Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2013 Sep 12.
Published in final edited form as: Genomics. 2012 Jul 21;100(5):297–302. doi: 10.1016/j.ygeno.2012.07.009

Chromatin state and microRNA determine different gene expression dynamics responsive to TNF stimulation

Ruijuan Li a, Weilong Guo a, Jin Gu a, Michael Q Zhang a,b, Xiaowo Wang a,*
PMCID: PMC3771509  NIHMSID: NIHMS507365  PMID: 22824656

Abstract

Gene expression is a dynamic process, and what factors influence gene expression changes upon external stimulus have not been clearly understood. We studied gene expression profiles in human umbilical vein endothelial cells (HUVEC) after the Tumor Necrosis Factor (TNF) stimulus, and found that: the promoters of fast-response up-regulated genes were enriched with several “active” chromatin markers like H3K27ac and H3K4me3, and also preferentially bound by Pol II and c-Myc; the core-promoter regions of slow-response up-regulated genes were frequently occupied by nucleosomes; down-regulated genes were more intensively regulated by microRNAs. Moreover, the Gene Ontology and motif analysis of the promoter regions revealed that gene clusters with different response behaviors had different functions and were regulated by different sets of transcription factors. Our observations suggested that the different gene expression patterns upon external stimulus were regulated by a combination of multi-layer regulators.

Keywords: TNF, Gene expression profiles, Chromatin, Histone code, MicroRNAs

1. Introduction

Gene expression is a complex process and influenced by both cell internal status and external environments. Previous studies have found varied gene expression patterns induced by external stimulus [1–4]. However, how the transcriptional and posttranscriptional regulators, such as chromatin states, transcription factors and microRNAs, are involved in the selective regulation of induced genes after external stimulus is still not fully understood.

The chromatin state, which is marked largely by different histone modifications, is recently found to play key roles in regulating gene transcription [5]. Previous studies used histone modification patterns to locate the positions of gene promoters [6] and predict gene transcription levels successfully [7]. Certain chromatin states have been demonstrated to form binding surfaces for other regulators (e.g. transcription factors) to activate or repress transcription [5]. It has also been reported that the chromatin state may influence gene expression dynamics. For example, some of the rapidly inducible genes were reported to associate with high levels of H3K4me3 modification at their promoter regions [8], and bivalent genes whose promoters were marked by both active (H3K4me3) and repressive (H3K27me3) histone makers were found to change their expression levels rapidly during differentiation [9]. The mRNA expression level is determined by both the transcription and degradation rate of the transcripts. MicroRNAs are another important class of regulators, which could bind to their mRNA targets by sequence complementary, leading to translational repression and/or accelerated degradation [10]. It has been shown that microRNA could function to ensure the preciseness and fidelity of dynamic and spatially restricted gene expression [11], and some microRNAs were reported to regulate the expression of TNF-induced genes [12–14]. However, it is not clear how chromatin state and microRNA regulation could affect the gene expression dynamics upon external stimulus.

Tumor necrosis factor (TNF) is an important external stimulus involving in many biological processes, such as cell differentiation, proliferation, inflammation, immune responses, and tissue development. The effects of TNF are mainly mediated by a cluster of transcription factors (TFs), including NF-κB, AP-1 and interferon-regulatory factors, among which NF-κB is a key factor [2,3,15,16]. Previous studies have found that the genes induced by TNF showed varied kinetics, and mRNA stability was found to contribute to the temporal order of gene response [2,3].

In this study, we set out to explore the mechanisms regulating the kinetics of gene expression after external stimulus. This work started from analyzing the relationships between chromatin states and responding speeds of the TNF induced genes. By coupling histone modification profiles and time serial microarray data in HUVEC, we showed that the histone modification state of gene promoter regions was significantly correlated with the response speeds of induced gene expression changes. Though having similar average expression values before stimulus, the up-regulated genes with different kinetics showed different histone modification patterns and transcription factors binding preferences. And the core-promoter regions of fast-responding up-regulated genes were nucleosome depleted and showed high levels of Pol II binding. While, the down-regulated genes showed less pronounced histone modification profile difference, but were more intensively regulated by microRNAs. Moreover, the Gene Ontology (GO) and motif analysis displayed different characteristics among gene clusters with different kinetic properties. Our work indicated that the different expression kinetics of induced genes could be influenced by a combination of histone modifications, transcription factors and microRNAs, and different mechanisms were involved in regulating the up- and down-regulated genes.

2. Materials and methods

2.1. Gene expression data

The time-serial gene expression microarray data after TNF stimulation in HUVEC were downloaded from the NCBI GEO database (GSE9055). The data had 25 time points, from 0 to 8 h after the stimulus, with 0.25 h interval before 4 h, and 0.5 h interval after 4 h. The raw data was normalized, log2-transformed and processed by the dChip software [17] (version 2010_01). Genes with standard deviation more than 0.5 over the time course and expression level no less than 4.00 in at least 50% samples were kept for further analysis. The genes were considered as up-regulated if log (average expression level)−log (original expression level)>0.4, and were considered as down-regulated if log (average expression level)−log (start expression level)<−0.4. Then, the up-regulated and down-regulated genes were further subdivided into three sub-clusters based on the k-means clustering. The proper category numbers for the clustering were decided by the elbow point of the curve of criterion function calculated by the k-means algorithm [18] (Supplementary Fig. S1). Finally, 66, 133, and 208 genes fell into three up-regulated gene clusters: up1 (fast), up2 (mid), up3 (slow); and 33, 111, and 162 genes fell into three down-regulated gene clusters: down1 (fast), down2 (mid), down3 (slow) respectively.

2.2. Sequencing data

We downloaded the 14 kinds of deep sequencing data of HUVEC from ENCODE [19,20]. The data included nine ChIP-seq data for histone modifications: H3K4me1, H3K4me2, H3K4me3 and H3K9ac, H3K9me1, H3K27ac, H3K27me3, H3K36me3 and H4K20me1; three ChIP-seq data sets for RNA Polymerase II (Pol II), c-Myc (MYC) and CTCF; two types of data for identifying open chromatin regions: DNaseI HS (DNaseI hypersensitivity signifies chromatin accessibility following binding of trans-acting factors in place of a canonical nucleosome) and FAIRE (formaldehyde Assisted Isolation of Regulatory Elements identifies nucleosome-depleted regions of the genome). For each sequencing data, we calculated the read numbers along the regions [−1500 bp, 1500 bp] relative to the transcription start site (TSS) according to the RefSeq annotation for hg18 [20], and normalized by the total read numbers in each library.

2.3. Regression analysis

The linear regression analysis was done between the gene reaction rate after TNF stimulus (Y) and the chromatin state (X) for the 407 up-regulated genes. All the 14 kinds of sequencing data were used. Suppose there was a linear relationship between Y and Xi, we constructed the model for them (Y=(Et−E0)/t, t was the time when the expression level reached half of the maximum/minimum for the up/down regulated genes, Et was the expression value at time t, E0 was the expression value at time 0, before TNF stimulus; Xi (i=1, 2,…, 14) was the normalized reads count for the ith sequencing data set in the regions [−1500 bp, 1500 bp] relative to the TSSs).

2.4. Motif and gene ontology analysis

The human genome sequences and RefSeq gene annotations (hg18, NCBI build 36) were downloaded from UCSC Genome Browser [20]. The known Position Weight Matrices for TF binding sites were downloaded from the TRANSFAC [21]. The motif analysis was performed with the motifclass program in CREAD package [22]. For each gene, we collected its promoter sequences from the region [−800 bp, 200 bp] relative to the TSS. The promoter sequences of each cluster of genes were set as the foreground respectively, and these randomly chosen from non-induced genes were set as the background.

GO and KEGG pathway enrichment analysis was performed with DAVID (version 6.7) [23,24]. A threshold of false discovery rate<0.05 and Benjamini–Hochberg corrected p-value <0.01 was used to find the enriched terms.

2.5. MicroRNA targeting analysis

The microRNA target predictions were downloaded from TargetScan (version 5.2) [25], and only the highly confident and conserved ones were used in our analysis. For each microRNA, we counted the number of its targets in a gene sub-cluster, and compared it with the number of its targets among all RefSeq genes. A microRNA was called to be over-represented in a gene cluster only when the Chi-square test p-value <1e−4 after Bonferroni multi-test correction.

3. Results

3.1. TNF induced gene-expression patterns in HUVEC

The time-serial microarray data of gene expression after TNF stimulus in HUVEC were filtered and clustered. The results showed that 407 genes were up-regulated and 307 were down-regulated after the stimulation (Supplementary Table S1).

To further investigate the mechanisms contributing to the dynamic patterns, we categorized each of the up- and down-regulated genes into three sub-clusters respectively, based on k-means clustering (66, 133 and 208 genes for the three up-regulated gene sub-clusters; 33, 111 and 162 genes for the three down-regulated gene sub-clusters). Each sub-cluster of genes showed a clear difference in the responding speeds (fast, middle and slow). The expression level of the fast responding clusters rose or fell rapidly after the stimulus (reached the half of their final static value within 1 h), the expression level of the middle clusters took about 3 h to reach the half of their static value, and the expression level of the slow responding clusters kept on increasing or descending during the whole 8 h (Fig. 1).

Fig. 1.

Fig. 1

Open in a new tab

Clustering of TNF-induced genes. The TNF induced genes were classified into two main clusters depending on whether they were up or down regulated. In each cluster, the genes were further subdivided into three sub-clusters based on k-means clustering. (a) The average expression pattern of the three down-regulated gene sub-clusters (down1: fast response down-regulated gene cluster, contained 33 genes; down2: middle response down-regulated gene cluster, contained 111 genes; down3: slow response down-regulated gene cluster, contained 162 genes). (b) The average expression pattern of the three up-regulated gene sub-clusters (up1: fast response up-regulated gene cluster, contained 66 genes; up2: middle response up-regulated gene cluster, contained 133 genes; up3: slow response up-regulated gene cluster, contained 208 genes).

3.2. Chromatin states were different for genes with different response speeds

We studied the relationships between the chromatin state of the gene promoter regions before TNF stimulus and the response speeds of the corresponding genes after stimulus. For the up-regulated genes, significant differences were observed between the three sub-clusters. Five (H3K27ac, H3K4me1, H3K4me2, H3K4me3 and H3K9ac) histone modifications showed significant different patterns. And the faster the response, the higher these histone modification levels (Fig. 2 and Supplementary Fig. S2). There were two distinct peaks located upstream and downstream of the genes’ transcriptional start sites (TSSs) (where the trough presumably reflected the nucleosome depletion at the core-promoter) in the profiles of up1 and up2 sub-clusters, and the main difference between the up1 and up2 sub-clusters lied in the enrichment levels of histone modifications. But for the genes in up3 cluster (slow cluster), the profile only showed one peak around TSS, which was quite different from the other two sub-clusters. While, no significant histone modification level difference (t-test p-value>0.1 for any pairwise comparison) was observed among the down-regulated clusters (Supplementary Fig. S3, Table S2).

Fig. 2.

Fig. 2

Open in a new tab

Specific histone modifications of the three up-regulated sub-clusters. (a) H3K27ac, H3K4me3 and H3K9ac profiles across the promoter regions of the genes in up-regulated sub-clusters. The x-axis was the relative distance to the TSS. The y-axis was the normalized average reads number. (b) The y-axis was the log2 reads count at [−1500 bp, 1500 bp] region relative to TSS. The p-value was calculated by t-test.

Previous work reported that gene expression levels were related to the histone modification levels in their promoter regions [6]. Then we asked whether the different histone modification levels between the three up-regulated clusters were correlated with the different expression levels before TNF simulation. We checked the average expression levels of genes in the sub-clusters before stimulus. Our results showed no significant difference among the three up-regulated sub-clusters (t-test p-value>0.1 for any two clusters of the three. Supplementary Fig. S4). Therefore, those histone modification diversities were more likely to be correlated with the response speeds of the genes rather than their original expression levels.

Previous work demonstrated that histone modifications could affect the recruit of transcription factors [5], and the histone modification marker H3K4me3 was found to be enriched at the promoters of the rapidly inducible genes [8]. Therefore, higher levels of some histone modifications (H3K27ac, H3K4me1, H3K4me2, H3K4me3 and H3K9ac in our study) might ease the binding of the transcription factors that mediated TNF stimulus, and resulted in faster reaction rate of their target genes. While, the single peak around the TSSs of genes in up3 cluster seemed to be related to the nucleosome positioning at core-promoter regions, which might inhibit the binding of transcription factors and RNA polymerase [26–28].

Next, we analyzed the patterns of Pol II binding, CTCF insulator binding, c-Myc binding and chromatin accessibility (measured by DNaseI hypersensitivity assay and FAIRE assay) [19] around the TSSs of the TNF induced genes (Fig. 3 and Supplementary Figs. S2, S3). The signals of these data were all strong for the up1 cluster, weak for the up2 cluster, and weakest for the up3 cluster. The Pol II signal on average was much higher for the up1 cluster genes compared with the up2 and up3, which might be able to explain the fast responding speeds of these genes. The Pol II signal was higher in down1 than the down2 and down3, while the overall average level of down-regulated clusters was higher than up-regulated clusters. Interestingly, we also found higher CTCF and c-Myc binding signals around the promoters of the up1 and up2 clusters comparing to the up3 cluster. Pol II is reported to be stalled at the CTCF binding sites [16], and c-Myc could regulate transcriptional pause release [29]. Considering the similar expression levels of the genes in three up-regulated clusters before TNF stimulus, the CTCF binding around the TSS might cause the Pol II to be paused or poised on the promoters of up1 gene cluster and c-Myc might be prepared to release the paused Pol II after the stimulation. On the contrary, down-regulated genes showed lower CTCF but higher c-Myc signal at their promoters than the up-regulated ones, which seemed to be correlated with their higher original expression levels. Moreover, the DNaseI HS (Figs. 3e–f) and FAIRE (Supplementary Figs. S2, S3) data proved that the core-promoter regions around TSSs for the up1, up2 and all down-regulated genes were nucleosome depleted, while the chromatin regions around TSS for the up3 genes were less accessible. This observation supported the notion that the up3 gene promoters were more frequently occupied by nucleosome that might lead to their slow response after TNF stimulus.

Fig. 3.

Fig. 3

Open in a new tab

The Pol II, CTCF, DNaseI and c-Myc patterns in the promoter regions of up- and down-regulated genes. The x-axis was the promoter region at [−1500 bp, 1500 bp] around the TSS. The y-axis was the normalized average reads count. (a, c, e, g) Showed the four kinds of signals for three up-regulated gene sub-clusters; (b, d, f, h) revealed the different patterns for the three down-regulated gene sub-clusters. (a, b) Pol II binding signal for up1 cluster was much stronger than up2 and up3 clusters. CTCF was found to be preferentially bound to the TSSs of up1 and up2 genes but not up3 and down-regulated genes by (c) and (d). The promoter regions of up1 and up2 cluster of genes were more accessible than up3 cluster measured by (e) DNaseI and FAIRE (Supplementary Fig. 1–2) assay. The promoter regions of down-regulated genes were all nucleosome-depleted compared to the up-regulated ones by (e) and (f).

To test whether the chromatin state before TNF stimulus could predict the response speeds of the induced genes, we build a linear regression model on the 407 up-regulated genes to predict the response speeds, by using the profile intensities of histone modifications, Pol II, CTCF, c-Myc, DNaseI and FAIRE at the promoter regions as input (See Materials and methods). This linear model resulted in an R-square of 0.197 and F-statistic p-value of 5.251e-11 for the up-regulated genes. We randomly shuffled the reaction rates of 407 up-regulated genes for 1000 times, and got R-square no more than 0.112 and p-values no smaller than 1.231e–05 (Supplementary Fig. S5). This result suggested that the chromatin state significantly contributed to the response speeds of TNF induced up-regulated genes. But for the down regulated genes, no significant correlation was observed (R-square 0.0758 and p-value 0.137).

3.3. Motif and gene ontology analysis

Next, we checked whether the genes with similar response behavior in each cluster might be regulated by a set of similar transcription factors. We collected the known TF binding consensus sequences from TRANSFAC [21] and tried to find the enriched motifs for the promoter regions of each cluster of genes using the CREAD package [22].

Several significant motifs were sought out for the three up-regulated genes. For the up1 cluster, NF-κB and REL were found to be the most significant transcription factors, both of which were known as important factors mediating the effects of TNF [3,30]. For the up2 cluster, NF-κB binding sites were found to be significantly enriched, but the enrichment was less pronounced than in up1 cluster (The p-value of NF-κB for the up1 cluster was 0.007, and it was 0.07 for the up2 cluster). For the up3 cluster, no significant motif was found (Supplementary Table S3). This result implied that the fast response genes were activated by transcription factors which directly responded to the TNF signal, while slow response genes were controlled by a more diverse set of regulators. While for the down-regulated gene, no significant motif was found in any of the three clusters.

Previous work has found that some fast responding genes which encode regulatory proteins could regulate genes responding slower [31,32]. We did a GO analysis for each cluster of genes using DAVID [23]. The GO terms related to response to external stimulus, inflammatory response, chemokine activity, receptor and protein binding were found to be significantly enriched in the up1 cluster (Supplementary Table S4). For up2 cluster, the terms of regulation of immune system process, the regulation of immune response and regulation of effecter process were enriched, while immune response was enriched for the up3 cluster. For the down regulated genes, less enriched terms were found compared with the up regulated genes. Organ development and anatomical structure morphogenesis were the only two enriched terms for the down1 cluster. For down2 cluster, the terms of regulation of transcription, the regulation of RNA and nucleic acid metabolic process, and transcription regulator and transcription factor activity, DNA binding were significantly enriched, while no term was enriched for the down3 cluster. These observations suggested that the functions of response genes were correlated with the time-serial regulation of gene expression after TNF stimulus: fast-responding genes were directly responded to TNF stimulus, while the genes of up2 cluster contained more regulators, and the slow-responding clusters were more enriched in effectors with diverse functions.

3.4. microRNA contributes to gene repression

Finally, we checked whether microRNAs could contribute to the different dynamic patterns of TNF induced genes. We obtained the microRNA target predictions from TargetScan [25]. For each microRNA, we counted the number of its target genes in each sub-cluster. The microRNA was called to be over-represented if its targets were significantly overrepresented in that cluster of genes comparing with the portion of its targets among all RefSeq genes. There were 2, 46 and 45 microRNAs found to be over-represented for down1, down2 and down3 clusters, respectively (p-value <1e−4, Supplementary Table S5). To check the significance of these numbers, we randomly chose the same number of genes for each cluster from all the RefSeq genes for 1000 times, and recorded the numbers of over-represented microRNAs each time. Except for down1 cluster, there were significantly more microRNAs regulating down2 and down3 clusters (p-value <0.0001) (Fig. 4). We also calculated the enriched microRNAs for the up-regulated genes, only 1, 7 and 0 over-represented microRNAs were found for up1, up2 and up3 clusters, which were not significant at all (Supplementary Fig. S6). Among the enriched microRNAs we identified, miR-27b was reported to post-transcriptionally regulate the expression of adenosine 2B receptor induced by TNF in colonic epithelial cells [12]. And miR-25 could protect cells against TNF-related apoptosis-inducing ligand (TRAIL)-induced apoptosis [13]. These observations suggested that microRNA regulation might be an important factor that mediated the repression of the down-regulated genes after TNF stimuli.

Fig. 4.

Fig. 4

Open in a new tab

Over-represented microRNAs for the down-regulated genes. Histogram of the numbers of over-represented microRNAs for 1000 times random sampling. For each down-regulated cluster, we randomly chose the same number of RefSeq genes, and then calculated the number of the over-represented microRNAs for the random clusters. The x-axis indicates the number of over-represented microRNAs in the random sampling experiments. The label showed the number of enriched microRNAs found in each cluster and the p-value of the random sampling test.

4. Discussion

Gene expression is a complicated dynamic process, which is regulated by a combination of many factors. The expression of a wide range of genes could be changed after the external stimulus. In this study, we coupled time-serial gene expression microarray data and high-resolution histone modification/TF binding data to investigate how the original state of the cell could influence its response to an external stimulus. The TNF-induced genes in HUVEC were classified into two major clusters, up-regulated genes and down-regulated genes, and each major cluster was further divided into three sub-clusters according to their reaction speeds.

We observed that up-regulated genes showed distinct chromatin states in their promoter regions. The promoters of fast-responding up-regulated genes were enriched in several active histone modification markers, such as H3K27ac, H3K4me1, H3K4me2, H3K4me3 and H3K9ac. This indicated that higher levels of these histone modifications might facilitate the recruitment of necessary transcription factors and proteins, and thereby enabled the fast response. The fast responding genes also showed higher Pol II, CTCF and c-Myc binding signals, suggesting that Pol II might be paused/poised by the CTCF at promoters of these genes and high levels of c-Myc binding could promote transcriptional pause release after TNF simulation. In the contrary, the slow-responding up-regulated genes lacked these makers at their promoter regions, and their core-promoters were more frequently occupied by nucleosomes. Our analysis revealed that different transcription factor binding motifs were enriched at the promoters of up-regulated gene clusters. The fast-responding genes were more frequently regulated by the known TNF related TFs, such as NF-κB and c-Rel, and more NF-κB binding sites with high affinity were found in the up1 cluster promoters than the other clusters. The absence of enriched motif for up3 cluster indicated that these genes might be regulated by a more diverse or secondary set of TFs. And GO analysis on the up-regulated clusters demonstrated that gene clusters with different reaction speeds were coincident with different functions. The fast responding genes were more directly connected with the known TNF responding pathways. Genes in up1 cluster, such as Cxcl1 and Cxcl2, encoded neutrophil chemoattractants which could initiate the response to infection [1,2]. And it has been reported that the transcription factor Irf1 in up1 cluster could activate downstream pathways involving genes like Cxcl10, Cxcl11 and Ccl5 in the slower response sub-clusters [1,2].

For the down-regulated genes, the chromatin states at their promoter regions did not show significant correlation with their response speeds, and no significant consensus TF binding motif could be identified. However, we found that the down-regulated genes were intensively regulated by microRNAs, which suggested that post-transcriptional regulation instead of transcriptional regulation might contribute more to their down-regulation.

Several previous reports have studied gene-activation kinetics induced by various stimuli, including growth factors [1], LPS [4], and TNF [2,3], etc. Those studies focused mainly on the different kinetic patterns of the induced genes, and the down-regulated genes were largely ignored. The factors regulating the temporal order of gene expression after TNF stimulus were still unclear. Our work classified the induced genes by the unsupervised clustering method, and tried to find the different regulatory mechanisms for the down- and up-regulated genes. Our analysis suggested that the dynamic property of up-regulated genes was correlated with transcriptional regulation factors, while for the down-regulated genes, they might be influenced by the microRNA mediated post-transcriptional regulation. These results implied that genes’ response speeds were regulated by multi-layer factors, and different mechanisms were involved in regulating the up- and down-responding genes.

Our observations suggested that chromatin state and microRNA regulation could significantly contribute to the different responding speeds of TNF induced genes. However, the accumulation of transcripts is related not only to transcription and degradation rates at the original time but also to the changes over time. We expect that by further experimental tests of our results and deeper integrating more time-serial microRNA expression data and histone modification profiles under external stimulus, the regulatory mechanism would be better understood.

Supplementary Material

Supp data and figures

Acknowledgments

This work is supported by the NBRPC grant 2012CB316503, NSFC grant (60905013, 60934004, 91019016, and 31061160497), and Doctoral Fund of Ministry of Education of China (No. 20090002120023). MQZ is also supported by NIH grant HG001696. We thank the ENCODE project consortium for the sequencing data.

Appendix A. Supplementary data

Supplementary data to this article can be found online at http://dx.doi.org/10.1016/j.ygeno.2012.07.009.

References

  • 1.Tullai JW, et al. Immediate-early and delayed primary response genes are distinct in function and genomic architecture. J Biol Chem. 2007;282:23981–23995. doi: 10.1074/jbc.M702044200. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2.Hao S, Baltimore D. The stability of mRNA influences the temporal order of the induction of genes encoding inflammatory molecules. Nat Immunol. 2009;10:281–288. doi: 10.1038/ni.1699. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.Yarilina A, Park-Min KH, Antoniv T, Hu X, Ivashkiv LB. TNF activates an IRF1-dependent autocrine loop leading to sustained expression of chemokines and STAT1-dependent type I interferon-response genes. Nat Immunol. 2008;9:378–387. doi: 10.1038/ni1576. [DOI] [PubMed] [Google Scholar]
  • 4.Doyle S, et al. IRF3 mediates a TLR3/TLR4-specific antiviral gene program. Immunity. 2002;17:251–263. doi: 10.1016/s1074-7613(02)00390-4. [DOI] [PubMed] [Google Scholar]
  • 5.Kouzarides T. Chromatin modifications and their function. Cell. 2007;128:693–705. doi: 10.1016/j.cell.2007.02.005. [DOI] [PubMed] [Google Scholar]
  • 6.Wang X, Xuan Z, Zhao X, Li Y, Zhang MQ. High-resolution human core-promoter prediction with CoreBoost_HM. Genome Res. 2009;19:266–275. doi: 10.1101/gr.081638.108. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Karlic R, Chung HR, Lasserre J, Vlahovicek K, Vingron M. Histone modification levels are predictive for gene expression. Proc Natl Acad Sci U S A. 2010;107:2926–2931. doi: 10.1073/pnas.0909344107. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Roh TY, Cuddapah S, Cui K, Zhao K. The genomic landscape of histone modifications in human T cells. Proc Natl Acad Sci U S A. 2006;103:15782–15787. doi: 10.1073/pnas.0607617103. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Pan G, et al. Whole-genome analysis of histone H3 lysine 4 and lysine 27 methylation in human embryonic stem cells. Cell Stem Cell. 2007;1:299–312. doi: 10.1016/j.stem.2007.08.003. [DOI] [PubMed] [Google Scholar]
  • 10.Bartel DP. MicroRNAs: target recognition and regulatory functions. Cell. 2009;136:215–233. doi: 10.1016/j.cell.2009.01.002. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Zhao Y, Srivastava D. A developmental view of microRNA function. Trends Biochem Sci. 2007;32:189–197. doi: 10.1016/j.tibs.2007.02.006. [DOI] [PubMed] [Google Scholar]
  • 12.Kolachala VL, et al. Adenosine 2B receptor expression is post-transcriptionally regulated by microRNA. J Biol Chem. 2010;285:18184–18190. doi: 10.1074/jbc.M109.066555. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Razumilava N, et al. miR-25 targets TNF-related apoptosis inducing ligand (TRAIL) death receptor-4 and promotes apoptosis resistance in cholangiocarcinoma. Hepatology. 2012;55:465–475. doi: 10.1002/hep.24698. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Suarez Y, Wang C, Manes TD, Pober JS. Cutting edge: TNF-induced microRNAs regulate TNF-induced expression of E-selectin and intercellular adhesion molecule-1 on human endothelial cells: feedback control of inflammation. J Immunol. 2010;184:21–25. doi: 10.4049/jimmunol.0902369. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Hoffmann A, Leung TH, Baltimore D. Genetic analysis of NF-kappaB/Rel transcription factors defines functional specificities. EMBO J. 2003;22:5530–5539. doi: 10.1093/emboj/cdg534. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.Wada Y, et al. A wave of nascent transcription on activated human genes. Proc Natl Acad Sci U S A. 2009;106:18357–18361. doi: 10.1073/pnas.0902573106. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Li C, Wong WH. Model-based analysis of oligonucleotide arrays: expression index computation and outlier detection. Proc Natl Acad Sci U S A. 2001;98:31–36. doi: 10.1073/pnas.011404098. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.Bian Z, Zhang X. Pattern Recognition. 2. TsingHua University Press; Beijing: 1999. [Google Scholar]
  • 19.Rosenbloom KR, et al. ENCODE whole-genome data in the UCSC Genome Browser. Nucleic Acids Res. 2010;38:D620–D625. doi: 10.1093/nar/gkp961. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Fujita PA, et al. The UCSC Genome Browser database: update 2011. Nucleic Acids Res. 2011;39:D876–D882. doi: 10.1093/nar/gkq963. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.Wingender E, Dietze P, Karas H, Knuppel R. TRANSFAC: a database on transcription factors and their DNA binding sites. Nucleic Acids Res. 1996;24:238–241. doi: 10.1093/nar/24.1.238. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Schones DE, Sumazin P, Zhang MQ. Similarity of position frequency matrices for transcription factor binding sites. Bioinformatics. 2005;21:307–313. doi: 10.1093/bioinformatics/bth480. [DOI] [PubMed] [Google Scholar]
  • 23.Huang da W, Sherman BT, Lempicki RA. Systematic and integrative analysis of large gene lists using DAVID bioinformatics resources. Nat Protoc. 2009;4:44–57. doi: 10.1038/nprot.2008.211. [DOI] [PubMed] [Google Scholar]
  • 24.Huang da W, Sherman BT, Lempicki RA. Bioinformatics enrichment tools: paths toward the comprehensive functional analysis of large gene lists. Nucleic Acids Res. 2009;37:1–13. doi: 10.1093/nar/gkn923. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25.Lewis BP, Burge CB, Bartel DP. Conserved seed pairing, often flanked by adenosines, indicates that thousands of human genes are microRNA targets. Cell. 2005;120:15–20. doi: 10.1016/j.cell.2004.12.035. [DOI] [PubMed] [Google Scholar]
  • 26.Bai L, Morozov AV. Gene regulation by nucleosome positioning. Trends Genet. 2010;26:476–483. doi: 10.1016/j.tig.2010.08.003. [DOI] [PubMed] [Google Scholar]
  • 27.Segal E, et al. A genomic code for nucleosome positioning. Nature. 2006;442:772–778. doi: 10.1038/nature04979. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Ramirez-Carrozzi VR, et al. Selective and antagonistic functions of SWI/SNF and Mi-2beta nucleosome remodeling complexes during an inflammatory response. Genes Dev. 2006;20:282–296. doi: 10.1101/gad.1383206. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29.Rahl PB, et al. c-Myc regulates transcriptional pause release. Cell. 2010;141:432–445. doi: 10.1016/j.cell.2010.03.030. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.Tian B, Nowak DE, Jamaluddin M, Wang S, Brasier AR. Identification of direct genomic targets downstream of the nuclear factor-kappaB transcription factor mediating tumor necrosis factor signaling. J Biol Chem. 2005;280:17435–17448. doi: 10.1074/jbc.M500437200. [DOI] [PubMed] [Google Scholar]
  • 31.Altin JG, et al. Differential induction of primary-response (TIS) genes in PC12 pheochromocytoma cells and the unresponsive variant PC12nnr5. J Biol Chem. 1991;266:5401–5406. [PubMed] [Google Scholar]
  • 32.Winkles JA. Serum- and polypeptide growth factor-inducible gene expression in mouse fibroblasts. Prog Nucleic Acid Res Mol Biol. 1998;58:41–78. doi: 10.1016/s0079-6603(08)60033-1. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supp data and figures
NIHMS507365-supplement-Supp_data_and_figures.docx (198.7KB, docx)

RESOURCES