IKKα and IKKβ Each Function to Regulate NF-κB Activation in the TNF-Induced/Canonical Pathway

Background Activation of the transcription factor NF-κB by cytokines is rapid, mediated through the activation of the IKK complex with subsequent phosphorylation and degradation of the inhibitory IκB proteins. The IKK complex is comprised of two catalytic subunits, IKKα and IKKβ, and a regulatory protein known as NEMO. Using cells from mice that are genetically deficient in IKKβ or IKKα, or using a kinase inactive mutant of IKKβ, it has been proposed that IKKβ is critical for TNF-induced IκB phosphorylation/degradation through the canonical pathway while IKKα has been shown to be involved in the non-canonical pathway for NF-κB activation. These conclusions have led to a focus on development of IKKβ inhibitors for potential use in inflammatory disorders and cancer. Methodology Analysis of NF-κB activation in response to TNF in MEFs reveals that IKKβ is essential for efficient phosphorylation and subsequent degradation of IκBα, yet IKKα contributes to the NF-κB activation response in these cells as measured via DNA binding assays. In HeLa cells, both IKKα and IKKβ contribute to IκBα phosphorylation and NF-κB activation. A kinase inactive mutant of IKKβ, which has been used as evidence for the critical importance of IKKβ in TNF-induced signaling, blocks activation of NF-κB induced by IKKα, even in cells that are deficient in IKKβ. Conclusions These results demonstrate the importance of IKKα in canonical NF-κB activation, downstream of cytokine treatment of cells. The experiments suggest that IKKα will be a therapeutic target in inflammatory disorders.

IkB phosphorylation by the high molecular weight IkB kinase (IKK) complex (approximately 700 kDa) is a critical regulatory step in the NF-kB activation pathway [1][2][3][4][5]. This kinase complex was partially identified initially in unstimulated Hela cells and was later found to be activated in cells treated with TNFa [6]. Subsequently several groups identified two highly related kinases named IKK1/IKKa and IKK2/IKKb as the catalytic components of this complex [6][7][8]. Both of these kinases have been shown to have specificity for serines 32 and 36 in the N-terminus of IkBa with phosphorylation leading to ubiquitination and degradation of this inhibitory protein [9]. In addition to IKKa and IKKb, a noncatalytic, regulatory component of IKK was also identified and called NF-kB Essential modifier (NEMO) or IKKc [10,11]. Additionally, it has been reported that both IKKa and IKKb can phosphorylate the RelA/p65 subunit to promote transactivation potential [12].
Insight into the physiological roles of the two catalytic IKK subunits comes from gene targeting studies. IKKb knockout mice display a phenotype similar or identical to knockout of RelA, namely embryonic lethal with severe liver apoptosis [13][14][15]. A similar phenotype was seen in the NEMO/IKKc knockout animal [16]. Mouse embryonic fibroblast cells that were isolated from IKKb deficient embryos showed a marked reduction in TNFaand interleukin-1alpha-induced NF-kB activity, as measured by EMSA and by effects on IkB degradation. The IKKb 2/2 knockout cells exhibit significantly enhanced apoptosis in response to TNFa [13][14][15]. Importantly, IKK activity directed to phosphorylation of IkB in vitro was essentially lost in IKKb null cells [13][14][15]. A role of IKKa in classical NF-kB signaling is less clear compared to IKKb. IKKa deficient mice exhibit abnormal morphogenesis and developmental defects [17][18][19]. Consistent with conclusions derived using IKKb 2/2 fibroblasts, IKKa does not seem to have a significant influence on cytokine-induced IKK activity directed to IkBa [17,18]. However, IKKa-deficient mouse embryonic fibroblast (MEF) cells exhibited reduced NF-kB activation as measured by EMSA in response to cytokine treatment [17,18]. Another group did not find reduced cytokineinduced NF-kB DNA binding activity in IKKa 2/2 MEFs [19]. In the light of these genetic studies and additional biochemical studies, it has been generally assumed that IKKb but not IKKa is the primary regulator of NF-kB dependent proinflammatory signal transduction [1][2][3][4][5]. On the other hand, IKKa is known to be essential in non-canonical NF-kB activation by regulating p100 precursor processing and activation of the p52/RelB heterodimer [1][2][3][4][5]. Recently, we and others have demonstrated that IKKa has an important nuclear function by regulating the control of target genes at the level of histone phosphorylation [20,21]. Interestingly, the observation that hepatocyte-specific ablation of IKKb did not lead to impaired activation of NF-kB by TNF as measured by gel shift assay and IkB degradation [22] suggests the involvement of another kinase in the canonical pathway at least in adult hepatocytes. Here we have explored individual roles of IKKa and IKKb in canonical NF-kB activation in MEF cells as well as cancer cells. Our results suggest that IKKa, like IKKb, is critical for efficient cytokine-induced NF-kB activation. In fibroblasts IKKa is not significantly involved in IkBa phosphorylation/ degradation, yet contributes to activation of NF-kB through an unknown mechanism in these cells. In HeLa cells, IKKa and IKKb each contribute to IKK activity directed to IkBa to control its phosphorylation and subsequent degradation. Expression of a kinase inactive variant of IKKb, which has been used previously to provide evidence for the importance of IKKb in the canonical pathway, is shown here to block IKKa activity. These studies suggest that inhibition of IKKa is a rational approach in blocking inflammatory disorders.

Reagents and Materials
Mouse embryonic fibroblast (MEF) and HeLa cells were cultured in Dulbecco's modified Eagle's medium (DMEM), complemented with 10% fetal calf serum (FCS), 100 units/ml penicillin, 100 mg/ml streptomycin. SKBr3 cells were cultured in McCoy's 5A medium complemented with 10% fetal calf serum (FCS), 100 units/ml penicillin and 100 mg/ml streptomycin. Wild type, IKKa, IKKb single and IKKa/b double knockout cells (DKO) were the kind gift from Dr. Inder Verma. Antibodies to phospho-specific NF-kB p65 (Ser-536) and IkBa (Ser 32/36) were obtained from Cell Signaling. Antibodies to b-tubulin and to IkBa were obtained from Santa Cruz. Antibodies to IKKa and IKKb were obtained from Upstate Biotechnology Inc. RhTNF-a (Promega) was used at a final concentration of 10 ng/ml.

Western Blot
After stimulation, cultured cells were lysed on ice for 5 min in RIPA lysis buffer with freshly added protease and phosphatase inhibitor cocktails. Lysates were cleared by centrifugation at 4 uC for 15 min at 13,000 g. The amount of total protein was measured and equal amounts (20 mg) were fractionated by NuPAGE Novex 4-12% Bis Tris gels (Invitrogen) and electro-transferred to polyvinylidene difluoride membranes. Membranes were blotted with the indicated antibodies, and proteins were detected using an enhanced chemiluminescence detection system (Amersham Biosciences, Freiburg, Germany). Where indicated, membranes were stripped and re-probed with the indicated antibody.

Electrophoretic Mobility Shift Assay (EMSA)
EMSAs were performed as previously described [23]. Briefly, 425 mg of nuclear extracts, prepared following cell stimulation, were incubated with a radiolabeled DNA probe containing an NF-kB consensus site. For supershifts, 1 ml of anti-p65 antibody (Rockland) or 2 ml of anti-p50 antibody (Santa Cruz, SC-7178) was added and the binding reaction was allowed to proceed for an additional 15 min. Protein2DNA complexes were resolved on a non-denaturing polyacrylamide gel and visualized by autoradiography.

siRNA Knockdown Experiments
IKKa and IKKb mRNA were knocked down with siRNA obtained from Dharmacon. Dharmafect 1 (Dharmacon Company) transfection reagent was used for all si-RNA transfection as described in the manufacturer's protocol. SiRNA was transfected for 72 hrs, and lysate preparation and westerns were performed as described [24].

Luciferase Assays
SKBr3 cells stably expressing the 3x-kB plasmid were plated in equal number in triplicate in 24-well plates and transfected with siRNA for 72 hours. Cells were lysed in MPER and luciferase activity was measured with Promega Luciferase Assay System (Promega). Luciferase levels were normalized by protein concentration using a Bradford assay. MEF cells were seeded in 24-well plates at 30-50% density and transfected the next day with the indicated expression vectors and 3x-kB Luciferase reporter gene for 48 h using Effectene (Qiagen) transfection reagent according to the manufacturer's instruction. b-galactosidase reporter gene was used as an internal control. The total amount of transfected DNA (500 ng of DNA) in each well was adjusted by adding empty plasmid vector (pcDNA3.1). Where indicated, 100 ng and 200 ng of IKKb KM vector has been used. Luciferase activity of whole cell lysates was measured by using a luciferase assay kit (Promega). b-galactosidase activity was measured by liquid -galactosidase assay with chlorophenolred-b-D-galactopyranoside substrate. Relative luciferase activity was calculated by normalizing the assay results to b-galactosidase expression values. Luciferase-fold induction was calculated by normalizing the results to control treatment, which was assumed as 1-fold induction. HeLa cells, seeded in 24-well plates were transiently transfected with the indicated siRNAs for 48 hr. Media was then replaced and cells were further transfected with the NF-kB response 3X-kB luciferase reporter and a control Renilla luciferase construct. 24 hr later, cells were lysed and dual luciferase assays were performed. Luciferase readings in untreated and control vector transfected cells were normalized to 1.

TNF-Induced NF-kB Activity Is Diminished in IKKa as Well as IKKb Deficient MEF Cells
Experiments were initiated to assay roles of IKKa and IKKb in inducing IkBa phosphorylation and activation of NF-kB in response to a well-studied NF-kB inducer, TNFa. For this purpose, mouse embryonic fibroblast cells (MEFs) that are deficient for IKKa or for IKKb singly as well as IKKa/b double knock-out cells have been utilized. As shown in Figure 1, TNFa induces expected p65 phosphorylation at Ser 536 position as well as IkBa degradation in as early as 5 minutes post-stimulation. Importantly, there is diminished p65 phosphorylation in both IKKa and IKKb deficient MEF cells. Interestingly, lack of IKKa delayed IkBa degradation while lack of IKKb significantly suppressed the TNF-induced degradation of IkBa. Relative to the IKKb deficient cells, IkBa appears weakly degraded at the 30 minute time point (Fig. 1) but by 60 minutes these levels return (data not shown, and also see ref. 15). IKKa/b DKO MEFs have near complete loss of p65 phosphorylation and IkBa degradation, as expected (note lower levels of IkBa in these cells, indicating significantly reduced NF-kB-dependent transcription of its inhibitor). These results (Figure 1) demonstrate that both IKKa and IKKb are required for efficient NF-kB activation in MEFs as measured by p65 phosphorylation at Ser536, yet IKKb appears to be significantly more important in the IkBa phosphorylation/ degradation response in fibroblasts. This work is consistent with previous work [13][14][15][17][18][19] which showed that IKK in vitro activity, directed to recombinant IkBa, is not diminished in IKK1 (IKKa) null MEFs, yet is significantly reduced in IKK2/IKKb null cells.

NF-kB DNA Binding Activity Is Diminished in both IKKa and IKKb Deficient MEF Cells
In addition to IkBa degradation and p65 phosphorylation, we have studied whether IKKa and IKKb differentially affect NF-kB DNA binding activity in response to TNF as measured by EMSA/ gel shift assay. NF-kB DNA binding activity was investigated in WT, IKKa 2/2, and IKKb 2/2 cells. As shown in Figure 2, there is significant induction of NF-kB (p50/p65) DNA binding activity in response to TNFa in WT MEF cells. However, this DNA binding activity is diminished in both IKKa and IKKb deficient cells. The level and the kinetics of NF-kB DNA binding activity is comparable in IKKa and IKKb deficient MEFs cells. This data suggests that IKKa, as well IKKb, is essential for optimal NF-kB DNA binding activity, potentially through different mechanisms (see Discussion). Promoter studies (see below) confirm a functional role for IKKa in TNF-induced NF-kB activation in MEF cells.

Similar Roles for IKKa and IKKb in Response to TNFa Induced NF-kB Activation in Hela Cells
Most studies regarding the roles of IKKa and IKKb have been performed in MEFs null for either subunit. To expand these studies, we have analyzed the differential roles of IKKa and IKKb in response to TNF in Hela cells (Figure 3). For this purpose we have utilized siRNA to knockdown IKKa, IKKb and IKKa and IKKb together in Hela cells. After 3 days of siRNA transfection, knockdown of IKKa and of IKKb was highly effective. HeLa cells were then treated with TNFa for the indicated times and NF-kB activity was examined through analysis of IkB phosphorylation and degradation. Importantly, kinetics of IkB phosphorylation and degradation in IKKa and IKKb knock-down cells are both impaired compared to control siRNA treated cells (Figure 3). For instance, 5 min after TNF treatment, there is significant degradation of IkBa in the control cells, while there is little or no loss at that time point in the IKKa or IKKb knocked-down cells. Additionally, phosphorylation of IkBa is reduced in the IKKa and IKKb knockdown cells, which is more dramatic given that there are elevated levels of IkBa in these cells at the 5 minute time point. Degradation of IkBa is nearly lost with doubleknockdown ( Figure 3). To determine the individual roles of IKKa and IKKb in regulating NF-kB transcriptional activity, knockdown experiments in HeLa cells were combined with transfection of an NF-kB-dependent luciferase reporter (Figure 4). In response to TNF treatment, IKKa and IKKb each contribute to NF-kB transcriptional activity as measured through reporter assays ( Figure 4). These results indicate that IKKa contributes signifi-

Knockdown of IKKa or IKKb Diminish TNF-Induced NF-kB Activity in Breast Cancer Cells
To further analyze the roles of individual IKK kinases on NF-kB activity and to analyze another cell type, we utilized siRNA knockdown of IKKa, IKKb, and IKKa/b in SKBR3 breast cancer cells ( Figure 5). These cells were engineered to stably express an NF-kB-dependent luciferase reporter. siRNA-transfected cells were either left untreated or were treated with TNF. As shown in Figure 5, knockdown of IKKa significantly reduced NF-kB dependent luciferase activity in response to TNF. Comparable reduction was observed with IKKb knock-down. Importantly, knockdown of IKKa and IKKb together further reduced the luciferase activity in response to TNF. These results indicate that both IKKa and IKKb are required for efficient TNF-induced NF-kB activity in breast cancer cells.

Kinase Inactive IKKb Inhibits IKKa Activity
The data presented so far indicate that IKKa as well as IKKb have significant roles in canonical NF-kB activation. Previous results derived from expression of an IKKb kinase inactive mutant have suggested that IKKb activity is highly dominant in canonical NF-kB activation. In order to further examine this hypothesis, we have utilized WT and IKKb KO MEF cells for transfection studies. WT and IKKb 2/2 MEF cells were transfected with empty vector or with an expression vector encoding IKKa along with an NF-kB luciferase reporter plasmid. Results shown in Fig. 6 indicate that IKKa expression activates the NF-kB-dependent reporter in both WT and IKKb 2/2 cells, consistent with a role for IKKa in the canonical pathway and demonstrating that IKKa can activate NF-kB in the absence of IKKb. Co-transfection of the IKKa expression vector with low and higher levels of the IKKb kinase inactive mutant demonstrates that the kinase inactive form of IKKb blocks NF-kB-dependent reporter activity in WT and, interestingly, in IKKb 2/2 cells. These findings demonstrate that a kinase inactive version of IKKb inhibits the activity of IKKa. TNF treatment of WT and IKKb 2/2 cells showed that cytokine stimulation led to an approximate 4-fold increase in NF-kBdependent luciferase activity and this response was reduced to approximately 2-fold with the loss of IKKb ( Figure 6). This result is consistent with the findings presented above for reporter activity in cells knocked down for IKKa or IKKb, and indicate that IKKa plays a key role driving NF-kB activity in the TNF-responsive (canonical) pathway. Expression of the kinase inactive mutant of IKKb strongly suppressed TNF-induced NF-kB activity in both WT and IKKb 2/2 cells. These results further demonstrate that the kinase inactive form of IKKb suppresses both IKKb as well as IKKa activity. Therefore studies utilizing IKKb KM need to be interpreted carefully as the effects observed from IKKb KM expression will be derived from effects on both IKKb (as expected) and IKKa (and see discussion). The results from these experiments  support the hypothesis that IKKa plays an important role in controlling NF-kB-activity in the canonical pathway.

Discussion
Based on the phenotypes of IKKa and IKKb animals, and on results utilizing IKKa 2/2 and IKKb 2/2 MEFs, it has been concluded that IKKb is the more important IKK catalytic subunit relative to the control of NF-kB activation in the canonical pathway [see [1][2][3][4][5]. Additionally, results using a kinase inactive version of IKKb have supported these findings since expression of this mutant strongly suppresses NF-kB activation in several cell types. A variety of experiments have implicated NF-kB as a key regulator of human cancer and of diseases associated with inflammation [1][2][3][4][5]. Thus, interest in inhibiting NF-kB activation has focused on the development of drugs that block IKKb. In fact, IKKb inhibitors have shown efficacy in different models of disease [25,26]. It is noted that blocking recruitment of IKKc (NEMO) to the core IKK complex, which blocks canonical IKK activity, has shown broad efficacy in animal models of inflammatory disorders . IKKa and IKKb each contribute to TNF-induced NF-kB activity in HeLa cells. HeLa cells, seeded in 24-well plates were transiently transfected with indicated siRNA constructs for 48 hr. Then media was replaced and cells were further transfected with NF-kB responsive 3x-kB luciferase and a control Renilla luciferase contructs. TNF was added (as indicated) and 24 hr later cells were lysed and dual lucifearse assay was performed. Luciferase readings in untreated and control vector transfected cells were normalized as 1. doi:10.1371/journal.pone.0009428.g004 Figure 5. Knockdown of IKKa or IKKb blocks basal and TNF-a induced NF-kB luciferase activity in breast cancer cells. SKBR3 cells stably expressing 4x-kB firefly luciferase reporter gene were transfected with 100 nM siRNA against IKKa, IKKb or both. Cells were treated with PBS (black) or 10 ng/ml TNF (gray striped) for 12 hr. Luciferase activity was measured and normalized to total protein levels. doi:10.1371/journal.pone.0009428.g005 [27]. The experiments presented here indicate that targeting IKKa (alone) or in combination with IKKb inhibition (via use of distinct IKKa/IKKb inhibitors, or through blocking IKKc interaction with the catalytic IKK components) will generate an anti-inflammatory approach. Additionally, inhibiting IKKa alone may have distinct advantages over inhibiting IKKb since IKKb inhibition is associated with enhanced release of IL-1 [28].
Original data using knockout MEFs indicated that IKKb is the critical kinase downstream of TNF in inducing IkBa phosphorylation and degradation [see [13][14][15][17][18][19]. While our data completely agree with those results, loss of IKKa significantly reduced NF-kB activation induced by TNF in MEFs as measured through EMSA (Fig. 2) and reporter assays (Fig. 4-6). Additionally, NF-kB-dependent reporter activity is only partly suppressed in IKKb2/2 cells (Fig. 6), indicating the involvement of IKKa in controlling NF-kB activity in MEFs. The mechanism of IKKaregulated NF-kB activation is unclear, but may involve the phosphorylation response on p65 where IKKa is clearly involved (see Fig. 1). For the IKKa-controlled pathway, phosphorylation of p65 at ser536 may control DNA binding activity or release from IkB. Interestingly, it was reported that phosphorylation of p65 at ser536 does in fact induce release from IkB without degradation [29]. Additionally, IKKa could potentially induce degradation of IkBb or IkBe but our analysis did not reveal evidence of this mechanism (data not shown). It has also been reported that IKKa can control IKKb activity [30,31], which may contribute to TNFinduced activity in wild-type cells but this cannot explain IKKa activity in IKKb2/2 cells (Fig. 6). Future experiments will address the specific effect whereby IKKa regulates NF-kB activity.
Based on the results obtained in MEFs, we extended our studies to HeLa cells. Using siRNA knockdown of IKKa or IKKb in these cells, we demonstrate that loss of either IKK subunit suppresses IkBa phosphorylation, and delays IkB degradation (see Fig. 3). These results indicate that in HeLa cells both IKKa and IKKb are important for IkBa phosphorylation downstream of TNF-induced signaling. The reason that IKKa is not involved in IkBa phosphorylation/degradation in MEFs is unclear at the present time, but is not related to lower relative levels of IKKa in these cells, as determined by immunoblot analysis (see Fig. 1). To analyze another cell type, we utilized SKBR3 human breast cancer cells. Knockdown of IKKa or IKKb suppressed TNFinduced NF-kB-dependent reporter levels (see Figs. 4 and 5), again supporting the hypothesis that IKKa and IKKb are both important for TNF-induced NF-kB activation.
We analyzed the effect of expression of a kinase-inactive form of IKKb on NF-kB-driven reporter gene expression (see Fig. 6). Previously, results derived from utilization of this mutant form of IKKb have been used to argue the selective involvement of IKKb in canonical signaling. IKKa expression in WT and in IKKb 2/2 cells induces NF-kB reporter activity, which is blocked by expression of IKKb KM (Fig. 6). The ability of TNF to activate the NF-kB-dependent luciferase reporter is only partly inhibited in IKKb 2/2 cells, indicating the involvement of IKKa in the response. Interestingly, expression of the IKKb kinase mutant strongly suppresses TNF-induced reporter activity (below that seen in IKKb 2/2 cells) and blocks TNF-induced in IKKb 2/2 cells, indicating that the IKKb mutant blocks IKKa activity. Thus these results indicate that IKKa is important in the NF-kB-dependent gene expression response to TNF, and that the kinase inactive IKKb blocks IKKa activity, potentially through engaging a key regulatory molecule upstream of both IKKa and IKKb or through dimerization with a wild-type IKK subunit and inhibition of the IKK complex.
Why MEFs and HeLa cells appear to utilize IKKa and IKKb differently regarding effects on IkBa phosphorylation is unclear. This observation may indicate species differences or that different cells/tissues utilize IKKa and IKKb differently, a concept that should be considered in potential approaches to disease therapy. This latter point may relate to different levels of key upstream regulators of IKK. In this regard, it was reported that knockout of IKKb in adult hepatocytes did not significantly suppress the ability of TNF to activate NF-kB in these cells, with activity presumably derived from IKKa [22]. This is in contrast to embryonic RelA 2/2 or IKKb 2/2 hepatocytes which are sensitive to TNF-induced killing due to poor activation of NF-kB. Also, this group reported that IKK1/a and IKK2/b cooperate in the canonical pathway in hepatocytes [32]. Furthermore, it was reported that loss of IKKb leads to a compensatory activation of IKKa [33], but that does not explain why loss of IKKa leads to suppression of NF-kB activity in our studies.
IKKb inhibitors have been developed and have shown therapeutic responses in different animal models of diseases and are in early clinical trials [25,26]. These inhibitors show significant preference to IKKb over IKKa when tested against recombinant proteins. The results presented here indicate that IKKa inhibitors should be developed and tested using animal models of inflammatory diseases. Additionally, the results indicate that dual inhibition of IKKa/b would appear to be an optimal approach to block NF-kB activity downstream of TNF and other inflammatory cytokines [also see 33]. In summary, the data presented here demonstrate that IKKa and IKKb are both functionally important and cooperate in optimal TNF-induced (canonical) NF-kB activation, with evidence that different cells may utilize IKKa and IKKb differently.