Fig 1.
Inducible transcription and RELA binding in activated B lymphoma cells.
BJAB human B lymphoma cells were treated with P+I for 1 or 4 h, followed by analyses of inducible transcription and RELA binding by RNA-Seq and ChIP-Seq, respectively. (A) Gene expression changes at either 1 or 4 h post activation compared to unactivated cells. Two biological replicate experiments were carried out, and differential gene expression was analyzed using EBSeq. (B) Inducibly up- or down-regulated transcripts were partitioned into 6 patterns (middle panel, top, or bottom, respectively) by k-means clustering (see also S1B–S1E Fig). Each graph represents the gene expression pattern at 0, 1, and 4 h. Numbers of genes in each pattern are indicated below in black. The letter “A” after a pattern number indicates genes that are activated by P+I, and the letter “R” indicates genes that are repressed by P+I. Red numbers identify putative RELA target genes according to criteria discussed in the text and summarized in parts D and E below. (C) Representative examples of RNA-Seq profile of genes corresponding to patterns 1A and 5A of up-regulated genes (right panel, top, and middle) and from pattern 1R of down-regulated genes (right panel, bottom). The y axis represents normalized reads per million. Chromosomal locations of genes in hg19 are shown above RNA-Seq tracks. (D) Two replicate RELA ChIP-Seq experiments were carried out as described in the Methods section. Peaks were called using MACS2 with input DNA as control at FDR ≤ 0.05. Peaks with peak score ≥ 100 that were common to both biological replicates were used for further analysis. ChIP-Seq experiments were annotated with respect to genomic location using HOMER. Total numbers of RELA peaks are noted above the bars, and activation times are indicated below. (E) Relationship of inducible transcription and RELA binding. Genes were categorized based on the level of inducible activation (top 2 rows) or repression (bottom 2 rows). Genes in each category (second column) that bound RELA (third column) were assigned based on criteria noted in the text. The majority of RELA binding was induced upon cell activation (fourth column). Newly identified RELA-bound up- and down-regulated genes are listed in S1 Table. (F) Representative browser tracks of RELA binding to genes whose transcriptional responses are shown in part C. The y axis represents normalized reads per 10 million. Chromosomal locations of each gene in hg19 are shown above ChIP-Seq tracks. RNA-Seq and ChIP-Seq data are available on the GEO website (http://www.ncbi.nlm.nih.gov/geo/) (Accession number GSE117259). ChIP-Seq, chromatin immunoprecipitation and sequencing; FDR, false discovery rate; GEO, Gene Expression Omnibus; P+I, phorbol 12-myristate 13-acetate and ionomycin; RNA-Seq, RNA sequencing; TSS, transcription start site; TTS, transcription termination site; UTR, untranslated region.
Fig 2.
Attenuation of classical NF-κB-dependent transcription by dnIκBα expression.
Two clones of BJAB cells were generated to perturb canonical NF-κB signaling by Tet-inducible dnIκBα expression. (A) Western blot analysis of parent BJAB cells and 1 clone treated with Tet for the indicated times (h); whole-cell extracts were fractionated by SDS-PAGE and transferred to NC membranes, which were probed with anti-IκBα or anti-β-actin (loading control) antibodies. (B) RELA induction in dnIκBα-expressing BJAB cells. Clone 1 cells were activated with P+I for 1 or 4 h either in the absence of Tet (-Tet) or after Tet pretreatment for 24 h (+Tet). Nuclear extracts were fractionated with SDS-PAGE, transferred to membranes, and probed with anti-RELA or anti-hnRNP antibodies (loading control). Results with 1 representative clone are shown. (C) Relationship of dnIκBα-responsive transcripts to RELA binding. Total RNA isolated from each BJAB clone activated in the presence or absence of Tet was subject to RNA-Seq (2 biological replicates from each clone). Genes that were differentially expressed in each clone because of dnIκBα expression were identified based on FDR ≤ 0.05 in EBSeq. Inducible activation of 806 genes was reduced by dnIκBα expression in both clones; of these, 304 bound RELA (middle panel, green), and 502 (middle panel, gray) did not bind RELA. Time-dependent expression patterns of RELA-binding genes in the absence (blue lines) or presence (green lines) of Tet-induced dnIκBα were determined by k-means clustering (right). The letters “Ad” after the pattern number indicate direct target genes that were activated by RELA, and the letters “Ai” indicate indirect target genes that were RELA dependent. Numbers of genes in each category are shown in black below the graphs. Numbers in red identify genes that were changed more than 2-fold in the absence of Tet. Time-dependent expression patterns of genes that were down-regulated by IκBα but did not bind RELA are shown to the left. Numbers and color coding are as described above. (D) Representative RNA-Seq and ChIP-Seq tracks for RELA target genes identified in C (green). Two previously known NF-κB target genes (TNFAIP3 and STAT5) with different expression kinetics are shown on the top line, and 2 newly identified RELA target genes (HERPUD1 and NR1D1) are shown on the bottom line. RNA-Seq tracks show the effects of dnIκBα expression (+Tet) at the time of maximum expression (1 h for pattern 3Ad and 4 h for pattern 2Ad). RNA-Seq tracks for all time points +/− Tet are provided in S2G Fig. All time points are shown for RELA ChIP-Seq tracks to visualize the dynamics of RELA recruitment and loss from the genome over the experimental time course. Numbers above the tracks refer to gene location in hg19. A complete list of all 304 RELA target genes identified in this study is provided in S2 Table. Genome scale datasets are available on the GEO website (http://www.ncbi.nlm.nih.gov/geo/) (Accession number GSE117259). ChIP-Seq, chromatin immunoprecipitation and sequencing; dnIκBα, dominant negative mutated IκBα; FDR, false discovery rate; GEO, Gene Expression Omnibus; hnRNP, heterogeneous nuclear ribonucleoprotein; IκBα, NFKB inhibitor alpha; NC, nitrocellulose; NF-κB, nuclear factor kappa B; P+I, phorbol 12-myristate 13-acetate and ionomycin; RNA-Seq, RNA sequencing; Tet, tetracycline.
Fig 3.
Modes of transcriptional activation by RELA.
(A) Selected RELA target genes were assayed for effects of ERK inhibition on inducible transcription. qRT-PCR was carried out using RNA extracted from BJAB cells activated with P+I for indicated times in the absence (blue lines) or presence (red lines) of PD0325901. The top line shows representative genes whose maximal activation occurs between 1 and 4 h, during the period when nuclear RELA levels diminish. The bottom line shows representative genes that are transiently induced by P+I; data shown are the average of 2 independent experiments with qRT-PCR carried out in duplicate; error bars represent the standard error of the mean between experiments. (B) Representative examples of transcription factors identified as direct RELA targets in activated BJAB cells. Two lines at the top of each panel show RNA-Seq tracks obtained from cells that were pretreated with Tet (+Tet) or not (-Tet) for 24 h, followed by activation with P+I. Only time points at which RNA expression changed maximally are shown; complete time courses for each gene are provided in S3H Fig. Y axis numbers denote normalized reads per million. The next 4 lines show representative RELA ChIP-Seq tracks over the entire time course. Peak calling was carried out with MACS2, using input DNA as the control. Transcription orientation (arrows) and gene organization in hg19 are noted below each set. The y axis denotes normalized reads per 10 million. ChIP followed by quantitative PCR validation and NF-κB dependence of these genes is shown in S3E Fig. IRF and KLF family members have been previously proposed as NF-κB target genes. HES1 and ZNF267 were identified in this study. (C) Kinetic patterns of gene induction of direct (304) and indirect (502) RELA target genes in activated BJAB cells. Cells were activated with P+I for the indicated times, followed by RNA-Seq. Mean normalized reads for direct and indirect target genes from 2 independent experiments are plotted for each time point. Error bars represent the standard error of the mean. Genome scale datasets are available on the GEO website (http://www.ncbi.nlm.nih.gov/geo/) (Accession number GSE117259). Underlying data for Fig 3A and C are provided in S1A and S1B Data. ChIP-Seq, chromatin immunoprecipitation and sequencing; ERK, extracellular signal–regulated kinase; GAPDH, glyceraldehyde 3-phosphate dehydrogenase; GEO, Gene Expression Omnibus; NF-κB, nuclear factor kappa B; P+I, phorbol 12-myristate 13-acetate and ionomycin; qRT-PCR, quantitative real-time PCR; RNA-Seq, RNA sequencing; Tet, tetracycline.
Fig 4.
Repression of gene expression by NF-κB.
Genes whose expression was increased by dnIκBα in both Tet-responsive clones were identified by RNA-Seq. (A) Genes that bound RELA (middle, green) and their time-dependent expression patterns in the absence (blue) or presence (green) of Tet-induced dnIκBα are shown on the right. See also S4A Fig. The letters “Rd” after a pattern number indicate genes that were down-regulated by RELA binding, and the letters “Ri” indicate genes whose down-regulation was RELA dependent but did not bind RELA. Numbers of genes in each category are shown in black below the graphs. Red numbers correspond to genes whose expression changed more than 2-fold in the absence of Tet. Expression patterns of 178 dnIκBα-up-regulated genes that did not bind RELA (middle, gray) are shown on the left, with numbers and color coding as described above. (B) Representative examples of dnIκBα-up-regulated genes that bind RELA. Top 2 tracks show RNA-Seq tracks in the presence or absence of Tet-induced dnIκBα at the time of maximal RNA expression; the y axis denotes normalized reads per million. Complete time courses for each gene are provided in S4B Fig. The bottom part shows RELA ChIP-Seq tracks over the entire time course in activated BJAB cells. Genomic organization and transcription orientation (arrows) are shown below the input DNA track. Numbers above the tracks refer to gene location in hg19. The y axis denotes normalized reads per 10 million. (C) Representative examples of dnIκBα-up-regulated genes that do not bind RELA. These genes are proposed to be modulated by RELA-responsive factors (see text). RNA-Seq tracks over the complete activation time course in the absence (-Tet) or presence (+Tet) of dnIκBα are shown. Genome scale datasets are available on the GEO website (http://www.ncbi.nlm.nih.gov/geo/) (Accession number GSE117259). ChIP-Seq, chromatin immunoprecipitation and sequencing; dnIκBα, dominant negative NFKB inhibitor alpha; GEO, Gene Expression Omnibus; NF-κB, nuclear factor kappa B; RNA-Seq, RNA sequencing; Tet, tetracycline.
Fig 5.
Kinetic patterning of NF-κB-dependent gene expression.
(A) Relationship of Pol II–associated chromatin looping to NF-κB-dependent gene expression. Unactivated BJAB cells were used to carry out ChIA-PET after immunoprecipitation of cross-linked chromatin with anti-Pol II antibodies. Resulting sequence data were analyzed and categorized according to the type of promoter status. Category I = genes with no Pol II (no loops); II = genes with Pol II at promoter (no loops); III = genes with single-gene-based loops, in which only 1 gene was involved; IV = genes with multi-gene-based loops that included promoter–promoter interactions in which at least 2 genes were involved. (B) Analysis of 130 direct RELA target genes and (C) 78 “indirect” RELA target genes that are induced ≥2-fold by P+I. Yellow and blue circles denote genes that have or do not have Pol II–associated loops, respectively, in unactivated BJAB cells. Average expression profiles of genes within each group were obtained from RNA-Seq analysis of Tet-inducible BJAB clones (right). The letters “Ad” after a pattern number indicate target genes that were activated by RELA binding, and the letters “Ai” indicate target genes that were indirectly activated by RELA. Genes in each category are listed in S6B Table. (D) RELA and Pol II binding over the activation time course on representative direct and indirect RELA target genes. The effect of dnIκBα expression on RNA expression is shown on the top 2 lines, followed by RELA ChIP-Seq and Pol II ChIP-Seq tracks. The bottom line summarizes loops identified by ChIA-PET. Complete RNA time courses for each gene are shown in (S6J Fig). ChIA-PET data are available on the GEO website (http://www.ncbi.nlm.nih.gov/geo/) (Accession number GSE117259). Underlying data for Fig 5B and 5C are provided in S1C and S1D Data. ChIA-PET, chromatin interaction analysis by paired-end tag sequencing; ChIP-Seq, chromatin immunoprecipitation and sequencing; dnIκBα, dominant negative NFKB inhibitor alpha; GEO, Gene Expression Omnibus; NF-κB, nuclear factor kappa B; P+I, phorbol 12-myristate 13-acetate and ionomycin; Pol II, RNA polymerase II; RNA-Seq, RNA sequencing; Tet, tetracycline.