Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2015 Mar 20.
Published in final edited form as: Mol Cell. 2014 Feb 13;53(6):867–879. doi: 10.1016/j.molcel.2014.01.026

Fold-change of nuclear NF-κB determines TNF-induced transcription in single cells

Robin E C Lee 1,2, Sarah R Walker 3,4,5, Kate Savery 1, David A Frank 3,4,5, Suzanne Gaudet 1,2,*
PMCID: PMC3977799  NIHMSID: NIHMS567925  PMID: 24530305

SUMMARY

In response to TNF, NF-κB enters the nucleus and promotes inflammatory and stress-responsive gene transcription. Because NF-κB deregulation is associated with many diseases, one might expect strict control of NF-κB localization. However, nuclear NF-κB levels exhibit considerable cell-to-cell variability, even in unstimulated cells. To resolve this paradox and determine how transcription-inducing signals are encoded, we quantified single-cell NF-κB translocation dynamics and transcriptional responses in the same cells. We show that TNF-induced transcription correlates best with fold-change in nuclear NF-κB, not absolute nuclear NF-κB abundance. Using computational modeling, we find that an incoherent feed-forward loop, from competition for binding to κB motifs, could provide ‘memory’ of pre-ligand state necessary for fold-change detection. Experimentally, we observed three gene-specific patterns of transcription that are recapitulated in our model by modulating competition strength alone. Fold-change detection buffers against stochastic variation in signaling molecules and explains how cells tolerate variability in NF-κB abundance and localization.

INTRODUCTION

TNF binding to TNF receptor 1 (TNFR1) initiates a network of intracellular signals via the sequential formation of protein complexes (Micheau and Tschopp, 2003). This network encompasses the canonical NF-κB pathway, stress kinases, and in some cells, the apoptotic caspase pathway (Wajant et al., 2003). TNF-dependent activation of NF-κB promotes transcription of anti-apoptotic and proinflammatory genes (Barkett and Gilmore, 1999; Karin and Lin, 2002; Micheau et al., 2001). Well regulated responses to TNF and NF-κB activation are important to normal physiology (Aggarwal, 2003; Li and Schwartz, 2001), and persistent low-levels of TNF lead to misregulated expression of NF-κB target genes, contributing to inflammatory diseases and inflammation-associated cancers (Lewis and Pollard, 2006; Marx, 2004; Schottenfeld and Beebe-Dimmer, 2006).

The NF-κB family of transcription regulators consists of five related proteins: RelA/p65, RelB, c-Rel, p50 and p52 (Oeckinghaus and Ghosh, 2009). All five proteins contain a conserved N-terminal Rel homology domain required for DNA binding and dimerization with other NF-κB family members (Baldwin, 1996; Hayden and Ghosh, 2004; O’Dea and Hoffmann, 2010; Oeckinghaus and Ghosh, 2009). NF-κB proteins can be subdivided based on transactivation potential: only RelA, RelB and c-Rel contain a transactivation domain (TAD) required to recruit transcriptional machinery, the p50 and p52 subunits do not. Although nearly all homo- and heterodimer pairs of NF-κB family proteins are predicted to exist, most are rarely observed. The predominant NF-κB dimers in most cell types are the RelA-p50 heterodimer and the p50-p50 homodimer (O’Dea and Hoffmann, 2010; Oeckinghaus and Ghosh, 2009; Phelps et al., 2000). Because p50-p50 and p52-p52 homodimers lack a TAD, they do not have the intrinsic ability to drive transcription and can instead repress transcription when bound to κB sites of target genes (Cheng et al., 2011; Zhong et al., 2002). For these reasons, “NF-κB” commonly refers to the transcriptionally competent RelA-p50 heterodimer.

In the absence of extracellular signals, RelA-p50 heterodimers interact with NF-κB inhibitor proteins (IκB), the most prevalent and best studied of which is IκBα. IκBα/RelA-p50 complexes are actively exported from the nucleus leading to the predominantly cytoplasmic cellular localization of RelA in unstimulated cells (Baeuerle and Baltimore, 1988). TNF and other inducers of the canonical NF-κB pathway promote phosphorylation and degradation of IκBα, releasing RelA-p50 and exposing its nuclear localization sequence (NLS) (Baeuerle and Baltimore, 1988; Hayden and Ghosh, 2008). The unmasked NLS of free RelA-p50 directs it to the nucleus, where it can access the promoter regions of its target genes. Based on this regulatory mechanism, the degree of NF-κB pathway activation is often equated with nuclear RelA abundance (e.g. (Cheong et al., 2011; Tay et al., 2010)).

When in the nucleus, the RelA-p50 NF-κB heterodimer drives transcription of hundreds of genes (http://www.bu.edu/nf-κB/gene-resources/target-genes/; (Pahl, 1999)). Early responding transcripts (as clustered based on temporal expression profiles in TNF-treated cells) encode cytokines (e.g. IL-8 and IL-6) as well as regulators of the NF-κB pathway itself (e.g. A20 and IκBα). Abundance of these transcripts peaks at around 1 hr after TNF addition (Tian et al., 2005a; Tian et al., 2005b). The expression of certain early response genes, such as NFKBIA that encodes for IκBα, must also be constitutive and scale with NF-κB abundance to maintain pathway homeostasis (Brown et al., 1993; Scott et al., 1993). In contrast, certain cytokine-encoding genes are strictly inducible, not expressed in untreated cells (e.g IL6 and IL8 (Tian et al., 2005a)). The mechanisms by which the same NF-κB dimer coordinates constitutive and scalable as well as inducible transcriptional responses are still unclear.

NF-κB-dependent transcription regulates critical cellular behaviors and must be tightly regulated to prevent disease. It is therefore surprising that the abundance of nuclear NF-κB, in particular the RelA subunit, is highly variable in unstimulated and TNF-treated cells (Cheong et al., 2011; Tay et al., 2010). We consequently hypothesized that nuclear RelA abundance alone does not determine transcription, but that transcription may be directed by aspects of RelA translocation dynamics. Here, we test this by measuring both RelA translocation and transcription of early response genes in the same cells, and thereby directly quantify the transcriptional impact of TNF-induced changes in RelA localization in single cells. We find that transcription is determined by fold-change, not absolute abundance, of nuclear RelA. Even though they are all responsive to fold-change signals, three different NF-κB-dependent genes exhibit distinct patterns of transcription ranging from strictly inducible to constitutive and scaling with RelA abundance. The diversity of transcription patterns suggests that this global regulator of transcription is also tunable for specific gene targets. Using a computational model of ligand-induced transcription we show that competition for binding to κB sites in the promoters of NF-κB-dependent genes is sufficient to recapitulate all of our experimental observations. Our study reveals that the NF-κB system is capable of fold-change detection and provides a framework for understanding how transcriptional networks interpret and act on dynamical protein signals.

RESULTS

Fold-change of nuclear RelA is Less Variable than Absolute RelA Abundance

We used fixed-cell immunofluorescence to characterize the heterogeneity of endogenous RelA localization in TNF-treated HeLa cells (Figure 1). As others have reported (Cheong et al., 2011; Hoffmann et al., 2002; Nelson et al., 2004; Tay et al., 2010), we observed that the timing and intensity of RelA translocation in response to TNF vary between cells (Figure 1 and see also Figure 2). This cell-to-cell variability is likely due, at least in part, to variation in levels of key regulatory proteins as described previously for other signaling systems (Cohen-Saidon et al., 2009; Feinerman et al., 2008; Gaudet et al., 2012; Spencer et al., 2009). The amount of nuclear RelA has been used as a measure of TNF-induced NF-κB pathway activation (Cheong et al., 2011; Tay et al., 2010) and we therefore quantified total nuclear amount and nuclear density of RelA from our images. We found that both the abundance and density of nuclear RelA showed substantial cell-to-cell variability in untreated and TNF-treated cells (Figure 1B and 1C). Many cells were included in the overlap between distributions from untreated and TNF-treated cells (31% for ‘total nuclear fluorescence’, 22% for ‘nuclear density’; Figure 1B and 1C). Consistent with a recent study (Cheong et al., 2011), our results indicate that nuclear RelA abundance cannot be used on its own to unambiguously determine if a particular cell has responded to TNF, although a response to TNF is clear at the cell population level (Figure S1A). Overall these results suggest that if single cells can indeed distinguish between the absence and the presence of TNF, more information may be encoded in the dynamics of RelA localization.

Figure 1. TNF-induced NF-κB sub-cellular localization is variable.

Figure 1

Open in a new tab

(A) Fixed-cell RelA immunofluorescence images of HeLa cells treated with 10 ng/mL TNF for indicated times; scale bar is 10 μm.

(B&C) Frequency histograms for total nuclear (B) and nuclear density (C) of endogenous RelA for HeLa cells treated (red) or not (blue) with 10 ng/mL TNF (t = 30 min, dose and time when nuclear translocation was maximal; Figure S1); n = 800–1000 cells.

Figure 2. TNF-induced NF-κB translocation varies in live cells.

Figure 2

Open in a new tab

(A) Time-lapse images of stable HeLa FP-RelA treated with 10 ng/mL TNF. Arrow and asterisk indicate nuclei of cells with different FP-RelA translocation dynamics; scale bar is 10 μm.

(B) Single-cell FP-RelA nuclear density time courses quantified from time-lapse images of HeLa FP-RelA treated with 10 ng/mL TNF. To reduce the influence of high frequency noise in mean fluorescence intensity, nuclear time courses were represented as 3-frame running averages (Figure S2). Inset defines time course descriptors.

(C) Bar graph of the coefficients of variation (CV) for select time course descriptor of FP-RelA nuclear density (descriptors are defined in Figure S2). Error bars represent the standard deviation of the mean (S.E.M.) for triplicate experiments.

At the cell population level, different dynamics of RelA localization yield different TNF-induced gene profiles (Ashall et al., 2009; Tian et al., 2005a). Properties of RelA localization should therefore encode information in single cells. To characterize single-cell RelA translocation dynamics in our cellular system, we generated HeLa cells stably expressing EGFP-RelA. We measured nuclear amounts of fluorescent protein-RelA (FP-RelA) by time-lapse imaging before and after adding a range of TNF concentrations. In the absence of TNF, nuclear NF-κB abundance in each cell fluctuated only marginally over time (Figure S2). In response to TNF, single cells showed transient and variable nuclear translocation of RelA (Figure 2A, B, and Movie S1), as others have reported in other systems (Ashall et al., 2009; Cohen-Saidon et al., 2009; Nelson et al., 2004; Tay et al., 2010). We analyzed single-cell data by calculating the values of quantitative descriptors (Figure 2B inset and S2; (Cohen-Saidon et al., 2009)) and quantified cell-to-cell variability by calculating the coefficient of variation (CV) for each descriptor. For all TNF concentrations, the CVs for nuclear FP-RelA abundance at single time points had values near 0.5 (Figure S2) whereas the CVs for certain time-course descriptors were substantially smaller (e.g., CV = 0.24±0.07 for Fmax/Fi, the ratio of maximal to initial nuclear RelA; Figures 2C and S2). These data show that the amount of cell-to-cell variability observed in response to TNF depends on the descriptor chosen to represent NF-κB activation.

Variability in Transcription and Sensitivity to RelA Abundance is Gene Specific

To investigate whether RelA abundance influences NF-κB-driven transcription in single cells, we examined cell-to-cell variability in transcription in both parental and FP-RelA expressing HeLa cells. We designed single-molecule fluorescent in situ hybridization (smFISH; (Raj et al., 2008)) probes targeting the mRNA products of IL8, TNFAIP3, and NFKBIA (which respectively encode IL-8, A20 and IκBα), three NF-κB-dependent genes strongly expressed 1 hr after exposure to TNF (Haskill et al., 1991; Krikos et al., 1992; Kunsch and Rosen, 1993; Tay et al., 2010; Tian et al., 2005a). The smFISH approach allowed us to unambiguously detect single mRNA molecules and we therefore quantified transcriptional responses by counting mRNAs in single cells (Figure 3A–E). In untreated cells, transcription of IL8 and TNFAIP3 was minimal while transcription of NFKBIA varied (3 to 61 mRNAs/cell in parental HeLa; Figure 3F, G; Table S1). Expression of FP-RelA increased baseline transcription only for NFKBIA (p < 10−12; Figures 3G and S3) possibly to allow cells to maintain a constant ratio of IKBα to RelA. In TNF-treated cells, transcription of all three genes was dose dependent but varied cell to cell (Figure 3G; Table S1). Expression of FP-RelA increased responsiveness to TNF for both NFKBIA and TNFAIP3, but not for IL8 (Figures 3G and S3; p = 10−18, 10−6 and 0.5 respectively, t-tests for 10 ng/ml TNF datasets).

Figure 3. Variability of TNF-induced NF-κB-dependent transcription is transcript specific.

Figure 3

Open in a new tab

(A) Transmitted light, (B) Hoechst channel and (C) maximum intensity projection images of a representative TNFAIP3 smFISH fluorescence z-stack. The perimeters of nuclei are marked (dashed yellow line) and side view projections of z-stack are depicted.

(D) Linescans from image in (C) demonstrate the high signal-to-noise ratio for typical mRNAs (blue line) and active transcription sites (ATS, red line) where nascent transcripts accumulate on the gene locus (Raj et al., 2008).

(E) Images merging Hoechst and fluorescent channels for control samples that were not exposed to smFISH probes but were otherwise treated identically to all other samples.

(F) Transmitted light and IL8, TNFAIP3 and NFKBIA smFISH images of untreated and TNF-treated HeLa. Nuclei were counterstained with Hoechst (blue).

(G) Boxplots of the IL8, TNFAIP3 and NFKBIA mRNAs/cell distributions for parental (P) and FP-RelA HeLa lines and indicated treatments. Red bars and notches indicate median and 95% confidence interval; statistical significance of differences were assessed by two-sample Kolmogorov-Smirnov test (n > 45, **p ≪0.01; Figure S3); scale bars are 10μm for all.

Overall, our data show that the three genes we assayed have distinct patterns of sensitivity to RelA abundance. Although transcript numbers are variable from cell to cell in all conditions, they were less variable than nuclear FP-RelA abundance measured at single time points (Table S2). These results suggest that if gene transcription is quantitatively controlled by NF-κB translocation, nuclear RelA abundance may not be an adequate descriptor of this transcription-inducing signal.

Fold-change of Nuclear RelA Determines the Transcription Output

To systematically evaluate which descriptors of RelA translocation best predict transcription, we explored correlations between same-cell FP-RelA translocation dynamics and transcript numbers (Figure 4A and Movie S2). For each cell tracked in these experiments, translocation of FP-RelA was again parameterized into a series of descriptors (Figure S2;(Cohen-Saidon et al., 2009)), and each descriptor was plotted against mRNA transcript number for IL8, TNFAIP3 and NFKBIA (Figures 4B and S4). Data from cells treated with various doses of TNF were combined for each plot to maximize the dynamic range examined for each descriptor (Table S3). We then calculated coefficients of determination (R2) to assess the extent to which each descriptor predicts the transcriptional response of each gene.

Figure 4. Transcriptional responses to TNF are determined by fold-change of nuclear NF-κB.

Figure 4

Open in a new tab

(A) Workflow connecting live-cell imaging of FP-RelA nuclear translocation to same-cell smFISH (Movie S2).

(B) Assessment of correlations between descriptor and mRNA number. Example weak and strong correlations are shown, corresponding to mean nuclear fluorescence at t=30′ (Ft=30′) and maximum nuclear fold-change (Fmax/Fi) respectively.

(C) Bar graph of the coefficient of determination (R2) of each nuclear FP-RelA descriptor for three NF-κB-dependent transcripts (see also Figure S4).

(D) Plots showing the R2 of listed descriptors at each time point for IL8, TNFAIP3 and NFKBIA. The straight red line indicates the R2 for maximum fold-change of nuclear NF-κB, the strongest single predictor of the transcriptional response for all three genes ((C) and Figure S4). The summary of cell numbers for same-cell FP-RelA translocation and transcription experiments is in Table S3.

In scanning entire nuclear FP-RelA time courses in response to TNF, we found that nuclear FP-RelA abundance at any time point maximally accounted for 30%, 15%, or 16% of observed IL8, TNFAIP3 or NFKBIA transcript number variability (Figure 4C and 4D, dashed light blue lines, and S4). Similar results were found for mean nuclear fluorescence, which normalizes the FP-RelA signal to nuclear area to account for variability in nuclear size or morphology (Figure 4C and 4D, light blue lines, and Figure S4). In contrast, descriptors that quantified change of nuclear FP-RelA over time were stronger predictors of the transcriptional response (Figure 4C and 4D, orange, purple and black lines, and S4). In particular, ‘maximum fold change’ of nuclear FP-RelA (Fmax/Fi) was the strongest single predictor of transcript number for all three genes (R2 = 0.52, 0.61 and 0.67 for NFKBIA, IL8 and TNFAIP3 respectively; Figures 4C and S4). Fold change of nuclear FP-RelA quantified at any time point between 35–50 min explained the variance in transcription as well as Fmax/Fi (Figure 4D). Taken together, relationships between same-cell FP-RelA translocation and transcript number suggest that the NF-κB transcription regulation system is capable of fold-change detection.

A Model with Competition on NF-κB Target Promoters is Capable of Fold-change Detection

NF-κB-driven transcription has been modeled as a circuit with two negative feedback loops via A20 and IκBα (D2F model (Ashall et al., 2009); see also Supplemental Model Information). We asked whether this architecture inherently encoded fold-change detection of nuclear NF-κB. In a series of simulations recapitulating sets of FP-RelA translocation time courses similar to those we observed in experiments, we found that the model generated a transcription pattern qualitatively different from our observations for IL8 and TNFAIP3 (but similar to the pattern for NFKBIA). The pattern of transcription generated by the D2F model was also less well determined by the maximal fold change in nuclear NF-κB (R2 = 0.21) than the patterns observed experimentally for all three genes assayed.

In the D2F model, simulated target gene transcription in absence of stimulus was non-negligible and proportional to RelA abundance (Figure S5B left column); this contrasts with very low and nearly invariant baseline transcription observed for IL8 and TNFAIP3 in untreated cells (Figure S5A left column). For the transcriptional responses to fold-change signals, again simulation results were qualitatively different from our IL8 and TNFAIP3 data (Figure S5, middle column). We scored this difference by calculating the ‘relative variance’ in transcript number over a series of bins of fold change in nuclear RelA. Relative variance was defined as variance normalized to the variance observed in the lowest fold-change bin (Figure S5, right column and Supplemental Model Information). This allowed us to distinguish qualitatively different patterns: where variance in transcript number increases, or decreases, with fold change in nuclear RelA. While our data showed that variance increased with fold change for both IL8 and TNFAIP3, this could not be reproduced in simulations of the D2F model (Figure S5). Overall, the D2F model failed in reproducing both fold-change detection and the transcription patterns of IL8 and TNFAIP3.

To detect fold-change signals, each cell must retain a ‘memory’ of previous states against which it can measure change in nuclear RelA. Because incoherent type 1 feed forward loop motifs (I1-FFL; Figure 5A) can impart memory and allow fold-change detection (Cohen-Saidon et al., 2009; Goentoro and Kirschner, 2009; Goentoro et al., 2009), we explored whether minimal and plausible changes to the model could generate such a motif. Conceptually, the D2F model can be separated into a pulse generator module, responsible for transient cytokine-dependent nuclear translocation of NF-κB, and a transcriptional module (Figure 5B). Viewed this way, it becomes clear how the D2F transcription module is directly influenced by nuclear NF-κB abundance. To the D2F model, we added an indirect inhibitory pathway by including a transcriptionally incompetent DNA-binding protein that competitively binds to NF-κB target promoters. As we describe more at length in our discussion, NF-κB-regulated transcriptionally incompetent p50-p50 and p52-p52 homodimers or BCL3, a protein that stabilizes homodimer complexes, are plausible competitor species. Using as a guide the protein-DNA interactions previously described for the NFKB1 promoter (which encodes p50; (Ten et al., 1992)), ‘competitor’ was set to be an NF-κB-dependent species, also subject to repression from its own protein product. Adding an NF-κB-dependent competitor species completes an I1-FFL-like motif, an architecture we hereafter refer to as the D2FC model (D2F with competition; Figure 5C and Supplemental Model Information). After setting the DNA-binding affinities of competitor and NF-κB for target promoters to the same value, our new model generated a transcription pattern qualitatively similar to those obtained experimentally for IL8 and TNFAIP3 (Figure S7, column 5). Importantly, with this new model, maximal fold change in nuclear NF-κB explained 82% of the variance in target transcript number. Therefore, embedding indirect negative regulation of gene transcription through competition on target promoters completes an I1-FFL-like motif within the NF-κB transcriptional network, and the resulting transcriptional system exhibits features of fold-change detection.

Figure 5. An I1-FFL model of NF-κB-mediated transcription recapitulate experimental transcriptional patterns.

Figure 5

Open in a new tab

(A) Schematic diagram of an I1-FFL network motif.

(B&C) Diagram of the NF-κB-induced transcriptional network showing the pulse-generator (I) and transcriptional (II) modules for direct (B) and I1-FFL-like (C) transcriptional models.

(D&E) Scatter plots of transcript numbers vs. total FP-p65 in untreated cells (left), and vs. maximal nuclear FP-RelA fold-change for TNF-treated cells (center, data from cells treated with 0.1, 1 and 10 ng/mL TNF are all plotted on the same graph). Bar graphs show the relative variance for all three genes at different fold-change levels (right; Figure S7). Results are shown for simulations with high (cyan), moderate (red) and low (yellow) affinity competition (D) and for experiments for IL8 (cyan), TNFAIP3 (red), and NFKBIA (yellow) (E).

A Model with Competition Predicts Experimental Patterns of Transcription

In initial D2FC model simulations the affinity of competitor for target promoters was set equal to that of the transcriptionally competent NF-κB species. However, differences in promoter sequences could alter the relative DNA-binding affinity of competitor vs. NF-κB (Siggers et al., 2012; Udalova et al., 2000). We found that by changing the relative affinity of competitor vs. NF-κB for a target promoter, simulations generated a spectrum of responses. With high competitor affinity, target gene transcription was ‘inducible’: repressed across RelA expression levels in the absence of TNF and scaling with fold-change of nuclear RelA in simulations of TNF treatment (Figures 5D, 7B and S7). The expression of such a gene strictly depends on changes in nuclear NF-κB over short time-scales. By contrast, in simulations where competitor affinity was set to a low value, the system behaved similarly to the original D2F model, defining a ‘constitutive’ gene class (Figures 5D, 7B and S7). Like D2F, the D2FC model then recapitulates well how the expression of NFKBIA (coding for IκBα) is constitutive, and its expression levels scale with RelA to effectively inhibit NF-κB in each cell (Brown et al., 1993; Scott et al., 1993). Transcription from promoters with moderate competitor affinity was inducible, but reached higher transcript number than that with high competitor affinity. Comparing model and experiments, our results suggest that a competitor species respectively binds at high, moderate, and low affinity on promoters for IL8, TNFAIP3 and NFKBIA (Figure 5D, E).

Figure 7. The model explains how transcription patterns are tuned by changes to competitor affinity and abundance.

Figure 7

Open in a new tab

(A) Hypothetical plots mapping the multifactorial system that regulates gene expression. ‘Hard-wired factors’ can be described using plots showing hypothetical affinity of competitor complex vs. RelA dimer affinity for a series of competitor complexes (left); shaded regions represent a hypothetical space occupied if plotting values for all κB sites and squares show that the binding affinity for different competitor complexes can be different for two hypothetical κB sites. Competitor affinity, RelA dimer affinity and competitor identity are only three axes in a larger multidimensional space representing all the factors that affect NF-κB driven gene expression (right).

(B) Schematics for repressed, inducible and constitutive patterns of transcription (also see Figure S7).

(C) Matrix of scatter plots showing transcript number vs. fold-change in nuclear RelA for simulations with increasing competitor abundance (from left to right) and increasing competitor affinity for target promoter (from bottom to top). In the model, competitor expression and affinity are lumped variables representing the aggregate abundance of all competitor complexes that regulate a gene, and their combined affinity for the κB sites in the promoter. Patterns were classified as ‘constitutive’, ‘inducible’ or ‘repressed’ (see Parameter Sweep section in the Supplementary Information for additional discussion).

Next, we investigated the effect of changes in competitor abundance by simulating competitor knockdown in a cell population with a RelA concentration range set to reflect RelA abundance in parental HeLa cells. Because fold change of nuclear RelA in parental HeLa cells is similar to that of FP-RelA in EGFP-RelA HeLa cells (Figure S6A and S6B), simulation conditions were selected so that the distribution of maximum fold-change in nuclear NF-κB matched that observed in experiments for FP-RelA (Figure S6C). When plotting population-average transcript number, simulations predicted that in the absence or presence of TNF, knockdown of the competitor should markedly increase transcription of target genes with high affinity for competitor (Figure 6A); the effect is more pronounced with higher efficiency knockdown. In contrast, for target genes with low affinity for competitor, competitor knockdown should only marginally increase transcription (up to ~3-fold for complete knockout).

Figure 6. Individual genes show different sensitivity to knockdown of candidate competitors.

Figure 6

Open in a new tab

(A) Graphs of the predicted change in transcript abundance (expressed as fold induction over the no-knockdown condition) as a function of competitor knockdown efficiency for baseline conditions (left) and TNF-treatment conditions (right). Simulations of the D2FC model were run using initial conditions mimicking parental HeLa cells (see Figure S6C and its legend for details). As affinity of the competitor for the target gene promoter was increased, the predicted change in abundance also increased (arrow). Regions of high and low competitor affinity, used to model baseline and TNF-induced transcription corresponding to IL8 and NFKBIA, are shown as yellow and blue regions respectively.

(B) Bar graphs showing fold induction in transcript abundance over control siRNA (si-ct) condition as measured by quantitative PCR for parental HeLa cells transfected with siRNA targeting three candidate competitors (p50, p52 and BCL3). IL8 (yellow bars) and NFKBIA (blue bars) mRNA levels were quantified in baseline condition (untreated cells; left) and in cells treated with 10 ng/ml TNF for 60 min (right). mRNA abundances that are significantly greater than in control siRNA conditions are marked with their p-values (grey lines, two-tailed t-test). Asterisks mark values where IL8 siRNA abundance changes significantly more than the abundance of NFKBIA mRNA (p < 0.05 in a one-tailed t-test). Error bars represent the standard deviation from at least four independent siRNA transfection experiments

(C&D) Scatter plots on the left show correlation between nuclear densities of RelA vs. p50 (C) or BCL3 (D). Histograms of total nuclear Hoechst intensity (middle) were used to categorize cells as G1 (smaller nuclei, less DNA content) or G2 (larger nuclei, more DNA content) and scatter plots on the right compare nuclear densities for cells with similar DNA content.

To test our model prediction, we measured IL8, TNFAIP3 and NFKBIA transcript abundance by quantitative PCR in parental HeLa cells treated with siRNA against candidate competitor proteins (si-p50, si-p52 and si-BCL3 which respectively target NFKB1, NFKB2 and BCL3) or control siRNA. In agreement with model predictions for a gene with high affinity for competitor, expression of IL8 was markedly increased by BCL3 knockdown both in the absence or presence of TNF (≥ 17-fold, p < 0.05 by t-test; Figure 6B). For p50 knockdown, transcription increased only in the absence of TNF (~6-fold, p = 0.08; Figure 6B), and for p52 knockdown, transcription increased only in the presence of TNF (~5-fold, p = 0.03; Figure 6B). This suggests that different competitor species inhibit transcription in different conditions. For NFKBIA, transcript abundance was consistent with DF2C model predictions for a gene with low affinity for competitor. Baseline and TNF-induced NFKBIA transcript number increased after knockdown of p50 and BCL3 (<3-fold; p < 0.05, t-test; Figure 6B) but much less so than IL8 (p < 0.05).

The effects of candidate competitor knockdown on TNFAIP3 transcript numbers are harder to interpret. In TNF-treated cells, we observed marginal increases for knockdown of all three candidate competitors, but si-p52 and si-BCL3 treatments reduced baseline transcript numbers (p < 0.02 by t-test; Figure S6D). The latter result may be explained in part by the reported observation that a p52:BCL3:Tip60-containing complex can bind to a κB site in the TNFAIP3 promoter and act as a co-activator (Wang et al., 2012). Because the single-cell transcriptional pattern of TNFAIP3 was most consistent with that of a gene with moderate affinity for competitor, the lack of a strong effect from competitor knockdown raises the possibility that another, yet unidentified, competitor protein acts on TNFAIP3. However, other possible confounding factors are that p50 knockdown may affect RelA-p50 heterodimer abundance, and that it may be difficult to isolate the influence of BCL3, p50 or p52 on transcription of a gene that is regulated by multiple competitors because knockdown of one candidate competitor affects the transcript abundance of the others (Figure S6E).

Overall, our results implicate p50, p52, and BCL3 as proteins that participate in competitor complexes and suggest that others exist. Furthermore, the experimental validation of the D2FC model-predicted effect of competitor knockdown for IL8 and NFKBIA show that the I1-FFL-like topology encoded in the model could exist in parental HeLa cells that do not overexpress RelA. Finally, in additional support of the I1-FFL-like structure of our working model, we observed by immunofluorescence that nuclear density of p50 and BCL3 correlated with that of RelA in single cells of the parental HeLa (Figures 5C and 6C, D).

Relative Abundance and Affinity of Competitor Complexes Tunes the Transcriptional Response

The affinities of transcriptionally competent and incompetent protein complexes for a promoter’s κB site should be ‘hard wired’ biochemical parameters – governed by protein-DNA interactions (Figure 7A). Because a promoter could have multiple κB sites, accurate predictions of transcript abundance for a specific gene will require knowing the identity and affinity of each complex that binds and how the activities on all κB sites are integrated (Figure 7A). Nevertheless, D2FC model simulations show that depending on the ‘apparent’ integrated relative affinities of these complexes for a target gene, its expression could be ‘constitutive’, ‘inducible’, or even ‘repressed’ (Figures 7B, 7C (top to bottom) and S7).

In addition, noise in protein expression and epigenetic changes to the promoter of the competitor species, could alter the competitor to RelA protein ratio between cells. We examined how the relative abundance of competitor and RelA-p50 dimers affects transcription patterns in the D2FC model. In these simulations, we observed that the transcriptional response from a single promoter, with hard-wired affinity to competitor(s) and transcriptionally competent NF-κB, can switch between ‘repressed’, ‘inducible’ and ‘constitutive’ behaviors depending on the ‘aggregate’ abundance of competitor(s) relative to RelA-p50 (left to right in Figure 7C; ‘Parameter sweep’ in Supplemental Model Information).

Taking advantage of cell-to-cell variability, and characterizing same-cell relationships between RelA translocation dynamics and transcriptional output in response to TNF, we have identified qualitatively different transcriptional response patterns. Although undoubtedly a simplification of a complex system, the I1-FFL-like architecture of our working model recapitulates all of our experimentally observed patterns by modifying ‘apparent’ competitor affinity, showing how indirect feed forward inhibition can fine-tune transcriptional regulation. Furthermore, simulations show that cell type-to-cell type and cell-to-cell variations in the relative expression of the competitor species, here modeled as single aggregated species, could alter the landscape of baseline and ligand-induced NF-κB-dependent transcription, contributing to the long-observed heterogeneity in cellular responses to TNF.

DISCUSSION

When in the nucleus, NF-κB family proteins regulate the expression of inflammatory and survival-associated genes central to human health. Because protein expression is a noisy process (Bar-Even et al., 2006) and baseline nuclear RelA amounts scale with protein abundance (Supplemental Model Information, 2nd figure), cell-to-cell variability in the abundance of NF-κB subunits could lead to spontaneous pathway activation and inflammation with potentially deleterious consequences. Our results demonstrate that basal and TNF-induced nuclear RelA abundance as well as NF-κB-dependent TNF-induced transcription are variable from cell to cell. However, through combined imaging of FP-RelA translocation dynamics and target gene transcripts in the same cell, we have determined that nuclear RelA amount is not a strong predictor of transcription. Instead, transcription of inducible early response NF-κB target genes in single cells gains robustness to variability in RelA expression through fold-change detection of nuclear RelA.

In contrast with our experimental findings, simulated transcription from the D2F NF-κB model (Ashall et al., 2009) was not well determined by fold-change in nuclear RelA. Our modified working model, D2FC, incorporating RelA-dependent expression of a competitor also binding on target promoters, reproduced the transcriptional patterns that we measured for IL8, TNFAIP3 and NFKBIA, and was capable of fold-change detection. The D2FC model also correctly predicted the effect of siRNA-mediated knockdown of candidate competitors on transcription of IL8 and NFKBIA in parental cells. This suggests that NF-κB-driven transcription may be controlled by a paradoxical circuit, one that both activates and represses transcription of the same gene. Because the indirect inhibition arm of the I1-FFL architecture in the D2FC model requires both transcription and translation of a competitor protein, there is a considerable delay before competitor abundance respond to changes in nuclear RelA. This could allow for a separation of time scales where TNF-induced early response genes are transcribed after rapid RelA translocation while the concentration of competitor protein adjusts to long term, or steady state, nuclear RelA abundance. In this regard, the competitor protein would provide a ‘memory’ of previous nuclear RelA amount in each cell and normalize transcriptional responsiveness across cells with differing RelA expression levels.

In the D2FC model, distinct transcriptional patterns for IL8, TNFAIP3 and NFKBIA were recapitulated by altering the affinity of competitor for gene promoters. A continuum of transcriptional responses, ranging from ‘constitutive’, to ‘inducible’ and ‘repressed’ can be produced by altering the relative strength of activating or inhibitory I1-FFL arms. Therefore, paradoxical transcriptional circuits such as this I1-FFL architecture can not only allow fold-change detection and improve robustness to noise in protein expression, but also provide tunability – in a single cell, different target genes can respond distinctly to the same activating signal.

The competitor protein or complex included in the D2FC model of NF-κB-driven transcription should fulfill three conditions to establish a ‘memory’ of baseline nuclear RelA. It should: 1) prevent RelA from binding to target κB sites or reduce the transcription-inducing ability of bound RelA, 2) have no, or low, transcriptional activity on κB sites compared to RelA, and 3) be expressed in a RelA-dependent manner at levels that scale with baseline nuclear RelA. Homodimers of p50 or p52, the mature products of the NFKB1 and NFKB2 genes, and BCL3, a protein that stabilizes repressive homodimer complexes on a subset of κB sites (Wang et al., 2012; Wessells et al., 2004), all satisfy these conditions and are plausible candidate competitors. The DNA-binding sequence specificity of p50 and p52 homodimers overlaps with that of RelA-p50 NF-κB heterodimers (Siggers et al., 2012); repressive homodimers suppress transcription of κB site-containing genes including TNF, IL6 and the HIV long terminal repeat (Lewin et al., 1997; Udalova et al., 2000; Wang et al., 2012; Zhong et al., 2002). Finally, the genes encoding p50, p52 and BCL3 are themselves regulated by NF-κB (Ten et al., 1992; Tian et al., 2005a) and we found that nuclear densities of p50 and BCL3 correlate with that of RelA.

Although we have included a single competitor species in the D2FC model, this is likely an oversimplification; our data for siRNA-mediated knockdown of candidate competitors support the idea that several competitor complexes exist. NF-κB-dependent expression of atypical IκB-family proteins (such as BCL3) may also regulate the stability or DNA-binding of NF-κB complexes at certain κB sites (Ghosh and Hayden, 2008; Marienfeld et al., 2003). In addition, the promoters of each NF-κB-driven gene can have multiple κB binding motifs, each with its own affinity profile for RelA-p50 heterodimers and competitor complexes. We speculate that the existence of multiple competitor proteins and many κB binding motifs enhances tunability in the regulation of specific target genes. As we accrue genome-scale and single-cell information about transcriptional responses and DNA-binding activities, we can begin to address the challenge of understanding how transcription-inducing and transcription repressing activities are integrated on promoters that contain multiple κB sites.

In a particular cell, which protein complex is bound to a specific κB motif will be determined by both affinity and concentration. Interestingly, it has been shown that in differentiating monocytes the susceptibility to HIV replication is anti-correlated with p50 homodimer abundance suggesting that changes to the relative abundance of different NF-κB dimers alters the transcriptional state of cells (Lewin et al., 1997). The relative abundance of NF-κB dimers and different competitive complexes in a cell, or cell line, may therefore be an important control point that governs its transcriptional response, and ultimately its cell fate, induced by NF-κB-activating cytokines.

In summary, our results demonstrate that the dynamics of NF-κB translocation to the nucleus encode transcription-inducing signals that produce gene-specific patterns of transcription. These observations can be recapitulated by including a competitor species in a model of NF-κB-dependent transcription. This would impart an I1-FFL structure to the system which is capable of fold-change detection, buffering against slow fluctuations in steady-state levels of nuclear RelA that could otherwise be misinterpreted as initiating signals. Regulated expression of system-specific competitors may be a common mechanism to allow transcriptional diversity or tunability between cell types, while reducing effects of cell-to-cell variability in protein expression through memory and fold-change detection.

EXPERIMENTAL PROCEDURES

The Supplemental Experimental Procedures section of our Supplemental Information provides additional details for each procedure.

Cell Culture and Treatment of Cells with TNF

Parental HeLa cells (ATCC) and HeLa cells stably transfected with a pEGFP-RelA plasmid (gift of Dr. Mollie Meffert) were cultured in Dulbecco’s Modified Eagle Medium (DMEM) supplemented with 10% FBS, 100 U/mL penicillin, 100 μg/mL streptomycin and 0.2 mM L-glutamine (Invitrogen) at 37°C and 5% CO2. Cells were plated on 96-well imaging plates (BD Biosciences) or Matriplate 96-well No.1.5 glass bottom plates for smFISH (Matrical Bioscience). After 24 hrs, when cell density reached ~4×104 cells/well, cells were treated with indicated concentrations of recombinant human TNF (Peprotech). For all experiments, TNF was diluted into serum free DMEM and 10 μL of the TNF-DMEM mix was spiked into wells containing 150 μL of growth media to yield the indicated final TNF concentrations. TNF containing media remained on the cells for the duration of the experiment.

Fixed-Cell Immunofluorescence Imaging and Analysis

For fixed-cell immunofluorescence, parental HeLa cells grown in 96-well imaging plates (BD Biosciences) were treated with the indicated concentrations of recombinant human TNF (Peprotech). Cells were then fixed and labeled using antibodies targeting RelA, p50 or BCL3 (as detailed in the Supplemental Experimental Procedures). Nuclei were counterlabeled with 200 ng/mL Hoechst in wash buffer (PBS with 0.1% Tween-20). The cells were imaged on a BD Pathway 855 Bioimager (BD Biosciences) with an UAPO/340 20x objective (0.75 NA; Olympus, USA). Images were flat-field corrected, background was subtracted custom scripts in ImageJ (NIH) and fluorescence signal was quantified using CellProfiler image analysis software (Carpenter et al., 2006). To quantify ‘nuclear density’ and ‘nuclear amount’ of RelA from fixed-cell images, a thresholding algorithm was applied to images of Hoechst-stained nuclei to identify the nuclear region of interest (ROI).

Live-Cell Imaging and Analysis

For live-cell experiments, HeLa cells stably expressing FP-RelA were imaged for 1–2 hrs before addition of TNF. For experiments that included smFISH, cells were then fixed and labeled as below. Wide-field epifluorescence and transmitted light live-cell imaging was done in an environmentally controlled chamber (37°C, 5% CO 2) on the BD Pathway 855 Bioimager using a UAPO/340 20x objective (0.75 NA; Olympus), capturing images at 3 min intervals. Data was extracted from flat-field and background corrected time-lapse movies in ImageJ. Mean fluorescence intensity (MFI) of nuclear FP-RelA was collected for each cell at each time point using custom scripts.

smFISH microscopy and image analysis

Following TNF-treatment, parental or FP-RelA HeLa in 96-well Matriplates were fixed using RNase free reagents (Ambion). For any live-cell to fixed-cell experiments, the field of view imaged by time-lapse microscopy was marked and then imaged post-fixation to record any changes to the distribution of cells during fixation. Cells then were hybridized overnight with smFISH probes as described previously (Raj et al., 2008) with optimized probe-set specific conditions (see Supplemental Experimental Procedures). Hybridized cells were imaged using a Deltavision microscope (Applied Precision) with a 60x objective (1.42 NA; Olympus) and a CoolSNAP HQ2 camera using temperature-matched oil (n = 1.516, 23°C). For each field, we a cquired z-stacks of 30–60 images with 0.3 μm intervals; images were deconvolved with SoftWoRx 5.0 software (Applied Precision). Quantitative analysis of mRNA content was performed using scripts in ImageJ (NIH): 1) cells were manually segmented, 2) mRNA molecules were reconstituted as 3D objects (filtered for objects larger than 10 pixels across the z-stack) and counted using the ImageJ 3D object counter plugin (Bolte and Cordelieres, 2006). For each experiment, the minimum intensity thresholds required by the ImageJ 3D object counter were determined from non-hybridized control images. Overlapping cells, cells only partly in the field of view and multinucleated cells were discarded from mRNA content analysis.

Computational modeling

The computational models describing TNF-induced NF-κB-driven transcription were encoded as ordinary differential equations in MatLab (Mathworks; Supplemental Model Information). Cell-to-cell variability in protein expression was modeled by sampling from appropriate distributions of parameter values (Supplemental Model Information).

Competitor gene knockdown and qPCR

Briefly, Hela cells were reverse transfected with 10 nM siRNA (see Supplemental Experimental Procedures) using Lipofectamine RNAi/MAX (Invitrogen). After 48 hr, cells were stimulated with 10 ng/ml TNF (Peprotech) for 1 hr. RNA was isolated using an RNeasy kit (Qiagen) and cDNA was generated using a Taqman reverse transcription kit (Applied Biosystems). qPCR was performed using Sybr Select Master Mix (Applied Biosystems) on a 7500 qPCR machine (Applied Biosystems) and data were normalized to actin and expressed relative to si-control.

Supplementary Material

01
02
Download video file (679.7KB, avi)
03
Download video file (7.1MB, avi)

HIGHLIGHTS.

  • Transcript numbers for NF-κB-dependent genes vary less than nuclear NF-κB abundance

  • Fold-change of nuclear NF-κB is a quantitative predictor of transcript number

  • Individual target genes interpret fold-change signals distinctly

  • An I1-FFL model recapitulates patterns of transcription and fold-change detection

Acknowledgments

We thank B. B. Aldridge, W. W. Chen, M. S. Owen, X. Xia and A. Aref for advice on the manuscript and many helpful discussions and L. Hill for technical assistance. We also thank Dr. Max Heiman for the use of his microscope. This work was funded by NIH grant CA139980, R01-GM104247 and a Barr investigator award to SG, and NCI grant R01-CA160979 and funding from the DeGregorio Family Foundation to DAF. S.G. is a Kimmel Scholar and R.E.C.L is a CIHR research fellow.

Footnotes

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

References

  1. Aggarwal BB. Signalling pathways of the TNF superfamily: a double-edged sword. Nat Rev Immunol. 2003;3:745–756. doi: 10.1038/nri1184. [DOI] [PubMed] [Google Scholar]
  2. Ashall L, Horton CA, Nelson DE, Paszek P, Harper CV, Sillitoe K, Ryan S, Spiller DG, Unitt JF, Broomhead DS, et al. Pulsatile stimulation determines timing and specificity of NF-kappaB-dependent transcription. Science. 2009;324:242–246. doi: 10.1126/science.1164860. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Baeuerle PA, Baltimore D. I kappa B: a specific inhibitor of the NF-kappa B transcription factor. Science. 1988;242:540–546. doi: 10.1126/science.3140380. [DOI] [PubMed] [Google Scholar]
  4. Baldwin AS., Jr The NF-kappa B and I kappa B proteins: new discoveries and insights. Annu Rev Immunol. 1996;14:649–683. doi: 10.1146/annurev.immunol.14.1.649. [DOI] [PubMed] [Google Scholar]
  5. Bar-Even A, Paulsson J, Maheshri N, Carmi M, O’Shea E, Pilpel Y, Barkai N. Noise in protein expression scales with natural protein abundance. Nat Genet. 2006;38:636–643. doi: 10.1038/ng1807. [DOI] [PubMed] [Google Scholar]
  6. Barkett M, Gilmore TD. Control of apoptosis by Rel/NF-kappaB transcription factors. Oncogene. 1999;18:6910–6924. doi: 10.1038/sj.onc.1203238. [DOI] [PubMed] [Google Scholar]
  7. Bolte S, Cordelieres FP. A guided tour into subcellular colocalization analysis in light microscopy. J Microsc. 2006;224:213–232. doi: 10.1111/j.1365-2818.2006.01706.x. [DOI] [PubMed] [Google Scholar]
  8. Brown K, Park S, Kanno T, Franzoso G, Siebenlist U. Mutual regulation of the transcriptional activator NF-kappa B and its inhibitor, I kappa B-alpha. Proc Natl Acad Sci U S A. 1993;90:2532–2536. doi: 10.1073/pnas.90.6.2532. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Carpenter AE, Jones TR, Lamprecht MR, Clarke C, Kang IH, Friman O, Guertin DA, Chang JH, Lindquist RA, Moffat J, et al. CellProfiler: image analysis software for identifying and quantifying cell phenotypes. Genome Biol. 2006;7:R100. doi: 10.1186/gb-2006-7-10-r100. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Cheng CS, Feldman KE, Lee J, Verma S, Huang DB, Huynh K, Chang M, Ponomarenko JV, Sun SC, Benedict CA, et al. The specificity of innate immune responses is enforced by repression of interferon response elements by NF-kappaB p50. Sci Signal. 2011;4:ra11. doi: 10.1126/scisignal.2001501. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Cheong R, Rhee A, Wang CJ, Nemenman I, Levchenko A. Information transduction capacity of noisy biochemical signaling networks. Science. 2011;334:354–358. doi: 10.1126/science.1204553. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Cohen-Saidon C, Cohen AA, Sigal A, Liron Y, Alon U. Dynamics and variability of ERK2 response to EGF in individual living cells. Mol Cell. 2009;36:885–893. doi: 10.1016/j.molcel.2009.11.025. [DOI] [PubMed] [Google Scholar]
  13. Feinerman O, Veiga J, Dorfman JR, Germain RN, Altan-Bonnet G. Variability and robustness in T cell activation from regulated heterogeneity in protein levels. Science. 2008;321:1081–1084. doi: 10.1126/science.1158013. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Gaudet S, Spencer SL, Chen WW, Sorger PK. Exploring the contextual sensitivity of factors that determine cell-to-cell variability in receptor-mediated apoptosis. PLoS Comput Biol. 2012;8:e1002482. doi: 10.1371/journal.pcbi.1002482. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Ghosh S, Hayden MS. New regulators of NF-kappaB in inflammation. Nat Rev Immunol. 2008;8:837–848. doi: 10.1038/nri2423. [DOI] [PubMed] [Google Scholar]
  16. Goentoro L, Kirschner MW. Evidence that fold-change, and not absolute level, of beta-catenin dictates Wnt signaling. Mol Cell. 2009;36:872–884. doi: 10.1016/j.molcel.2009.11.017. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Goentoro L, Shoval O, Kirschner MW, Alon U. The incoherent feedforward loop can provide fold-change detection in gene regulation. Mol Cell. 2009;36:894–899. doi: 10.1016/j.molcel.2009.11.018. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Haskill S, Beg AA, Tompkins SM, Morris JS, Yurochko AD, Sampson-Johannes A, Mondal K, Ralph P, Baldwin AS., Jr Characterization of an immediate-early gene induced in adherent monocytes that encodes I kappa B-like activity. Cell. 1991;65:1281–1289. doi: 10.1016/0092-8674(91)90022-q. [DOI] [PubMed] [Google Scholar]
  19. Hayden MS, Ghosh S. Signaling to NF-kappaB. Genes Dev. 2004;18:2195–2224. doi: 10.1101/gad.1228704. [DOI] [PubMed] [Google Scholar]
  20. Hayden MS, Ghosh S. Shared principles in NF-kappaB signaling. Cell. 2008;132:344–362. doi: 10.1016/j.cell.2008.01.020. [DOI] [PubMed] [Google Scholar]
  21. Hoffmann A, Levchenko A, Scott ML, Baltimore D. The IkappaB-NF-kappaB signaling module: temporal control and selective gene activation. Science. 2002;298:1241–1245. doi: 10.1126/science.1071914. [DOI] [PubMed] [Google Scholar]
  22. Karin M, Lin A. NF-kappaB at the crossroads of life and death. Nat Immunol. 2002;3:221–227. doi: 10.1038/ni0302-221. [DOI] [PubMed] [Google Scholar]
  23. Krikos A, Laherty CD, Dixit VM. Transcriptional activation of the tumor necrosis factor alpha-inducible zinc finger protein, A20, is mediated by kappa B elements. J Biol Chem. 1992;267:17971–17976. [PubMed] [Google Scholar]
  24. Kunsch C, Rosen CA. NF-kappa B subunit-specific regulation of the interleukin-8 promoter. Mol Cell Biol. 1993;13:6137–6146. doi: 10.1128/mcb.13.10.6137. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Lewin SR, Lambert P, Deacon NJ, Mills J, Crowe SM. Constitutive expression of p50 homodimer in freshly isolated human monocytes decreases with in vitro and in vivo differentiation: a possible mechanism influencing human immunodeficiency virus replication in monocytes and mature macrophages. J Virol. 1997;71:2114–2119. doi: 10.1128/jvi.71.3.2114-2119.1997. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Lewis CE, Pollard JW. Distinct role of macrophages in different tumor microenvironments. Cancer Res. 2006;66:605–612. doi: 10.1158/0008-5472.CAN-05-4005. [DOI] [PubMed] [Google Scholar]
  27. Li YP, Schwartz RJ. TNF-alpha regulates early differentiation of C2C12 myoblasts in an autocrine fashion. FASEB J. 2001;15:1413–1415. doi: 10.1096/fj.00-0632fje. [DOI] [PubMed] [Google Scholar]
  28. Marienfeld R, May MJ, Berberich I, Serfling E, Ghosh S, Neumann M. RelB forms transcriptionally inactive complexes with RelA/p65. J Biol Chem. 2003;278:19852–19860. doi: 10.1074/jbc.M301945200. [DOI] [PubMed] [Google Scholar]
  29. Marx J. Cancer research. Inflammation and cancer: the link grows stronger. Science. 2004;306:966–968. doi: 10.1126/science.306.5698.966. [DOI] [PubMed] [Google Scholar]
  30. Micheau O, Lens S, Gaide O, Alevizopoulos K, Tschopp J. NF-kappaB signals induce the expression of c-FLIP. Mol Cell Biol. 2001;21:5299–5305. doi: 10.1128/MCB.21.16.5299-5305.2001. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Micheau O, Tschopp J. Induction of TNF receptor I-mediated apoptosis via two sequential signaling complexes. Cell. 2003;114:181–190. doi: 10.1016/s0092-8674(03)00521-x. [DOI] [PubMed] [Google Scholar]
  32. Nelson DE, Ihekwaba AE, Elliott M, Johnson JR, Gibney CA, Foreman BE, Nelson G, See V, Horton CA, Spiller DG, et al. Oscillations in NF-kappaB signaling control the dynamics of gene expression. Science. 2004;306:704–708. doi: 10.1126/science.1099962. [DOI] [PubMed] [Google Scholar]
  33. O’Dea E, Hoffmann A. The regulatory logic of the NF-kappaB signaling system. Cold Spring Harb Perspect Biol. 2010;2:a000216. doi: 10.1101/cshperspect.a000216. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Oeckinghaus A, Ghosh S. The NF-kappaB family of transcription factors and its regulation. Cold Spring Harb Perspect Biol. 2009;1:a000034. doi: 10.1101/cshperspect.a000034. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Pahl HL. Activators and target genes of Rel/NF-kappaB transcription factors. Oncogene. 1999;18:6853–6866. doi: 10.1038/sj.onc.1203239. [DOI] [PubMed] [Google Scholar]
  36. Phelps CB, Sengchanthalangsy LL, Malek S, Ghosh G. Mechanism of kappa B DNA binding by Rel/NF-kappa B dimers. J Biol Chem. 2000;275:24392–24399. doi: 10.1074/jbc.M003784200. [DOI] [PubMed] [Google Scholar]
  37. Raj A, van den Bogaard P, Rifkin SA, van Oudenaarden A, Tyagi S. Imaging individual mRNA molecules using multiple singly labeled probes. Nat Methods. 2008;5:877–879. doi: 10.1038/nmeth.1253. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Schottenfeld D, Beebe-Dimmer J. Chronic inflammation: a common and important factor in the pathogenesis of neoplasia. CA Cancer J Clin. 2006;56:69–83. doi: 10.3322/canjclin.56.2.69. [DOI] [PubMed] [Google Scholar]
  39. Scott ML, Fujita T, Liou HC, Nolan GP, Baltimore D. The p65 subunit of NF-kappa B regulates I kappa B by two distinct mechanisms. Genes Dev. 1993;7:1266–1276. doi: 10.1101/gad.7.7a.1266. [DOI] [PubMed] [Google Scholar]
  40. Siggers T, Chang AB, Teixeira A, Wong D, Williams KJ, Ahmed B, Ragoussis J, Udalova IA, Smale ST, Bulyk ML. Principles of dimer-specific gene regulation revealed by a comprehensive characterization of NF-kappaB family DNA binding. Nat Immunol. 2012;13:95–102. doi: 10.1038/ni.2151. [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Spencer SL, Gaudet S, Albeck JG, Burke JM, Sorger PK. Non-genetic origins of cell-to-cell variability in TRAIL-induced apoptosis. Nature. 2009;459:428–432. doi: 10.1038/nature08012. [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Tay S, Hughey JJ, Lee TK, Lipniacki T, Quake SR, Covert MW. Single-cell NF-kappaB dynamics reveal digital activation and analogue information processing. Nature. 2010;466:267–271. doi: 10.1038/nature09145. [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Ten RM, Paya CV, Israel N, Le Bail O, Mattei MG, Virelizier JL, Kourilsky P, Israel A. The characterization of the promoter of the gene encoding the p50 subunit of NF-kappa B indicates that it participates in its own regulation. EMBO J. 1992;11:195–203. doi: 10.1002/j.1460-2075.1992.tb05042.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Tian B, Nowak DE, Brasier AR. A TNF-induced gene expression program under oscillatory NF-kappaB control. BMC Genomics. 2005a;6:137. doi: 10.1186/1471-2164-6-137. [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. 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. 2005b;280:17435–17448. doi: 10.1074/jbc.M500437200. [DOI] [PubMed] [Google Scholar]
  46. Udalova IA, Richardson A, Denys A, Smith C, Ackerman H, Foxwell B, Kwiatkowski D. Functional consequences of a polymorphism affecting NF-kappaB p50-p50 binding to the TNF promoter region. Mol Cell Biol. 2000;20:9113–9119. doi: 10.1128/mcb.20.24.9113-9119.2000. [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Wajant H, Pfizenmaier K, Scheurich P. Tumor necrosis factor signaling. Cell Death Differ. 2003;10:45–65. doi: 10.1038/sj.cdd.4401189. [DOI] [PubMed] [Google Scholar]
  48. Wang VY, Huang W, Asagiri M, Spann N, Hoffmann A, Glass C, Ghosh G. The transcriptional specificity of NF-kappaB dimers is coded within the kappaB DNA response elements. Cell Rep. 2012;2:824–839. doi: 10.1016/j.celrep.2012.08.042. [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Wessells J, Baer M, Young HA, Claudio E, Brown K, Siebenlist U, Johnson PF. BCL-3 and NF-kappaB p50 attenuate lipopolysaccharide-induced inflammatory responses in macrophages. J Biol Chem. 2004;279:49995–50003. doi: 10.1074/jbc.M404246200. [DOI] [PubMed] [Google Scholar]
  50. Zhong H, May MJ, Jimi E, Ghosh S. The phosphorylation status of nuclear NF-kappa B determines its association with CBP/p300 or HDAC-1. Mol Cell. 2002;9:625–636. doi: 10.1016/s1097-2765(02)00477-x. [DOI] [PubMed] [Google Scholar]

Associated Data

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

Supplementary Materials

01
NIHMS567925-supplement-01.pdf (3.1MB, pdf)
02
Download video file (679.7KB, avi)
03
Download video file (7.1MB, avi)

RESOURCES