIP by quantifying background binding to mRNAs and lncRNAs
© Friedersdorf and Keene; licensee BioMed Central Ltd. 2014
- Received: 31 July 2013
- Accepted: 7 January 2014
- Published: 7 January 2014
Sequence specific RNA binding proteins are important regulators of gene expression. Several related crosslinking-based, high-throughput sequencing methods, including PAR-CLIP, have recently been developed to determine direct binding sites of global protein-RNA interactions. However, no studies have quantitatively addressed the contribution of background binding to datasets produced by these methods.
We measured non-specific RNA background in PAR-CLIP data, demonstrating that covalently crosslinked background binding is common, reproducible and apparently universal among laboratories. We show that quantitative determination of background is essential for identifying targets of most RNA-binding proteins and can substantially improve motif analysis. We also demonstrate that by applying background correction to an RNA binding protein of unknown binding specificity, Caprin1, we can identify a previously unrecognized RNA recognition element not otherwise apparent in a PAR-CLIP study.
Empirical background measurements of global RNA-protein crosslinking are a necessary addendum to other experimental controls, such as performing replicates, because covalently crosslinked background signals are reproducible and otherwise unavoidable. Recognizing and quantifying the contribution of background extends the utility of PAR-CLIP and can improve mechanistic understanding of protein-RNA specificity, protein-RNA affinity and protein-RNA association dynamics.
- Background Correction
- Background Sample
- Background Binding
- Crosslinking Procedure
- Reproducible Background
RNA binding proteins (RBPs) control many important and interconnected steps in posttranscriptional gene expression, including co-transcriptional regulation, epigenetics, pre-RNA processing, mRNA splicing, nuclear export, quality control, subcellular localization, stability and translation . Furthermore, RBPs can coordinate production of functionally related proteins through posttranscriptional operons/regulons [2, 3]. RBPs achieve these complex regulatory functions of interconnection and coordination by binding to specific RNA recognition elements. The ultimate regulatory fate of a transcript is defined by its unique “RNP code”, a medley of RNA recognition elements in a single RNA that is combinatorially regulated by a network of RBPs [2, 4, 5]. Moreover, RNA-protein interactions play critical roles in both normal and diseased states [6–10].
To identify global in vivo RNA-protein interactions, numerous methods have been applied including immunopurification (IP) of RBPs followed by microarray analysis (RIP-Chip) or sequencing (RIP-seq). Essential to the identification of bound RNAs by RIP-Chip is the empirical measurement of background binding by using a control IP with isotype matched IgG or other mock proteins [11, 12]. The determination of background has proved highly valuable by allowing widespread and reproducible use of standardized RIP-Chip procedures . Furthermore, background measurements have been used as a reference for quantitative determination of bound transcripts and to calculate fold changes following cellular perturbations, such as T cell activation .
To identify individual bindings sites of RBPs within a RNA target, several distinct but related techniques have been developed based on covalent UV crosslinking followed by IP and high-throughput sequencing (CLIP) (reviewed in ). Most CLIP-related procedures assume that background binding is biochemically eliminated through rigorous and stringent washes given that non-covalent bonds have been replaced by UV-induced covalent bonds. This approach to dealing with background binding has produced notable difficulties that have limited the widespread and standardized use of these techniques, evidence for which can be seen in the number of new, distinct crosslinking techniques that have been devised of necessity (reviewed in [16, 17]). For example, PAR-CLIP is a useful and novel crosslinking approach devised by Tuschl and coworkers that addresses the issue of indirect targets arising from non-crosslinked noise, and discriminates indirect from direct RNA-protein interactions by identifying diagnostic conversions resulting from covalent adducts at sites of crosslinking of the RBP to specific nucleotides . However, none of the CLIP-related studies has quantitatively addressed the issue of background generated by adventitious covalent crosslinking of RNAs to proteins that are not the RBP of interest.
In this study, we demonstrate that covalently crosslinked background binding during the PAR-CLIP procedure is common, reproducible and likely universal to crosslinking procedures, and that this can have serious implications for understanding protein-RNA specificity as well as protein-RNA affinity. Given the relative inefficiency of UV crosslinking procedures in general, characteristic sequence biases and the exquisite sensitivity of high-throughput sequencing, false binding targets should be expected [12, 19–22]. We find that low affinity RBPs are especially affected by false crosslinking events that can limit the ability of global analysis to determine authentic binding sites. These problems are especially acute when attempting to study the effects of mutations in RNA-binding domains on underlying RNA recognition mechanisms or, to discern dynamic changes in RNA targets following induced perturbations or varied growth conditions. In this study, we show that by quantifying and accounting for background, as is standard in RIP-Chip protocols, we are able to both improve the specificity of PAR-CLIP target identification and remove erroneous and misleading results. Among several examples of the utility of these improvements, we reveal false binding sites in XIST and MALAT1 lncRNAs and use a previously published PAR-CLIP dataset to identify a novel A-rich RNA motif in the RBP, Caprin1.
PAR-CLIP background contains many uniquely mapping sequence reads
High-throughput sequencing detected reads from PAR-CLIP background samples are abundant
Total reads (PARalyzer utilized)
Unique sites (Locations)
Background reads contain T > C conversions
Background reads tend to be G-rich and represent cellular RNAs
To determine if the background reads were mostly from a cellular source or were being introduced at one of the other steps, such as adapter ligation, we compared the reads mapping to annotated genomic regions versus those represented in total RNA. To control for methodological biases that may influence mapping of PAR-CLIP reads we modified the PAR-CLIP protocol for isolation of total RNA. To do this we made lysates from cells treated with 4SU and irradiated at UV 365 nm, then the lysates were proteinase K digested followed by rRNA depletion with RiboZero and finally the RNA was partially digested with RNase T1. The library preparation, sequencing and mapping parameters were identical for PAR-CLIP and for the total RNA. We observed that the percent of background PAR-CLIP reads mapping to coding, 3’utr, 5’utr, intron and lincRNA were similar to the values for total RNA reads (Figure 2B). The only significant differences between PAR-CLIP background reads and total RNA reads were that background reads mapped to repeat regions less frequently and to intergenic regions more frequently. However, when comparing background reads to reads from HuR, each mapped to drastically different regions. This indicates that the background reads are from a cellular RNA source and not simply the result of sequencing artifacts.
Next, to identify sequences and motifs that were enriched in background reads we used a kmer approach to build motifs from the union of all three GFP background libraries. The highest enriched motifs were extremely G-rich, including 24 of the top 25 most abundant 8mer sequences. After guanosine the next most common nucleoside in these motifs was adenosine followed by cytidine and uridine (Figure 2C). The enrichment of guanosine in these motifs is surprising considering the use of RNase T1, which cleaves at single stranded guanosines, furthermore the relative scarcity of uridines in the motifs is also surprising despite the use of 4SU as a specific cosslinking agent. This G-rich motif is much more abundant in background samples than in the transcriptome suggesting that this motif is the result of biases of the procedure.
High abundance background sites are common and reproducible between different molecular weight gel slices of PAR-CLIP background
Background binding sites are prevalent in PAR-CLIPs of weak affinity RBPs
To compare the background PAR-CLIP binding sites to the binding sites of HuR PAR-CLIP, we focused on the reads at three representative sites, a non-coding RNA site in MALAT1, a coding sequence site and the 3’ UTR of the ELAVL1 mRNA (the HuR gene transcript). MALAT1 is a highly conserved, abundantly expressed non-coding RNA that is primarily located in the nucleus. Several recent global protein-RNA crosslinking studies have identified MALAT1 as a target of Sfrs1, Tardbp and Dgcr8 RBPs [24–26]. Elavl1, also known as HuR, is an abundantly expressed member of the ELAV/Hu family of proteins that are well known to bind and autoregulate their own messages through 3’UTR binding sites [27–30].
To investigate how frequently PAR-CLIP data sets overlap with background we used the same re-analyzed set of PAR-CLIP data sets described above and determined the percent of reads that overlapped with reads from any one of the three background samples. We observed that over a range of 45% to less than 10% of sites in an RBP PAR-CLIP overlapped with background sites (Figure 5B and Additional file 68: Table S1). Interestingly, we noticed that the PAR-CLIP experiments with a higher percent overlap with background were frequently either from PAR-CLIP studies on putative RBPs, previously uncharacterized RBPs, or PAR-CLIP libraries that had produced very few mapped reads. This demonstrates that the degree of a particular PAR-CLIP library overlapping with background varies greatly depending on the RBP; and weak binding (or weakly crosslinked) RBPs in particular contain a larger fraction of background binding. In many cases, these RBPs are considered putative based upon the absence of a known RNA-binding motif, and have been characterized as non-professional RBPs [38, 39]. Moreover, this can be exceedingly important when investigating defined mutations in professional RNA-binding domains, as well as dynamic in vivo binding events that may involve sequential low-affinity maturation such as occurs with cooperative binding experiments.
Background correction markedly improves motif finding
We have shown above that background binding is a common and reproducible feature of PAR-CLIP, and it is essential to assess its utility in data interpretation. To this end, we investigated how background correction can be used to improve motif finding. Therefore, we analyzed Pum2, a member of the Puf family of RBPs that are widely known for their highly specific and evolutionarily conserved recognition of RNA elements [40, 41]. In particular, mammalian Pum proteins are known to bind to mRNAs containing one or more UGUAHAUA sequences [42–44]. Hafner and colleagues validated the specificity of PAR-CLIP by showing that Pum2 PAR-CLIP can enrich for Pum motifs (UGUAHAUA), especially half-site Pum motifs (UGUA), and revealed positional information of binding . To correct for background binding, we eliminated sites from Pum2 PAR-CLIP that overlapped by one or more nucleotides with sites from at least one of the three PAR-CLIP background samples. Background correction dramatically increased the percent of sites that contained either Pum motifs or half-site Pum motifs, with increases in specificity of up to 1.2 fold (Figure 7A). When compared to HuR, an RBP that is not known to bind to Pum motifs, background correction did not show any noticeable improvement in enrichment of the full motif and a negative enrichment of the half motif. This lack of Pum motif enrichment was also observed when background correction was applied to the total library. An increase of 20% in specificity can be highly significant, even crucial, for assigning confidence values to motifs, most of which are not as definitive as the Puf motif. For example, computational searches for motifs are most precise when the motif itself is definitive. Overall, our data demonstrate that background correction significantly improves PAR-CLIP specificity by eliminating non-specific sequences (i.e. those not containing Pum motifs) in preference over sequences that do contain Pum motifs.
While background correction significantly improved the specificity for identifying Pum motifs, it was already shown that PAR-CLIP, prior to background correction, could also identify these highly specific motifs. Thus, to further test our approach, we examined whether background correction can offer significant improvement in identifying a truly novel motif. To do this we attempted to identify a candidate motif from published PAR-CLIP data of Caprin1, a known RBP that has been shown by using PAR-CLIP to bind to both coding regions and 3’UTRs but for which no motif has been identified [38, 45, 46]. To identify possible motifs for Caprin1, we counted the abundance of 7mer sequences within the PAR-CLIP identified sites (Figure 7B). Among the most common 7mers were two motifs; one motif was U-rich with occasional adenines and cytosines, while the other motif was A-rich with occasional uridines. A third, less common, G-rich motif was also observed. We then performed a matched pairs analysis on the 7mers before and after background correction to identify sequences that were enriched or depleted by background correction. The matched pair analysis allowed us to investigate the more abundant motifs, plotted along the x-axis, on a continuum without setting arbitrary cutoffs. We observed that background correction strongly enriched for the A-rich motifs while moderately depleting the U-rich motifs and strongly depleting the G-rich motifs (Figure 7B). As expected, the depleted G-rich motifs were similar to the G-rich motifs that were found most frequently in the background PAR-CLIPs. Among the strongly enriched A-rich motifs were a subset of sequences that contained canonical polyA signal sequences. The similarity of this motif to the polyA signal sequence led us to investigate the enrichment of Caprin1 RNA-binding sites at the end of transcripts, and we found that 3’ termini were enriched in background corrected Caprin1 PAR-CLIP sites. This suggests that recognition of motifs containing polyA signal sequences by Caprin1 maybe functional and may indicate a role for Caprin1 in polyadenylation. Our data suggests that we have identified a candidate A-rich motif, which includes canonical polyA signal sequences, for Caprin1. This motif needs to be confirmed biochemically, but it was not obvious from the uncorrected PAR-CLIP data (Additional file 70: Figure S7). Therefore, by using background correction we were able to distinctly identify novel candidate motifs for an RBP, demonstrating the utility of our approach.
Background correction can remove misleading results from PAR-CLIP data
In this study we have expanded the utility of PAR-CLIP by developing a method to quantify background binding. We have shown that PAR-CLIP generates readily detectable, T-to-C containing background reads, and that PAR-CLIP data still contains background reads in spite of the assumption of detecting only direct protein-RNA interaction. We have also shown that background reads are reproducibly common in PAR-CLIP libraries, often with identical sequences appearing in libraries from many different RBPs and from experiments performed in several different laboratories. Despite the rigorous and stringent purification conditions of the PAR-CLIP procedure, background reads are easily detectable due to the extreme sensitivity of high-throughput sequencing . Moreover, it should be noted that in none of the CLIP or PAR-CLIP studies has there been a formal demonstration that a covalent chemical crosslink has actually formed; rather, an operational assumption of crosslinking is asserted based upon survival after stringent washing. Taken together, these observations suggest that investigators using any crosslinking procedures should be cautious about the presence of background, since low efficiency conversion of a set of noncovalent bonds to a single covalent bond is not a guarantee of perfection, even when combined with rigorous and stringent purification conditions.
Reproducible background reads and replicates
Background reads are found nearly universally across PAR-CLIP experiments from different RBPs for many sites and are often reproducible among background samples of different molecular weights, thus demonstrating that these background reads are an inherent characteristic of the process. This implies that if one only addresses non-specific binding by requiring reproducibility in replicates one will incorrectly regard these common and reproducible background reads as positives. However, requiring reproducibility in replicate samples is likely better at removing other background sources resulting from random events than using only quantitative measurement of background. Therefore, we recommend a combined approach of empirical background measurement and biological replicates to optimally minimize contribution of non-specific background binding events.
PAR-CLIP background can derive from multiple sources
While the precise source of these background reads are unknown, it is likely that at least some of the reads containing T-to-C conversions are from non-specific protein-RNA interactions. These may arise in two categories: 1) RNAs that are not specifically recognized by the protein of interest, but that during the course of crosslinking are randomly proximal to reactive amino acids of the RBP of interest and become covalently linked; or 2) RNPs that do not contain the RBP of interest, yet are not entirely removed during the procedure. Non-specific protein-RNA interactions of the first type would not be expected to produce sites with much read depth or reproducibility because the interactions are random and must be physically juxtaposed to be crosslinked. Some of these non-specific interactions may include sites with one read that are uniquely present in gel slices. Of the second variety of non-specific protein-RNA interactions, these should be more numerous and more reproducible across experiments, and thus, are likely represented by some of the more abundant RNA sites we observed. Interestingly, these sites with deeper reads were frequently found in multiple background gel slices suggesting that they were not focused in any single location on the gel. There are several possible interpretations for this outcome; first gels may be overloaded in PAR-CLIP experiments so that they do not properly resolve the most abundant complexes. Secondly, these may represent sites with variable RNP composition, and thus, migrate at multiple locations on the gel. Additionally, there are still many other potential sources of background in crosslinking procedures that may or may not account for the high rate of T-to-C conversions. For example, RNAs can fold into conformations that can increase their “stickiness” or even potentially mimic protein epitopes . Thus, these aptameric RNAs can be greatly enriched in background reads as both non-specific, sticky RNAs or as off-target RNAs such as epitope mimics.
Radiochemical and radiophysical effects on nucleic acids
Alternatively, it is possible that damage caused by UV irradiation causes changes in migration patterns of nucleic acids as is known to occur in comet assays used to measure DNA damage . In this scenario, electrophoresis may be less successful at focusing and separating signal RNPs from background RNPs, potentially leading to the detection of the same background reads in all three gel slices. On the other hand, irradiation induced damage is much less likely for PAR-CLIP which irradiates at 365 nm than for other crosslinking methods that use UV 254 nm, which is know to cause damage to nucleic acids and to affect translation or to generate cellular RNP aggregates . Finally, it is possible that some of the abundant and reproducible background T-to-C conversions may not actually represent non-specific protein-RNA interactions but may instead be non-crosslinked, 4SU-containing RNA that still converts to cytosine following cDNA synthesis .
Background binding is characteristic of all biochemical enrichments
Of the many potential background sources mentioned above, it is likely that they are not mutually exclusive and any combination, as well as any number of other unforeseen ones, could contribute to the measured background reads. The numerous and diverse sources of potential background combined with the extreme sensitivity of high-throughput sequencing demonstrate the impracticality of attempting to entirely eliminate background biochemically. Furthermore, inherent limitations of the procedures may prevent ideal optimization, as does the fact that the mRNAs in any given subset from an IP are likely to have a range of crosslinked efficiencies . For example, trying to combat potential over-loading of gels by using less starting material or splitting samples over more wells may help lower background, but given the low efficiency of crosslinking, especially of UV 254 nm, this may reduce signal below acceptable levels . Instead of attempting to optimize the elimination of biochemical background, we offer a more practical solution of empirically measuring background binding events in every experiment. We show that this quantified background can be used to correct for non-signal or potentially erroneous events in PAR-CLIP, thus sidestepping the impractical task of identifying all sources of background and reducing the optimization to reasonable and achievable levels.
In addition to being a practical solution, empirically measuring background can also offer substantial benefits, as seen with the improved detection of the Pum motifs and discovering novel regulatory elements such as the A-rich motif for Caprin1. The amount of improvement provided by background correction will likely vary depending on the RBP and the fastidious quality of the experiment. Given the large percent overlap of weak or under sequenced RBPs with background (up to 45%), this background correction will be especially important for defining and validating RNA targets of weak binding RBPs, RBP mutants or candidate novel RBPs identified by global approaches [21, 38, 39, 51–55].
Interestingly, a recent global crosslinking study identified numerous RNAs associated with Ago2 in Dicer null cells despite the fact that the cells were lacking processed miRNAs . These Ago2-associated RNAs were enriched for a G-rich motif that the authors suggest indicates a preference for Ago2 to bind to G’s in the absence of processed miRNAs. However, background binding was not directly measured in that study as well. Given our observation of a highly similar G-rich motif in PAR-CLIP background reads, a potential alternative explanation is that Ago2 doesn’t bind RNA in Dicer-/- cells and that the G-rich RNAs simply represent background binding.
Background measurements for global RNA dynamics studies with PAR-CLIP
The quantitative measurement of background in PAR-CLIP may also have benefits beyond improving percent signal and removing misleading results, it may also be used as a reference in determining fold enrichments, affinities for different RNAs or fold changes in binding during changes in cellular conditions, such as during immune activation. In these cases, measured background may prove more useful than adjusting for the level of expressed RNA because it will naturally incorporate the inherent biases of the procedure. This approach of using background as a reference has already been shown to be useful for detecting dynamic changes in RNP association during T-cell activation when applied to RIP-Chip . This approach may prove especially beneficial when combined with computational approaches that model the quantitatively discrete nature of sequencing data .
Matching background controls to experiments
We have shown that applying background correction to several published PAR-CLIP data sets of various RBPs can substantially improve results. This is surprising considering that these data sets were generated in different laboratories using different RBPs that migrate at different sizes. The reproducibility and universality of background reads across each of these independent experiments suggest that the background data we generated here may prove useful in future PAR-CLIP studies. However, it should be cautioned that with a few exceptions, the studies compared in this manuscript were from HEK293 lysates and the protocols were carried out with nearly identical conditions. Therefore, it remains to be seen whether the universality of these background reads hold up for PAR-CLIPs performed with different lysates or modifications to the protocol. This becomes especially obvious when considering cell-type-specific or condition-specific transcripts that would be present in experimental samples but not in the background samples preformed in this manuscript. One extreme example of condition-specific transcripts is Ago PAR-CLIP performed on HIV infected cells, where it was recently reported that Ago PAR-CLIP showed evidence of binding to a miR-29a site in the HIV genome but that this site was shown to be non-functional using RIP and reporter assays . Since our background study was performed in cells that were not HIV infected, we have no way of knowing whether this site represents background binding or is an Ago-bound, non-functional site. Therefore, we would recommend performing and sequencing appropriate background controls matched with individual experiments whenever possible. We also recommend validation of targets through other methods including assays for functional responses.
Much in the way PAR-CLIP has improved on UV 254 crosslinking methods by measuring T-to-C conversions we have improved upon PAR-CLIP by accounting for crosslinked background. We achieved this by borrowing an approach from RIP-Chip, namely, measuring background empirically. This approach is a key feature of RIP-Chip that allows for quantitative measurement of protein-RNA interactions, and more importantly, to compare RIP-Chip data from different conditions to determine protein-RNA dynamics. For PAR-CLIP and other global crosslinking techniques to reliably achieve these quantitative and dynamic measurements several considerations will still need to be addressed. One such issue is the presence of PCR amplification artifacts that can limit the quantitative analysis of protein-RNA interactions. This issue has already been addressed in several related crosslinking methods by the introduction of “randomer” barcodes into the library making process [15, 59, 60]. Another consideration is whether high-throughput sequencing of binding sites has reached saturation, and thus, whether sequencing depth is in the dynamic range for quantification of all binding sites. We are unaware of any global protein-RNA sequencing studies to date that have demonstrated full saturation. In the present study we failed to reach saturation despite pushing the current limits of sequencing technologies with approximately 250 million raw reads per library (Additional file 71: Figure S8). Future developments and refinements to global protein-RNA interaction studies, like PAR-CLIP, will lay the foundation for unraveling the “RNP code” and understanding the organizational and mechanistic properties underlying the dynamics of posttranscriptional RNA operons and regulons.
Human embryonic kidney 293 cells (HEK 293) stably expressing Dox-inducible HuR or GFP were plated at 7.5 × 10^5 cells/ml and grown for 24 hours in normal growth media (DMEM with 10% Tet-reduced FBS). The cells were then grown overnight in media supplemented with 100 uM 4SU and 1 uM Doxycycline.
Procedure is similar to Hafner et al, with minor adjustments. Specifically, both RNase digestion steps used less RNase T1. In the first digestion step a final concentration of 0.5 U/ul of RNase T1 was used and in the second digestion step, the one after immunoprecipitation, a final concentration of 0.005 U/ul RNase T1 was used.
Mapping, processing and analysis of sequencing data
50 bp single read libraries were run with a single sample per lane on the Illumina HiSeq 2000 instrument. Adapter sequences were removed, and reads containing fewer than 10 bp, were eliminated from further analysis. Reads were then mapped to the human genome (hg19) using bowtie with parameters suggested for use with PARalyzer, no more than 2 mismatches and 10 multi-mappers (-v 2 –m 10 --all --best --strata) . Bowtie output was then further refined using the PARalyzer algorithm using standard parameters which restricted reads to only those containing 0, 1 or 2 T-to-C mismatches mapping uniquely to the genome . PARalyzer “groups” were defined as sites with any read evidence and “clusters” were defined as sites with at least 2 T-to-C conversion locations and at least 5 overlapping reads.
K-mer length motifs were generated by quantifying the occurrence of each oligomer of length k in all PARalyzer utilized reads for a given library. The abundance of each k-mer was counted and rank ordered. Caprin1 motifs were identified by grouping similar sequences in the matched pairs analysis followed by alignment of the sequences by PhyloGibbs analysis. For background motifs the top 25 8-mers were used to make motif logos using enoLOGOS . Significance of enrichment for motifs was determined by comparing observed frequencies to expected from shuffling the libraries while preserving di-nucleotide frequencies.
To remove reads from PAR-CLIP RBP libraries that were also present in the union of PAR-CLIP backgrounds, entire sites were removed if they overlapped by one or more bp between both libraries by using BEDTools .
Saturation analysis was performed by randomly sampling reads prior to processing. 5 independent sets were sampled to depths to match 10, 30, 50, 70 and 90% of all reads. These sampled sets were then processed and analyzed as described above. The number of sites or clusters (defined as sites with 5 or more reads and two or more T-to-C conversion locations) for each sampled set were counted and reported as a fraction of all sites (or clusters) identified in the whole, un-sampled set.
Raw and processed data for the background data sets (G45, G35 and G20), two HuR replicates and total crosslinked RNA samples have been submitted to GEO (Accession number: GSE50989).
We gratefully acknowledge Tom Tuschl, Jeff Blackinton, Kyle Mansfield and Neel Mukherjee and other members of the Keene lab for their advice and comments. This work was supported by National Science Foundation grant 0842621 (J.D.K.) and National Institutes of Health grant R01CA157268 (J.D.K.).
- Moore MJ: From birth to death: the complex lives of eukaryotic mRNAs. Science. 2005, 309: 1514-1518. 10.1126/science.1111443.View ArticleGoogle Scholar
- Keene JD, Tenenbaum SA: Eukaryotic mRNPs may represent posttranscriptional operons. Mol Cell. 2002, 9: 1161-1167. 10.1016/S1097-2765(02)00559-2.View ArticleGoogle Scholar
- Keene JD: RNA regulons: coordination of post-transcriptional events. Nat Rev Genet. 2007, 8: 533-543. 10.1038/nrg2111.View ArticleGoogle Scholar
- Barash Y, Calarco JA, Gao W, Pan Q, Wang X, Shai O, Blencowe BJ, Frey BJ: Deciphering the splicing code. Nature. 2010, 465: 53-59. 10.1038/nature09000.View ArticleGoogle Scholar
- Irimia M, Blencowe BJ: Alternative splicing: decoding an expansive regulatory layer. Curr Opin Cell Biol. 2012, 24: 323-332. 10.1016/j.ceb.2012.03.005.View ArticleGoogle Scholar
- Lukong KE, Chang KW, Khandjian EW, Richard S: RNA-binding proteins in human genetic disease. Trends Genet. 2008, 24: 416-425. 10.1016/j.tig.2008.05.004.View ArticleGoogle Scholar
- Khalil AM, Rinn JL: RNA-protein interactions in human health and disease. Semin Cell Dev Biol. 2011, 22: 359-365. 10.1016/j.semcdb.2011.02.016.View ArticleGoogle Scholar
- Srikantan S, Gorospe M: HuR function in disease. Front Biosci. 2012, 17: 189-205. 10.2741/3921.View ArticleGoogle Scholar
- Castello A, Fischer B, Hentze MW, Preiss T: RNA-binding proteins in Mendelian disease. Trends Genet. 2013, 29: 318-327. 10.1016/j.tig.2013.01.004.View ArticleGoogle Scholar
- Ule J: Ribonucleoprotein complexes in neurologic diseases. Curr Opin Neurobiol. 2008, 18: 516-523. 10.1016/j.conb.2008.09.018.View ArticleGoogle Scholar
- Tenenbaum SA, Carson CC, Lager PJ, Keene JD: Identifying mRNA subsets in messenger ribonucleoprotein complexes by using cDNA arrays. Proc Natl Acad Sci U S A. 2000, 97: 14085-14090. 10.1073/pnas.97.26.14085.View ArticleGoogle Scholar
- 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: 302-307. 10.1038/nprot.2006.47.View ArticleGoogle Scholar
- Morris AR, Mukherjee N, Keene JD: Systematic analysis of posttranscriptional gene expression. Wiley Interdiscip Rev Syst Biol Med. 2010, 2: 162-180. 10.1002/wsbm.54.View ArticleGoogle Scholar
- Mukherjee N, Lager PJ, Friedersdorf MB, Thompson MA, Keene JD: Coordinated posttranscriptional mRNA population dynamics during T-cell activation. Mol Syst Biol. 2009, 5: 288-View ArticleGoogle Scholar
- Konig J, Zarnack K, Luscombe NM, Ule J: Protein-RNA interactions: new genomic technologies and perspectives. Nat Rev Genet. 2011, 13: 77-83.View ArticleGoogle Scholar
- Milek M, Wyler E, Landthaler M: Transcriptome-wide analysis of protein-RNA interactions using high-throughput sequencing. Semin Cell Dev Biol. 2012, 23: 206-212. 10.1016/j.semcdb.2011.12.001.View ArticleGoogle Scholar
- Ascano M, Hafner M, Cekan P, Gerstberger S, Tuschl T: Identification of RNA-protein interaction networks using PAR-CLIP. Wiley Interdiscip Rev RNA. 2012, 3: 159-177. 10.1002/wrna.1103.View ArticleGoogle Scholar
- Hafner M, Landthaler M, Burger L, Khorshid M, Hausser J, Berninger P, Rothballer A, Ascano M, Jungkamp AC, Munschauer M, et al: Transcriptome-wide identification of RNA-binding protein and microRNA target sites by PAR-CLIP. Cell. 2010, 141: 129-141. 10.1016/j.cell.2010.03.009.View ArticleGoogle Scholar
- Smith KC: Photochemical addition of amino acids to 14C-uracil. Biochem Biophys Res Commun. 1969, 34: 354-357. 10.1016/0006-291X(69)90840-7.View ArticleGoogle Scholar
- Schott HN, Shetlar MD: Photochemical addition of amino acids to thymine. Biochem Biophys Res Commun. 1974, 59: 1112-1116. 10.1016/S0006-291X(74)80093-8.View ArticleGoogle Scholar
- Klass DM, Scheibe M, Butter F, Hogan GJ, Mann M, Brown PO: Quantitative proteomic analysis reveals concurrent RNA-protein interactions and identifies new RNA-binding proteins in Saccharomyces cerevisiae. Genome Res. 2013, 23: 1028-1038. 10.1101/gr.153031.112.View ArticleGoogle Scholar
- Sugimoto Y, Konig J, Hussain S, Zupan B, Curk T, Frye M, Ule J: Analysis of CLIP and iCLIP methods for nucleotide-resolution studies of protein-RNA interactions. Genome Biol. 2012, 13: R67-10.1186/gb-2012-13-8-r67.View ArticleGoogle Scholar
- Darnell JC, Van Driesche SJ, Zhang C, Hung KY, Mele A, Fraser CE, Stone EF, Chen C, Fak JJ, Chi SW, et al: FMRP stalls ribosomal translocation on mRNAs linked to synaptic function and autism. Cell. 2011, 146: 247-261. 10.1016/j.cell.2011.06.013.View ArticleGoogle Scholar
- Sanford JR, Wang X, Mort M, Vanduyn N, Cooper DN, Mooney SD, Edenberg HJ, Liu Y: Splicing factor SFRS1 recognizes a functionally diverse landscape of RNA transcripts. Genome Res. 2009, 19: 381-394.View ArticleGoogle Scholar
- Tollervey JR, Curk T, Rogelj B, Briese M, Cereda M, Kayikci M, Konig J, Hortobagyi T, Nishimura AL, Zupunski V, et al: Characterizing the RNA targets and position-dependent splicing regulation by TDP-43. Nat Neurosci. 2011, 14: 452-458. 10.1038/nn.2778.View ArticleGoogle Scholar
- Macias S, Plass M, Stajuda A, Michlewski G, Eyras E, Caceres JF: DGCR8 HITS-CLIP reveals novel functions for the Microprocessor. Nat Struct Mol Biol. 2012, 19: 760-766. 10.1038/nsmb.2344.View ArticleGoogle Scholar
- Samson ML: Evidence for 3' untranslated region-dependent autoregulation of the Drosophila gene encoding the neuronal nuclear RNA-binding protein ELAV. Genetics. 1998, 150: 723-733.Google Scholar
- Pullmann R, Kim HH, Abdelmohsen K, Lal A, Martindale JL, Yang X, Gorospe M: Analysis of turnover and translation regulatory RNA-binding protein expression through binding to cognate mRNAs. Mol Cell Biol. 2007, 27: 6265-6278. 10.1128/MCB.00500-07.View ArticleGoogle Scholar
- Mansfield KD, Keene JD: Neuron-specific ELAV/Hu proteins suppress HuR mRNA during neuronal differentiation by alternative polyadenylation. Nucleic Acids Res. 2012, 40: 2734-2746. 10.1093/nar/gkr1114.View ArticleGoogle Scholar
- Dai W, Zhang G, Makeyev EV: RNA-binding protein HuR autoregulates its expression by promoting alternative polyadenylation site usage. Nucleic Acids Res. 2012, 40: 787-800. 10.1093/nar/gkr783.View ArticleGoogle Scholar
- Corcoran DL, Georgiev S, Mukherjee N, Gottwein E, Skalsky RL, Keene JD, Ohler U: PARalyzer: definition of RNA binding sites from PAR-CLIP short-read sequence data. Genome Biol. 2011, 12: R79-10.1186/gb-2011-12-8-r79.View ArticleGoogle Scholar
- Anders G, Mackowiak SD, Jens M, Maaskola J, Kuntzagk A, Rajewsky N, Landthaler M, Dieterich C: doRiNA: a database of RNA interactions in post-transcriptional regulation. Nucleic Acids Res. 2012, 40: D180-D186. 10.1093/nar/gkr1007.View ArticleGoogle Scholar
- Brown CJ, Hendrich BD, Rupert JL, Lafreniere RG, Xing Y, Lawrence J, Willard HF: The human XIST gene: analysis of a 17 kb inactive X-specific RNA that contains conserved repeats and is highly localized within the nucleus. Cell. 1992, 71: 527-542. 10.1016/0092-8674(92)90520-M.View ArticleGoogle Scholar
- Ascano M, Mukherjee N, Bandaru P, Miller JB, Nusbaum JD, Corcoran DL, Langlois C, Munschauer M, Dewell S, Hafner M, et al: FMRP targets distinct mRNA sequence elements to regulate protein expression. Nature. 2012, 492: 382-386. 10.1038/nature11737.View ArticleGoogle Scholar
- Levine TD, Gao F, King PH, Andrews LG, Keene JD: Hel-N1: an autoimmune RNA-binding protein with specificity for 3' uridylate-rich untranslated regions of growth factor mRNAs. Mol Cell Biol. 1993, 13: 3494-3504.View ArticleGoogle Scholar
- Mukherjee N, Corcoran DL, Nusbaum JD, Reid DW, Georgiev S, Hafner M, Ascano M, Tuschl T, Ohler U, Keene JD: Integrative regulatory mapping indicates that the RNA-binding protein HuR couples pre-mRNA processing and mRNA stability. Mol Cell. 2011, 43: 327-339. 10.1016/j.molcel.2011.06.007.View ArticleGoogle Scholar
- Lebedeva S, Jens M, Theil K, Schwanhausser B, Selbach M, Landthaler M, Rajewsky N: Transcriptome-wide analysis of regulatory interactions of the RNA-binding protein HuR. Mol Cell. 2011, 43: 340-352. 10.1016/j.molcel.2011.06.008.View ArticleGoogle Scholar
- Baltz AG, Munschauer M, Schwanhausser B, Vasile A, Murakawa Y, Schueler M, Youngs N, Penfold-Brown D, Drew K, Milek M, et al: The mRNA-bound proteome and its global occupancy profile on protein-coding transcripts. Mol Cell. 2012, 46: 674-690. 10.1016/j.molcel.2012.05.021.View ArticleGoogle Scholar
- Castello A, Fischer B, Eichelbaum K, Horos R, Beckmann BM, Strein C, Davey NE, Humphreys DT, Preiss T, Steinmetz LM, et al: Insights into RNA biology from an atlas of mammalian mRNA-binding proteins. Cell. 2012, 149: 1393-1406. 10.1016/j.cell.2012.04.031.View ArticleGoogle Scholar
- Wang X, McLachlan J, Zamore PD, Hall TM: Modular recognition of RNA by a human pumilio-homology domain. Cell. 2002, 110: 501-512. 10.1016/S0092-8674(02)00873-5.View ArticleGoogle Scholar
- Gerber AP, Herschlag D, Brown PO: Extensive association of functionally and cytotopically related mRNAs with Puf family RNA-binding proteins in yeast. PLoS Biol. 2004, 2: E79-10.1371/journal.pbio.0020079.View ArticleGoogle Scholar
- White EK, Moore-Jarrett T, Ruley HE: PUM2, a novel murine puf protein, and its consensus RNA-binding site. RNA. 2001, 7: 1855-1866.Google Scholar
- Morris AR, Mukherjee N, Keene JD: Ribonomic analysis of human Pum1 reveals cis-trans conservation across species despite evolution of diverse mRNA target sets. Mol Cell Biol. 2008, 28: 4093-4103. 10.1128/MCB.00155-08.View ArticleGoogle Scholar
- Galgano A, Forrer M, Jaskiewicz L, Kanitz A, Zavolan M, Gerber AP: Comparative analysis of mRNA targets for human PUF-family proteins suggests extensive interaction with the miRNA regulatory system. PLoS One. 2008, 3: e3164-10.1371/journal.pone.0003164.View ArticleGoogle Scholar
- Shiina N, Shinkura K, Tokunaga M: A novel RNA-binding protein in neuronal RNA granules: regulatory machinery for local translation. J Neurosci. 2005, 25: 4420-4434. 10.1523/JNEUROSCI.0382-05.2005.View ArticleGoogle Scholar
- Solomon S, Xu Y, Wang B, David MD, Schubert P, Kennedy D, Schrader JW: Distinct structural features of caprin-1 mediate its interaction with G3BP-1 and its induction of phosphorylation of eukaryotic translation initiation factor 2alpha, entry to cytoplasmic stress granules, and selective interaction with a subset of mRNAs. Mol Cell Biol. 2007, 27: 2324-2342. 10.1128/MCB.02300-06.View ArticleGoogle Scholar
- Tang F, Barbacioru C, Wang Y, Nordman E, Lee C, Xu N, Wang X, Bodeau J, Tuch BB, Siddiqui A, et al: mRNA-Seq whole-transcriptome analysis of a single cell. Nat Methods. 2009, 6: 377-382. 10.1038/nmeth.1315.View ArticleGoogle Scholar
- Tsai DE, Kenan DJ, Keene JD: In vitro selection of an RNA epitope immunologically cross-reactive with a peptide. Proc Natl Acad Sci U S A. 1992, 89: 8864-8868. 10.1073/pnas.89.19.8864.View ArticleGoogle Scholar
- Blasiak J, Trzeciak A, Malecka-Panas E, Drzewoski J, Wojewodzka M: In vitro genotoxicity of ethanol and acetaldehyde in human lymphocytes and the gastrointestinal tract mucosa cells. Toxicol In Vitro. 2000, 14: 287-295. 10.1016/S0887-2333(00)00022-9.View ArticleGoogle Scholar
- Gaillard H, Aguilera A: A novel class of mRNA-containing cytoplasmic granules are produced in response to UV-irradiation. Mol Biol Cell. 2008, 19: 4980-4992. 10.1091/mbc.E08-02-0193.View ArticleGoogle Scholar
- Butter F, Scheibe M, Morl M, Mann M: Unbiased RNA-protein interaction screen by quantitative proteomics. Proc Natl Acad Sci U S A. 2009, 106: 10626-10631. 10.1073/pnas.0812099106.View ArticleGoogle Scholar
- Tsvetanova NG, Klass DM, Salzman J, Brown PO: Proteome-wide search reveals unexpected RNA-binding proteins in Saccharomyces cerevisiae. PLoS One. 2010, 5: 12671-10.1371/journal.pone.0012671.View ArticleGoogle Scholar
- Scherrer T, Mittal N, Janga SC, Gerber AP: A screen for RNA-binding proteins in yeast indicates dual functions for many enzymes. PLoS One. 2010, 5: e15499-10.1371/journal.pone.0015499.View ArticleGoogle Scholar
- Scheibe M, Butter F, Hafner M, Tuschl T, Mann M: Quantitative mass spectrometry and PAR-CLIP to identify RNA-protein interactions. Nucleic Acids Res. 2012, 40: 9897-9902. 10.1093/nar/gks746.View ArticleGoogle Scholar
- Mitchell SF, Jain S, She M, Parker R: Global analysis of yeast mRNPs. Nat Struct Mol Biol. 2013, 20: 127-133.View ArticleGoogle Scholar
- Leung AK, Young AG, Bhutkar A, Zheng GX, Bosson AD, Nielsen CB, Sharp PA: Genome-wide identification of Ago2 binding sites from mouse embryonic stem cells with and without mature microRNAs. Nat Struct Mol Biol. 2011, 18: 237-244. 10.1038/nsmb.1991.View ArticleGoogle Scholar
- Uren PJ, Bahrami-Samani E, Burns SC, Qiao M, Karginov FV, Hodges E, Hannon GJ, Sanford JR, Penalva LO, Smith AD: Site identification in high-throughput RNA-protein interaction data. Bioinformatics. 2012, 28: 3013-3020. 10.1093/bioinformatics/bts569.View ArticleGoogle Scholar
- Whisnant AW, Bogerd HP, Flores O, Ho P, Powers JG, Sharova N, Stevenson M, Chen CH, Cullen BR: In-depth analysis of the interaction of HIV-1 with cellular microRNA biogenesis and effector mechanisms. MBio. 2013, 4: e000193-View ArticleGoogle Scholar
- Chi SW, Zang JB, Mele A, Darnell RB: Argonaute HITS-CLIP decodes microRNA-mRNA interaction maps. Nature. 2009, 460: 479-486.Google Scholar
- Konig J, Zarnack K, Rot G, Curk T, Kayikci M, Zupan B, Turner DJ, Luscombe NM, Ule J: iCLIP reveals the function of hnRNP particles in splicing at individual nucleotide resolution. Nat Struct Mol Biol. 2010, 17: 909-915. 10.1038/nsmb.1838.View ArticleGoogle Scholar
- Langmead B, Trapnell C, Pop M, Salzberg SL: Ultrafast and memory-efficient alignment of short DNA sequences to the human genome. Genome Biol. 2009, 10: R25-10.1186/gb-2009-10-3-r25.View ArticleGoogle Scholar
- Workman CT, Yin Y, Corcoran DL, Ideker T, Stormo GD, Benos PV: enoLOGOS: a versatile web tool for energy normalized sequence logos. Nucleic Acids Res. 2005, 33: W389-W392. 10.1093/nar/gki439.View ArticleGoogle Scholar
- Quinlan AR, Hall IM: BEDTools: a flexible suite of utilities for comparing genomic features. Bioinformatics. 2010, 26: 841-842. 10.1093/bioinformatics/btq033.View ArticleGoogle Scholar
- Skalsky RL, Corcoran DL, Gottwein E, Frank CL, Kang D, Hafner M, Nusbaum JD, Feederle R, Delecluse HJ, Luftig MA, et al: The viral and cellular microRNA targetome in lymphoblastoid cell lines. PLoS Pathog. 2012, 8: e1002484-10.1371/journal.ppat.1002484.View ArticleGoogle Scholar
This article is published under license to BioMed Central Ltd. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.