Acetyl-Boswellic Acids Inhibit Lipopolysaccharide-Mediated TNF-α Induction in Monocytes by Direct Interaction with IκB Kinases¹ Free

Despite the substantial progress in understanding the molecular mechanisms underlying chronic inflammatory diseases and the intensive search for new drugs, the current treatment options are still not satisfactory. Therefore, the quest continues for both novel targets and compounds that combine therapeutic efficacy with a low profile of side effects.

The nuclear transcription factor NF-κB is a key player in the development and progression of chronic inflammatory diseases, such as rheumatoid arthritis, inflammatory bowel disease, asthma, and atherosclerosis (1). NF-κB is therefore considered a promising target for anti-inflammatory intervention (1, 2, 3). The available treatment regimens for chronic inflammatory diseases already include several drugs such as glucocorticosteroids, sulfasalazine, aspirin, and gold compounds that may in very high concentrations and mostly unspecifically mediate part of their effects through inhibition of NF-κB activity (1, 2, 3).

Molecular biological approaches blocking NF-κB by adenoviral transfer of IκBα or by NF-κB decoy inhibited joint inflammation in vitro and in animal models of arthritis (1, 3). Similarly, molecular biological interventions also reduced inflammation in models of chronic intestinal inflammation and myocardial infarction (1, 3). Inhibition of NF-κB leads to a reduction of the inflammatory response, because NF-κB takes the center stage in the regulation of a wide range of genes involved in chronic inflammatory diseases including cytokines such as TNF and IL-1, but also inducible NO synthase and the adhesion molecule ICAM-1 (1, 4). Therapeutic approaches targeting these effector proteins have also been developed. For example, inhibition of TNF-α and its signaling with Abs or soluble receptors has been recognized as a highly successful strategy for the treatment of chronic inflammatory diseases, such as rheumatoid arthritis (5, 6).

Only recently, resveratrol, a polyphenolic phytoalexin present in the skin of red grapes and in several other plants, was found to inhibit NF-κB activation (7). In addition, some other plant-derived compounds have also been reported to interfere with the NF-κB signaling pathway (8, 9). Indeed, plants harbor a plethora of secondary metabolites that might serve as lead compounds for the development of novel therapeutic approaches. In traditional Ayurvedic medicine, extracts from the gum resin from Boswellia serrata, commonly termed Indian frankincense, have been used as anti-inflammatory remedies. Such extracts, which are marketed in the United States, have already been used in small clinical pilot studies for the treatment of rheumatoid arthritis and inflammatory bowel diseases (10, 11, 12, 13). After purification to chemical homogeneity, we have previously characterized the structural configuration of acetyl-boswellic acids (ABAs)³ belonging to the family of pentacyclic triterpenes (14, 15); these compounds are believed to represent the active principle of the aforementioned phytopharmaceuticals (10, 16).

Monocytes and macrophages represent essential effector cells in both chronic inflammation and in the host defense against bacterial infection. Using LPS as a potent activator of human monocytes, we found that acetyl-α-boswellic acid (AαBA) and acetyl-11-keto-β-boswellic acid (AKβBA) inhibit NF-κB signaling. We succeeded in identifying specific inhibitory effects of ABAs on IκBα kinase (IKK), which is pivotal for the degradation of the NF-κB inhibitor IκB, as well as the phosphorylation of p65, two steps essential for NF-κB activation and the subsequent cytokine expression. Using purified human recombinant GST-IKKα and His-IKKβ, we positively confirmed the direct effect of the AαBA and AKβA on the IKK complex. The direct inhibition of IKK places AαBA and AKβBA apart from other plant-derived compounds, such as flavopiridol, and ursolic and betulinic acid, which seem to exert their effects upstream of the IKK (8, 9, 17). Against this background, AαBA and AKβBA could be used either as tools or as lead compounds for the development of novel therapeutic approaches in inflammation research.

AαBA and AKβBA were isolated from African frankincense, purified by reverse-phase HPLC to chemical homogeneity (i.e., >99% purity), and characterized by mass spectrometry and one- and two-dimensional nuclear magnetic resonance spectroscopy (14, 15, 18). The compounds were dissolved in DMSO, and controls contained equivalent amounts of solvent. Abs against p65, IκBα, IKK (SC-7607), and the subunit IKKα were from Santa Cruz Biotechnology, and those against ERK2 (used for immunoprecipitation) were from BD Pharmingen. Abs against the phosphorylated form of p65, and ERK1/2, as well as Elk-1 fusion protein, were from Cell Signaling Technology. Human recombinant GST-IKKα and His-IKKβ were from Upstate. Ab against α-tubulin were from Sigma-Aldrich. Monoclonal anti-CD14 and anti-CD41 Abs were purchased from Immunotech. Rhodamine Red-X-conjugated donkey anti-rabbit IgG F(ab)₂ was purchased from Dianova. Human recombinant p50 and c-Rel were from Promega, p65 was from Active Motif, tagged IκBα fusion protein was from Santa Cruz Biotechnology, and LPS (Escherichia coli serotype 055:B5) was obtained from Sigma-Aldrich. Percoll and poly(dI:dC) were from Amersham Biosciences, TRIzol was from Invitrogen Life Technologies, and neutravidin was from Pierce. Recombinant human TNF-α was obtained from R&D Systems. RPMI 1640, DMEM, and FCS were from Invitrogen Life Technologies. Other chemicals were of analytical grade; all reagents were LPS-free as measured by the Limulus amebocyte lysate assay (Sigma-Aldrich).

Monocytes were isolated by autologous plasma-Percoll gradient centrifugation as described (19, 20, 21). Preparations with ≥94% CD14⁺ cells were used. Contaminating cells were lymphocytes. Flow cytometric analysis (FACScan; BD Biosciences) with anti-CD41 mAb did not reveal any platelets associated with monocytes.

Monocyte and human embryonic kidney epithelial cell line 293 (HEK293; American Type Culture Collection) cell viability was measured, according to the manufacturer’s instructions, either by a modified formazan assay (XTT assay) (22) after 8 h of treatment with AαBA or AKβBA using the Cell Proliferation kit II, or the Cytotoxicity Detection kit (lactate dehydrogenase; Roche Diagnostics). Briefly, 0.25 × 10⁶ monocytes were seeded in 300 μl of phenol red-free RPMI 1640 supplemented with 1% FCS (FCS) in 96-well plate, and treated for 8 h either with DMSO or ABAs (each at 10 μM). For the analysis of the HEK293 viability, the cells were seeded at a density of 5000 cells/300 μl phenol red-free DMEM with 0.5% FCS, treated with AαBA, AKβBA (each at 10 μM), or DMSO, and analyzed 8 h later.

Monocytes (0.5 × 10⁶) were resuspended in 200 μl of RPMI 1640 with 1% FCS, and after preincubation with AαBA, AKβBA, or the solvent DMSO for 60 min, they were stimulated with LPS (100 ng/ml) for 6 h; in previous experiments, this LPS concentration was found to trigger a near-maximum release of TNF-α into the medium (21, 23). Total RNA isolated with TRIzol (Invitrogen Life Technologies) was analyzed with specific primers for TNF-α and GAPDH as an internal standard (19). PCR did not reach the saturation phase. Control experiments showed no DNA contaminations. The amplification products were identified by direct sequencing (Prism 310; Applied Biosystems).

The supernatants of cells treated with AαBA, AKβBA, or solvent, as described above, were used for TNF-α measurements with an ELISA (R&D Systems).

The HEK293 was transiently transfected with the vector pNFκB-Luc containing four tandem copies of the κB enhancer element upstream of the firefly luciferase reporter gene (Clontech). One day before transfection, 0.2 × 10⁶ cells/500-μl DMEM with 10% FCS were seeded into 24-well plates. The vector pNFκB-Luc (0.5 μg) was transfected with Superfect reagent (Qiagen), according to the manufacturer’s instructions. Media were changed 3 h after transfection. Twenty-four hours after transfection, media were replaced by DMEM with 0.5% FCS for 6 h to reduce background NF-κB activation due to FCS. Subsequently, cells were treated with AαBA, AKβBA, or the solvent DMSO for 1 h, followed by stimulation with TNF-α (100 ng/ml). After 4 h, the cells were washed, harvested in 200 μl of 0.1 M potassium phosphate buffer (pH 7.8), lysed by three freezing/thawing cycles, and analyzed for protein contents with the BCA kit (Pierce). Aliquots of 20 μl were measured in 96-well microtiter plates in a PlateLumino luminometer (Stratec) with 10-s integration time of the luciferase reaction. The luciferase activities were normalized to the protein contents. Results are expressed as fold change from the nonstimulated promoter activity. Lysates from each transfection were assayed in triplicate from at least three independent transfection experiments. Control cells were transfected with pTal-Luc vector (Clontech) and treated with TNF-α (100 ng/ml) in the same way as samples from pNF-κB-Luc-transfected cells. The control cells showed no increase in luciferase activity, indicating that the effects observed were due to NF-κB activation.

Alternatively, the cells were transfected with an IFN-stimulated response element (ISRE) luciferase reporter gene (Stratagene) either alone, or together with the constitutively active form of IFN regulatory factor (IRF)-3 (IRF-3 5D) (24). Twenty-four hours posttransfection, the medium was replaced with DMEM containing 0.5% FCS, and cells were treated with ABAs or solvent for 6 h. Luciferase expression was analyzed as above.

Freshly isolated human monocytes (5 × 10⁶) resuspended in 500 μl of RPMI 1640 supplemented with 1% FCS were cultured on hydrophobic PetriPerm membranes (Vivascience); they were preincubated in the presence of AαBA, AKβBA (3 and 10 μM each), or the solvent DMSO for 60 min, followed by stimulation with LPS (100 ng/ml) for 60 min. Nuclear extracts (5 μg) were subjected to EMSA as previously described (21, 23, 25). For competition experiments, nuclear extracts were incubated for 30 min with a 100-fold excess of unlabeled specific NF-κB or AP-2 oligonucleotides.

Binding of p50/c-Rel and p50/p65 heterodimers to NF-κB binding sites was measured by SPR using a CMD-20 B2 sensor chip. Double-stranded biotinylated DNA containing the NF-κB binding site (AGTTGAGGGGACTTTCCCAGGC) was immobilized on a SPR sensor chip (XanTec Analysensysteme) (23, 26). Mixtures of human recombinant p50 and c-Rel, or p50 and p65 (200 nM each) in 60 μl of NF-κB binding buffer (10 mM Tris-HCl (pH 7.5), 150 mM NaCl, 1 mM MgCl₂, 0.5 mM EDTA, 0.5 mM DTT, 0.1% BSA, 0.1% Nonidet P40, and 3 μg of poly(dI:dC)) were preincubated for 30 min at 20°C with 100 μM of either AαBA or AKβBA, or equivalent amounts of DMSO. The mixture was applied to the sensor chip, and binding to DNA carrying the NF-κB binding site was analyzed with a dual-channel ESPRIT optical sensor device (Autolab). The binding of the recombinant proteins to DNA containing the AP-1 consensus binding sequence (CGCTTGATGAGTCAGCCGGAA) was used as a control. Alternatively, human recombinant GST-IKKα and His-IKKβ were immobilized on the surface of the SPR sensor chip. AαBA or AKβBA (100 μM each) in kinase buffer (20 mM HEPES (pH 7.5), 10 mM MgCl₂, 20 mM β-glycerophosphate, 100 μM Na-orthovanadate, 1 mM DTT, and 0.01% Nonidet P40) was applied onto the sensor chip surface. Pretreatment for 20 min at 20°C with 100 nM of either GST-IKKα or His-IKKβ before the analysis of the binding to immobilized kinase was used to ensure the specificity of the binding.

Monocytes (2 × 10⁶) resuspended in 800 μl of RPMI 1640 with 1% FCS, were permitted to adhere to chamber slides (Nalgene Nunc) for 30 min, and were treated for 60 min with the solvent DMSO or 10 μM of either AαBA or AKβBA, followed by stimulation with LPS (100 ng/ml) for an additional 60 min. Monocytes were fixed, permeabilized with 1% Triton X-100, and stained with Hoechst 33342 (DNA marker), FITC-labeled anti-α-tubulin Ab (cytosol marker), and rabbit anti-p65 visualized with anti-rabbit Rhodamine Red-labeled secondary Ab. The cells were analyzed with a Leica DM IRBE confocal laser-scanning microscope (Leica Microsystems).

Monocytes (1–2 × 10⁶ cells/sample) were treated with the indicated concentrations of AαBA or AKβBA for 60 min, and were subsequently stimulated with LPS (100 ng/ml) for an additional 60 min. Whole-cell lysates, and cytosolic and nuclear fractions were prepared and analyzed as described (21, 25). p65 was analyzed in cytosolic and nuclear fractions. Phosphorylation of p65 was analyzed in monocyte nuclear extracts. Abs against IκBα, p65, and the phosphorylated form of p65 (Ser⁵³⁶) were used. For the control of equal protein loading, blots were reprobed with ERK1/2, p65, or topoisomerase I Ab.

Monocytes (20 × 10⁶) were resuspended in 2 ml of RPMI 1640 and stimulated with LPS (1 μg/ml) for 30 min. Monocytes were lysed with buffer containing 0.1% Nonidet P-40. Lysates were precleared with rabbit IgG and protein agarose beads. The IKK complex or ERK2 were immunoprecipitated from the precleared cell lysates with appropriate rabbit Abs and protein A-agarose beads. After extensive washing of immunoprecipitated IKKs, equal amounts of kinases in terms of protein were pretreated with different concentrations of AαBA or AKβBA at 20°C for 15 min and used for kinase assays with rIκBα-tagged fusion protein corresponding to full-length IκBα (aa 1–317) of human origin, or recombinant p65 in the presence of ³²P-labeled ATP, at 30°C for 20 min. ERK2 kinase assay was performed in analogy using a rElk-1 fusion protein as substrate. Samples were separated by SDS-PAGE and blotted onto nitrocellulose membranes. Phosphorylated IκBα, p65, and Elk-1 were visualized and quantified using a PhosphorImager (Molecular Dynamics) (21, 25). Alternatively, 30 nM human recombinant GST-IKKα or His-IKKβ were treated with 0.1–10 μM of either AαBA or AKβBA, or solvent and analyzed as indicated above.

For in vitro knockdown of IKKs, phosphorothioate oligodeoxynucleotides (ODN; ThermoHybaid) were used. The ODN were selected on the basis of the major predicted secondary structures, i.e., loops (27). The antisense ODN used against IKKα and IKKβ were 5′-CAATTATTTTATGTATT-3′ and 5′-GTCGACGGTCACTGTGT-3′, respectively; control ODN was 5′-AAACAGAATCATCCATC-3′.

Monocytes were treated with 0.1 μM IKK-specific antisense or control phosphorothioate ODNs in DMEM supplemented with 10% FCS for 28 h (28). After treatment, media were replaced by RPMI 1640, and the cells were allowed to recover for 12 h to reduce any putative procedure-induced signaling. Subsequently, the cells were incubated in RPMI 1640 supplemented with 1% human AB serum and were used for additional experiments.

Values shown represent mean ± SEM where applicable. Statistical significance was calculated with the Newman-Keuls test. Differences were considered significant for p < 0.05.

Macrophages and monocytes are considered to be the main source of TNF-α (29), which as a prototypical proinflammatory cytokine plays a key role not only in chronic inflammatory diseases but also in innate immunity (6, 30). In this study, we have investigated whether AαBA or AKβBA is able to affect the TNF-α generation in LPS-stimulated human peripheral monocytes. TNF-α is synthesized as a precursor, which is processed and released from the membrane (31), implying that regulation can occur at any of those steps. Therefore, we have measured TNF-α expression at both the mRNA and the protein level. Stimulation of monocytes with 100 ng/ml LPS triggered an increased expression of TNF-α mRNA, which was concentration-dependently inhibited by AαBA and AKβBA (Fig. 1 A). At a concentration of 10 μM, AKβBA was a more potent inhibitor of TNF-α mRNA expression than AαBA.

We further analyzed the effects of the ABAs on the release of mature TNF-α into the medium. Stimulation of monocytes with 100 ng/ml LPS induced the expected TNF-α release that was concentration-dependently inhibited by both ABAs (Fig. 1,B). Similar to the inhibition of the mRNA expression, AKβBA at a concentration of 10 μM was more potent than AαBA. Treatment of monocytes with 10 μM AαBA or AKβBA alone significantly affected neither the expression of TNF-α (Fig. 1 B) nor the monocyte viability as measured by the XTT assay (data not shown).

NF-κB activation is essential for the expression of various proinflammatory genes, including the TNF-α gene, which contains several binding sites for NF-κB in its promoter region (32). Only recently has it been shown that the plant-derived compound resveratrol exerts its effects at least partially through inhibition of the NF-κB activation (7). Against the background of the observed down-regulation of TNF-α mRNA, we decided to analyze the effects of ABAs on NF-κB activation in a luciferase gene reporter assay, where the amount of the luciferase gene product reflects the extent of NF-κB activation. The low concentrations of AαBA and AKβBA, i.e., not >10 μM, did not affect the viability of HEK293 cells as judged by the lactate dehydrogenase release assay (data not shown). TNF-α-mediated stimulation of cells transfected with NF-κB reporter vector, but not with the control vector, resulted in a 17-fold increase of the luciferase activity. Both AαBA and AKβBA concentration-dependently inhibited the NF-κB activation in transfected HEK293 cells (Fig. 2 A); pretreatment with AαBA and AKβBA (10 μM) inhibited the NF-κB activity by 40.9 ± 9.8 and 76.9 ± 7.6%, respectively.

To prove the specificity of the observed inhibitory effects of ABAs, we analyzed their effects on the ISRE-mediated luciferase expression. ABAs did not affect the basal luciferase expression by pISRE-luciferase-transfected HEK293 cells. Cotransfection of pISRE-luciferase with the constitutively active form of IRF-3, IRF-3 5D, resulted in a >100-fold increase in luciferase expression, which was not affected by pretreatment with 10 μM of either AαBA or AKβBA (Fig. 2 B). This indicated that ABAs specifically inhibit NF-κB.

To confirm the inhibitory effects of AαBA and AKβBA on NF-κB in activated human monocytes, we stimulated monocytes with 100 ng/ml LPS and analyzed the NF-κB activity by EMSA. LPS stimulation led to the expected NF-κB activation, which was concentration-dependently inhibited by AαBA and AKβBA (3 and 10 μM), and almost abolished by preincubation with 10 μM AKβBA (Fig. 2 C). Inhibition of the NF-κB activation by EMSA may reflect either a lack of NF-κB proteins in the nuclei, i.e., inhibition of translocation, or inhibition of DNA binding, or inhibition of transactivation, or a combination thereof.

We have previously shown that, in an enzymatic assay system, high concentrations of ABAs are able to reduce binding of human topoisomerases to substrate DNA (26). This prompted us to investigate whether AαBA and AKβBA affect NF-κB activity through a similar mechanism. Binding of human recombinant p50/c-Rel and p50/p65 heterodimers was analyzed in vitro by SPR. dsDNA containing NF-κB binding sites was immobilized on a SPR sensor chip. The addition of rNF-κB proteins to the liquid phase resulted in an increase in the SPR signal reflecting binding of the proteins to the immobilized DNA. There was no binding of rNF-κB proteins to dsDNA containing the AP-1 binding sequence, which was linked to the sensor chip of the reference channel. Preincubation of p50/c-Rel (Fig. 3 A) or p50/p65 (B) with up to 100 μM AαBA or AKβBA did not affect their binding to DNA carrying NF-κB binding sites. Thus, ABAs do not interfere with the binding of NF-κB to DNA. These data indicated that ABAs inhibit NF-κB activation upstream of DNA binding.

After phosphorylation and degradation of the inhibitors, NF-κB proteins translocate to the nucleus and activate NF-κB-dependent genes. Through laser-scanning microscopy of monocytes stained with fluorescence-labeled Ab against p65, we analyzed the nuclear accumulation of p65 in the presence of AαBA and AKβBA. To distinguish between nucleus and cytosol, monocytes were stained with DNA-specific Hoechst 33342 and with anti-α-tubulin Ab. In nonstimulated monocytes, p65 was diffusely distributed throughout the cytosol and the nucleus (Fig. 4,A). p50 and p65 contain a nuclear-export sequence, which is effectively masked by IκBα only at p65 (4). Therefore, it has been proposed that p65/p50 might be shuttling into the nucleus in nonstimulated cells, but, being bound to the inhibitor, is not able to activate genes. After stimulation with 100 ng/ml LPS, p65 nearly disappeared from the cytosol and strongly accumulated in the nucleus (Fig. 4,A), a distribution inhibited by both 10 μM AαBA and 10 μM AKβBA. After treatment of monocytes with the ABAs in the absence of LPS, the pattern of the p65 subcellular distribution was similar to that in the nonstimulated cells (data not shown). Similar results were obtained when the subcellular distribution of p65 was studied by Western blotting analysis of the cytosolic and nuclear fractions (Fig. 4 B); again, 10 μM AKβBA basically abolished and 10 μM AαBA severely hampered the nuclear localization of p65 in LPS-stimulated monocytes.

To investigate the mechanism of the NF-κB inhibition by ABAs, we analyzed the degradation of the NF-κB inhibitor, IκBα, in the presence of AαBA and AKβBA. IκBα is present in nontreated cells and is degraded upon stimulation with 100 ng/ml LPS (Fig. 5 A). However, preincubation of monocytes with the AαBA and AKβBA concentration-dependently inhibited the LPS-induced degradation of IκBα. AKβBA appeared to be more potent than AαBA.

The gene transactivation activity of NF-κB depends also on posttranslational modifications, as, for example, phosphorylation of the NF-κB proteins. Only recently has it been demonstrated that IKKβ is essential for the phosphorylation of p65 at Ser⁵³⁶ (33), although a role for IKKα in this process has also been suggested (34). The phosphorylation increases the transcriptional activity of p65. Therefore, we further analyzed whether ABAs in addition to the degradation of IκBα also affect the LPS-induced phosphorylation of p65. Treatment of monocytes with 100 ng/ml LPS led to p65 phosphorylation in nuclear extracts (Fig. 5 B). Pretreatment with AαBA concentration-dependently reduced, and pretreatment with AKβBA abolished the LPS-induced phosphorylation of p65.

Both IκBα and p65 are phosphorylated by IKK (1, 4, 33, 34, 35). Hence, we hypothesized that ABAs might inhibit IKK. The effects of AαBA and AKβBA on IKK were analyzed in an in vitro kinase assay using immunoprecipitated IKK. The Ab used recognizes both the IKKα and IKKβ subunits. As expected, LPS (1 μg/ml) triggered activation of IKK within 30 min (Fig. 6,A). We immunoprecipitated activated IKK complex from monocytes stimulated with 1 μg/ml LPS for 30 min and analyzed phosphorylation of recombinant IκBα and p65 in the presence of AαBA and AKβBA. In the presence of the solvent DMSO, the immunoprecipitated IKK complex phosphorylated rIκBα. This IKK-mediated phosphorylation was concentration-dependently inhibited by both AαBA and AKβBA (1–10 μM) (Fig. 6,B). By contrast, both AαBA and AKβBA did not affect ERK2 activity, supporting the view that the ABAs exert a specific inhibitory effect on the IKK complex. Inhibition of IKKs by ABAs was confirmed by performing an in vitro kinase assay with active human recombinant GST-IKKα and His-IKKβ. AαBA and, to a larger extent, AKβBA inhibited activity of both GST-IKKα and His-IKKβ. As little as 1 μM of AKβBA affected the phosphorylation of IκBα by either kinase (Fig. 6,C). We used the more potent compound AKβBA to prove direct interaction with rIKKs using SPR analysis. AKβBA evidently binds to GST-IKKα with higher affinity than to His-IKKβ (Fig. 6 D). This binding was specific, because it was inhibited by pretreatment of AKβBA with GST-IKKα or His-IKKβ before measurement. Thus, we have identified AαBA and AKβBA as distinct inhibitors of IKK activity.

It has previously been suggested that LPS-induced expression of proinflammatory cell activation involves activation of IKKα and IKKβ (36, 37, 38). To confirm the link between IKK and TNF-α induction in our LPS-stimulated monocytes, we used an in vitro knockdown approach, using IKK-specific antisense ODN. Through immunoblot analysis, we confirmed that treatment of monocytes with IKK-specific antisense, but not with control ODN, induces significant down-regulation of IKKα and IKKβ (Fig. 7 A). In line with these findings, the control ODN did not significantly affect the TNF-α release into the medium, whereas the antisense ODN against IKKα and IKKβ significantly inhibited the TNF-α release by 46.1 ± 12.4% (n = 4; p < 0.01) and 79.4 ± 3.7% (n = 4; p < 0.01), respectively. Thus, inhibition of IKKα and IKKβ clearly leads to inhibition of the TNF-α release in the LPS-stimulated monocytes.

In a number of clinical pilot studies, therapeutic benefits have been claimed when chronic inflammatory diseases, such as rheumatoid arthritis, inflammatory bowel diseases, or asthma, were treated with ABA-containing extracts (10, 11, 12, 13, 39). The anti-inflammatory mechanism of action of ABAs, specifically of AKβBA, was originally thought to involve inhibition of the 5-lipoxygenase pathway of arachidonic acid metabolism (16, 40). Although leukotrienes generated via this pathway were believed to act as important lipid mediators of inflammatory diseases, a number of potent leukotriene antagonists and inhibitors failed in clinical studies, casting doubts on the significance of these mediators. Despite enormous industrial efforts, currently only a few compounds with moderate therapeutic activity, such as montelukast, zafirlukast, or zileuton, are being used for the sole indication of asthma (41). In the light of the claimed therapeutic efficacy of ABA-containing therapeutics in chronic inflammatory diseases, we assumed that ABAs might target different proinflammatory mechanisms that might be worth therapeutic exploitation.

A common theme in chronic inflammation is stimulation of the proinflammatory transcription factors AP-1 and NF-κB (1, 3, 42). Having observed that ABAs inhibit TNF-α release, we tested the hypothesis that ABAs might exert their anti-inflammatory activity through inhibition of NF-κB.

We used endotoxic LPS, which is one of the most potent activators of monocytes (29, 36). LPS is first engaged by CD14 and then brought into direct contact with the LPS-responsive TLR4 and the coreceptor MD-2 (43, 44). Toll signaling to NF-κB originates from the conserved Toll-IL-1R (TIR) domain, which mediates recruitment of the TIR domain-containing adapter molecule MyD88 that is important for signaling through all TLRs. The recruitment of MyD88 to cytoplasmic TIR domains of activated TLR4 allows for the interaction and activation of the IL-1R-associated kinase family members and the subsequent activation of TNFR-associated factor-6 (45). A larger protein complex that contains TNFR-associated factor-6 activates TGF-β-activated kinase 1, which phosphorylates the IKK complex and thereby induces the activation of transcription factors NF-κB and AP-1 (44, 46). Apart from the MyD88-dependent pathway, in dendritic cells and macrophages TLR4 can also trigger signaling in a MyD88-independent fashion that requires interaction with additional adapter proteins, such as TIR domain-containing adapter-inducing IFN-β-related adapter molecule and TIR domain-containing adapter-inducing IFN-β (43, 47, 48). However, information on this signaling mechanism is still incomplete, and it has been suggested that the TLR4 signaling pathway might acquire activation of both the MyD88-dependent and -independent pathways to induce inflammatory cytokines, such as TNF-α (44).

Activation of NF-κB is a multistep process that involves activation of the IKK complex, phosphorylation of inhibitors and their degradation, transport of NF-κB proteins to the nucleus, binding to the NF-κB-consensus sequence, and activation of genes (1, 4). In unstimulated cells, mature NF-κB dimers are trapped in the cytoplasm by interaction with the inhibitory proteins termed IκBs, which mask the nuclear localization sequence of NF-κB proteins. In response to stimuli, the IκB proteins are phosphorylated, which enables NF-κB to enter the nucleus where it activates gene expression such as TNF (4, 49). It is the multi-subunit IKK complex, consisting of two catalytic subunits, IKKα and IKKβ, and a regulatory subunit IKKγ, that phosphorylates IκB proteins on two distinct serine residues, thereby targeting them to rapid ubiquitin-dependent proteolysis that initiates the activation of NF-κB (4, 49). Recently, additional posttranslational modifications of the NF-κB proteins, such as phosphorylation or acetylation, have been shown to be essential for the stimulatory activity of NF-κB (4). Inhibition of either step would lead to the impaired activation of target genes.

LPS stimulation triggers activation of IKK in monocytes and the monocytic THP-1 cell line. From transfection experiments with dominant-negative mutants of IKKα and IKKβ in THP-1 cells, it was concluded that IKKβ plays a dominant role in LPS-induced signaling (36). This finding contrasts with recent data from macrophages showing that, for the LPS-induced NF-κB activation and TNF-α production, IKKβ is not required (50). Others have shown that LPS stimulation of THP-1 cells leads to activation of IKKα and IKKβ (38). Furthermore, activation of both IKKα and IKKβ appears to be essential for CD14-independent LPS signaling (37). Using the IKK-specific antisense ODN approach, our data clearly indicate that both IKKα and IKKβ are engaged in the LPS-stimulated TNF-α production in human primary monocytes. In that regard, they are consistent with data from mouse embryonic fibroblasts, where it was shown that IKKα is just as critical as IKKβ and NEMO/IKKγ for the global activation of NF-κB-dependent, TNF-α- and IL-1-responsive genes (49), indicating that each IKK might be required for the NF-κB-mediated inflammatory response program.

Several plant-derived compounds have been shown to interfere with the NF-κB pathway (7, 8, 9, 17, 51). Despite the fact that ursolic and betulinic acids are structurally similar to the ABAs, they do not directly interfere with the IKK activity (8, 9). Similarly, the synthetic flavone flavopiridol exerts its activity, probably, via inhibition of Akt, and has no direct effect on the IKKs (17). Curcumin seems to inhibit the activity of immunoprecipitated IKKs; however, these data were not confirmed with purified enzymes, nor were effects on other kinases excluded (51).

With both native IKK complexes immunoprecipitated from LPS-activated monocytes, as well as recombinant active GST-IKKα and His-IKKβ, we demonstrate that AαBA and AKβBA inhibit the IKK activity. The inhibitory effect on IKK activity is reflected by the inhibition of phosphorylation of IκBα and of p65 on Ser⁵³⁶, resulting in reduced nuclear translocation of p65 and the inhibition of subsequent NF-κB-dependent expression of TNF-α. Phosphorylation of IκBα in conjunction with that of p65 on Ser⁵³⁶ located in the TA1 transactivation domain was originally identified in TNF-α-stimulated HeLa cells (35), a finding that has also been confirmed in other cell types, including LPS-stimulated mouse macrophages and human monocytic cell lines (33, 35, 38). The existing evidence from various in vitro studies, transfection, as well as from knockout experiments, indicates that p65 phosphorylation on Ser⁵³⁶ in the cytoplasm is catalyzed by both IKKα and IKKβ (34, 52). Although phosphorylation of p65 in general and on Ser⁵³⁶ (33) has been implicated in enhanced NF-κB transcriptional activity (53), the precise physiological role of the Ser⁵³⁶ phosphorylation is at present still unclear (52).

As to the potential pharmacotherapeutic use of IKK inhibitors, there are quite some reservations against the background that IKKβ knockout in mice is associated with embryonic lethality owing to TNF-α-driven massive liver necrosis (54). Similarly, ablation of IKKβ in mouse enterocytes prevented the systemic inflammatory response upon mesenteric ischemia-reperfusion injury, but also resulted in severe apoptotic damage to the reperfused intestinal mucosa (55). In contrast to IKKβ knockout mice, IKKα knockout mice show normal liver development, but skin and limb abnormalities (56). Furthermore, there are also concerns regarding side effects when IKK is inhibited. However, preliminary data from our laboratory show that mice treated with therapeutic doses of ABAs do not exhibit any major toxic effects. Similarly, patients using ABA-containing phytopharmaceuticals do not experience major side effects either, suggesting that inhibition of IKK might offer realistic therapeutic opportunities.

Oral administration of a single dose of 1200–1600 mg of ABA-containing extract preparations yielded effective plasma concentrations of 2–32 μM of various acetylated and nonacetylated boswellic acids (11, 57, 58), yet a high bioavailability of boswellic acids strongly depends on concomitant food intake (58). Treatment of glioblastoma patients with an extract from the gum resin of B. serrata, i.e., Indian frankincense, over 10 days led to a plasma levels of acetylated and nonacetylated forms of α-boswellic acid of 4 μM (18). Thus, oral intake of frankincense extracts may very well yield concentrations required for the inhibition of NF-κB signaling.

In conclusion, our data demonstrate that selective inhibition of IKK represents a potential therapeutic target for the suppression of NF-κB-dependent cytokine expression, and that both AαBA and AKβBA are novel selective inhibitors of IKK activity. Taken together, these findings offer new perspectives for novel therapeutic approaches.

We are grateful to Dr. M. Cathcart for valuable advice concerning the ODN experiments, and we thank J. Marinaci and W. Zugmaier for expert technical assistance.