Rocaglamide Derivatives Are Potent Inhibitors of NF-κB Activation in T-cells

Crude extracts from different Aglaiaspecies are used as anti-inflammatory remedies in the traditional medicine of several countries from Southeast Asia. Because NF-κB transcription factors represent key regulators of genes involved in immune and inflammatory responses, we supposed that the anti-inflammatory effects of Aglaia extracts are mediated by the inhibition of NF-κB activity. Purified compounds ofAglaia species, namely 1H-cyclopenta[b]benzofuran lignans of the rocaglamide type as well as one aglain congener were tested for their ability to inhibit NF-κB activity. We show that a group of rocaglamides represent highly potent and specific inhibitors of tumor necrosis factor-α (TNFα) and phorbol 12-myristate 13-acetate (PMA)-induced NF-κB-dependent reporter gene activity in Jurkat T cells with IC50 values in the nanomolar range. Some derivatives are less effective, and others are completely inactive. Rocaglamides are able to suppress the PMA-induced expression of NF-κB target genes and sensitize leukemic T cells to apoptosis induced by TNFα, cisplatin, and γ-irradiation. The suppression of NF-κB activation correlated with the inhibition of induced IκBα degradation and IκBα kinase activation. The level of interference was determined and found to be localized upstream of the IκB kinase complex but downstream of the TNF receptor-associated protein 2. Our data suggest that rocaglamide derivatives could serve as lead structures in the development of anti-inflammatory and tumoricidal drugs.

In recent years Aglaia species have attracted considerable interest due to their unique 1H-cyclopenta[b]benzofuran lignans, which have been isolated from more than ten Aglaia species so far and are exclusively confined to members of this genus (1)(2)(3)(4). Rocaglamide derivatives are potent natural insecticides that are comparable with regard to their activity to azadirachtin from the Neem tree Azadirachta indica (1)(2)(3)(4). Moreover, these compounds were found to have cytostatic ac-tivity in human cancer cell lines (5)(6)(7). These cytostatic effects were comparable to the established anticancer drugs such as vinblastine sulfate and actinomycin D (8,9). Furthermore, rocaglamides have been shown to block protein biosynthesis and to induce growth arrest in the G 2 /M phase in certain tumor cells lines (8). However, the one or more underlying molecular mechanisms of these rocaglamide mediated effects have not been identified.
The members of the Rel/NF-B transcription factor family bind to DNA as homo-and/or heterodimers. They are critically involved in the regulation of genes mediating inflammatory responses and cellular processes such as cell survival, apoptosis, development, differentiation, cell growth, and neoplastic transformation (reviewed in Ref. 10). In unstimulated cells, NF-B dimers are sequestered in the cytoplasm due to the interaction with proteins of the IB family. Stimulation of cells, e.g. by pro-inflammatory agents results in the rapid activation of the IB kinase (IKK) 1 complex. This complex consists of two kinases IKK1/␣ and IKK2/␤ as well as a regulatory component called NEMO/IKK-␥ (10,11). Its activation results in the de novo phosphorylation of conserved serine residues in the Nterminal domain of the IB proteins marking them for ubiquitination and subsequent degradation by the proteasome. This allows nuclear translocation of NF-B and binding to cognate DNA motifs in the promoter region of target genes, which subsequently initiates transcription of these genes and finally starts a genetic program responsible for e.g. inflammatory responses (12).
Numerous efforts have been initiated to develop or to identify specific low molecular weight compounds to inhibit this pathway (13,14). Substances that inhibit the proteasome as well as radical scavengers have been shown to block NF-B activation. These inhibitors have been valuable for many studies of NF-B functions in cell culture systems. However, because these compounds also affect multiple other cellular reactions, they are not useful as therapeutic agents in vivo (13). Substances that either block the action of the IB kinase complex or interfere with its activation are thought be more specific and should have fewer side effects. Recently, a novel reversible inhibitor of the proteasome, PS-341, a dipeptidyl boronic acid analogue, has been described to be a potent inhibitor of NF-B activation (15,16). This compound has excellent bioavailability and stability and has been shown to yield effective anti-cancer responses both in vitro and in vivo (17,18). Phase II trials in several hematological malignancies and solid tumor types are now in progress (19).
In the past, several natural occurring compounds like sesquiterpene lactones, curcumin, tetrandrine, and triptolide have been shown to inhibit NF-B activation (20 -27). However, their broad side effects and the relatively high doses required for NF-B inhibition make them less feasible as therapeutic drugs (13). The rationale for studying a series of rocaglamide derivatives for their influence on NF-B activity was based on the observation that leaves and flowers of Aglaia duperreana and Aglaia odorata are used in the traditional medicine of several countries from Southeast Asia (e.g. Vietnam) for the treatment of asthma and inflammatory skin diseases. Here we show that certain rocaglamide derivatives are efficient inhibitors of NF-B activation and NF-B target gene expression, which could explain the anti-inflammatory function of these herbal remedies. Interestingly, they show a high degree of cell type specificity being much more active in T lymphocytes than in other cell types. Furthermore, inhibition of NF-B by rocaglamides is responsible for sensitization of resistant leukemic T cells toward cancer therapy induced apoptosis. The mechanism of NF-B inhibition by rocaglamides is predominantly localized upstream of the IB kinase complex.

EXPERIMENTAL PROCEDURES
Cells and Cell Culture-CEM-S (acute T cell leukemia) cells die rapidly in response to ␥-irradiation, treatment with chemotherapeutic agents, or direct triggering of death receptors. Subclones of CEM-S have been selected from the parental cells by periodical triggering of CD95 with ␣APO-1 (50 ng/ml to 10 g/ml) for at least 1 year (28). These subclones (CEM-R) do not die in response to ␣APO-1 and exhibit a strongly delayed percentage of apoptosis following ␥-irradiation or treatment with chemotherapeutic agents. These cell lines were grown in RPMI 1640 medium supplemented with 10% heat-inactivated fetal calf serum (Biochrom, Hamburg, Germany), 100 units/ml penicillin, 100 g/ml streptomycin, 25 mM HEPES, and 2 mM L-glutamine (all from Invitrogen, Karlsruhe, Germany). Jurkat T and Jurkat T cells deficient in IKK␥/NEMO expression (kindly provided by S.-C. Sun, Pennsylvania State University), EL-4, HeLa, S107, NIH3T3, and A549 cells were cultured in Dulbecco's modified Eagle's medium (Invitrogen, Paisley, Scotland) containing 10% fetal calf serum (PAN Systems, Aidenbach, Germany), 50 M ␤-mercaptoethanol, 100 units/ml penicillin, and 100 g/ml streptomycin (Invitrogen, Karlsruhe, Germany). PC12 cells were cultured in the same medium but containing 10% horse serum (PAN Systems, Aidenbach, Germany), 5% fetal calf serum, and no ␤-mercaptoethanol.
Stimulation of Cells-Cisplatin (Sigma, Deisenhofen, Germany) was dissolved in Me 2 SO at a concentration of 10 g/l, stored in aliquots at Ϫ80°C and further diluted in medium to the working solution of 1 M. PMA (Sigma, Deisenhofen, Germany), Ionomycin (Calbiochem, Germany) and the various rocaglamide derivatives were also dissolved in Me 2 SO and used at the indicated concentrations. Human recombinant TNF␣ (a gift of Dr. G. Adolf, Boehringer Ingelheim, Vienna, Austria) was dissolved in a buffer containing 10 mM sodium phosphate, pH 7, 200 mM sodium chloride, and 2 mg/ml bovine serum albumin and used at the indicated concentrations. Cells were ␥-irradiated (10 Gy) in their flasks using a cesium radiator.
Transfection Experiments-For transient transfections, 3 g of the NF-B-dependent or the AP-1-dependent luciferase reporter and 50 ng of Renilla luciferase reporter (under control of the ubiquitin promoter) or 1 g of Rous sarcoma virus LacZ were cotransfected with 20 g of empty vector (pcDNA3) or expression vectors for TRAF2, NIK, MEKK1, and IKK2-EE (29 -33). Cells (1 ϫ 10 7 ) were transfected in 300 l of culture medium by electroporation using a Bio-Rad gene pulser (Germany) at 975 microfarads (Fa) and 250 V (for Jurkat, A401, and EL-4) and at 250 Fa and 450 V (for HeLa). After electroporation cells were immediately resuspended in medium, seeded in six-well plates, and treated 16 h later as indicated. Cells were harvested 7 h later, and luciferase enzyme activity was measured. Renilla luciferase activity or ␤-galactosidase activity were used to normalize for differences in transfection efficiencies. For generation of stable transfectants of the Jurkat T and PC12 cell lines, cells were electroporated with 20 g of the 3xB.luc reporter plasmid together with 1 g of a pSV.puro vector at 975 Fa and 250 V and at 250 Fa and 450 V, respectively. Cell clones with an integrated reporter gene were selected in medium containing 2 g/ml (Jurkat) and 2.5 g/ml (PC12) puromycin.
Electrophoretic Mobility Shift Assay-Preparation of whole cell extracts was performed by the freeze-thaw method, and EMSAs were performed as described earlier (34). In all cases, whole cell extracts were incubated with radiolabeled double-stranded oligonucleotides containing a Ig-enhancer consensus NF-B site, a octamer-specific site (35), or an Sp-1-specific site (5Ј-attcgatcggggcggggcgagc-3Ј) for 20 min at room temperature, and the DNA-protein complexes formed were then separated from free oligonucleotides on a native 4% polyacrylamide gel.
Immune Complex Kinase Assays and Western Immunoblots-For IKK activity assays, cells were lysed in Nonidet P-40 lysis buffer (36), and the endogenous IKK complex was immunoprecipitated from 1 mg of extract using IKK1/2-specific antibodies (Santa Cruz Biotechnology, sc-7607). The precipitated IKK complex was incubated in kinase assay mixture containing 25 mM HEPES (pH 7.5), 150 mM NaCl, 25 mM ␤-glycerophosphate, 10 mM MgCl, 1 mM dithiothreitol, 10 Ci of [␥-32 ]ATP and 600 ng of GST-IB␣ substrate. Coupled in vitro kinase assays were performed in kinase assay mixture supplemented with 1 M unlabeled ATP. After 20 min (30°C) the reaction was terminated by boiling with SDS sample puffer, and the proteins were separated on 10% polyacrylamide gels. Finally, the gel was either dried or the proteins were electrotransferred to a polyvinylidene difluoride membrane, and radioactive bands were visualized by phosphorimaging or autoradiography. Kinase activities were quantified by PhosphorImager (Amersham Biosciences) analysis. The membranes were used for immunoblot analysis as described elsewhere (37) and were labeled with antibodies specific for IKK1 (Santa Cruz Biotechnology, sc-7183) to determine total amounts of immunoprecipitated IKK1. Western immunoblot analysis for monitoring IB␣ degradation and RelA expression were performed as described earlier (37) using IB␣ (Cell Signaling, #9242)-and RelA (Santa Cruz Biotechnology, sc-372)-specific antibodies.
Measurement of Apoptosis-Early apoptotic changes were identified by staining of cells with fluorescein thiocyanate-conjugated annexin V and propidium iodide (Becton Dickinson, Heidelberg, Germany) and analyzed by flow cytometry (FACScan, Becton Dickinson) as described earlier (38).

Dose-dependent Inhibition of NF-B by
Rocaglamide Derivatives-Sixteen naturally occurring 1H-cyclopenta[b]benzofuran lignans of the rocaglamide type as well as one naturally occurring aglain congener were tested for their ability to inhibit NF-B function. The different substances were isolated from various Aglaia species as described earlier (3). All compounds were prepared to at least 98% purity as assayed by highperformance liquid chromatography. These compounds (named 1-16) differed mainly in their amide side chains and/or in the substitution patterns of aromatic rings A and B (Fig. 1).
Jurkat T cells bearing an integrated NF-B-dependent luciferase reporter (3xB.luc) served as a test system for NF-B activity. Jurkat T cells were pre-treated with the different rocaglamide compounds at a concentration of 100 nM. One hour later the cells were stimulated with TNF, and at 6 h poststimulation luciferase activities were determined. TNF stimulation resulted in a potent activation of NF-B-dependent transcription (Ͼ30-fold). Interestingly we found that pretreatment of the Jurkat T cells with several of the different rocaglamide compounds led to inhibition of TNF-induced NF-B-dependent luciferase activity ( Fig. 2A and data not shown). The most efficient inhibitors were N,N-didesmethyl-N-4-hydroxybutylrocaglamide (compound 8), didesmethyl-rocaglamide (synonym with RocB, compound 1), and compounds 3 and 4. These substances show an inhibition activity ranging from 70% up to complete inhibition. Compounds 2, 5-7, 9 -12, 14, and 15 are less powerful inhibitors that repress TNF-induced NF-Bdependent luciferase activity to levels less than 50% ( Fig. 2A and data not shown). The aglain derivative (compound 17), 8␤-methoxyrocaglaol (compound 13), and compound 16 were completely inactive even at concentrations up to 2 mM ( Fig. 2A and data not shown).
We next investigated the dose dependence of the inhibition and determined the IC 50 values for several rocaglamides following TNF treatment. All investigated active compounds showed a dose-dependent inhibition of TNF-induced NF-Bdriven gene expression. An example of the dose-dependent inhibition of the transcription factor NF-B by RocB is shown in Fig. 2B. A complete inhibition of the transcription factor NF-B-dependent gene expression in Jurkat T cells was observed, when the T-cells were treated with RocB at a concentration of 200 nM. The IC 50 value was determined to be 58 nM, and several other compounds showed IC 50 values in the range of 200 nM (Table I).
A wide variety of stimuli has been shown to induce NF-B activity. We therefore asked whether rocaglamides would specifically inhibit TNF␣-induced NF-B or whether they might also interfere with PMA-induced NF-B activation. RocB (compound 1) was able to inhibit PMA stimulation in the same concentration range as shown for TNF␣ stimulation. Other rocaglamide congeners were somewhat less efficient in their inhibitory action for PMA-induced NF-B activity. IC 50 values of five rocaglamide derivatives were determined for their inhibition of PMA stimulation (Table I). The aglain derivative and 8␤-methoxyrocaglaol were again completely inactive up to the final concentration tested. Our results demonstrate that, depending on their chemical structure, rocaglamide derivatives represent potent inhibitors of the NF-B pathway.
We next asked whether rocaglamide derivatives specifically interfere with the NF-B pathway or whether they affect the activity of other transcription factors, too. For this purpose we analyzed the influence of rocaglamides on AP-1 activity. AP-1 is a dimeric transcription factor, which regulates gene expression in response to a great variety of stimuli, including TNF and PMA (39). Jurkat T cells transiently transfected with an AP-1-dependent reporter gene (5xTRE-tk-luc) showed a strong basal activity already (in comparison to tk-luc). This activity was only marginally induced by TNF or PMA. We investigated the influence of rocaglamides on this basal AP-1 activity in Jurkat T cells. We found that AP-1 activity was not altered in the presence of active rocaglamide derivatives 1-5 and the previously inactive aglain derivative (compound 17). Interestingly, compound 8 strongly reduced reporter gene activity (Fig.  2C). Similar results were obtained when we investigated the PMA/ionomycin-induced activity of the Oct-coactivator BOB.1/ OBF.1 (40,41). PMA/ionomycin-induced Oct-dependent activity was not inhibited by rocaglamides (data not shown), but again compound 8 showed a dramatic reduction of Oct-dependent activity. From this and several other experiments we concluded that the effects of compound 8 were unspecific. This rocaglamide derivative also showed strong cytotoxic effects (data not shown) and was therefore excluded from further experiments. In conclusion, rocaglamide derivatives are specific inhibitors of the NF-B system and do not block the activity of the AP-1 and Oct transcription factors.
Inhibition of NF-B Activity by Rocaglamides Is Cell Typespecific-We next tested the inhibition properties of rocaglamides in other cell lines to address the question whether the inhibition of the NF-B-dependent gene expression might be due to a cell-type-specific mechanism. First we analyzed the A301 (human) and EL-4 (mouse) T-cell lines, which were transiently transfected with an NF-B-dependent reporter construct. Similar to the results obtained with Jurkat T cells, the TNF (Fig. 3A) or PMA (Fig. 3B) stimulated activation of the NF-B-driven reporter gene could be inhibited by RocB in EL-4 cells and in A301 cells (data not shown). The concentrations of RocB were in the same range as those needed for the inhibition of the NF-B activation in Jurkat T cells. Again, the inactive compounds (the aglain derivative and 8␤-methoxyrocaglaol) showed no inhibitory activity. Thus, we conclude that rocaglamides inhibit NF-B induction in mouse and human T cell lines.
Surprisingly, a remarkable resistance of several other cell lines toward inhibition of NF-B by rocaglamide congeners was observed. In PC-12 cells bearing an integrated NF-B-dependent reporter gene and transiently transfected human HeLa cells, rocaglamide derivatives were not able to block NF-B activation. Even RocB concentrations up to 800 nM failed to inhibit NF-B activation by PMA (Fig. 4A) or TNF in PC-12 cells (data not shown). In HeLa cells several rocaglamides used at a concentration of 200 nM showed no inhibitory effect on TNF and PMA-induced NF-B activity (Fig. 4, B and C). In contrast, when PC-12 and HeLa cells were treated with highly active rocaglamide derivatives, an increase of the NF-B-dependent gene expression was found. The aglain derivative and 8␤methoxyrocaglaol showed no signs of inhibition or induction on the NF-B system in these cells. Similar results were obtained in NIH3T3 fibroblasts and A549 human lung epithelial cells bearing an integrated NF-B-dependent reporter gene (data not shown). We therefore conclude that rocaglamides do not block NF-B in these non-T cells.
All the cell lines that exhibited no rocaglamide-dependent inhibition of NF-B were of non-lymphoid origin. We therefore asked whether B lymphocytes might respond to rocaglamides. For this purpose we used the variant plasmacytoma cell line S107, which lacks constitutive NF-B activity typical for mature B cells but is still highly responsive to TNF (42). NF-Bdependent gene expression was investigated using S107 cells with an integrated reporter gene. Although the required rocaglamide concentrations for inhibiting TNF-induced NF-B activity in S107 cells were ϳ5 to 7 times higher than in Jurkat T cells, inhibition could be observed (Fig. 4D). Again, 8␤methoxyrocaglaol showed no inhibitory activity. From these results we conclude that NF-B activation in lymphocytes, but not in several other cell types, is controlled by a rocaglamidesensitive pathway.
RocB Suppresses PMA-induced Expression of NF-B-dependent Target Genes-All of the NF-B inhibitory effects of rocaglamides described here were observed in reporter gene assays.

FIG. 2. Dose-dependent inhibition of NF-B dependent gene expression by didesmethylrocaglamide. A, Jurkat T cells containing an
integrated B-dependent reporter (3xB.luc) were pretreated with the indicated rocaglamide derivatives (100 nM). After 1 h, cells were stimulated with TNF␣ (80 ng/ml). Six hours post-stimulation cells were harvested and luciferase activity was determined. The luciferase activity for unstimulated cells was defined as 1. All other activities are given relative to this standard. B, Jurkat T cells containing the integrated 3xB.luc construct were incubated with didesmethylrocaglamide (RocB) at the indicated concentrations (nM). After 1 h, cells were induced with TNF␣ (80 ng/ml). Six hours post-stimulation luciferase activity was determined. C, Jurkat T cells were transiently transfected with an AP-1-dependent reporter gene (5xTRE-tk.luc, five copies of the consensus TRE motif upstream of a minimal thymidine kinase promoter followed by the luciferase gene). 16 h after transfection cells were incubated with the indicated rocaglamide derivatives (200 nM/ml). Luciferase activity was measured 7 h after treatment and normalized on the basis of ␤-galactosidase expression. The expression level of the 5xTRE-tk.luc construct (20 g) was used as a reference luciferase activity and defined as 1.  [9] 161.3 197.4 [11] 208.0 499.0 [14] 205.0 319.4 [15] 159.9 243.4 Therefore, we next investigated whether rocaglamides also influence the activation of endogenous NF-B target genes. Jurkat T cells were pretreated with RocB, a highly active rocaglamide derivative and subsequently exposed to PMA. Total RNAs were then prepared and analyzed by RNase protection assay (RPA) analysis for the expression of known NF-B target genes. To prove the NF-B dependence of the respective genes, Jurkat cells deficient in IKK␥ (43), the essential component of the NF-B signaling pathway, were used in parallel for RNase protection assay analysis. These mutant Jurkat cells are unable to activate NF-B in response to various NF-B inducers, including PMA and TNF, but show normal activation of NF-AT and AP-1 (44). We found a strong up-regulation of LT-␤ (Fig. 5A), Bfl1/A1 (Fig. 5B), and TRAIL (Fig. 5C) RocB Interferes with DNA-binding Activity of NF-B and IB␣ Degradation-We next asked where the one or more molecular targets of rocaglamide action are localized in the NF-B signaling pathways induced by PMA and TNF. We first investigated the effects of RocB on NF-B DNA-binding activity. For this purpose we performed EMSA experiments using extracts from Jurkat T cells, which were pretreated with RocB and subsequently stimulated with TNF␣ and PMA. In the absence of RocB maximal NF-B DNA-binding activity was detected after 10 min of TNF treatment, remained at this level until 1 h, and decreased after 2 h (Fig. 6A). PMA stimulation resulted in maximal NF-B DNA-binding activity after 30 min, and this activity decayed from 2 to 8 h. RocB pretreatment resulted in a virtually complete block of the NF-B induction for both TNF and PMA stimulation (Fig. 6A). Also a second wave of NF-B activation, which was seen after 4 and 8 h of TNF treatment, was inhibited by RocB. At the same time no significant alterations in the DNA-binding activity of the Sp1 transcription factor were detected in the presence of RocB, suggesting that RocB specifically interferes with induced bind-

FIG. 3. Inhibition of NF-B activity by RocB in EL-4 T-cells.
EL-4 cells were transiently transfected with the 3xB.luc reporter. Sixteen hours post-transfection cells were treated with RocB at the indicated concentrations (nM) and with derivative 13 at a concentration of 800 nM. After 1 h, cells were stimulated with TNF␣ (A, 80 ng/ml) and PMA (B, 25 ng/ml). 6 h post-stimulation cells were harvested, and luciferase activity was determined and normalized on the basis of ␤-galactosidase expression.

FIG. 4. RocB does not interfere with NF-B activation in PC12 and HeLa cells. A, PC12 cells containing an integrated B-dependent
reporter (3xB.luc) were pretreated with RocB at the indicated concentrations (nM) and with derivative 13 at a concentration of 800 nM. After 1 h, cells were stimulated with PMA (50 ng/ml), and after 6 h luciferase activity was measured. B and C, HeLa cells were transiently transfected with the 3xB.luc reporter. Sixteen hours post-transfection, cells were treated with the indicated rocaglamide derivatives (200 nM). After 1 h, cells were stimulated with TNF␣ (B, 80 ng/ml) and PMA (C, 25 ng/ml). 6 h post-stimulation cells were harvested, and luciferase activity was measured and normalized on the basis of ␤-galactosidase expression. D, the murine plasmacytoma cell line S107 containing an integrated B-dependent reporter (3xB.luc) was incubated with RocB at the indicated concentrations (nM) and with derivative 13 at a concentration of 800 nM. After 1 h, cells were stimulated with TNF␣ (80 ng/ml), and 6 h post-stimulation luciferase activity was determined.
ing of NF-B. The inactive 8␤-methoxyrocaglaol had no influence on the DNA-binding activity (data not shown).
Some plant-derived inhibitors of NF-B have been shown to directly modify NF-B dimers thereby interfering with DNA binding (52)(53)(54)(55)(56). We therefore analyzed the effect of RocB on overexpressed RelA. Overexpression of RelA leads to constitutive levels of nuclear NF-B complexes. This level is further increased in the presence of TNF (Fig. 6B). In the presence of RocB we found that RelA-dependent complexes are not influenced, only the complexes arising additionally upon TNF induction can be blocked by RocB pretreatment. This result demonstrates that nuclear translocation of non-IB-complexed NF-B and DNA binding of NF-B (RelA) are not altered by rocaglamides. It also suggests that rocaglamides do not act directly by modifying NF-B proteins.
The translocation of NF-B dimers to the nucleus is preceded by the induced proteolytic degradation of IB␣. To prove whether inhibition of TNF and PMA-induced NF-B activation was due to inhibition of IB␣ degradation, the expression of this protein was monitored in Western blot analysis. As shown in Fig. 6C, TNF and PMA induce IB␣ degradation in a timedependent manner and with a delayed kinetic in the case of PMA stimulation. Endogenous IB␣ was degraded within 10 min of TNF exposure and within 30 min of PMA induction. Resynthesis of IB␣, a target gene dependent upon NF-B, starts already after 1 h of stimulation. Upon RocB pretreatment the TNF and PMA-induced IB␣ degradation was strongly repressed and the resynthesis of IB␣ was suppressed in comparison to cells without RocB treatment. At the same time RelA expression was not affected by RocB. From these results we suggest that the mechanism of NF-B inhibition by rocaglamides is localized on the level of cellular factors either involved in the induced degradation of IB␣ or signaling components upstream of this process.
Rocaglamides Inhibit NF-B Activation "Upstream" of the IKK Complex-An essential prerequisite for induced IB␣ degradation is the de novo phosphorylation of conserved serine residues on the N-terminal domain of IB␣ catalyzed by the IB kinase complex. The IKK1 and IKK2 kinases present in the complex are both capable of phosphorylating IB␣ in response to extracellular signals. We therefore asked whether the IB kinase complex is a target of rocaglamide action. To address this question, immune complex kinase assays were performed. Consistent with the induced degradation of IB␣, stimulation of Jurkat T cells with TNF and PMA resulted in an activation of the IKK complex monitored via phosphorylation of the GST-IB␣ substrate (Fig. 7A). However, in the presence of RocB the PMA-induced phosphorylation of the substrate was completely blocked, and in the case of TNF induction a significant reduction was observed. Immunoblot analysis revealed equal protein levels of precipitated IKK1. These data argue for a rocaglamide inhibition mechanism upstream of the IKK complex or at the level of the IKK complex.
To distinguish between a direct inhibition of IKK1 and IKK2 enzyme activity by RocB or effects on upstream factors of the IKK complex, we performed in vitro kinase assays with purified recombinant IKK proteins (57). In a coupled kinase assay with the GST-IB␣ substrate we observed no repression of IKK1 and IKK2 kinase activities in the presence of 100, 200, or 1000 nM RocB. Both IKK proteins showed some level of autophosphorylation, which also was not affected by the addition of RocB (Fig. 7B). Even very high RocB concentrations of up to 330 M did not result in any alteration of IKK1 and IKK2 kinase activity (data not shown). These observations suggest that rocaglamides have no direct influence on IKK1 and IKK2 enzyme activity but rather act on factors regulating or activating the kinase activity.
To investigate whether upstream signaling events in the signal transduction cascade initiated by TNF␣ and PMA are likely to be targeted by rocaglamides, we performed transient transfections with known components of the TNF signal transduction cascade, namely TRAF2, NIK, MEKK1, and IKK2. Upon overexpression, these proteins induce NF-B activity measured via activation of a B-dependent luciferase reporter (Fig. 8). We could observe that, in the presence of RocB the NF-B activation induced by TRAF2, NIK, MEKK1, and IKK2 is affected to different degrees, whereas in all cases the inactive derivative aglain (compound 17) had no inhibitory effect. TRAF2-mediated NF-B activation was almost completely blocked similar to the inhibition observed after TNF␣ stimulation. MEKK1 and NIK, kinases suggested to be involved in the signal transduction cascade leading to NF-B activation in response to TNF, were inhibited to a lower degree (30 -50% of stimulated activity) by RocB action. Likewise a component of the IKK complex, the IKK2 kinase (a constitutive active version, IKK2-EE) or the wild type version (data not shown), was inhibited about 2-fold. Interestingly, such a 2-fold reduction of activity was even seen in experiments, where RelA was overexpressed (data not shown). This suggests that rocaglamides predominantly affect a activation step upstream of the IKK complex and downstream of TRAF2, which primarily prevents nuclear translocation of NF-B dimers, but they also weakly reduce the transcriptional activity of NF-B complexes probably regulated by the same signaling component.
Rocaglamides Promote Apoptosis in Therapy-resistant Leukemic T Cells-The role of NF-B in apoptosis has been well established over the past years. In most cell types, NF-B has an anti-apoptotic function by up-regulation the expression of protective factors such as cIAP1 and 2, Bfl-1, Bcl-XL, and c-FLIP (58 -64). Some tumor cell lines show increased resistance to the induction of apoptosis, which contributes to the tumorigenic potential of these cells (65,66). Interestingly, suppression of NF-B activation increases the apoptotic response to chemotherapy (67,68). In two variants of the CEM T cell leukemia line, which were either sensitive (CEM-S) or resistant (CEM-R) toward therapy-induced apoptosis, we could recently show that inhibition of NF-B activation by proteasome inhibitors and dominant-negative IB␣ augments induction of apoptosis and abrogates apoptosis resistance (69). We therefore asked whether rocaglamide derivatives could affect apoptosis induction in these cell lines. CEM-S cells and the derived resistant cell line CEM-R were ␥-irradiated or treated with cisplatin and TNF␣. 48 h later apoptosis was determined by annexin-V staining and fluorescence-activated cell sorting analysis. Strongly enhanced apoptosis was detected in ␥-irradiated CEM-S cells and to a lower extent after treatment with cisplatin and TNF␣. In contrast, apoptosis due to these insults was not significantly altered in CEM-R cells (Fig. 9). However, cotreatment with RocB increased apoptosis in CEM-R cells, whereas the high percentage of apoptosis in CEM-S cells was not further enhanced. RocB also enhanced basal apoptosis in sensitive and resistant CEM cells. The inactive aglain congener failed to induce apoptosis at any concentration (data not shown). Thus, whereas RocB does not influence cancer therapyinduced apoptosis in sensitive cells, the therapeutic effect in resistant tumor cells is increased by RocB. These data suggest RocB as being a potentially helpful therapeutic agent that may overcome the apoptosis defect of therapy-resistant cancer cells by blocking the anti-apoptotic effect of NF-B.

DISCUSSION
In this study we analyzed the effect of plant-derived natural products on NF-B activity. We show that rocaglamides are potent inhibitors of NF-B activity. They block inducible NF-B DNA-binding activity and IB␣ degradation as well as expression of NF-B target genes in T lymphocytes. We characterized the NF-B inhibition properties of 16 rocaglamide derivatives and one naturally occurring aglain congener. We could define a group (compounds 1, 3, and 4) of rocaglamide derivatives, which show a dose-dependent inhibition of PMAand TNF-induced NF-B activation in Jurkat T cells and which mediate an almost complete inhibition at a final concentration of 200 nM. The results correspond well with earlier experiments, in which the insecticidal properties of rocaglamides (2) and their anti-proliferative activities were determined (8,9). RocB (didesmethylrocaglamide) was found to be the most active NF-B-specific compound with IC 50 values of 58 nM for TNF-induced NF-B activity and 44 nM for PMA-induced NF-B activity, respectively. RocB, compounds 3 and 4 have a polar aminoacyl side chain at carbon C-2 and an unsubstituted C-3Ј at the phenyl ring system B in common. A replacement of the polar amide group at carbon C-2 either by a polar COOCH 3 group (compounds 9, 10, 14, and 15) or an non-polar hydrogen (compounds 11 and 12) led to a reduction of NF-B inhibitory potential in Jurkat T cells. Furthermore, an insertion of an OH group (compounds 2, 5, and 7) or a methoxy substituent (compound 6) at C-3Ј of the rocaglamide skeleton diminished NF-B inhibition. An introduction of either a 3Ј,4Ј-methylendioxy group (compound 14) or a 6,7-methylendioxy group (compound 15) resulted in a decrease of NF-B inhibition. The most dramatic loss in NF-B inhibition was observed when the hydroxyl group at position C-8b was replaced by a methoxy substituent (compound 13). Also, the replacement of the typical benzofuran backbone of the rocaglamide derivatives by a benzopyran skeleton in the case of the aglain conger (compound 17) and the introduction of a methoxyl group at C-3Ј and a 6,7-methylendioxy group in the case of compound 16 led to a total loss of activity.
Interestingly, the RocB also showed the highest insecticidal activity against larvae of the pest insect Spodoptera littoralis and the highest antiproliferative activity on Mono Mac 6 cells in vitro (3,8). The molecular mechanism responsible for insecticidal activity of rocaglamides has not been resolved yet. Interestingly, the NF-B system is conserved in insects not only at the level of the Rel/NF-B transcription factors (Dorsal, Dif, and Relish in Drosophila) but also with respect to the signaling components involved in regulation of the function of these factors (70). This cascade is involved in the regulation of embryo development and "innate immune responses" in insects. Therefore, it seems possible that inhibition of this pathway by rocaglamide derivatives contributes to its insecticidal effects. The mechanism of the rocaglamide-induced block of cell cycle progression observed in some cell lines may also be mediated by NF-B inhibition. It was shown that expression of cyclin D1, a key factor in the regulation of the cell cycle, is regulated by NF-B (71)(72)(73)(74).
When the active concentrations of rocaglamide derivatives 1, 3, und 4 are compared with other NF-B inhibitors, it becomes clear that these rocaglamides are highly active compounds (13). They are active in concentrations of 50 -100 nM, whereas most other inhibitors work in the micromolar to millimolar range. For example, aspirin and sulfasalazine inhibit NF-B activity at concentrations of 20 -50 mM, proteasome inhibitors, and inhibitors of the anti-oxidant class act at 20 -100 M, curcumin at 40 -60 M and herbimycin A at 200 -2000 nM. Recently, another family of plant-derived NF-B inhibitors, namely sesquiterpene lactones, were described previously (20,21). These compounds were shown to block NF-B activation induced by PMA and TNF at a concentration of 5-20 M, which is still 5-to 20-fold higher than the active concentration of rocaglamide derivatives. So far only gliotoxin, a metabolite produced by Aspergillis fumigatus and other pathogenic fungi, was shown to inhibit NF-B in nanomolar concentrations (50 -1000 nM (75)). Glucocorticoids such as dexamethasone are also potent inhibitors of NF-B activity, which act in a range of 100 nM (76,77). From this comparison we suggest that rocaglamide derivatives act in a concentration range that is of future therapeutic relevance. In addition, the observed reduction of the inhibitory capacity found in several compounds due to variations in the chemical structure may help to develop structural modifications that allow the generation of derivatives with optimized features.
An interesting outcome of our study was the obvious cell type specificity of NF-B inhibition. The reason for this specificity is not clear, yet, but two different hypotheses could account for this result. One possibility would be that TNF and PMA initiated signal transduction cascades differ between T cells and other cell types. Although the signaling components from the cell surface to the IKK complex have been studied extensively, not all intermediates have been unequivocally identified. Indeed more than 20 protein kinases are implicated in the acti-vation and the regulation of the IKK complex. These kinases, including NIK, NAK, MEKK1, -2, and -3, Akt1, HPK-1, Raf, several members of the protein kinase C family, TPL-2/Cot, MLK3, BTK, and PAK1 may form signaling cascades that differ in their composition depending on the NF-B inducer (reviewed in Refs. 11 and 14). It is therefore conceivable that different cell types may utilize different components/isoforms for the activation and/or regulation of the IKK complex and the component used in T cells is susceptible to rocaglamide inhibition. It is not clear whether there is a common factor in TNFand PMA-induced signaling pathway, which is affected by rocaglamide action. In that respect, rocaglamide derivatives may turn out to be helpful tools to identify such pathway and cell type-specific signaling intermediates.
A second explanation for the cell type specificity could be that rocaglamide derivatives are only active after an obligatory metabolizing step. Only certain cell types might possess the machinery for rocaglamide uptake and/or for the enzymatic conversion of rocaglamides into its active derivate. Evidence in support of this explanation comes from insecticidal (2) and cell proliferationassayswithRocB (8,9).Intheseassaysrocaglamidedependent effects were only observed at late time points (48 -96 h), and this finding was explained with a potential metabolizing step. Also in Jurkat T cells conversion of rocaglamides seemed to be a rate-or time-limiting step. In EMSA experiments it turned out that preincubation with RocB for at least 2 h before stimulation with TNF or PMA was necessary for complete inhibition at early induction time points.
Our data virtually rule out the possibility that the IB kinases themselves are the target for inhibition by rocaglamides. First of all we show that active IKK2 is only slightly affected by rocaglamides in transient transfection assays. More important, we have analyzed IKK activity in vitro using purified recombinant IKK1 and IKK2. In both cases rocaglamides had no effect on the IB kinase activity measured in vitro. Furthermore, constitutive NF-B activity in mature B cells was not affected by RocB (data not shown) also suggesting that RocB does not interfere directly with NF-B activity but, rather, blocks a step in the signaling cascades leading to the activation of NF-B.
The NF-B transcription factor family functions broadly in the host control of immunoregulatory gene expression, inflammation, and apoptosis. Our results demonstrate that the expression of NF-B-dependent target genes involved in these processes such as Bfl-1/A1, lymphotoxin-␤ (LT-␤), and TRAIL can be efficiently blocked by rocaglamides. Bfl-1/A1 is an antiapoptotic protein of the bcl-2 family, whose preferential expression in hematopoietic, leukemic, and endothelial cells is controlled by inflammatory stimuli. Bfl-1/A1 is a direct transcriptional target of NF-B that is able to block TNF-induced apoptosis and mediates resistance to chemotherapy-induced apoptosis (47,78). Interestingly, in the presence of RocB, TNF and chemotherapy-induced apoptosis is strongly enhanced in otherwise TNF-resistant leukemic T cells (CEM-R). In this context it is possible that up-regulation of Bfl-1/A1 in response to TNF and other NF-B inducers like ␥-irradiation or cisplatin confers resistance to these inducers, which is overcome by rocaglamide action. However, it remains open whether the different stimuli induce expression of Bfl-1/A1 in CEM cells and/or whether there are differences in the expression level in CEM-S and CEM-R cells.
Lymphotoxin-␤ (LT-␤) is a tumor necrosis factor-related membrane-bound cytokine that forms a heterotrimeric complex with LT-␣ on the surface of lymphoid cells. LT-␤ is produced by activated lymphocytes and functions as a mediator of inflammatory processes. The PMA-induced up-regulation of LT-␤ mRNA in Jurkat T cells is consistent with previous findings (45) demonstrating that PMA-inducible promoter activity depends on the integrity of a conserved NF-B binding site. The inhibition of PMA-induced expression of LT-␤ by RocB offers a possible explanation for the use of rocaglamide-containing plant extracts as anti-inflammatory remedies in traditional medicine, although the effects on other pro-inflammatory cytokines require further detailed investigation. TNF-related apoptosis-inducing ligand (TRAIL) induces apoptosis in a broad range of tumor cells with apparently no cytotoxic effects on most non-transformed cells. It was recently shown that the regulation of TRAIL expression depends critically on NF-B function (48,49). Thus, our data strongly argue for a specific NF-B inhibition mechanism of rocaglamides in PMA-induced TRAIL expression.
In summary, we have identified rocaglamide derivatives as a novel class of plant-derived compounds that are potent inhibitors of NF-B-dependent gene activation in the nanomolar range. Although the specific molecular mechanism of inhibition has not been elucidated, rocaglamides are therapeutic candidates for various human diseases involving inflammatory diseases and cancer.