Interleukin (IL) 1β Induction of IL-6 Is Mediated by a Novel Phosphatidylinositol 3-Kinase-dependent AKT/IκB Kinase α Pathway Targeting Activator Protein-1*

Here we describe a novel role for the phosphatidylinositol 3-kinase/AKT pathway in mediating induction of interleukin-6 (IL-6) in response to IL-1. Pharmacological inhibition of phosphatidylinositol 3-kinase (PI3K) inhibited IL-6 mRNA and protein production. Overexpression of either dominant-negative AKT or IκB kinase α mutant, IKKαT23A, containing a mutation in a functional AKT phosphorylation site, shown previously to be important for NFκB activation, completely abrogated IL-6 promoter activation in response to IL-1. However, mutation of the consensus NFκB site on the IL-6 promoter did not abrogate promoter activation by IL-1 in contrast to the AP-1 site mutation. IL-1 induces phosphorylation of IKKα on the NFκB inducing kinase (NIK) phosphorylation sites Ser176/Ser180 and on the Thr23 site, and although phosphorylation of IKKαT23 is inhibited both by LY294002 and wortmannin, phosphorylation of Ser176/Ser180 is not. Neither inhibition of PI 3-kinase/AKT nor IKKαT23A overexpression affected IκBα degradation in response to IL-1. Only partial inhibition by dominant-negative AKT and no inhibitory effect of IKKαT23A was observed on an IL-6 promoter-specific NFκB site in contrast to significant inhibitory effects on the AP-1 site. Taken together, we have discovered a novel PI 3-kinase/AKT-dependent pathway in response to IL-1, encompassing PI 3-kinase/AKT/IKKαT23 upstream of AP-1. This novel pathway is a parallel pathway to the PI 3-kinase/AKT upstream of NFκB and both are involved in IL-6 gene transcription in response to IL-1.

aging (1,2). Sometimes viewed as an anti-inflammatory cytokine, prolonged IL-6 production may also cause disease and injury. Inhibition of IL-6 production in males reduces the risk of chemically induced hepatocellular carcinoma (3). IL-6 is a recognized modifier gene of intestinal tumorigenesis (4) and serves as a growth factor for pre-malignant enterocytes that give rise to colitis-induced cancer (5)(6)(7).
Increased mucosal IL-6 production was shown to cause local inflammation associated with Crohn disease and ulcerative colitis (inflammatory bowel disease) (8 -11). Although the intestinal epithelium is not the main source of IL-6, intestinal macrophages and CD4ϩT-cells also increase their production of IL-6 and its soluble receptor leading to IL-6 trans-signaling via glycoprotein 130 during inflammatory bowel disease (10). In the intestinal mucosa during sepsis, endotoxemia, and severe injury, the enterocyte increases its production of IL-6 (12). Although in the short term the beneficial effects of IL-6 includes enterocyte acute phase protein induction, mucosal protein synthesis and IgA production in Peyers patch B cells elevated IL-6 may impair mucosal integrity (13,14).
The primary inflammatory cytokine, interleukin-1 (IL-1), tumor necrosis factor (TNF), platelet-derived growth factor, bacterial lipopolysaccharide, acute viral infections, and transforming growth factor (TGF␤) each induce IL-6 expression. IL-1 is the major proinflammatory cytokine responsible for mediating several physiological responses such as fever, activation of lymphocytes, and induction of acute phase protein synthesis (15). Recent findings suggest that cellular responses to IL-1 are mediated by cascades of intracellular events including activation of mitogen-activated protein kinases (MAPKs) involved in the activation of AP-1 and IB kinases (IKKs) involved in the activation of NF-B (16, 17) (Fig. 9).
Induction of IL-6 by lipopolysaccharide, from Gram-negative bacteria, lipoteichoic, LTA, from Gram-positive bacteria, and lysophosphatidic acid, a naturally occurring phospholipid have been shown to be phosphatidylinositol 3-kinase (PI3K)dependent (18,19). Little attention, however, has focused on PI 3-kinase as a downstream effector of IL-1 (20,21). The Ser/Thr kinase AKT/protein kinase B has been identified as an important target of PI 3-kinase. AKT provides a potent cell survival signal that is likely involved in its transformation and growthpromoting properties (22,23). NFB is an important downstream target of AKT (24). NFB is typically composed of a dimer between p50 and the transactivating subunit p65 (RelA).
IKK␣ and IKK␤ are two important kinases required for NFB activation. The canonical NFB pathway, triggered in response to microbial and viral infections as well as proinflammatory cytokines, involves IKK␣ or IKK␤-mediated phosphorylation of the inhibitor, IB␣, followed by its subsequent ubiquitination, degradation, and entry of p50/p65 into the nucleus (25). This pathway is cell type-specific and depends on the levels of IKK␣ or -␤ within the cell (26).
IKK␣ is the predominant form of the IKK complex activated in response to IL-1 (27). PI 3-kinase/AKT is involved in the activation of IKK␣. AKT binds to and increases the activity of IKK␣ (28). The PI 3-kinase/AKT-mediated phosphorylation of IKK␣ on Thr 23 in response to TNF␣ results in the activation of canonical NFB through liberation of IB␣ (29). IKK␣ is also activated by autophosphorylation and phosphorylation on Ser 176 and Ser 180 by its upstream activator, NIK (30,31). Both AKT and NIK acting in parallel contribute to NFB activation in response to TNF␣ (29). IKK␣ also mediates activation of the non-canonical pathway, which is involved in B cell maturation and lymphoid organogenesis, through AKT-dependent phosphorylation of the p100 NFB precursor (32).
Activation of AP-1 occurs primarily through signaling pathways terminating in a group of serine-threonine kinases, MAP kinases that act separately on its components. The predominant form of AP-1 in most cells are heterodimers of fos and jun, which have high affinity for binding to an AP-1 site (33). The MAP kinase, ERK, phosphorylates Elk-1, which is part of a complex that binds to the serum response element in the fos promoter. JNK phosphorylates c-jun, Elk-1, and ATF-2 and is regulated itself by MAP/ERK kinase-4 (MEK4/SEK1) and MEK7, which are dual specificity kinases mediating phosphorylation on the TPY motif (34). p38 kinase phosphorylates Elk-1 and ATF-2 (35). AP-1 activates a broad range of genes designed to protect cells from adverse environmental conditions. Activation of PI 3-kinase by IL-1 is sufficient for full activation of AP-1 but not NFB (21). AP-1 activation by epidermal growth factor, 12-O-tetradecanoylphorbol-13-acetate (TPA), as well as by 5-MCDE another tumor promoter, and the tax oncoprotein is PI 3-kinase dependent (36 -39). Furthermore, it has been demonstrated that the PTEN phosphatase down-regulates AP-1 via PI 3-kinase/AKT inhibition (40).
This investigation addresses the role of the PI 3-kinase signaling pathway in IL-1 induction of the IL-6 gene in the Caco-2 cell, a colon carcinoma with enterocyte-like characteristics. We provide evidence for two PI 3-kinase/AKT-dependent pathways to induction of the IL-6 gene in response to IL-1. The first is a newly discovered pathway encompassing PI3K/AKT/IKK␣ upstream of AP-1. This IL-1 responsive pathway targets the IL-6 promoter AP-1 site and mutation of this site and not the NFB site significantly reduces activation of the gene by IL-1. The second PI 3-kinase/AKT-dependent pathway is upstream of NFB and both pathways, in parallel, are necessary for induction of IL-6 gene transcription in response to IL-1. A novel PI 3-kinase-dependent pathway involving AKT/IKK␣ upstream of AP-1 suggests a more widespread effect of IKK␣ on gene expression than can be attributed to activation of NFB and is further evidence of cross-talk between two transcription factors known to be involved in growth and tumor development.
With increasing association of IL-6 and chronic disease, discovery of novel signaling pathways can lead to the design of therapeutic interventions.
Co-immunoprecipitation of IKK␣ and AKT-Lysates from control and IL-1-treated cells were incubated overnight with control rabbit IgG, anti-IKK␣ antibody (Cell Signaling), or AKT antibody. After several washes, immunoprecipitated proteins were run on a 10% SDS gel followed by Western blotting (41). IKK␣ immunoprecipitates were blotted for the presence of both AKT and IKK␣ and AKT immunoprecipitates were blotted for the presence of both IKK␣ and AKT.
mRNA Isolation and Reverse Transcriptase-PCR-Total RNA was isolated using Stat 60 (AMS biotechnology). Reverse transcriptase-PCR was carried out as previously described in detail (42) using the following primers and probes: hIL-6 -87F, CCAGTACCCCCAGGAGAAGAT; hIL-6-157R, CGTTCT-GAAGAGGTGAGTGGC; and TaqMan probe: hIL-6-110T, CAAAGATGTAGCCGCCCCACACAGAC. Amplification of 18 S RNA as an internal standard was performed in the same reaction with the alternatively labeled Vic probe. IL-6 mRNA was normalized to the 18 S mRNA levels. Samples were assayed in duplicate.
Plasmids and Transfections-All AKT expression plasmids were a gift from Dr. J. R. Woodgett, Department of Molecular and Medical Genetics, Ontario Cancer Institute, University of Toronto. The dominant-negative AKT (dnAKT) (AKT AAA) plasmid in pcDNA3 has Thr 179 , Thr 308 , and Ser 473 residues all substituted by alanine rendering it unable to bind ATP or be activated. The constitutively active AKT (caAKT) in pcDNA3 is linked to the viral gag sequence, which targets it to the cell membrane. The pIL-6-luc651 plasmid, containing a 651-bp fragment of the human IL-6 gene promoter located directly upstream of the transcription start site and the 3 mutated plasmids, pIL-6-651mAP-1, pIL-6-651mC/EBP-␤, and pIL-6-651mNFB were a generous gift from Dr. Oliver Eickelberg, University Hospital, Basel, Switzerland. Transcription factor binding site mutations were as described below.
Transient Transfections-Caco-2 cells were plated to 50% confluence in transfection media (growth media minus antibiotics) in 24-well plates. After overnight attachment, cells were transfected with reporter plasmid (0.25 g) together with ␤-galactosidase reporter plasmid (0.25 g) for transfection efficiency, in the presence or absence of expression plasmids (0.125 g/well). 18 h following transfection, plates were either left untreated (C) or treated overnight with IL-1␤ (0.5 g/ml). All plates were harvested 18 h after treatments and assayed for luciferase activity (Invitrogen, Bright Glo) and ␤-galactosidase (Invitrogen). Results are expressed as normalized values, luciferase/␤-galactosidase (41).
Trans-AM NFB ELISA and Inhibitors-p50 and p65 binding to the WT IgB binding site (GGGACTTTCC) was measured by ELISA using antibodies to the activated form of the proteins. This site differs from the WT IL-6-kB site by one base at position 5, which is replaced by a C in the IL-6-B site (GGGACTTTCC). This nucleotide is crucial for the binding of a repressor protein RBP-J (44).
Statistical Analysis-Analysis of variance was used to compare control versus IL-1 treated with Tukey test comparisons of all groups. t test (paired) two-tailed analysis was also used in parallel to compare 2 treatment groups. Experiments were performed at least 3 times with high reproducibility.

Interleukin-1 Induction of IL-6 Production in Caco-2 Cells Is PI 3-Kinase Dependent-
To determine the optimal dose of IL-1 needed to induce maximal IL-6 secretion, Caco-2 cells were exposed to dose levels of IL-1 ranging from 0 to 10 ng/ml. Maximum IL-6 levels in the media was observed with dose levels between 0.5 and 5 ng/ml (Fig. 1A), demonstrating a 10-fold increase from 2 to 20 pg/ml. A dose level of 0.5 ng/ml of IL-1 was subsequently used. A time course is presented in Fig. 1B demonstrating increasing levels of IL-6 secreted into the tissue culture media from 2 to 4 h. Because both IL-1 and IL-6 are mediators of the acute phase response, including fever production, and elevated temperature has previously been demonstrated to induce IL-6 gene expression (46), we compared the effects of heat shock (HS) to IL-1 induction of IL-6. A 43°C heat shock by itself increased IL-6 levels from 2 to 5.8 pg/ml with significant increases observed only at the end of the 4-h time course (Fig. 1B). IL-1 induces 6-fold more IL-6 mRNA than HS, reflecting the larger increase in IL-6 protein production by IL-1 ( Fig. 1, B and C).
One of the downstream kinases stimulated by IL-1 is PI 3-kinase. We next investigated if this pathway was important in the induction of IL-6 by IL-1. The PI 3-kinase inhibitor LY294002 inhibited IL-6 mRNA in response to IL-1 and HS and significantly inhibited IL-1 induction of IL-6 secretion into the media in a dose responsive manner ( Fig. 1, C and D). Significant inhibition was observed even at the low dose of 3.12 M. In addition, another highly specific PI 3-kinase inhibitor, wortmannin, significantly inhibited IL-1 induction of IL-6 secretion, further supporting a role for the PI 3-kinase pathway in IL-6 induction by IL-1.
AKT Is Activated by Interleukin-1 and Heat Shock in a PI 3-Kinase-dependent Manner-One of the major downstream targets of PI 3-kinase is the AKT kinase. Stress stimuli, including heat shock, as well as growth factor stimulation have been shown to activate AKT in a PI 3-kinase dependent manner (47). AKT activation is followed by phosphorylation on its 3 major phosphorylation sites, Thr 179 , on its ATP site, Thr 308 in the activation loop, and Ser 473 in the kinase domain (23). We therefore investigated whether IL-1 and HS induced phosphorylation of the serine 473 AKT site, the site most commonly used to demonstrate activated AKT.
A short time course (0 -60 min) was performed to examine the pattern of AKT Ser 473 phosphorylation by IL-1 ( Fig. 2A). Significant activation of AKT as detected by phosphorylation on Ser 473 is detected by 5 min and is maximal by 30 min to 1 h representing a 2-fold increase in molar phosphorylation (ratio of phosphorylated to non-phosphorylated AKT) (Fig. 2B). In Fig. 2C an extended time course up to 4 h is shown in the absence and presence of the PI 3-kinase inhibitor LY294002. AKT Ser 473 phosphorylation and kinase activation, as measured by GSK3␤ S21/9 phosphorylation, was sustained up to 4 h. At the end of the 4-h time course the molar phosphorylation ratio for AKT was 0.71 Ϯ 0.16 (n ϭ 4) (Fig. 2C). Near complete inhibition of AKT Ser 473 and GSK3␤ phosphorylation, by LY294002, over the entire period of the time courses was observed demonstrating that activation of AKT in response to IL-1 is PI 3-kinase dependent.
Lysates from cells exposed to a heat shock of 43°C were next blotted for activated AKT (Ser 473 phosphorylation) and total AKT. A HS of 43°C induced a rapid onset of AKT activation, within 5 min, representing a 3-fold increase in molar phosphorylation by 1 h (Fig. 2, D and E). This was sustained over a 5-h time course (Fig. 2F) with an ending molar phosphorylation ratio of 1.33 Ϯ 0.28 (n ϭ 4). AKT activation by HS was also blocked by LY294002 and therefore is PI 3-kinase dependent.
Taken together, the data shows that activation of AKT by IL-1 is maximal by 1 h and correlates with the initiation of IL-6 secretion into the media. Inhibition of PI 3-kinase by LY294002 inhibited both AKT activation and IL-6 secretion. Similar increases in AKT activation in response to HS were insufficient to induce comparable IL-6 secretion to that induced by IL-1.
Canonical NFB Activation Occurs following AKT Activation in Response to IL-1, the IKK Complex Is Required for IL-1 Induction of IL-6-Because NFB is a well known downstream target of AKT in response to IL-1 we next examined the time course of activation of NFB in terms of IB␣ degradation and DNA binding of p50/p65. In Fig. 3A degradation of IB␣ in response to IL-1␤ occurs between 15 and 30 min following stimulation

FIGURE 2. Interleukin-1 and heat shock activate AKT in a PI 3-kinase dependent manner.
A, Caco-2 cells were treated with IL-1 (0.5 ng/ml) and harvested at the indicated time points. Lysates (70 g) were run on 10% SDS-PAGE followed by immunoblotting with phospho-specific AKT Ser 473 . The lysate blots were stripped using 5 ml of Pierce stripping reagent and reprobed using an anti-AKT antibody and an antibody for ␤-actin. B, the intensity of phosphorylation of AKT over a 1-h time course was estimated using NIH Image software and is expressed as the ratio of phosphorylated protein over total AKT. C, Caco-2 cells were treated with IL-1 (0.5 ng/ml) for up to 4 h in the presence or absence of the PI 3-kinase inhibitor (LY294002, 25 M) incubated 30 min prior to IL-1 treatment. Cells were harvested and lysed at the indicated hourly time points. AKT was immunoprecipitated and the kinase assay performed using GSK3␤ peptide. The product of the reaction and the initial lysate were blotted as in A. D, cells were heat shocked (43°C for 1 h) and harvested at the indicated time points during HS. Cells were lysed, and blotted for AKT Ser 473 as in A. E, the intensity of phosphorylation of AKT over the 1-h time course for HS treatment was estimated using NIH Image software as in B. Mean Ϯ S.E. (n ϭ 3) from four separate experiments are presented. F, cells were HS at 43°C for 1 h in the presence or absence of the PI 3-kinase inhibitor (LY294002, 25 M) incubated 30 min prior to HS and returned to 37°C for an additional 4 h. Cells were harvested and lysed at the indicated hourly time points. AKT Ser 473 phosphorylation was detected as in A. and this was unaffected by PI 3-kinase inhibition by either LY294002 or wortmannin (see Fig. 7B). This was followed by a 3-fold increase in the binding of p50 by 1 h and a 4-fold induction in the binding of p65. Binding of p65 decreased thereafter, whereas p50 binding was sustained for up to 4 h (Fig. 3B). Taken together with Fig. 2A degradation of IB␣ occurs immediately following initial AKT activation at 5 min. There is also a correlation between the initial and sustained p50 subunit of NFB binding to DNA and the pattern of AKT activation.
In Fig. 3C, a cell permeable aminoacetamide compound and potent inhibitor of IKK complex activity and NFB activation was used. At dose levels between 5 and 25 M this inhibitor lowered IL-6 secretion into the media by 50% in response to IL-1, suggesting that the NFB pathway contributes to IL-6 production in response to IL-1.
To further explore the possibility that AKT might be involved in the activation of the NFB pathway in response to IL-1 we examined the time course of activation of IKK␣ in the presence and absence of the PI 3-kinase inhibitors. Phosphorylation of Ser 176 /Ser 180 (NIK phosphorylation sites) and Thr 23 (AKT phosphorylation site) have been shown previously to indicate IKK␣ activation that precedes IB␣ degradation and NFB activation (25, 29 -31). Phosphorylation of IKK␣ on Ser 176 /Ser 180 and Thr 23 in response to IL-1 as detected by Western analysis using phosphospecific antibodies was maximal by 15-30 min and decreased thereafter to baseline levels (Fig. 3D). Phosphorylation of IKK␣ at 15-30 min correlates with the degradation of IB␣ at the same time point in Fig.  3A. A comparison of molar phosphorylation ratios of AKT activation on Ser 473 and IKK␣ on Ser 176 / Ser 180 and Thr 23 (Fig. 3D, lower) demonstrates that increases in AKT phosphorylation at 15 min coincides with maximal IKK␣ phosphorylation and that although AKT molar phosphorylation is maximal at 30 min, thereafter there is a similar decline in the molar phosphorylation of the 2 kinases. We next demonstrate in Fig. 3E that neither wortmannin nor LY294002, two potent PI 3-kinase inhibitors, at concentrations that inhibit PI 3-kinase in cell based assays (57), affect the phosphorylation of IKK␣S176/180, while significantly inhibiting the  phosphorylation of IKK␣T23 and, as expected, the phosphorylation of AKT Ser 473 . Taken together, this suggests that PI 3-kinase/AKT is upstream of IKK␣T23 (and not IKK␣S176/180) and, as is the case with TNF␣, is involved in a parallel pathway with NIK to activate IKK␣ in response to IL-1.
AKT and IKK␣ Are Necessary for IL-6 Promoter Induction in Response to IL-1-Expression of IL-6 is tightly regulated at the level of transcription. To further explore the role of AKT in the induction of IL-6 by IL-1, the wild type IL-6 promoter reporter construct (pIL-6-luc651), containing a 651-base pair fragment of the IL-6 promoter, in front of the luciferase gene was used (Fig. 4A). This contains all the elements necessary for its induction by a variety of stimuli. There is a multiple response element consisting of a CRE followed by a consensus binding site for C/EBP␤/NF-IL-6 at Ϫ173 to Ϫ145, which is involved in tissuespecific transcription of the IL-6 gene (87). There is an NFB binding site from Ϫ73 to Ϫ63, and an AP-1 site from Ϫ283 to Ϫ277 (48).
NFB is a well known downstream target of AKT, and one of its activating kinases, IKK␣, is the predominant form of the IKK complex activated in response to IL-1 (27). IKK␣ contains a functional AKT phosphorylation site on Thr 23 (29) and this is a necessary step for the degradation of IB␣ and NFB activation in response to TNF␣. We therefore investigated if co-transfection of either dnAKT or IKK␣T23A would affect IL-6 promoter induction by IL-1. Caco-2 cells were transfected with the IL-6 promoter together with, caAKT, dnAKT (mutated in the 3 phosphorylation sites, Thr 179 , Thr 308 , and Ser 473 ), IKK␣WT, or IKK␣T23A (containing the AKT site mutation at Thr 23 ).
The pIL-6-luc651 construct was activated 2-fold by IL-1 (Fig.  4B). Co-transfection of caAKT only slightly enhanced baseline and IL-1 induction of the IL-6 promoter, whereas dnAKT completely abrogated its induction by IL-1 (compare the seventh to ninth bars). Although IKK␣WT did not, by itself, activate the IL-6 promoter (fourth bar) nor did it enhance its induction by IL-1 (compare the seventh to 10th bars), IKK␣T23A, inhibited the induction by IL-1 similar to that observed by dnAKT (compare the seventh bar to the ninth and 11th bars). In Fig. 4C, overexpression of IKK␣WT was sufficient to reverse the inhibitory effects of dnAKT on IL-6 promoter activation (compare the sixth bar to the seventh and eighth bars). Taken together, this demonstrates that AKT and IKK␣ are part of a common pathway to IL-6 gene transcription in response to IL-1.
To control for transfection efficiency and to ensure equal expression of plasmids in transfected cells, Western blots were performed to compare levels of expression. Western blotting confirmed similar levels of expression of IKK␣WT and IKK␣T23A in the transfected cells (Fig. 4D). Elevated and comparable levels of expression in transfected cells of both caAKT and dnAKT with an increasing dose of plasmid were also observed (Fig. 4E).
To determine whether AKT and IKK␣ associate in vivo in response to IL-1 we performed cross-coimmunoprecipitation experiments in the presence or absence of IL-1. These demonstrated that AKT and IKK␣, as is the case in response to TNF␣ (29), associated in vivo in the cell in response to IL-1 with maximal association at 30 min (Fig. 4F). Significant decreases in association of AKT and IKK␣ were observed by 1 h in both pull-downs correlating with the decrease in phosphorylation of IKK␣ observed in Fig. 3D. No detectable AKT or IKK␣ was detectable in immunoprecipitates using control rabbit IgG (see Fig. 4F).

Mutation of the NFB Site Affects Baseline IL-6 Promoter Activation While Its Induction by IL-1 Is Dependent on an
Intact AP-1 Site-To distinguish which of the promoter elements contained within the pIL-6-luc651 was most important in baseline or IL-1 activation of the gene, we compared baseline and IL-1 activation of three mutant IL-6 promoter constructs to the WT pIL-6-luc651. The mutated promoters, pIL-6-luc651mNFB, pIL-6-luc651mC/EBP, and pIL-6-luc651mAP-1, contained mutations in their respective transcription factor binding sites that have been shown previously to abrogate their binding (43) (Fig. 4A). In one representative of at least three experiments (Fig. 5A), mutation of the NFB site significantly reduced baseline IL-6 promoter activation with no significant baseline effect when the C/EBP␤ site was mutated (Fig. 5A). Together with WT pIL-6-luc651, both the mNFB and mC/EBP IL-6 promoter constructs were significantly activated by IL-1 in marked contrast to the near absence of activation of the pIL-6-luc651mAP-1 (AP-1 mutant) IL-6 promoter reporter.
In an average of three separate experiments shown in Fig. 5B and expressed as -fold activation by IL-1, the AP-1 site mutant was consistently less responsive to IL-1 compared with the WT, pIL-6-luc651mNFB (NFB mutant), or pIL-6-luc651mC/EBP (C/EBP mutant) plasmids (Fig. 5B). These experiments demonstrate that AP-1 is the primary target site for activation of the IL-6 gene in response to IL-1. Mutation of the consensus C/EBP site had no effect on either baseline or IL-1 induction of the pIL-6-luc651 construct. As the IL-6 promoter mutated in the NFB site does not bind NFB, the 2-fold IL-1 induction is due to other factors that bind to the remaining intact sites on this mutated promoter of which AP-1 is the most likely, as the AP-1 site is the only transcription factor binding site (of the three transcription factor sites investigated (NFB, C/EBP, and AP-1 sites) (Fig. 5B) that is necessary for full IL-1 induction. Taken together, these experiments demonstrated that baseline activation depends on intact NFB and AP-1 sites and that maximal IL-1 induction depends only on an intact AP-1 site.

An Intact AP-1 Site and Not an Intact NFB Site Is Necessary
for Inhibition by dnAKT and IKK␣T23A-To further explore the downstream target of the AKT/IKK␣ pathway on IL-6 gene transcription we next investigated the effect of overexpression of dnAKT and IKK␣T23A on two of the IL-6 promoter constructs employed above, the pIL-6-luc651mNFB (NFB mutant) and pIL-6-luc651mAP-1 (AP-1 mutant). Similar to the WT promoter (Fig. 4B) the pIL-6-luc651mNFB, containing the NFB site mutation (Fig. 6A), demonstrated a 2-fold induction by IL-1 as well as almost complete inhibition of the IL-1 response by either dnAKT or IKK␣T23A (compare the seventh bar to the ninth and 11th bars). This suggests that an intact NFB site on the IL-6 promoter is not necessary for inhibition by either dnAKT or IKK␣T23A.
The IL-6 promoter containing the AP-1 site mutation, pIL-6-luc651mAP-1 (Fig. 6B), demonstrated near complete loss of promoter induction by IL-1. There was no significant inhibition of baseline activity (Fig. 6B) either by dnAKT or IKK␣T23A (compare the first bar to the third and fifth bars). Furthermore, IL-1 induction of pIL-6-luc651mAP-1 activation was not inhibited by dnAKT but was inhibited by IKK␣T23A suggesting that an intact AP-1 site was necessary for inhibition by dnAKT but not IKK␣T23A. Taken together, this suggests that an intact AP-1 site is necessary for inhibition by at least dnAKT and IKK␣T23A may, in addition to AP-1, be targeting another transcription factor necessary for IL-1 induction of the IL-6 promoter.
Interestingly, we observe a significant 2-fold baseline activation of the pIL-6luc651mAP-1 promoter by IKK␣WT but not by IKK␣T23A, demonstrating that IKK␣ overexpression can compensate for the loss of an IL-1 effect from AP-1 site mutation. This suggests involvement of the AKT/IKK␣ pathway in the activation of another transcription factor that may be negatively regulated by AP-1 binding. These results suggest that AP-1 is a likely downstream target of AKT/IKK␣ in response to IL-1 on the intact IL-6 promoter and that NFB may yet be the target of AKT/IKK␣ on promoters not carrying an AP-1 site.
Requirement for AKT but Not the IKK␣T23 AKT Phosphorylation Site for NFB Activation in Response to IL-1-In Fig. 7A we investigated the effect of either pharmacological inhibition of PI 3-kinase/AKT, by LY294002 or IKK␣T23A overexpression on canonical NFB activation by Western blotting for the presence of IB␣. Neither IKK␣T23A overexpression nor exposure to LY294002 prevented significant loss of IB␣, suggesting that AKT/IKK␣ is not upstream of IB␣ degradation in the NFB activation pathway. This is further confirmed in Fig.  7B (in the absence of cycloheximide) where again no significant inhibition of the degradation of IB␣ was observed in the presence of either LY294002 or wortmannin.
To further define the role of NFB and determine whether AKT and/or IKK␣T23 are necessary for NFB activation downstream of IB␣ degradation we performed luciferase reporter assays employing two separate reporter plasmids, each containing three NFB binding sites. The first reporter plasmid contains a 3ϫ NFB binding site from the IgB gene (Fig. 7C), the second reporter plasmid contained a 3ϫ NFB site from the IL-6 gene itself (Fig. 7D). A representative experiment from three separate experiments is presented, each with consistent results. As expected, IL-1 strongly activated the NFB reporters. A 15-fold induction of the IgB NFB reporter, and a 7-fold induction of the IL-6-specific NFB reporter was observed in response to IL-1 (Fig. 7, C and D). Although caAKT did not, by itself, activate the NFB reporter, dnAKT significantly reduced baseline (compare the first bar to the third bar) and its induction by IL-1 (compare the sixth bar to the eighth bar). This suggests that AKT is necessary for NFB activation in response to IL-1. This is also reflected in Fig. 7D employing the IL-6 promoter-specific NFB site (compare the sixth bar to the eighth bar).
This experiment also demonstrated that IKK␣WT overexpression alone is capable of activating the NFB luciferase reporters up to 3-fold (compare the first bar to the fourth bar, Fig. 7, C and D) and this is independent of the Thr 23 phosphorylation site as IKK␣T23A is also capable of activation. Similarly, and contrary to its inhibitory effect on IL-1 induction of pIL-6-luc651WT, IKK␣T23A overexpression had no significant effect on IL-1 induction of the NFB reporter plasmids (compare the sixth bars to the ninth and 10th bars, Fig. 7, C and  D). Taken together, this suggests that although IKK␣ is capable of activating NFB by itself, AKT is necessary for NFB activation in response to IL-1 independent of IKK␣.
PI 3-Kinase AKT/IKK␣ Thr 23 Is Upstream of AP-1-To determine whether AP-1 is the downstream target of AKT/ IKK␣T23 on the IL-6 promoter, we employed an AP-1 luciferase reporter plasmid containing 5 copies of the IL-6 AP-1 site (Fig. 8). We observed a significant dose-dependent activation of this AP-1 reporter in response to IL-1 with increasing activation up to 5 ng/ml (Fig. 8A). In the HT29 colon carcinoma cell  SEPTEMBER 19, 2008 • VOLUME 283 • NUMBER 38

IL-6 Gene Regulation by PI 3-Kinase/AKT/IKK␣/AP-1
line the AP-1 reporter could be activated by PI 3-kinase catalytic domain p110 overexpression (Fig. 8B). The AP-1 reporter displayed elevated constitutive activation in Caco-2 cells, which was inhibited by the PI 3-kinase inhibitor LY294002 (Fig. 8C). In Fig. 8D in the presence of IL-1 (0.5 ng/ml), the dose used to induce maximal IL-6 secretion in Caco-2 cells, dnAKT overexpression abrogated the induction by IL-1, and lowered constitutive activation by over 50% (compare the first bar to the third bar). Similar effects were observed with IKK␣T23A overexpression (compare the first bar to the fifth bar). Overexpression of either caAKT or IKK␣ WT alone had no significant effect on AP-1 luciferase activation, whereas both together significantly enhanced constitutive activation (data not shown) suggesting a common pathway. This is further supported by the demonstration that overexpression of IKK␣ WT reversed the inhibitory effect of dnAKT as also observed on the intact IL-6 promoter (Figs. 4C and 8D). This supports a functional role for AKT/ IKK␣ in the regulation of constitutive and IL-1-stimulated AP-1. The pattern of inhibition of the AP-1 reporter gene both by dnAKT and IKK␣T23A reflects the pattern observed on the WT IL-6 promoter as well as that observed on the IL-6 pro-moter containing a mutated NFB site. Taken together, this suggests that AKT/IKK␣ targets AP-1 on the IL-6 gene and unravels a novel signaling pathway encompassing PI 3-kinase/AKT/IKK␣ and AP-1. Fig. 9 is a model of the PI 3-kinase/ AKT-dependent pathways involved in induction of the IL-6 gene in response to IL-1. The first (1) involves AKT directly targeting IKK␣ on its Thr 23 phosphorylation site, a necessary step toward AP-1mediated transcription. In the second pathway (2) AKT directly targets NFB, which is indirectly involved in the induction of the IL-6 gene, likely by binding to other NFB responsive promoters and inducing the transcription of other factors with a transcriptional role such as fos/jun, which are involved in IL-6 gene expression in response to IL-1.

DISCUSSION
We demonstrate for the first time the requirement for PI 3-kinase/ AKT in the induction of the IL-6 gene in response to IL-1. This pathway has 2 components. The first pathway targets IKK␣, one of the kinases involved in the activation of NFB and is upstream, not of NFB, but of AP-1. The second pathway directly targets NFB, which is likely binding to and activating not the consensus NFB site on the IL-6 promoter but the promoters of other genes involved directly or indirectly in the induction of IL-6 by IL-1.
This is the first demonstration that AP-1 is a downstream target of IKK␣ mediated by AKT and defines a new IL-1 responsive pathway. This novel pathway is important not only in terms of the regulation of the IL-6 gene by IL-1, but also in terms of further evidence of cross-talk between two transcription factors known to be involved in growth and tumor development. It also lends support to recent findings showing evidence for a wider role for IKK␣ in gene regulation other than in the activation of non-canonical NFB (49 -52).
Future investigations will determine which components of the AP-1 complex are activated by IL-1 to bind to the AP-1 site in the IL-6 promoter in the Caco-2 cell line, and whether IKK␣ targets one of these directly. The role of the MAP kinase pathway in IL-6 induction has been extensively investigated in the Caco-2 cell line (53). With the use of specific pharmacological inhibitors it has been shown that all three MAP kinases (ERK, JNK, and p38) are activated in response to IL-1 and are involved in the induction of IL-6 (54). Interestingly, however, the JNK inhibitor SP 600125 was reported to be significantly less potent than the other MAP kinase inhibitors in the inhibition of IL-6 secretion in the Caco-2 cell (53). As JNK is one of the major kinases that phosphorylates c-Jun leading to the activation of AP-1, there are likely physiological circumstances where its activation is compromised. For example, binding of vaccinia-related kinase 2 to the JIP scaffolding protein prevented the recruitment of JNK and caused downregulated IL-1 responsive AP-1 transcription (55). Additionally it has been reported that AKT phosphorylates kinases upstream of JNK activation, i.e. SEK1 (MKK4) and ASK1 leading to inactivation of JNK (85,86). Therefore if PI 3-kinase/AKT can lead to the loss of JNK activation an alternative pathway to the activation of AP-1 might be via IKK␣.
Other studies have demonstrated that although JNK is the most important MAPK involved in IL-6 production by renal epithelial cells, the regulation of IL-6 gene transcription by JNK is independent of the AP-1 binding site but rather involves interference with other signaling pathways such as NFB and ERK (56).
Inhibition of IL-6 mRNA and protein production by LY294002 suggested the involvement of PI 3-kinase in the induction of IL-6 in response to IL-1. To confirm the involvement of PI 3-kinase the more specific inhibitor wortmannin was also shown to inhibit IL-6 production at concentrations previously shown to inhibit PI 3-kinase in cell based assays (57). Maximal activation of AKT by 1 h was followed by elevations in IL-6 mRNA and increasing IL-6 secretion into the media from 2 to 4 h (Fig. 1B). The IL-6 promoter contains, among others, binding sites for several AKT responsive transcription factors including NFB, CREB, and AP-1 (43,48,58). Consistent with previous investigators, a highly consistent and statistically significant 2-3-fold activation of the WT IL-6 promoter by IL-1 was observed in Caco-2 cells (44,59). Although caAKT overexpression, by itself, was not sufficient to activate the IL-6 promoter, dnAKT completely abrogated the promoter response to IL-1 demonstrating for the first time that AKT is necessary for IL-1 activation of the gene.
A well characterized and commercially available IKK complex inhibitor that inhibits the classical NFB pathway significantly reduced IL-6 secretion in response to IL-1 implying a role for NFB. We explored the possibility that NFB might be the downstream target of PI 3-kinase/AKT. In this regard, the PI 3-kinase/AKT-mediated transactivation of the p65 and p50 subunits of NFB in response to IL-1 has previously been demonstrated (24,36,60,61). IKK␣, previously shown to be the predominant form of the IKK complex activated in response to IL-1, contains 2 serine residues previously shown to be phosphorylated by NIK, Ser 176 /Ser 180 (25), as well as a functional AKT phosphorylation site at Thr 23 shown to be necessary for NFB activation in response to TNF␣ (29). This phosphorylation site had not previously been tested on any signaling pathway downstream of IL-1. Maximal phosphorylation of these sites in response to IL-1 followed the initial activation of AKT in response to IL-1 but only the Thr 23 site and not the Ser 176 / Ser 180 site was inhibited by LY294002 and wortmannin suggesting that AKT was likely upstream of IKK␣T23 activation with parallel activation by NIK.
Unexpectedly, mutation of the consensus NFB site did not abrogate induction of the IL-6 promoter in response to IL-1 nor did it abrogate the inhibitory effects of either dnAKT or IKK␣T23A (Fig. 6A), however, baseline promoter activation was significantly impaired (Fig. 5A). Inhibition of the NFB reporter plasmids (Fig. 7, C and D) by dnAKT, but not IKK␣T23A, suggested that AKT was necessary for NFB activation but that it was not upstream of IKK␣T23. The lack of an effect of LY294002 or wortmannin on the degradation of IB␣ or on the phosphorylation of IKK␣ Ser 176 /Ser 180 together with the sustained pattern of AKT phosphorylation correlating with the pattern of p50 binding to the NFB response element suggested that AKT might have an effect on p50 binding, this awaits further investigation.
IKK␣ overexpression by itself, either WT or IKK␣T23A, was sufficient to activate the NFB reporters 2-3-fold suggesting that activation of the kinase by autophosphorylation was sufficient to activate NFB without any requirement for AKT. IKK␣ has been detected both in the nucleus and cytoplasm of cells Taken together, this suggested that there was likely another transcription factor target of the AKT/IKK␣ pathway on the IL-6 promoter. Our data supports the model that the role of NFB in the induction of the IL-6 gene in response to IL-1 is likely indirect, perhaps involving activation of NFB responsive AP-1 family members. AKT is also likely involved in this pathway. Thus a positive feedback loop of canonical NFB activation on AP-1 family members to increase IL-1 responsive IL-6 gene transcription might occur. At least one AP-1 family mem-ber, JunB has an NFB responsive promoter. Furthermore Elk-1, a TCF family member and an important regulator of c-fos transcription has an NFB site on its promoter. In fact through NFB-dependent AP-1 activation, NFB could indirectly control the expression of an AP-1 target gene such as IL-6, as it does vascular endothelial growth factor, by increasing the levels of family members, AP-1 making them more available for up-regulation by MAP kinase (see model in Fig. 9).
Mutation of the IL-6 promoter AP-1 site significantly reduced induction of the IL-6 promoter by IL-1. Consistent with this, other investigators have found that the IL-6 promoter AP-1 site is necessary for promoter activation in response to several stimuli including TGF␤ and the Kaposi sarcoma herpes virus (43,62). In our study, mutation of the IL-6 promoter AP-1 site unmasks additional regulation of NFB not observed in the WT IL-6 promoter (Fig. 6B). In this regard a 2-fold activation of this mutant IL-6 promoter by IKK␣WT, not seen in the intact promoter, is also seen on the NFB reporters (Fig. 7, B and C) and suggests negative cross-talk between IKK␣/AP-1 and IKK␣/ NFB pathways. In addition, the IL-6 promoter AP-1 site demonstrated high constitutive activation in Caco-2 cells, suggesting that the AP-1 site placed in the context of the IL-6 promoter may be subject to negative regulation (63). Negative cross-talk has also been described in liver tumor cells in response to TGF␤ where transient NFB activation inhibits AP-1 DNA binding and signaling with important consequences for tumor progression (64). There is evidence in the literature of both positive and negative cross-talk between the NFB and AP-1 pathways and many genes require the simultaneous activation of both transcription factors working cooperatively (64 -66).
the inhibitory effect of dnAKT. Taken together, this is direct evidence that AP-1 is a major downstream target of AKT/IKK␣ and is likely part of a novel IL-1 responsive signaling pathway to the induction of IL-6 gene transcription. Although other investigators have found evidence for the requirement of the IKK complex, as a whole, including IB␣ in AP-1 function, in response to lipopolysaccharide (66), this is the first report of a direct role for AKT/IKK␣ in AP-1 function. Interestingly in IKK␣ knock-out mouse embryonic fibroblasts there was evidence of decreased induction by serum stimulation of both JunB and JunD (59). A link between IKK␣ and c-fos activation in response to epidermal growth factor has been demonstrated to involve the phosphorylation of histone H3 on the c-fos promoter (67). Future work will address if IKK␣ directly phosphorylates AP-1 family members or is upstream of another kinase in AP-1 activation.
In our study there was 20-fold more IL-6 mRNA produced than was translated into protein, suggesting a translational block on IL-6 mRNA in Caco-2 cells. However, previous studies have shown that the 3Ј-untranslated region of the IL-6 mRNA is rich in AU sequences involved in mRNA stability by IL-1, lipopolysaccharide, and TNF␣ in osteoblasts (68). These sequences may not be functional in the Caco-2 cell line where at 4 h post-IL-1 treatment most of the IL-6 mRNA was downregulated suggesting rapid degradation. A role for the AKT/ IKK␣ pathway in regulation of Tor kinase has recently been established with resulting increases in protein synthesis rates (52). It will be of interest to determine whether AKT/IKK␣ might also be involved in IL-6 translational regulation in response to IL-1.
IKK␣ has been shown recently to phosphorylate the CREB co-activator CBP and mediates cytokine-induced phosphorylation and subsequent acetylation of specific residues in histone H3 on NFB responsive promoters (69). IKK␣ phosphorylation of CBP, which increases its histone acetyltransferase activity can switch its binding preference from one transcription factor to another with consequences for cell growth (51). Enhanced IL-6 promoter activation is associated with CBP/p300 binding to p65/NFB as well as to CREB with associated increases in histone acetylation (70). One might speculate that IL-1 signaling via AKT/IKK␣ could switch the binding preference of p300 from NFB to AP-1. Future investigations will address this as well as the role of AKT/IKK␣ and CREB in IL-1 induction of IL- 6. There have been several studies describing a causative role for IL-6 in colon tumor development and progression (71)(72)(73)(74). One of these involves an IL-6 gene variant with high sensitivity to IL-1␤ (75,76). In colon cancer cells harboring this variant IL-6 may be particularly effective in advancing the adenoma/ carcinoma sequence. Interestingly, a dominant-negative mutant of c-Jun exhibited a significant antitumor effect in colon cancer demonstrating the possibility of AP-1-based gene therapy as a novel treatment of colorectal cancer (77)(78)(79)(80). Understanding the regulation of NFB and AP-1 and their cross-talk in the regulation of their target genes such as IL-6 may lead to the development of novel therapeutics for the control of inflammatory diseases of the mucosa.