Many different physiological stimuli, including engagement of cytokine and T-cell receptors, DNA damage stress responses, and extracellular matrix/integrin attachment, activate a plethora of signaling cascades that mediate NF-κB transcription (22
). Despite the numerous avenues for NF-κB activation, many of these pathways converge on the IKK signalosome complex, indicating the importance of IKK activation for NF-κB transcription and biological responsiveness. NF-κB becomes activated through at least two different IKK signalosome complexes. In the classical NF-κB pathway, the IKK complex is composed of IKKα, IKKβ, and IKKγ, while the alternative NF-κB pathway involves the NIK-dependent activation of the IKKα homodimer (25
). In the classical pathway, IKKβ is critical for phosphorylation of IκBα and IκBβ, which governs nuclear translocation of the RelA/p65-p50 heterodimer. Unlike IKKβ, IKKα has distinct functions as a nuclear kinase capable of phosphorylating SRC-3 (66
), CBP (67
), histone H3, and the SMRT corepressor (4
), events linked to transcriptional activation of both NF-κB-regulated and non-NF-κB-regulated gene targets (4
Our laboratory recently demonstrated that IKKα is critical for derepressing NF-κB-regulated promoters by phosphorylating and inactivating the SMRT-HDAC3 repressor complex from the chromatin-bound p50 homodimer (26
). This event is critical for NF-κB transcription, because derepression of SMRT and HDAC3 relieves the basal repression complex, allowing recruitment of the transcriptionally competent p50-RelA/p65 heterodimer (Fig. ). Since IKKα has been shown to phosphorylate RelA/p65(S536) (30
), in the current study we examined whether IKKα potentiates NF-κB transcription at the chromatin level by phosphorylating both RelA/p65 and SMRT. This is an important question that examines whether IKKα controls NF-κB transcription by regulating derepression of the SMRT-HDAC3 complex from both basal and active components of NF-κB. Using ChIP and re-ChIP analysis, we found that RelA/p65 phosphorylation at S536 corresponds with IKKα activity on the cIAP-2
promoter upon attachment to laminin or following the addition of TNF-α. In an extended time course, our studies demonstrate that chromatin-bound IKKα but not IKKβ corresponds with RelA/p65(S536) phosphorylation during times of maximal NF-κB transcription. However, based on previous studies that support a role of IKKβ-induced phosphorylation of RelA/p65 (68
), we cannot exclude the possibility that IKKβ-mediated phosphorylation of RelA/p65(S536) occurs in the cytoplasm and is important during initial recruitment of RelA/p65 to chromatin. Importantly, we also found that IKKα mediates a similar phosphorylation event on the SMRT corepressor at S2410 when bound to the transcriptionally active RelA/p65 component of NF-κB. Re-ChIP analysis confirms that chromatin-bound IKKα coordinates the simultaneous phosphorylation of RelA/p65(S536) and SMRT(S2410). Although phosphorylated SMRT remains bound to RelA/p65, derepression of SMRT is evidenced by the loss of chromatin-associated HDAC3 activity. The resulting consequences are that phosphorylation of RelA/p65(S536) and SMRT(S2410) and the loss of HDAC3 activity occur prior to acetylation of RelA/p65 at K310. These results support the hypothesis that IKKα-induced phosphorylation of RelA/p65(S536) displaces corepressor activity, allowing p300 to induce acetylation of RelA/p65 (Fig. ). Simultaneous phosphorylation of RelA/p65 and SMRT is critical for NF-κB transcription, since disruption of IKKα activity blocked phosphorylation of these molecules, disrupted HDAC3 derepression, and prevented p300-mediated acetylation of RelA/p65. Our work indicates that IKKα-induced phosphorylation within the transactivation domain of RelA/p65(S536) deactivates SMRT-HDAC3 repressor complexes to allow RelA/p65 to become acetylated by p300.
FIG. 6. IKKα-mediated derepression of NF-κB-regulated genes occurs in distinct phases. Following cellular stimulation, IKKα is responsible for the removal of the SMRT-HDAC3 complex tethered by p50 homodimers. The SMRT-HDAC3 complex is (more ...)
Consistent with our results, Chen and colleagues recently demonstrate that IKK-mediated RelA/p65(S536) phosphorylation regulates K310 acetylation (17
). The authors provide evidence that the phosphorylation status of RelA/p65 at either S276 or S536 is required for recruitment of p300 and subsequent acetylation of RelA/p65(K310). Although we have not evaluated the phosphorylation status of RelA/p65(S276) in our model system, the use of nonphosphorylatable mutants of the RelA/p65(S336A) and SMRT(S2028,2410A) proteins or the inhibition of IKK activity results in active repression of NF-κB promoters by tethering the SMRT-HDAC3 complex. The interpretation of our results is similar to work published by Zhong and colleagues (72
), which demonstrated that phosphorylation of RelA/p65(S276) governs an exchange in which HDAC1 becomes derepressed and CBP/p300 is recruited to active RelA/p65. Therefore, in agreement with previously published work (17
), it is likely that phosphorylation of RelA/p65 at S276, S536, and possibly S311 and S529 activates RelA/p65 by first derepressing HDAC1 and SMRT-HDAC3 repression complexes, allowing subsequent recruitment of CBP/p300 and acetylation of RelA/p65(K310). This argument suggests that there are multiple phosphorylation sites within RelA/p65 that seem to function in a similar manner to mediate derepression of HDAC-containing complexes and recruitment of CBP/p300. This argument is supported by the fact that NF-κB is a ubiquitous transcription factor that must respond to many different physiological stimuli. Unlike NF-κB, many corepressors and HDAC enzymes display tissue-specific differences in protein expression (19
). Therefore, NF-κB has evolved such that different NF-κB stimuli activate selective combinations of serine/threonine kinases capable of phosphorylating RelA/p65 within the RHD at S276 or within the transactivation domain at S536 to derepress HDAC1 or SMRT-HDAC3 complexes. Derepression of HDAC1 or SMRT-HDAC3 from RelA/p65 has the same physiological response in the end: CBP/p300 recruitment and RelA/p65(K310) acetylation. However, even the specificities of these two HDAC complexes can be quite different. For example, our laboratory and others have found that HDAC1 directly interacts with RelA/p65 and that this occurs even in the absence of the SMRT or N-CoR corepressor (6
). Since HDAC1 is associated with many different corepressor complexes, including the Sin3-SAP, NURD, NURD-related, and CoREST-HDAC complexes (29
), it is possible that HDAC1 has the potential to directly recruit these corepressor complexes to NF-κB. In contrast, HDAC3 does not directly interact with RelA/p65 but rather is tethered by either SMRT or N-CoR (7
). However, unlike HDAC1, HDAC3 has the potential to recruit members of the class II HDACs to the repression complex (21
). Finally, not only are there multiple corepressor complexes within the same cell that bind and regulate NF-κB activity, but there seem to be promoter-specific preferences for certain repression complexes (7
). Therefore, derepression of various corepressor complexes associated with NF-κB probably depends on multiple parameters, including stimulus-dependent activation of serine/threonine kinases, promoter occupancy, and tissue specificity.
Although the current study and recently published work (17
) help to address the significance of RelA/p65 S536 phosphorylation, there are still many unanswered questions about SMRT derepression from active NF-κB transcriptional complex. Re-ChIP analysis indicates that phosphorylated SMRT(S2410) remains tethered to chromatin-bound phosphorylated RelA/p65(S536). Using in vitro kinase interaction assays, we demonstrated that phosphorylation of RelA/p65(S536) or RD I/II of SMRT(S2410) causes a disassociation between these two portions of the molecules. Since it is difficult to express full-length GST-SMRT proteins, we were unable to test whether a similar mechanism occurs with both of the full-length SMRT and RelA/p65 proteins. However, we have shown that RelA/p65 also interacts with SMRT in another region of the protein (amino acids 1031 to 1596) (Fig. ). IKKα is capable of phosphorylating SMRT within the same C-terminal domain that encompasses the RD I/II domain (26
). However, in in vitro kinase interaction assays, we were unable to displace the interaction of these two proteins (J. E. Hoberg and M. W. Mayo, data not shown). Based on this observation, we believe that endogenous phosphorylated SMRT may still interact with phosphorylated RelA/p65 through this domain.
Consistent with IKKα-induced phosphorylation and derepression of SMRT, HDAC3 is no longer associated with phosphorylated or acetylated RelA/p65. However, the preservation of the phosphorylated SMRT species on RelA/p65 at NF-κB-regulated promoters suggests either that this is a highly dynamic process where SMRT cycles between phosphorylation and derepression from NF-κB or that SMRT plays an ancillary role in the transcription process where it is bound in a derepressed state poised for reactivation. During our initial characterization of IKKα-mediated derepression of the SMRT corepressor complex (26
), we found that SMRT returns to chromatin prior to HDAC3 binding. Moreover, chromatin-associated SMRT is associated with the recruitment of the TCP-1 protein, a member of the TCP-1 ring complex (TRiC) responsible for loading HDAC3 onto SMRT (24
). These results suggest that the TRiC may assemble HDAC3 onto chromatin-bound SMRT. Therefore, it is possible either that the phosphorylated SMRT that we detect bound to RelA/p65 is not yet targeted for HDAC3 reloading by the TRiC or that IKKα-induced phosphorylation prevents the reestablishment of the repressor complex on NF-κB promoters.
In addition to HDAC3 loading issues, it is important to understand better how IKKα phosphorylation alters SMRT such that it no longer is capable of binding HDAC3. It is possible that IKKα-induced phosphorylation of RelA/p65 and potentially histone H3 alters the binding and/or conformation of the SMRT corepressor, thus potentiating derepression. SMRT contains two HDAC3 interaction domains. Perhaps the more important of these two regions resides in the SANT (SWI3/ADA2/N-CoR/TFIIB)-like domain of SMRT. SMRT contains a pair of SANT motifs; the N-terminal-most domain, called the deacetylase activation domain (DAD), which is required to bind and activate HDAC3 (23
), and the histone interaction domain (HID). The significance of the HID is that it has been shown to increase the enzymatic activity of HDCA3 by lowering the Km
for histone substrates (71
). Therefore, the DAD and HID motifs work together synergistically to promote acetylation of core histones. Interestingly, the HID binds preferentially to unacetylated histone substrates and is inhibited by acetylated histone substrates. Based on this understanding, the HID has been proposed to function in a “feed-forward” mechanism that may interpret the histone code (71
). The presence of the hydrophobic groove in the DAD is required to bind and activate HDAC3 in conjunction with the HID (18
); it is conceivable that posttranslational modifications to SMRT alter the structure of the SANT domains such that they can no longer recruit HDAC3 or bind core histones. This change in the structure of SMRT may also be regulated by the acetylation status of both RelA/p65 and histone tails associated with nucleosome surrounding the NF-κB-regulated promoter. This hypothesis is supported by several observations. First, the HID domain does not preferentially recognize acetylated core histones (71
), which would be predicted to affect greatly the ability of DAD to bind and activate HDAC3. Second, as expected, during full NF-κB-mediated transcription, histone H3 and H4 are hyperacetylated (4
), as is RelA/p65 (17 and Fig. ). Finally, in this study we provide evidence that IKKα-induced phosphorylation of SMRT within RD I/II displaces interaction with RelA/p65 (Fig. ). Together these observations provide support for the idea that phosphorylated SMRT and RelA/p65, as well as the acetylation of core histones, may change the secondary or tertiary structure of SMRT such that the DAD no longer binds and activates HDAC3. Work currently ongoing in the laboratory will address this hypothesis.
Finally, another interesting observation from our studies is that the RelA/p65 and IKKα proteins are not static but rather cycle on and off of chromatin during NF-κB transcription. It is well established that both NF-κB and SMRT cycle between the cytoplasm and the nucleus in response to physiological stimuli (15
). One potential explanation is that NF-κB becomes downregulated by the resynthesis of the IκBα proteins, which bind back to chromatin-associated RelA/p65 and facilitate the nuclear export of NF-κB (15
). Although it is clear that resynthesized IκBα does help redistribute NF-κB into the cytoplasm (16
), in this study chromatin-bound IκBα does not correspond with the cyclic phases of RelA/p65 binding to the cIAP-2
promoter. Perhaps a better explanation is that promoter-bound RelA/p65 is posttranslationally modified directly on chromatin and stimulated for degradation through a proteasome-dependent mechanism. RelA/p65 is known to be degraded through ubiquitin-dependent mechanisms (54
), and NF-κB transcription has been proposed to be terminated through ubiquitin-mediated targeting of chromatin-bound RelA/p65 (55
). The turnover of transcriptionally active NF-κB through a ubiquitin-dependent process is an attractive mechanism, since we have shown that SMRT-mediated turnover is TBL1/TBLR1/Ubc5 dependent (26
). Thus, it will be important to identify the posttranslational mechanisms governing chromatin-bound RelA/p65 and to elucidate how ubiquitin and/or ubiquitin-like processes regulate the turnover of RelA/p65 and IKKα to terminate NF-κB transcription.