Nencki Genomics Database--Ensembl funcgen enhanced with intersections, user data and genome-wide TFBS motifs

The hypoxia-induced changes in miRNA-mRNA in RNA-induced silencing complexes and HIF-2 induced miRNAs in human endothelial cells

The cellular adaptive response to hypoxia relies on the expression of hypoxia-inducible factors (HIFs), HIF-1 and HIF-2. HIFs regulate global gene expression changes during hypoxia that are necessary for restoring oxygen homeostasis and promoting cell survival. In the early stages of hypoxia, HIF-1 is elevated, whereas at the later stages, HIF-2 becomes the predominant form. What governs the transition between the two HIFs (the HIF switch) and the role of miRNAs in this regulation are not completely clear. Genome-wide expression studies on the miRNA content of RNA-induced silencing complexes (RISC) in HUVECs exposed to hypoxia compared to the global miRNA-Seq analysis revealed very specific differences between these two populations. We analyzed the miRNA and mRNA composition of RISC at 2 h (mainly HIF-1 driven), 8 h (HIF-1 and HIF-2 elevated), and 16 h (mainly HIF-2 driven) in a gene ontology context. This allowed for determining the direct impact of the miRNAs in modulating the cellular signaling pathways involved in the hypoxic adaptive response. Our results indicate that the miRNA-mRNA RISC components control the adaptive responses, and this does not always rely on the miRNA transcriptional elevations during hypoxia. Furthermore, we demonstrate that the hypoxic levels of the vast majority of HIF-1-dependent miRNAs (including miR-210-3p) are also HIF-2 dependent and that HIF-2 governs the expression of 11 specific miRNAs. In summary, the switch from HIF-1 to HIF-2 during hypoxia provides an important level of miRNA-driven control in the adaptive pathways in endothelial cells.

Functional and HRE motifs count analysis of induction of selected hypoxia-responsive genes by HIF-1 and HIF-2 in human umbilical endothelial cells

We analyzed the effects of selective knockdown of either HIF-1α or HIF-2α on the transcriptional response to hypoxia of human umbilical endothelial cells at two time-points (2 h and 8 h) of hypoxia. We focused on 13 previously identified hypoxia-responsive genes, pre-selected to have different activation kinetics and different proportions of HRE motifs annotated to either HIF-1 or HIF-2 in open promoters - open chromatin DNase-hypersensitive sites (DHS) regions within ±1 kb of the gene start. We report that genes activated by both HIF-1 and 2 tend to be activated earlier than genes activated by HIF-1 only, which, in turn, tend to be activated earlier than genes activated by HIF-2 only. Moreover, for the 13 analyzed genes, we found that the effect of silencing HIF1A on the gene induction by hypoxia is greater for the genes with more HRE motifs annotated to HIF-1 in their promoter open chromatin DHS regions within ±1 kb and also within ±10 kb of the gene start. We corroborated and extended this finding by showing that among 232 genes previously identified as activated by hypoxia, the genes with ChIP-seq peak(s) for HIF-1α within a ±10 kb flank of the gene start contain more HRE motifs annotated to HIF-1 in the DHS regions within this flank than the genes with no ChIP-seq peaks. Also in the whole genome, the DHS regions intersecting ChIP-seq peaks for HIF-1α contain more HRE motifs annotated to HIF-1 than the DHS regions not intersecting the ChIP-seq peaks. This suggests a mechanism, by which higher promoter content of HRE motifs in DHS regions increases HIF-1 binding, which in turn increases gene induction by hypoxia.

Genome-wide mRNA profiling identifies X-box-binding protein 1 (XBP1) as an IRE1 and PUMA repressor

Accumulation of misfolded proteins in ER activates the unfolded protein response (UPR), a multifunctional signaling pathway that is important for cell survival. The UPR is regulated by three ER transmembrane sensors, one of which is inositol-requiring protein 1 (IRE1). IRE1 activates a transcription factor, X-box-binding protein 1 (XBP1), by removing a 26-base intron from XBP1 mRNA that generates spliced XBP1 mRNA (XBP1s). To search for XBP1 transcriptional targets, we utilized an XBP1s-inducible human cell line to limit XBP1 expression in a controlled manner. We also verified the identified XBP1-dependent genes with specific silencing of this transcription factor during pharmacological ER stress induction with both an N-linked glycosylation inhibitor (tunicamycin) and a non-competitive inhibitor of the sarco/endoplasmic reticulum Ca²⁺ ATPase (SERCA) (thapsigargin). We then compared those results to the XBP1s-induced cell line without pharmacological ER stress induction. Using next‐generation sequencing followed by bioinformatic analysis of XBP1-binding motifs, we defined an XBP1 regulatory network and identified XBP1 as a repressor of PUMA (a proapoptotic gene) and IRE1 mRNA expression during the UPR. Our results indicate impairing IRE1 activity during ER stress conditions accelerates cell death in ER-stressed cells, whereas elevating XBP1 expression during ER stress using an inducible cell line correlated with a clear prosurvival effect and reduced PUMA protein expression. Although further studies will be required to test the underlying molecular mechanisms involved in the relationship between these genes with XBP1, these studies identify a novel repressive role of XBP1 during the UPR.

Transcription factors Krüppel-like factor 4 and paired box 5 regulate the expression of the Grainyhead-like genes

Genes from the Grainyhead-like (GRHL) family code for transcription factors necessary for the development and maintenance of various epithelia. These genes are also very important in the development of many types of cancer. However, little is known about the regulation of expression of GRHL genes. Previously, there were no systematic analyses of the promoters of GRHL genes or transcription factors that bind to these promoters. Here we report that the Krüppel-like factor 4 (KLF4) and the paired box 5 factor (PAX5) bind to the regulatory regions of the GRHL genes and regulate their expression. Ectopic expression of KLF4 or PAX5 alters the expression of GRHL genes. In KLF4-overexpressing HEK293 cells, the expression of GRHL1 and GRHL3 genes was upregulated by 32% and 60%, respectively, whereas the mRNA level of GRHL2 gene was lowered by 28% when compared to the respective controls. The levels of GRHL1 and GRHL3 expression were decreased by 30% or 33% in PAX5-overexpressing HEK293 cells. The presence of minor frequency allele of single nucleotide polymorphism rs115898376 in the promoter of the GRHL1 gene affected the binding of KLF4 to this site. The evidence presented here suggests an important role of KLF4 and PAX5 in the regulation of expression of GRHL1-3 genes.

Searching for improvements in predicting human eye colour from DNA

Increasing understanding of human genome variability allows for better use of the predictive potential of DNA. An obvious direct application is the prediction of the physical phenotypes. Significant success has been achieved, especially in predicting pigmentation characteristics, but the inference of some phenotypes is still challenging. In search of further improvements in predicting human eye colour, we conducted whole-exome (enriched in regulome) sequencing of 150 Polish samples to discover new markers. For this, we adopted quantitative characterization of eye colour phenotypes using high-resolution photographic images of the iris in combination with DIAT software analysis. An independent set of 849 samples was used for subsequent predictive modelling. Newly identified candidates and 114 additional literature-based selected SNPs, previously associated with pigmentation, and advanced machine learning algorithms were used. Whole-exome sequencing analysis found 27 previously unreported candidate SNP markers for eye colour. The highest overall prediction accuracies were achieved with LASSO-regularized and BIC-based selected regression models. A new candidate variant, rs2253104, located in the ARFIP2 gene and identified with the HyperLasso method, revealed predictive potential and was included in the best-performing regression models. Advanced machine learning approaches showed a significant increase in sensitivity of intermediate eye colour prediction (up to 39%) compared to 0% obtained for the original IrisPlex model. We identified a new potential predictor of eye colour and evaluated several widely used advanced machine learning algorithms in predictive analysis of this trait. Our results provide useful hints for developing future predictive models for eye colour in forensic and anthropological studies.

TGFβR-SMAD3 Signaling Induces Resistance to PARP Inhibitors in the Bone Marrow Microenvironment

Synthetic lethality triggered by PARP inhibitor (PARPi) yields promising therapeutic results. Unfortunately, tumor cells acquire PARPi resistance, which is usually associated with the restoration of homologous recombination, loss of PARP1 expression, and/or loss of DNA double-strand break (DSB) end resection regulation. Here, we identify a constitutive mechanism of resistance to PARPi. We report that the bone marrow microenvironment (BMM) facilitates DSB repair activity in leukemia cells to protect them against PARPi-mediated synthetic lethality. This effect depends on the hypoxia-induced overexpression of transforming growth factor beta receptor (TGFβR) kinase on malignant cells, which is activated by bone marrow stromal cells-derived transforming growth factor beta 1 (TGF-β1). Genetic and/or pharmacological targeting of the TGF-β1-TGFβR kinase axis results in the restoration of the sensitivity of malignant cells to PARPi in BMM and prolongs the survival of leukemia-bearing mice. Our finding may lead to the therapeutic application of the TGFβR inhibitor in patients receiving PARPis.

Exploring the possibility of predicting human head hair greying from DNA using whole-exome and targeted NGS data

BACKGROUND

Greying of the hair is an obvious sign of human aging. In addition to age, sex- and ancestry-specific patterns of hair greying are also observed and the progression of greying may be affected by environmental factors. However, little is known about the genetic control of this process. This study aimed to assess the potential of genetic data to predict hair greying in a population of nearly 1000 individuals from Poland.

RESULTS

The study involved whole-exome sequencing followed by targeted analysis of 378 exome-wide and literature-based selected SNPs. For the selection of predictors, the minimum redundancy maximum relevance (mRMRe) method was used, and then two prediction models were developed. The models included age, sex and 13 unique SNPs. Two SNPs of the highest mRMRe score included whole-exome identified KIF1A rs59733750 and previously linked with hair loss FGF5 rs7680591. The model for greying vs. no greying prediction achieved accuracy of cross-validated AUC = 0.873. In the 3-grade classification cross-validated AUC equalled 0.864 for no greying, 0.791 for mild greying and 0.875 for severe greying. Although these values present fairly accurate prediction, most of the prediction information was brought by age alone. Genetic variants explained < 10% of hair greying variation and the impact of particular SNPs on prediction accuracy was found to be small.

CONCLUSIONS

The rate of changes in human progressive traits shows inter-individual variation, therefore they are perceived as biomarkers of the biological age of the organism. The knowledge on the mechanisms underlying phenotypic aging can be of special interest to the medicine, cosmetics industry and forensics. Our study improves the knowledge on the genetics underlying hair greying processes, presents prototype models for prediction and proves hair greying being genetically a very complex trait. Finally, we propose a four-step approach based on genetic and epigenetic data analysis allowing for i) sex determination; ii) genetic ancestry inference; iii) greying-associated SNPs assignment and iv) epigenetic age estimation, all needed for a final prediction of greying.

Genome-wide mRNA profiling identifies RCAN1 and GADD45A as regulators of the transitional switch from survival to apoptosis during ER stress

Endoplasmic reticulum (ER) stress conditions promote a cellular adaptive mechanism called the unfolded protein response (UPR) that utilizes three stress sensors, inositol-requiring protein 1, protein kinase RNA-like ER kinase, and activating transcription factor 6. These sensors activate a number of pathways to reduce the stress and facilitate cell survival. While much is known about the mechanisms involved that modulate apoptosis during chronic stress, less is known about the transition between the prosurvival and proapoptotic factors that determine cell fate. Here, we employed a genetic screen that utilized three different pharmacological stressors to induce ER stress in a human-immortalized airway epithelial cell line, immortalized human bronchial epithelial cells. We followed the stress responses over an 18-h time course and utilized real-time monitoring of cell survival, next-generation sequencing, and quantitative real-time PCR to identify and validate genes that were upregulated with all three commonly employed ER stressors, inhibitor of calpain 1, tunicamycin, and thapsigargin. growth arrest and DNA damage-inducible alpha (GADD45A), a proapoptotic factor, and regulator of calcineurin 1 (RCAN1) mRNAs were identified and verified by showing that small interfering RNA (siRNA) knockdown of GADD45A decreased CCAAT-enhancer-binding protein homologous protein (a.k.a DDIT3), BCL2-binding component 3 (a.k.a. BBC3), and phorbol-12-myristate-13-acetate-induced protein 1 expression, 3 proapoptotic factors, and increased cell viability during ER stress conditions, whereas siRNA knockdown of RCAN1 dramatically decreased cell viability. These results suggest that the relative levels of these two genes regulate cell fate decisions during ER stress independent of the type of ER stressor.

Search for novel STAT3-dependent genes reveals SERPINA3 as a new STAT3 target that regulates invasion of human melanoma cells

Transcription factor signal transducer and activator of transcription 3 (STAT3) is constitutively activated in many cancers and promotes uncontrolled tumor growth and progression through multiple mechanisms. Compelling evidence shows tissue and cell-specific sets of STAT3 targets. Transcriptional targets of STAT3 in melanoma cells are largely unknown. Malignant melanoma is a deadly disease with highly aggressive and drug-resistant behavior. Less than 10% of patients with advanced melanomas reach the 5-year survival, partly due to the aggressive character of the tumor and ineffectiveness of current therapeutics for treating metastatic melanoma. STAT3 is constitutively activated in melanoma cells and plays important roles in its growth and angiogenesis in tumor xenograft studies. Moreover, highly metastatic melanoma cells have higher levels of active STAT3 than poorly metastatic ones. To identify genes that are driven by STAT3 in human melanoma cells, we performed JAK/STAT signaling specific and global gene expression profiling of human melanoma cells with silenced STAT3 expression. For selected genes, we performed computational identification of putative STAT3-binding sites and validated direct interactions STAT3 with defined promoters by using chromatin immunoprecipitation followed by qPCR. We found that STAT3 knockdown does not affect human melanoma cell viability, proliferation, or response to chemotherapeutics. We show that STAT3 regulates a discrete set of genes in melanoma cells, including SERPINA3, a novel STAT3 target gene, which is functionally involved in regulation of melanoma migration and invasion. Knockdown of STAT3 impaired cell migration and invasion, in part via regulation of its transcriptional target SERPINA3. Our results present novel targets and functions of STAT3 in melanoma cells.

miR-34c-5p modulates X-box-binding protein 1 (XBP1) expression during the adaptive phase of the unfolded protein response

During endoplasmic reticulum (ER) stress conditions, an adaptive signaling network termed the unfolded protein response (UPR) is activated. The UPR's function is to increase ER protein-folding capacity in order to attenuate ER stress, restore ER homeostasis, and, most importantly, promote cell survival. X-box-binding protein 1 (XBP1) is one component of the UPR and is a proadaptive transcription factor that is subject to transcriptional, post-transcriptional, and post-translational control. In the present study, we identified a post-transcriptional mechanism mediated by miR-34c-5p that governs the expression of both the spliced (active) and unspliced (latent) forms of XBP1 mRNAs. We showed that miR-34c-5p directly attenuates spliced XBP1 (XBP1s) mRNA levels during ER stress and thus regulates the proadaptive component of the UPR that is mediated by XBP1s without interfering with the induction of apoptotic responses.-Bartoszewska, S., Cabaj, A., Dąbrowski, M., Collawn, J. F., Bartoszewski, R. miR-34c-5p modulates X-box-binding protein 1 (XBP1) expression during the adaptive phase of the unfolded protein response.

Primary endothelial cell-specific regulation of hypoxia-inducible factor (HIF)-1 and HIF-2 and their target gene expression profiles during hypoxia

During hypoxia, a cellular adaptive response activates hypoxia-inducible factors (HIFs; HIF-1 and HIF-2) that respond to low tissue-oxygen levels and induce the expression of a number of genes that promote angiogenesis, energy metabolism, and cell survival. HIF-1 and HIF-2 regulate endothelial cell (EC) adaptation by activating gene-signaling cascades that promote endothelial migration, growth, and differentiation. An HIF-1 to HIF-2 transition or switch governs this process from acute to prolonged hypoxia. In the present study, we evaluated the mechanisms governing the HIF switch in 10 different primary human ECs from different vascular beds during the early stages of hypoxia. The studies demonstrate that the switch from HIF-1 to HIF-2 constitutes a universal mechanism of cellular adaptation to hypoxic stress and that HIF1A and HIF2A mRNA stability differences contribute to HIF switch. Furthermore, using 4 genome-wide mRNA expression arrays of HUVECs during normoxia and after 2, 8, and 16 h of hypoxia, we show using bioinformatics analyses that, although a number of genes appeared to be regulated exclusively by HIF-1 or HIF-2, the largest number of genes appeared to be regulated by both.-Bartoszewski, R., Moszyńska, A., Serocki, M., Cabaj, A., Polten, A., Ochocka, R., Dell'Italia, L., Bartoszewska, S., Króliczewski, J., Dąbrowski, M., Collawn, J. F. Primary endothelial cell-specific regulation of hypoxia-inducible factor (HIF)-1 and HIF-2 and their target gene expression profiles during hypoxia.

PIWI proteins contribute to apoptosis during the UPR in human airway epithelial cells

Small noncoding microRNAs (miRNAs) post-transcriptionally regulate a large portion of the human transcriptome. miRNAs have been shown to play an important role in the unfolded protein response (UPR), a cellular adaptive mechanism that is important in alleviating endoplasmic reticulum (ER) stress and promoting cell recovery. Another class of small noncoding RNAs, the Piwi-interacting RNAs (piRNAs) together with PIWI proteins, was originally shown to play a role as repressors of germline transposable elements. More recent studies, however, indicate that P-element induced WImpy proteins (PIWI proteins) and piRNAs also regulate mRNA levels in somatic tissues. Using genome-wide small RNA next generation sequencing, cell viability assays, and caspase activity assays in human airway epithelial cells, we demonstrate that ER stress specifically up-regulates total piRNA expression profiles, and these changes correlate with UPR-induced apoptosis as shown by up-regulation of two pro-apoptotic factor mRNAs, CHOP and NOXA. Furthermore, siRNA knockdown of PIWIL2 and PIWIL4, two proteins involved in piRNA function, attenuates UPR-related cell death, inhibits piRNA expression, and inhibits the up-regulation of CHOP and NOXA mRNA expression. Hence, we provide evidence that PIWIL2 and PIWIL4 proteins, and potentially the up-regulated piRNAs, constitute a novel epigenetic mechanism that control cellular fate during the UPR.

Increased phosphorylation of eIF2α in chronic myeloid leukemia cells stimulates secretion of matrix modifying enzymes

Recent studies underscore the role of the microenvironment in therapy resistance of chronic myeloid leukemia (CML) cells and leukemia progression. We previously showed that sustained mild activation of endoplasmic reticulum (ER) stress in CML cells supports their survival and resistance to chemotherapy. We now demonstrate, using dominant negative non-phosphorylable mutant of eukaryotic initiation factor 2 α subunit (eIF2α), that phosphorylation of eIF2α (eIF2α-P), which is a hallmark of ER stress in CML cells, substantially enhances their invasive potential and modifies their ability to secrete extracellular components, including the matrix-modifying enzymes cathepsins and matrix metalloproteinases. These changes are dependent on the induction of activating transcription factor-4 (ATF4) and facilitate extracellular matrix degradation by CML cells. Conditioned media from CML cells with constitutive activation of the eIF2α-P/ATF4 pathway induces invasiveness of bone marrow stromal fibroblasts, suggesting that eIF2α-P may be important for extracellular matrix remodeling and thus leukemia cells-stroma interactions. Our data show that activation of stress response in CML cells may contribute to the disruption of bone marrow niche components by cancer cells and in this way support CML progression.

Coordinated expression and genetic polymorphisms in Grainyhead-like genes in human non-melanoma skin cancers

BACKGROUND

The Grainyhead-like (GRHL) transcription factors have been linked to many different types of cancer. However, no previous study has attempted to investigate potential correlations in expression of different GRHL genes in this context. Furthermore, there is very little information concerning damaging mutations and/or single nucleotide polymorphisms in GRHL genes that may be linked to cancer.

METHODS

DNA and RNA were extracted from human non-melanoma skin cancers (NMSC) and adjacent normal tissues (n = 33 pairs of samples). The expression of GRHL genes was measured by quantitative real time PCR. Regulation of GRHL expression by miRNA was studied using cell transfection methods and dual-luciferase reporter system. Targeted deep sequencing of GRHL genes in tumor samples and control tissues were employed to search for mutations and single nucleotide polymorphisms. Single marker rs141193530 was genotyped with pyrosequencing in additional NMSC replication cohort (n = 176). Appropriate statistical and bioinformatic methods were used to analyze and interpret results.

RESULTS

We discovered that the expression of two genes - GRHL1 and GRHL3 - is reduced in a coordinated manner in tumor samples, in comparison to the control healthy skin samples obtained from the same individuals. It is possible that both GRHL1 and GRHL3 are regulated, at least to some extent, by different strands of the same oncogenic microRNA - miR-21, what would at least partially explain observed correlation. No de novo mutations in the GRHL genes were detected in the examined tumor samples. However, some single nucleotide polymorphisms in the GRHL genes occur at significantly altered frequencies in the examined group of NMSC patients.

CONCLUSIONS

Non-melanoma skin cancer growth is accompanied by coordinated reduced expression of epidermal differentiation genes: GRHL1 and GRHL3, which may be regulated by miR-21-3p and -5p, respectively. Some potentially damaging single nucleotide polymorphisms in GRHL genes occur with altered frequencies in NMSC patients, and they may in particular impair the expression of GRHL3 gene or functioning of encoded protein. The presence of these polymorphisms may indicate an increased risk of NMSC development in affected people.

The embryonic type of SPP1 transcriptional regulation is re-activated in glioblastoma

Osteopontin (SPP1, a secreted phosphoprotein 1) is primarily involved in immune responses, tissue remodelling and biomineralization. However, it is also overexpressed in many cancers and regulates tumour progression by increasing migration, invasion and cancer stem cell self-renewal. Mechanisms of SPP1 overexpression in gliomas are poorly understood. We demonstrate overexpression of two out of five SPP1 isoforms in glioblastoma (GBM) and differential isoform expression in glioma cell lines. Up-regulated SPP1 expression is associated with binding of the GLI1 transcription factor to the promoter and OCT4 (octamer-binding transcription factor 4) to the first SPP1 intron of the SPP1 gene in human glioma cells but not in non-transformed astrocytes. GLI1 knockdown reduced SPP1 mRNA and protein levels in glioma cells. GLI1 and OCT4 are known regulators of stem cell pluripotency. GBMs contain rare cells that express stem cell markers and display ability to self-renew. We reveal that SPP1 is overexpressed in glioma initiating cells defined by high rhodamine 123 efflux, sphere forming capacity and stemness marker expression. Forced differentiation of human glioma spheres reduced SPP1 expression. Knockdown of SPP1, GLI1 or CD44 with siRNAs diminished sphere formation. C6 glioma cells stably depleted of Spp1 displayed reduced sphere forming capacity and downregulated stemness marker expression. Overexpression of the wild type Spp1, but not Spp1 lacking a Cd44 binding domain, rescued cell ability to form spheres. Our findings show re-activation of the embryonic-type transcriptional regulation of SPP1 in malignant gliomas and point to the importance of SPP1-CD44 interactions in self-renewal and pluripotency glioma initiating cells.

Adult Deletion of SRF Increases Epileptogenesis and Decreases Activity-Induced Gene Expression

Although the transcription factor serum response factor (SRF) has been suggested to play a role in activity-dependent gene expression and mediate plasticity-associated structural changes in the hippocampus, no unequivocal evidence has been provided for its role in brain pathology, such as epilepsy. A genome-wide program of activity-induced genes that are regulated by SRF also remains unknown. In the present study, we show that the inducible and conditional deletion of SRF in the adult mouse hippocampus increases the epileptic phenotype in the kainic acid model of epilepsy, reflected by more severe and frequent seizures. Moreover, we observe a robust decrease in activity-induced gene transcription in SRF knockout mice. We characterize the genetic program controlled by SRF in neurons and using functional annotation, we find that SRF target genes are associated with synaptic plasticity and epilepsy. Several of these SRF targets function as regulators of inhibitory or excitatory balance and the structural plasticity of neurons. Interestingly, mutations in those SRF targets have found to be associated with such human neuropsychiatric disorders, as autism and intellectual disability. We also identify novel direct SRF targets in hippocampus: Npas4, Gadd45g, and Zfp36. Altogether, our data indicate that proteins that are highly upregulated by neuronal stimulation, identified in the present study as SRF targets, may function as endogenous protectors against overactivation. Thus, the lack of these effector proteins in SRF knockout animals may lead to uncontrolled excitation and eventually epilepsy.

Emerging role of transcription factor-microRNA-target gene feed-forward loops in cancer

Transcriptional regulatory networks are biological network motifs that act in accordance with each other to play decisive roles in the pathological processes of cancer. One of the most common types, the feed-forward loop (FFL), has recently attracted interest. Three connected deregulated nodes, a transcription factor (TF), its downstream microRNA (miRNA) and their shared target gene can make up a class of cancer-involved FFLs as ≥1 of the 3 can act individually as a bona fide oncogene or a tumor suppressor. Numerous notable elements, such as p53, miR-17-92 cluster and cyclins, are proven members of their respective FFLs. Databases of interaction prediction, verification of experimental methods and confirmation of loops have been continually emerging during recent years. Development of TF-miRNA-target loops may help understand the mechanism of tumorgenesis at a higher level and explain the discovery and screening of the therapeutic target for drug exploitation.

Optimally choosing PWM motif databases and sequence scanning approaches based on ChIP-seq data

BACKGROUND

For many years now, binding preferences of Transcription Factors have been described by so called motifs, usually mathematically defined by position weight matrices or similar models, for the purpose of predicting potential binding sites. However, despite the availability of thousands of motif models in public and commercial databases, a researcher who wants to use them is left with many competing methods of identifying potential binding sites in a genome of interest and there is little published information regarding the optimality of different choices. Thanks to the availability of large number of different motif models as well as a number of experimental datasets describing actual binding of TFs in hundreds of TF-ChIP-seq pairs, we set out to perform a comprehensive analysis of this matter.

RESULTS

We focus on the task of identifying potential transcription factor binding sites in the human genome. Firstly, we provide a comprehensive comparison of the coverage and quality of models available in different databases, showing that the public databases have comparable TFs coverage and better motif performance than commercial databases. Secondly, we compare different motif scanners showing that, regardless of the database used, the tools developed by the scientific community outperform the commercial tools. Thirdly, we calculate for each motif a detection threshold optimizing the accuracy of prediction. Finally, we provide an in-depth comparison of different methods of choosing thresholds for all motifs a priori. Surprisingly, we show that selecting a common false-positive rate gives results that are the least biased by the information content of the motif and therefore most uniformly accurate.

CONCLUSION

We provide a guide for researchers working with transcription factor motifs. It is supplemented with detailed results of the analysis and the benchmark datasets at http://bioputer.mimuw.edu.pl/papers/motifs/ .