We incorrectly named the products of two genes, gamma actin and betaAPP, as FMRP ligands because of a misreading of Table 1 from . BetaAPP is incorrectly referred to as an FMRP ligand in Tables 2 and S9, gamma actin in Table S9 only. This does not change the use of FMRP binding as a way to support the target predictions, as the statistics remain virtually the same; the enrichment of FMRP targets within the predicted microRNA target set changes from 5.31 to 5.26 as a result of our misclassiﬁ cation. In addition, it does not change the fact that betaAPP is a predicted microRNA target. Our Web site at http://www.microrna.org/ contains the updated information with changes ﬂ agged. We thank Dr. Denman for pointing out this error to us.
Xiao and colleagues , for instance, have focused in a systematic analysis of features importance carried with a random forest model, whereas here our goal is to deeply explore the predictive power of the random forests algorithm and perform a comprehensive comparison with other popular classifiers in the field. In our previous work , we presented RFMirTarget and discussed its features and training data, as well as preliminary results of our model performance for the prediction of Human miRNAs targets. In this paper, we extend this study in several directions: i) we perform a feature relevance analysis and investigate the effects of feature selection over the predictive accuracy of our model, observing improvement in the overall performance of our classifier, ii) we investigate the impact of feature categories in classification, as well as apply techniques to interpret the ensemble model and the effects of features values over class probabilities, iii) we discuss biological insights provided by the resulting model, highlighting its suitability in identifying biologically relevant features in this classification problem, iv) we test several algorithmic variants such as definition of class weights and distinct permutation methods for the selection of tree’s nodes, showing that the ensemble approach underlying the random forest algorithm seems to reduce its sensitivity to the class imbalance issue, v) we carry a thorough comparison of RFMirTarget against other popular classifiers, proving that the proposed RF model is indeed robust and that its performance is superior than its counterparts methods with statistical significance, vi) we assess the performance of our method in a completely independent test data set of experimentally verified positive and negative examples of miRNA-target and observe a good overall performance and outstanding sensitivity when compared to other miRNA target prediction algorithms.
miRNA promoter targets [13,14,15] were detected and included in the database. The presence of putative miRNA targets on most of gene upstream sequences suggests that miRNA gene regulation through promoter targeting may be a more general event than expected. In stead of merely displaying all possible computational prediction results, microPIR supplies three major types of supporting genomic information, including genome map of experimentally verified AGO binding sites, ESTs and evolutionary conservation of miRNA target sites. It has been hypothesized that promoter-targeting miRNA regulated transcription occurs Figure 2. Search input and report output from the advanced search module. (A) The advanced search page is separated into three main parts. 1) A target gene or associated miRNA is put as a query. 2) The choices of binding site parameter settings can be adjusted to fit the user’s needs. 3) The overlap of target site with specified annotated sequence is allowed as additional search criteria. (B) The list of resulting target sites obtained from search page is displayed. The information of associated miRNA is presented with the direction, chromosomal location, strand, length, upstream location, MFE, and conservation score of each target site including the number of bases with available score data. The number of different annotated sequences overlapping with each predicted target site is also shown. The hypertext link-outs to original sources of gene/miRNA associated information are provided. More target detail which includes a link to primer design is provided on the detail page of each target.
Colorectal cancer (CRC) is the fourth leading cause of cancer-induced mortality. Histone deacetylase 2 (HDAC2) is involved in prognosis and therapy of CRC. This study aimed to explore novel therapeutic targets for CRC. The alteration of HDAC2 expression in CRC tissues was estimated by qRT-PCR. After lentivirus transfection, HDAC2 knockdown was conﬁrmed by western blot analysis. The effect of HDAC2 knockdown on cell proliferation was then assessed by 3-(4,5-dimethylthiazol-2-yl)-2, 5-diphenyltetrazolium bromide (MTT) assay. Screened by TargetScan, microRNA (miR)-455 was predicted to bind to 3 0 UTR of
In the biogenesis of miRNAs, the Argonaute proteins (Ago1-4) along with Gemin3 and Gemin4 selectively bind to the guide strand to facilitate the formation of an miRNA- RNA-induced silencing complex (RISC) (Slaby et al., 2012). Single nucleotide polymor- phisms (SNPs) may be present in miRNA-binding sites, and mature miRNAs negatively regulate the expression level of their target genes via two distinct mechanisms (Bartel, 2004). In the first mechanism, miRNAs block target gene expression at the translational level with imperfect complementarity. In the second mechanism, miRNAs bind to their mRNA targets with perfect (or nearly perfect) complementarity to induce the RNA- mediated interference pathway (Esquela-Kerscher & Slack, 2006) (Fig. 1). Alterations in the miRNA biosynthesis pathway can lead to global miRNA deregulation. Because miRNAs are involved in a wide range of developmental and physiological processes, deregulation of miRNA processing pathways could potentially impact the transcription and splicing of miRNAs as well as the transcriptional regulation of genes that play funda- mental roles in cancers and/or many other human diseases (Kim et al., 2010; Melo & Melo, 2014). Since the impairment of mature miRNAs is emerging as a feature of human cancers (Sonia et al., 2010), given the critical function of Gemin3, Gemin4 and Ago1-4 in miRNA biosynthetic pathway. The host genomic polymorphism of those genes may represent keydeterminants of cancers. SNPs that deregulate miRNAs may alter the expression level of genes related to disease susceptibility (Horikawa et al., 2008; Liu et al., 2012a). Although several studies have investigated the association between the Gemin3 rs197412 T > C, Gemin4 rs7813 T > C and rs2740348 G > C polymorphisms with cancer susceptibility, the results were contradictory and uncertain. Hence, a metaanalysis based on the Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) criteria (Moher et al., 2009) was imperative to assess the associations between cancer susceptibility and the Gemin3 rs197412, Gemin4 rs7813 and Gemin4 rs2740348 polymorphisms.
Most mRNAs targeted for repression by miRNAs are degraded . Therefore, using the results from cohort 1, the miRNAs with regulated expression were analyzed further to determine whether the expression of these miRNAs inversely correlated with the mRNA expression of their predicted targets. Using Partek Genomics Suite software, all targets predicted by either the MicroCosm or TargetScan algorithms for each miRNA were identified. Analysis of the expression for all predicted targets of each miRNA as calculated in the microarray experiments failed to reveal any examples in which a downregulated miRNA correlated with a global increase in its predicted targets, or conversely an figure 1 were also used in these TLDA assays (cohort 1). A) Smoker-to-nonsmoker expression ratios are represented by black circles in order from lowest to highest for the 481 detected miRNAs. The arrow indicates the point where miRNA expression ratios in smokers and nonsmokers = 1. B) The expression ratios are shown for three additional endogenous control options provided with the TLDA assay are shown. C) The expression ratios and p-values of the 481 detected miRNAs are shown using a volcano plot. The significantly upregulated (red) and downregulated (blue) miRNAs are indicated along with the endogenous controls (green). D) The 54 miRNAs with smokers-to-nonsmokers expression ratios greater than 2 are shown following principle component analysis (PCA) with MATLAB software. This analysis identified a representative miRNA within each cluster with the highest Pearson correlation between its expression profile and the first principal component from our PCA analysis. E) Clustering analysis of the 54 regulated miRNAs was performed using CiMminer based on DCt-values of the TLDA results. The 4 clusters identified by PCA are labeled. Upregulated miRNAs are designated by various shades of red and downregulated miRNAs by various shades of blue.
The functional consequence or ‘‘cost’’ of seed region nucleotide changes involves the loss of regulatory control over previously targeted mRNAs and/or the acquisition of novel regulatory control over previously untargeted mRNAs. To systematically explore this phenomenon computationally, we first determined the number of nucleotide changes needed to transform one seed region into another (Hamming distance) for each of the 249 miRNAs analyzed in this study. We then calculated the percent overlap (cosine similarity) of predicted targets for all pairs of miRNAs having identical seeds. For example, the percent target gene overlap for miR-25 and miR-32 (both having the seed sequence: 59AUUGCAC) predicted by miRanda-mirSVR is 81% (Fig. 1A). The 19% (100%–81%) divergence in non- overlapping targets is attributable to sequence variation mapping to the non-seed regions (see S1a Figure for an additional example). The median percent overlap between all pairs of conserved miRNAs with identical seeds is 88% (Fig. 1C; S1 Table ; n5144), with an average of 12% divergence in non-overlapping genes attributable to variation in non-seed regions.
Programming of RISC with a mature miRNA sequence directs the complex to target mRNAs resulting in translational repression and enhanced turnover of some of the target mRNAs [28,29]. The enhanced turnover of target mRNAs can result in lower steady state transcript levels that can be detected via microarray analysis [30–34]. As a preliminary investigation of possible host targets for TTV-tth8-miR-T1, microarray analysis for human transcripts was performed on RNA harvested from HEK293TT cells transfected with TTV-tth8, TTV-MUT, TTV-SV40, or pUC control vector. To identify candidate direct targets, the following criteria were applied: putative targets must be lower in the wildtype sample than the mutant or SV40 control and a seed complementary sequence must be present in the 3 9UTR sequence of the candidate. Using this approach, 16 candidates were identified that were differen- tially expressed between the samples and that possessed seed complementary sequences in their 39UTRs (Table S3). We caution that such analysis should be used for preliminary hypothesis generation, and subsequent confirmation experiments are required to establish direct effects. Examining the list for candidates with known functions in viral infection led us to the gene N-myc (and Figure 4. Generation of mutant and SV40 recombinant TTVs. (A) RNA secondary structure predictions of miRNA region from TTV-tth8 wildtype, TTV-tth8 miRNA mutant, and TTV-tth8 SV40 recombinant viruses. (B) Northern blot analysis of total RNA harvested from HEK293TT cells transfected with either pUC plasmid (Vector), wildtype TTV-tth8 (TTH8 WT), miRNA mutant TTV-tth8 (TTH8 MUT), or SV40 miRNA recombinant TTV- tth8 (TTH8 SV40). Northern blot analysis for host miRNA hsa-miR-19a and ethidium bromide stained low molecular weight RNA serve as load controls. doi:10.1371/journal.ppat.1003818.g004
ATP6V0E1, and SGMS2) that were substantially down-regulated with decreases in protein synthesis of 10-fold or greater. In addition, 45 targets were estimated to have greater than 4-fold changes in protein synthesis. Regardless of the magnitude of regulation, mRNA destabilization accounted for ,75% of the change in estimated protein synthesis. This range of regulation is in good accord with previous studies with genetically characterized endogenous miRNAs as well as with studies introducing exogenous miRNAs introduced into human tissue culture [7,9,16,17,33]. However, our observation that miR-124 had only modest effects on the translation of hundreds of targets contrasts dramatically with several previous studies in which miRNAs reduced protein expression by 5–25-fold while only modestly decreasing mRNA levels (1.1–2-fold), suggesting substantial inhibitory effects on translation [37,44,61,69,91]. The previous studies, however, measured the effect of a specific miRNA on reporter constructs in which the 39-UTRs of the encoded mRNAs were not derived from mammalian mRNAs, but were either short (,250 nts) modified viral sequences or artificial. In contrast, mammalian mRNA 3 9-UTRs tend to be much longer (on average ,1,000 nts) and include regulatory sites for RNA-binding proteins and regulatory RNAs that influence mRNA localization, transla- tion, and decay. The basis for the discrepancy in the results from these two experimental designs remains to be determined, and the answer is likely to provide useful mechanistic insights. One possibility is that mRNAs containing exogenous 3 9-UTRs might have anomalously long mRNA half-lives that obscure the normal contribution of mRNA degradation to the miRNA-directed inhibition of protein expression. The large magnitude of effects observed in reporter-based assays, compared to what we and others have observed with endogenous mRNAs, is likely to be partially due to the multiple (four to eight) engineered miRNA binding sites in the reporter constructs used in those studies [37,44,61,69,91]. Further, these sites were in close proximity, and adjacent miRNA binding sites have been reported to act cooperatively [36,93,98]. Indeed, two studies that measured the effects of specific miRNAs on protein and mRNA levels of reporters with endogenous mammalian 39-UTRs found more modest effects on translation, less than 2-fold on average [11,72]. Moreover, the magnitude of the effects we observed on translation of the mRNAs targeted by miR-124 were in agreement with two recent studies that inferred the repressive effect of miRNAs on translation by measuring miRNA-mediated effects on mRNA and protein abundance [16,17]. Those reports, based on directly measured changes in protein levels by quantitative mass spectrometry, concluded that the effects of miRNAs on translation were small (less than 2-fold for hundreds of target mRNAs).
The miRNA-like sequences we have identified in HIV-1 are unique in that they do not seem to be derived from cellular miRNA, nor do they appear to represent viral miRNAs or their targets. The hsa-miR-195-like sequence corresponds to the first 18 nucleotides of the mature hsa-miR-195, which has a length of 21 nucleotides. Any functional similarity of this sequence to the cellular hsa-miR-195 may be speculative. However, it should be noted that there are miRNAs as short as 17 nt; and that HIV-1 has been reported to encode a viral miRNA, designated TAR-3p, whose cloned length is 17 nt . Further, the potential action of a miRNA is mostly dependent on base pairing between the miRNA seed sequence and its target; positions 13–16 of the miRNA may aid in pairing as well [5,36]. The miR-195-like sequence we have identified in #GU216763 contains both the seed region and positions 13–16 of hsa-miR-195 and is 100% conserved in these regions. It has also been predicted computa- tionally that the cellular hsa-miR-195 may interact with the HIV-1 Nef in the3’ LTR region based on a perfect complementarity of a 7 nucleotide seed sequence with its viral target .
Despite the role of miR-155 in Treg survival, this mol- ecule is still classified as proinflammatory, as it precisely regulates the levels of its targets to promote the immune response. Conversely, miR-146a and miR-21 are nega- tive feedback regulators that mute the immune response . It has been demonstrated that miR-21 acts as a positive indirect regulator of FOXP3 expression; in con- trast, miR-31 negatively regulates FOXP3 expression by binding directly to its target site in the 3 ′UTR of FOXP3 mRNA. Comparing miRNA expression profiles between human naïve CD4 +
inconsistency in miRNA expression between them, which may be due to differences in the high-throughput platforms and methods used in different laboratories or due to differences among the cancer population. Also, a full characterization of the complex relationship between miRNAs and their target mRNAs in cervical malignant transformation has not yet been carried out. We present the results of miRNA expression profiling in cervical squamous cell carcinomas (SCC), low and high-grade intraepithe- lial cervical lesions and normal cervical epithelial tissues. As in other studies, we have observed high expression variability between samples, especially among normal cervical samples, which did not allow us to obtain a unique miRNA expression signature for this tumour type. We demonstrate using Taqman miRNA real-time PCR quantification method that such variability is biological rather than technical. We have tackled such biological variability by pooling the RNAs from the normal samples, which averaged miRNAs levels in the controls, and we demonstrate that this methodology is sufficiently robust to identify miRNAs that were deregulated between malignant, pre-malignant and normal cervical tissues, which may be involved in cervical carcinogenesis. We also identify possible gene targets of relevant candidate miRNAs.
effectively inhibit miRNAs while sponges with perfect MBS do not. A limited effect on cell growth was observed using two bulged MBS and similar strong effects were observed using sponge constructs containing six and twenty MBS. Thus, in WEHI-231 cells introduction of miR-19 sponges with six bulged MBS already suffices for maximal miRNA inhibition. Depending on the level of miRNA expression and the number and binding energy of endogenous miRNA targets more or less MBS may suffice for maximal miRNA inhibition . In the case of miR-17 inhibition in WEHI-231, sponges with twelve MBS showed a 2-fold faster reduction in GFP+ cells as compared to sponges with four MBS (data not shown). No effects on cell growth were observed for perfect MBS sponges regardless the number of MBS incorporated in the sponge construct. The lower effectiveness of perfect sponges compared to bulged sponges can at least in part be explained by the lower perfect sponge RNA transcripts levels. It has been reported that upon perfect miRNA binding, mRNAs are cleaved by the slicer activity of the RISC complex . Thus, although we do not show direct cleavage of perfect MBS sponge transcripts it is likely that the perfect MBS sponge transcript levels are lowered due to mRNA cleavage and therefore are less suitable for miRNA inhibition. Nevertheless, perfect MBS sequences are suitable for miRNA mediated downreg- ulation of a specific transcript [25,26] as we show in our luciferase reporter assay. Both PITA and STarMir predicted a more efficient binding of miRNAs to perfect MBS sponges compared to bulged MBS sponges consistent with the results obtained from the luciferase assay. However, neither PITA nor STarMir take into consideration the effects of sponge degradation upon binding of a perfectly matched antisense miRNA. Thus, when designing sponges for long- term miRNA inhibition one should not solely focus on the DDG/ DG total but also ensure that MBS are not perfect antisense.
There is no crystallographic structure presently avail- able for TRR1 and KRE2 of P. brasiliensis and also for the other pathogenic fungi. Therefore, the 3D struc- tures of TRR1 and KRE2 of P. brasiliensis were con- structed by homology modeling based on known structures with high percentage of identity in amino acid sequences. We have initially modeled P. brasilien- sis proteins but it will be similar for the other patho- genic fungi since the sequences of the proteins are highly conserved. The known template structures were searched in the PDB. There were two templates for TRR1 protein: 3ITJ (PDB ID) of S. cerevisiae and 1VDC (PDB ID) of A. thaliana. There was one tem- plate for KRE2 protein: 1S4N (PDB ID) of S. cerevisiae. The templates structures for ERG6 and RIM8 showed low sequence identity, then not allowed the construc- tion of 3D structures for these proteins by molecular modeling. The amino acid residue sequences of TRR1 and KRE2 were compared with the primary sequences of the structures deposited in the PDB using the BLAST program. The homologous sequences allowed the construction of a 3D model of TRR1 and KRE2 using the homology module of the Insight II software package (Biosym/MSI, San Diego, Accelrys Inc. 2001). Briefly, the target sequences were aligned with the template structures, and coordinates from the tem- plates were transferred to the targets TRR1 and KRE2. For model optimization, the backbone atoms of the structures were initially frozen and only the side chains were allowed to move for a selective minimization by conjugate gradient method. A second selective minimi- zation, also by conjugate gradient method, was per- formed with only atoms of the complementary determining region (CDR) loops moving. The last minimization was performed by Steepest-descent method with all atoms of the structure relaxed, result- ing in whole, refined 3D structures. The molecular visualization was performed by PyMOL open-source software version 0.99rc6 (Delano Scientific LLC, 2006).
84. Delegates made a number of suggestions both for enhancing PAHO’s approach and for improving the document. The Delegate of Costa Rica, alluding to the strategic priorities for PAHO listed in paragraph 35 of the document, suggested adding “individual and collective” to the strategic priority that read “Protect health as a public good and human right.” Otherwise, it might appear that the concern was only for the health of individuals, not health in its collective dimension. The Delegate of Canada underscored the importance of reflecting in the Organization’s MDG strategy the key role of sexual and reproductive health in the attainment of a number of the targets. He also suggested that PAHO should develop an implementation timetable for its MDG work, indicating major milestones and deadlines for activities. The Delegate of Mexico proposed specific changes and additions to paragraphs 11, 12, 33, and 38 and said that his delegation would submit additional comments in writing. In particular, he suggested that the Secretariat should take a careful look at paragraph 12 in order to ensure that it did not give the impression that PAHO would lead the process of building partnerships and strategic alliances at country level. PAHO’s function should be to support ministries of health in that process and to strengthen their capacity to forge such alliances in response to needs identified by the country itself.
Tuberculosis (TB) is the most important cause of human death from a curable infectious disease. It is estimated that, worldwide, one hundred million people are infected annually and about ten million develop the disease, with five million of those progressing to an infectious stage, culminating with approximately three million deaths. According to the World Health Organization , the overall incidence of TB increases approximately 0.3% per year. The resurgence of this health problem occurred mainly due to the proliferation of multi (MDR-TB), extensively (XDR-TB), and recently, totally-drug (TDR-TB) resistant Mt strains. Besides, the high susceptibility of HIV/AIDS infected patients to TB is also a health problem. Therefore, there is an urgent need for the discovery and development of new and better drugs for the TB treatment .
Resveratrol is also capable of increasing the levels of miR-663, a tumor-suppressor miRNA that targets TGF-β1 transcripts. The results of this study indicate that since, it is possible to manipulate the levels of important miRNAs, like miR-663. Resveratrol could have its anti-metastatic and anticancer effects increased. (9)
One of the genetic variants tested in mice is located in the 59 UTR FADS1 intergenic region (rs6413006). FADS1 and FADS2 share the same 5 9 UTR region both in mice and humans, however sequence comparison of human and mouse homologues regions using VISTA computational comparative tool, reveals that the FADS1 and FADS2 overlapping 59 UTR region is less conserved (identity #70%) (Figure S3, VISTA computational comparative tool) [43,44]. One possible explanations for these findings is that the human and mouse TF binding sites might not be in the same 5 9 UTR regions and in addition there are many cases where regulatory regions do not align nicely unless you allow for gaps. The other possibility is that rs6413006 and D19Mit42 genetic variants are not the functional variants interfering with FADS1/2 expression but are in high linkage disequilibrium with more conserved functional sites.
LncRNAs have key roles in gene regulation and, consequently, affect various aspects of cellular homeostasis, including proliferation, survival, migration or genomic stability. Cancer is primarily caused by genetic alterations that result in the deregulation of the gene networks that are responsible for the maintenance of cellular homeostasis. In addition, tumor suppressor genes and oncogenes may regulate the transcription of lncRNAs. For their application as biomarkers, lncRNAs should present tissue- and cell type-specific expression and, should be stable and easily detectable in body fluids to permit noninvasive diagnosis (43). For example, the lncRNA PCA3 is a specific and sensitive marker of prostate cancer in patient urine samples, which was the first Food and Drug Administration-approved test based on a lncRNA (46). Similarly, the lncRNA HULC is highly expressed in patients with hepatocellular carcinoma and can be detected in the blood by conventional PCR methods. Since lncRNAs regulate specific facets of protein activity, they also represent potential therapeutic targets. Of note, drugs that target lncRNAs can be more refined and less toxic than conventional protein-targeting drugs (43). Some therapeutic strategies utilize oligonucleotides for the knockdown of lncRNAs. siRNAs consist of an antisense (or guide) strand and a sense (or passenger) strand, forming a duplex 19 to 25 bp in length with 3' dinucleotide overhangs, and are mainly located in the cytoplasm. When bound by Argonaut 2 in the RNA-induced silencing complex, the passenger strand is discarded, leaving the guide strand free to bind an RNA target that is subsequently degraded (50,51). Alternatively, LNA TM GapmeRs