A multitude of studies have established the critical role of NF-κB in the control of key physiological and pathological states, from immunity and inflammation to cancer1
. Regulation of the so-called ‘canonical’ NF-κB transcription factor by the IKK complex involves its cytosolic-to-nuclear translocation mediated by the phosphorylation and subsequent proteasomal degradation of the inhibitory molecule IκBα1
. However, in this issue of Nature Immunology
, Levy and co-workers describe a previously unknown mechanism in which basal and activated NF-κB is controlled through the methyltransferase SETD6–mediated methylation of Lys310 of RelA (also called p65), the transactivating subunit of the NF-κB complex2
. This event represses chromatin in the region of several NF-κB-dependent genes and ensures that they are not transcribed unless cells are activated by the appropriate stimuli that trigger phosphorylation of Ser311 by the atypical protein kinase PKC-ζ2
. These are very important observations, because they demonstrate a new layer of complexity in the regulation of NF-κB transcriptional activity through chromatin remodeling and establish PKC-ζ as a critical mediator of the chromatin changes necessary for effective transcription of κB-dependent genes ().
Figure 1 PKC-ζ-mediated phosphorylation of RelA regulates transcription. Under basal conditions, RelA (p65) is methylated (Me) at Lys310 by SETD6, promoting recruitment of GLP and methylation of histones, which leads to closed chromatin and inhibited transcription (more ...)
It is not surprising that the nuclear actions of NF-κB, a critical transcriptional regulator that controls the expression of thousands of genes, can be modulated and fine-tuned by mechanisms in addition to its all-or-nothing nuclear translocation. It is also known that even in the absence of stimuli, there are NF-κB subunits in the nucleus that are bound to gene-regulatory elements in the chromatin and that must be kept in check to prevent undesired, uncontrolled activity. Indeed, pioneering experiments established that dimers of the NF-κB subunit p50 associated with histone deacetylase sit on κB elements, which serves to deacetylate histones, therefore leading to closed, repressive chromatin that prevents gene expression under basal conditions3
. After degradation of IκBα, p50-RelA heterodimers are released from the inhibitory complex and RelA is phosphorylated at Ser276 by protein kinase A and/or mitogen- and stress-activated protein kinase, promoting interaction of RelA with the transcriptional coactivator CBP3
. This interaction results in more CBP-mediated acetylation of Lys310, which is apparently also important for transcriptional activation4
and for the acetylation of histones by CBP; this results in the generation of an open, permissive chromatin structure, leading to full transcriptional activity of the NF-κB complex, which replaces the inactive p50-p50 homodimers in the κB-regulatory elements3
. However, phosphorylation of RelA at Ser276 is not sufficient to promote binding of CBP, as Ser311 also needs to be phosphorylated in response to tumor necrosis factor (TNF)5
. This serine residue is specifically targeted by PKC-ζ, which has been shown through genetic manipulations to be required for full NF-κB transcriptional activity in vivo
and in cell culture6
(). Interestingly, Ser276 and Ser311 both reside in the Rel-homology-dimerization domain. Ser311 is in close proximity to Lys310, and it is unclear whether phosphorylation of Ser311 by PKC-ζ modulates the acetylation of Lys310. Nonetheless, Levy et al.
show here that it affects Lys310 methylation, which has an effect on the transcriptional activity of many NF-κB target genes2
. Their study stems from an unbiased screening of human protein lysine methyltransferases, in which they discover that one of these, SETD6, monomethylates chromatin-associated RelA at Lys310. A series of elegant biochemical experiments shows that methylated RelA resides in a histone H3–rich region near the promoters of several NF-κB target genes. This finding is important because it suggests that methylated RelA might have a repressive role in the control of gene expression under basal conditions. Indeed, knockdown of SETD6 gives rise to more κB-dependent transcription under basal conditions and, notably, under stimulated conditions as well. Thus, TNF activation correlates with lower abundance of methylated Lys310, which indicates that demethylation of this residue is a prerequisite for the expression of NF-κB target genes under activating conditions. The pathophysiological relevance of these findings is confirmed by experiments in which a lower abundance of SETD6 in transformed cells enhances their tumorigenic potential in vitro
and in vivo2
, consistent with a role for NF-κB in cancer7
. Furthermore, Levy et al.
show that blood mononuclear cells from patients with rheumatoid arthritis or juvenile idiopathic arthritis have much lower expression of SETD6
mRNA, which suggests the importance of this newly identified pathway in human inflammatory disease2
Once these authors establish the functional relevance of this previously unknown covalent modification of RelA, they go on to determine the precise mechanism whereby RelA methylated at Lys310 controls transcription2
. They again do an unbiased screen, this time looking for protein motifs that could potentially interact with RelA methylated at Lys310. They find a positive ‘hit’ in the ankyrin-repeat domain of GLP, which selectively interacts with this modified RelA with an affinity similar to that of the positive control, histone H3 methylated at Lys9 (ref. 2
). Interestingly, GLP, along with its partner G9a, are known to methylate histone H3 at Lys9 in chromatin regions with repressed transcription8
. Consistent with a role for this event in NF-κB function, chromatin-immunoprecipitation assays show that GLP and methylated histone H3 are lower in abundance at the promoters of canonical κB-responsive genes after stimulation with TNF and that knockdown of SETD6 results in less interaction of GLP with RelA2
. Collectively, all these results suggest a model whereby activation of the NF-κB signaling cascade leads to the release of GLP, which results in the opening of chromatin, allowing efficient activation of κB-dependent gene transcription.
The next big issue that remains to be addressed is identification of the precise signal that triggers the whole process. Here, PKC-ζ enters the picture; PKC-ζ, along with PKC-λ and PKC-ι, forms the atypical PKC subfamily. In contrast to the classical and novel PKC isoforms, proteins of the atypical PKC subfamily are insensitive to lipids and Ca2+
but interact with protein regulators and adaptors through a newly described PB1 protein-protein interaction domain9
. Levy et al.
reason that the close proximity of Ser311, a target site for PKC-ζ phosphorylation in response to TNF5
, to Lys310 might mean that methylation of RelA could be influenced by the PKC-ζ-mediated phosphorylation of Ser311 (ref. 2
). Interestingly, through biochemical and cellular studies of PKC-ζ-deficient cells, they find that phosphorylation of Ser311 by PKC-ζ in response to TNF stimulation does in fact lead to displacement of GLP from RelA and therefore to more transcription of κB-dependent genes. Consistent with published results6
, Levy et al.
find that PKC-ζ-deficient cells have lower production of inflammatory cytokines2
. Therefore, phosphorylation of RelA at Ser311 by PKC-ζ, in addition to favoring histone acetylation by allowing the recruitment of CBP, also inhibits histone methylation by preventing the interaction of GLP with RelA. This finding highlights the importance of the ‘hot spot’ at positions 310–311 and the relevance of PKC-ζ in its control (). Interestingly, the regulation of histone methylation by NF-κB has been shown before to also occur as a consequence of activation of this transcription factor. Studies have shown that JMJD3, which demethylates histone H3 at Lys27, is expressed in macrophages activated by lipopolysaccharide10
. It is believed that this enzyme, by erasing a critical histone mark, controls macrophage differentiation during chronic inflammation10
. Therefore, a model emerges whereby histone-methylation interaction in NF-κB signaling takes place by two mechanisms: one in which the synthesis of at least one histone demethylase is promoted, and another in which RelA itself is methylated, thus recruiting GLP, which is antagonized by TNF-activated PKC-ζ.
Like every breakthrough, these new findings solve many mysteries but also raise several questions. For example, what is the subunit composition of the NF-κB complex that contains methylated RelA at the chromatin? Is there a methylated RelA-RelA homodimer or is it in fact a p50-RelA heterodimer? Is this mechanism applicable to all κB-dependent genes or just those with a more closed promoter conformation? Published results suggest that PKC-ζ directly phosphorylates IKKβ in vitro11
and that PKC-ζ is required for activation of IKKβ in vivo6
. It will be important to define the cell types in which PKC-ζ acts upstream of IKKβ and those in which it controls RelA transcriptional activity in vivo
. Knock-in mouse lines expressing loss-of-function mutations of the sequence encoding Lys310 and Ser311 would help identify the physiological relevance of these modifications, as well as their importance in various diseases to which NF-κB has been linked. In this context, it will be of great interest to analyze samples from human patients with cancer to determine whether there are alterations in gene expression and/or mutations in the genes encoding SETD6 or GLP and whether there are changes in the acetylation or methylation of Lys310 and/or phosphorylation of Ser311. All these analyses would help establish the pathway identified here as a potential new source of therapeutic targets in the treatment of cancer and inflammatory diseases.