Duration of Nuclear NF-κB Action Regulated by Reversible Acetylation

See allHide authors and affiliations

Science  31 Aug 2001:
Vol. 293, Issue 5535, pp. 1653-1657
DOI: 10.1126/science.1062374


The nuclear expression and action of the nuclear factor kappa B (NF-κB) transcription factor requires signal-coupled phosphorylation and degradation of the IκB inhibitors, which normally bind and sequester this pleiotropically active factor in the cytoplasm. The subsequent molecular events that regulate the termination of nuclear NF-κB action remain poorly defined, although the activation of de novo IκBα gene expression by NF-κB likely plays a key role. Our studies now demonstrate that the RelA subunit of NF-κB is subject to inducible acetylation and that acetylated forms of RelA interact weakly, if at all, with IκBα. Acetylated RelA is subsequently deacetylated through a specific interaction with histone deacetylase 3 (HDAC3). This deacetylation reaction promotes effective binding to IκBα and leads in turn to IκBα-dependent nuclear export of the complex through a chromosomal region maintenance-1 (CRM-1)–dependent pathway. Deacetylation of RelA by HDAC3 thus acts as an intranuclear molecular switch that both controls the duration of the NF-κB transcriptional response and contributes to the replenishment of the depleted cytoplasmic pool of latent NF-κB–IκBα complexes.

NF-κB corresponds to an inducible transcription factor complex that plays a pivotal role in regulating the inflammatory, immune, and anti-apoptotic responses in mammals (1, 2). The prototypical NF-κΒ complex, which corresponds to a heterodimer of p50 and RelA subunits, is sequestered in the cytoplasm by its assembly with a family of inhibitory proteins termed the IκBs (1). Stimulus-induced phosphorylation of two NH2-terminal serines in the IκBs, mediated by a macromolecular IκB kinase complex (IKK) (3), triggers the rapid ubiquitination and subsequent degradation of this inhibitor by the 26S proteasome complex. The liberated NF-κΒ heterodimer rapidly translocates into the nucleus, where it engages cognate κΒ enhancer elements and alters gene expression. The NF-κB complex also recruits the p300/CBP and P/CAF coactivators, which participate in the activation of target gene transcription (4–6). Phosphorylation of an NH2-terminal site in RelA by protein kinase A facilitates NF-κB assembly with CBP/p300 (7). Both CBP and p300 contain histone acetyltransferase activity that has been implicated in the regulation of gene expression. These effects involve acetylation of core histones leading to changes in chromatin structure (8–13) as well as direct acetylation of select host transcription factors like p53, GATA-1, and E2F, which alters their biological function (9,10).

Although much progress has been made in understanding the biochemical events that underlie NF-κB induction, much less is known about how activated nuclear forms of NF-κB are negatively regulated, ensuring a transient transcriptional response. Prior studies have demonstrated that the IκBα gene is induced by NF-κB (14–16) and that the de novo expression of IκBα proteins, which display nucleocytoplasmic shuttling properties, participates in a negative feedback system limiting the NF-κB transcriptional response (17, 18). We now demonstrate that nuclear RelA is subject to reversible acetylation and that this posttranslational modification plays a pivotal role in NF-κB regulation by governing IκBα binding to RelA and the nuclear export of the NF-κB complex.

We first observed that trichostatin A (TSA), a specific inhibitor of the multiple histone deacetylases (HDACs) (11), enhanced tumor necrosis factor (TNF)-α–induced, but not basal, expression of a κB-luciferase reporter gene (Fig. 1A). In addition, TSA enhanced nuclear NF-κB DNA binding after TNF-α stimulation but did not itself stimulate NF-κB binding (Fig. 1B). TSA also did not alter the binding of the constitutively expressed Sp1 transcription factor in either the presence or absence of TNF-α (Fig. 1B, lower panel). Immunoblotting of the nuclear extracts with antibodies to RelA (anti-RelA) revealed sustained nuclear RelA expression in the presence of TSA (Fig. 1C). Together, these results indicate that TSA enhances TNF-α–induced NF-κB DNA binding, likely by prolonging intranuclear expression of RelA.

Figure 1

(A) TSA enhances TNF-α–mediated activation of κΒ-luciferase gene expression. 293T cells were transfected with κB-luciferase reporter plasmid DNA (0.1 μg) and luciferase activity was measured (19) after treatment with TSA and TNF-α (10 ng/ml). Results represent cumulated data from three independent transfections. (B) TSA enhances TNF-α induction of nuclear NF-κB DNA-binding activity. NF-κB DNA-binding activity was assessed in electrophoretic mobility shift assays (EMSA) with a32P-radiolabeled consensus κB enhancer oligonucleotide (5′- AGTTGAGGGGACTTTCCCAGGC-3′) (upper panel). Supershifting this complex with anti-RelA (SC-109, Santa Cruz Biotechnology, Santa Cruz, CA) and anti-p50 (SC-1190, Santa Cruz) is shown in lanes 7 and 8. Comparability of the various nuclear extracts was assessed by EMSA with a32P-radiolabeled Sp1 probe (Promega) (lower panel). (C) RelA expression is sustained in the nuclei of TNF-α–stimulated cells in the presence of TSA. HeLa cells were incubated in the presence (lanes 7 to 12) or absence (lanes 1 to 6) of TSA (400 nM) for 1 hour before stimulation with TNF-α (20 ng/ml) for 30 min. The cells were chased for the indicated time periods with medium alone (lanes 1 to 6) or medium containing TSA (400 nM) (lanes 7 to 12). Nuclear extracts from each culture were immunoblotted with anti-RelA and visualized with an enhanced chemiluminescence reagent (Amersham). (D) HDAC3, but not HDAC1, 2, 4, 5, or 6, inhibits TNF-α activation of κB-luciferase activity. Expression levels of each HDAC determined by anti-Flag immunoblotting of the cell lysates are shown in the insets.

TSA broadly inhibits the action of the HDACs. HDACs function not only to deacetylate core histones leading to repressive changes in chromatin structure but also to deacetylate various host transcription factors, altering their transcriptional activity (10–13). We assessed potential inhibitory effects of various HDACs on NF-κB action by cotransfecting 293T cells with expression vectors encoding HDAC1, 2, 3, 4, 5, or 6 and a κB-luciferase reporter followed by stimulation with TNF-α for 5 hours (Fig. 1D). Although each of the HDACs was comparably expressed, only HDAC3 inhibited TNF-α–induced κB-luciferase activity (lanes 7 and 8). This inhibitory effect of HDAC3 was abrogated in the presence of TSA, indicating that the deacetylase function of HDAC3 was required for these biological effects (Web fig. 1) (19). Consistent with this action of HDAC3, we found that the RelA subunit of NF-κB is acetylated in vivo (Fig. 2A, lane 2, upper panel). RelA acetylation was abolished in the presence of HDAC3 (lane 5) but not HDAC1 (lane 3). The addition of TSA to the HDAC3-expressing cultures was associated with the detection of acetylated RelA (lane 4). When HeLa cells were incubated in [3H]sodium acetate and stimulated with TNF-α or medium for 30 min, signal-induced acetylation of endogenous RelA was similarly detected (Fig. 2B). Together, these studies reveal that NF-κB is subject to reversible acetylation and that HDAC3 plays a central role in its deacetylation.

Figure 2

(A) RelA is acetylated in vivo.COS-7 cells were cotransfected with expression vectors encoding T7-RelA and HDAC1 or HDAC3. Acetylation levels of RelA determined by [3H]sodium acetate labeling (19) are shown in the upper panel. The total amount of T7-RelA in the individual samples is shown in the lower panel. (B) Endogenous RelA is inducibly acetylated after TNF-α stimulation. Acetylation of endogenous RelA was assessed by [3H]sodium acetate labeling (19) and is shown in the upper panel. The level of RelA present in each of the two immunoprecipitates is shown in the lower panel. (C) RelA is inducibly acetylated by p300 in vivo. 293T cells were cotransfected with T7-RelA and IκBα in the presence or absence of expression plasmids encoding p300. Cell cultures were stimulated with TNF-α for the indicated time periods, and acetylation levels of RelA were analyzed by immunoblotting with antibody to acetylated lysine (Cell Signaling, Beverly, MA). The level of T7-RelA present in each of the immunoprecipitates is also shown in the lower panel.

The potential participation of the p300, CBP, and P/CAF acetyltransferases in the acetylation of RelA was next investigated. When 293T cells were cotransfected with RelA, IκBα, and p300, and stimulated with TNF-α, inducible acetylation of RelA was detected (Fig. 2C). Coexpression of either p300 or CBP, but not P/CAF, with RelA produced a dose-dependent acetylation of T7-RelA [Web fig. 2 (19)]. Together, these findings demonstrate that endogenous RelA is acetylated in a signal-coupled manner likely mediated by p300 and CBP.

Coimmunoprecipitation of T7-RelA and HDAC3 in COS-7 cells suggests a physical interaction between these two proteins in vivo (Fig. 3A). In contrast, no interaction between RelA and HDAC1 was detected despite comparable levels of both HDACs being expressed (Fig. 3A). Moreover, anti-HDAC3 effectively coimmunoprecipitated endogenous RelA from HeLa cells, whereas only trace amounts of RelA immunoprecipitated with anti-HDAC1 (Fig. 3B). By using coimmunoprecipitation assays and the mammalian two-hybrid system (20), we found that NH2-terminal regions of both HDAC3 and RelA are required for the assembly of these two proteins in vivo [Web fig. 3) (19)].

Figure 3

(A) HDAC3 and RelA physically assemble in vivo. COS-7 cells were transfected with the indicated combinations of cDNAs encoding Flag epitope–tagged HDAC1 or HDAC3 and T7-RelA, and anti-T7 immunoprecipitations were performed followed by anti-Flag immunoblotting (19). Levels of expression of HDAC1-Flag, HDAC3-Flag, and T7-RelA are shown in the lower two panels. (B) The physical association of endogenous RelA and endogenous HDAC3 in HeLa cells. Whole-cell lysates from HeLa cells were immunoprecipitated with polyclonal anti-HDAC1 or anti-HDAC3 followed by immunoblotting with monoclonal anti-RelA (F-6, Santa Cruz). (C) HDAC3 inhibits TNF-α–induced nuclear expression of RelA. HeLa cells were transfected with control (lanes 1 to 4) or HDAC3-expression vector DNA and cultured for 24 hours, followed by stimulation with TNF-α (10 ng/ml) for 0, 10, 30, or 90 min. Nuclear extracts were prepared and EMSAs performed as described in Fig. 1B (panels 1 and 2). TNF-α–induced degradation of IκΒα was analyzed in the cytoplasmic extracts of these cultures by immunoblotting with anti-IκBα (C-21, Santa Cruz) (panel 3). Levels of nuclear RelA were assessed by immunoblotting nuclear extracts with specific anti-RelA (panel 4). The faster migrating bands reactive with the anti-RelA likely correspond to RelA degradation products and show similar changes in levels of nuclear expression. (D) HDAC3 stimulates nuclear export of RelA via a leptomycin B–sensitive pathway. GFP-RelA principally localized in the nucleus when expression plasmid DNA was transfected into HeLa cells. Coexpression of HDAC3 and GFP-RelA resulted in a cytoplasmic expression pattern of GFP-RelA; HDAC3 and GFP-RelA plasmid DNA was transfected at a 6:1 ratio to ensure that all GFP-RelA–expressing cells also contained HDAC3. Coexpression of HDAC1 and GFP-RelA did not alter the nuclear pattern of GFP-RelA epifluorescence. Treatment of the GFP-RelA and HDAC3–expressing cell cultures with TSA (800 nM, 5 hours) resulted in a nuclear pattern of expression of GFP-RelA. Treatment of GFP-RelA and HDAC3–expressing cultures with leptomycin B (LMB; 20 nM, 2 hours) produced a nuclear pattern of GFP-RelA epifluorescence. Average percentages of cells displaying the depicted phenotype are shown in the lower right corner of each panel. Results are derived from inspection of at least 200 transfected cells present in multiple microscopic fields from two independent experiments.

Studies were next performed to define the molecular basis for HDAC3-mediated inhibition of TNF-α activation of NF-κB. Expression of HDAC3 in HeLa cells (∼50% transfection efficiency) diminished both NF-κB DNA-binding activity (Fig. 3C, panel 1) and levels of nuclear RelA (panel 4). In contrast, HDAC3 expression did not alter TNF-α–induced degradation of IκBα occurring in the cytoplasm (panel 3), nor did it markedly change the levels of Sp1 DNA-binding activity (panel 2). Together with the results presented in Fig. 1C, these findings raise the possibility that HDAC3 may regulate the intracellular trafficking of RelA. A precedent for acetylation influencing the intracellular trafficking of a transcription factor is provided by recent studies on hepatocyte nuclear factor-4 and class II transactivator (CIITA) (21, 22). To monitor potential effects of HDAC3 on the intracellular trafficking of RelA, we expressed green fluorescent protein–RelA fusion proteins (GFP-RelA) in HeLa cells in the presence and absence of HDAC3 (Fig. 3D). When expressed alone or with HDAC1, GFP-RelA localized to the nucleus, whereas the coexpression of HDAC3 resulted in a predominantly cytoplasmic pattern of GFP-RelA expression. This cytoplasmic relocalization of GFP-RelA induced by HDAC3 did not occur in the presence of TSA. These effects of HDAC3 appeared to result from nuclear export of GFP-RelA because the addition of leptomycin B, a known inhibitor of CRM-1/exportin-1–dependent nuclear export (23), preserved the nuclear pattern of GFP-RelA epifluorescence. These findings suggest that HDAC3-mediated deacetylation of RelA promotes its export from the nucleus to the cytoplasm.

Among the target genes activated by NF-κB is the IκBα gene (14–16). The resultant de novo synthesis of IκBα serves to replenish the intracellular stores of this inhibitor depleted during the course of NF-κB activation. IκBα also displays nucleocytoplasmic shuttling properties and likely retrieves nuclear NF-κB complexes, thereby contributing to the termination of the NF-κB transcriptional response (17,18). We next examined whether the acetylation status of RelA regulates its assembly with IκBα. The acetylation of RelA induced by the coexpression of increasing amounts of p300 was associated with markedly diminished binding of RelA to GST-IκBα matrices (Fig. 4A, lanes 3 and 4). However, the coexpression of increasing amounts of HDAC3 in the presence of p300 restored RelA binding to GST-IκBα (Fig. 4A, lanes 5 and 6). Immunoblotting of the T7-RelA proteins with antibodies to acetylated lysine confirmed dose-related increases in RelA acetylation by p300 in lanes 3 and 4 and dose-related deacetylation of RelA by HDAC3 in lanes 5 and 6 (24). These findings suggest that acetylation of RelA prevents IκBα binding, whereas deacetylation of RelA by HDAC3 stimulates IκBα binding.

Figure 4

(A) Deacetylated RelA displays greater IκBα-binding activity than acetylated RelA. 293T cells were cotransfected with T7-RelA and p300 in the presence or absence of HDAC3 expression vector DNA as indicated. Twenty-four hours later, 50 μl of each whole-cell lysate was incubated with GST-IκBα (1.0 μg). The levels of RelA captured by the GST-IκBα matrix under each condition were assessed by immunoblotting with anti-T7 (upper panel). The total amounts of GST-IκBα and RelA present in the reaction mixtures are shown in the lower two panels. (B) HDAC3 does not stimulate GFP-RelA nuclear export in IκBα-deficient MEFs. MEF cells from wild-type and IκBα–/– mice were transfected with GFP-RelA (0.3 μg) (panels a and c) or GFP-RelA (0.3 μg) and HDAC3 (1.7 μg) (panels b and d) expression vector DNA. The IκBα−/− MEFs were also reconstituted with limiting amounts of IκBα expression vector (0.1 μg) in the absence and presence of HDAC3 (panels e and f). Average percentages of cells displaying the depicted phenotype derived from inspection of at least 120 cells present in multiple microscopic fields from two independent experiments are shown in the lower right corner of each panel. (C) Schematic model for the role of HDAC3- mediated deacetylation of RelA as an intranuclear molecular switch promoting IκBα binding and IκBα-dependent nuclear export of the NF-κB complex. This deacetylation-controlled response both leads to the termination of the NF-κB transcriptional response and aids in reestablishing latent cytoplasmic forms of NF-κB bound to IκBα, thereby preparing the cell to respond to the next NF-κB–inducing stimulus.

To test whether the HDAC3-induced nuclear export of RelA is dependent on IκBα, we studied the subcellular localization of GFP-RelA in murine embryo fibroblasts (MEFs) isolated from wild-type or IκBα–/– mice produced by targeted gene disruption (25) (Fig. 4B). In wild-type MEFs, GFP-RelA was principally expressed in the nucleus (panel a), whereas the coexpression of HDAC3 induced a cytoplasmic pattern of GFP-RelA epifluorescence (panel b). However, a very different pattern was obtained in the IκBα–/– MEFs. Whereas GFP-RelA also exhibited a nuclear pattern of epifluorescence, the coexpression of HDAC3 in these IκBα–/– MEFs failed to induce a cytoplasmic redistribution of RelA (panel d). Reconstitution of these IκBα–/– cells by transfection with small quantities of an IκBα expression vector restored HDAC3-induced cytoplasmic expression of the GFP-RelA protein (panel f). In the absence of HDAC3, GFP-RelA remained principally nuclear, indicating that the levels of IκBα expressed were not sufficient on their own to produce cytoplasmic sequestration of GFP-RelA in these IκBα–/– MEFs (panel e). These results indicate that IκBα is required for the nuclear export of deacetylated forms of RelA, which display increased binding of IκBα.

These findings reveal a new mechanism through which nuclear NF-κB function is regulated (Fig. 4C). RelA is subject to stimulus-coupled acetylation likely mediated through the p300 and CBP coactivators. One biological consequence of this modification is that acetylated RelA becomes a very poor substrate for binding by newly synthesized IκBα. Whether p50 or perhaps IκBα are similarly subject to biologically important acetylation/deacetylation reactions remains to be explored. Our studies identify acetylated RelA as a novel nonhistone substrate of HDAC3. As such, HDAC3-mediated deacetylation functions as an intranuclear molecular switch that when activated initiates a series of events culminating in the termination of the NF-κB transcriptional response. The IκBα-dependent nuclear export of the HDAC3-deacetylated RelA-containing complexes also serves to replenish the depleted cytoplasmic pool of latent NF-κB–IκBα complexes needed for the next inductive NF-κB response in these cells.

  • * To whom correspondence should be addressed. E-mail: wgreene{at}


Stay Connected to Science

Navigate This Article