Trichothecene Mycotoxins Trigger a Ribotoxic Stress Response That Activates c-Jun N-terminal Kinase and p38 Mitogen-activated Protein Kinase and Induces Apoptosis*

The trichothecene family of mycotoxins inhibit protein synthesis by binding to the ribosomal peptidyltransferase site. Inhibitors of the peptidyltransferase reaction (e.g.anisomycin) can trigger a ribotoxic stress response that activates c-Jun N-terminal kinase (JNK)/p38 mitogen-activated protein kinases, components of a signaling cascade that regulates cell survival in response to stress. We have found that selected trichothecenes strongly activate JNK/p38 kinases and induce rapid apoptosis in Jurkat T cells. Although the ability of individual trichothecenes to inhibit protein synthesis and activate JNK/p38 kinases are dissociable, both effects contribute to the induction of apoptosis. Among trichothecenes that strongly activate JNK/p38 kinases, induction of apoptosis increases linearly with inhibition of protein synthesis. Among trichothecenes that strongly inhibit protein synthesis, induction of apoptosis increases linearly with activation of JNK/p38 kinases. Trichothecenes that inhibit protein synthesis without activating JNK/p38 kinases inhibit the function (i.e. activation of JNK/p38 kinases and induction of apoptosis) of apoptotic trichothecenes and anisomycin. Harringtonine, a structurally unrelated protein synthesis inhibitor that competes with trichothecenes (and anisomycin) for ribosome binding, also inhibits the activation of JNK/p38 kinases and induction of apoptosis by trichothecenes and anisomycin. Taken together, these results implicate the peptidyltransferase site as a regulator of both JNK/p38 kinase activation and apoptosis.

The trichothecenes are a structurally related family of low molecular weight mycotoxins synthesized by various Fusarium species. The ability of trichothecenes to inhibit the growth of rapidly proliferating cells in vitro and to selectively target tissues with a high mitotic index (e.g. bone marrow, gastrointestinal epithelium) prompted the selection of a representative compound (i.e. diacetoxyscirpenol) for testing in phase I and phase II clinical trials in human cancers (1)(2)(3)(4)(5). Trichothecenes inhibit the peptidyltransferase reaction by binding to the 60 S ribosomal subunit in eukaryotic cells (6 -10), and their antiproliferative activity has been presumed to be a consequence of inhibition of protein synthesis. The recent discovery that peptidyltransferase inhibitors can trigger a ribotoxic stress re-sponse that activates JNK1 1 (11), a stress-activated MAP kinase that signals the cellular response to stress (12)(13)(14)(15)(16)(17), suggests that the toxicity of trichothecenes might not be a simple function of translational arrest. Anisomycin, a peptidyltransferase inhibitor that competes with trichothecenes for binding to the 60 S ribosomal subunit (6,7,9,10), is a strong activator of JNK and p38 MAP kinases (11)(12)(13)15). Anisomycin also induces rapid apoptosis in human lymphoid cells (18) (in marked contrast to the delayed apoptosis induced by many protein synthesis inhibitors that do not activate JNK and p38 MAP kinases, e.g. puromycin, emetine) (19,20), suggesting that the toxicity of compounds that target the peptidyltransferase site is multifactorial. The role of JNK/p38 kinase activation in the induction of apoptosis is controversial. Although JNK activation is required for the induction of apoptosis in some experimental systems (21)(22)(23)(24), it is not required in other systems (25)(26)(27)(28). It appears that the functional response to JNK activation can differ in different cell types exposed to different apoptotic stimuli. The precise relationship between anisomycin-induced translational arrest, JNK activation, and apoptosis is not known.
Deacetylation of anisomycin markedly inhibits its ability to bind to ribosomes, arrest translation (29,30), activate JNK/p38 kinases, and induce apoptosis (see below), suggesting that the functional effects of anisomycin may be related to ribosome binding. The rich structural diversity of the trichothecenes has the potential to further dissect the relationship between these ribosome initiated functional responses. Individual trichothecenes differ significantly in their ability to induce translational arrest (see below). The structural features that determine these functional differences are centered around chemical sidegroups that modify the C7 and C8 positions of a pentane ring common to all trichothecenes (see Fig. 1A and Table I). For example, diacetoxyscirpenol is a potent inhibitor of protein translation, whereas 3-acetyldiacetoxyscripentriol (produced by acetylation of the C6 position of the pentane ring) is not (Table I). Structural features that affect the ability of individual trichothecenes to interact with the ribosomal peptidyltransferase site are likely to determine the distinct functions of these compounds. As such, a structure:function analysis comparing the ability of individual trichothecenes to inhibit protein synthesis, activate JNK/p38 kinases and induce apoptosis might improve our understanding of how ribosome binding regulates these diverse functions. The results of this type of analysis suggest that cooperation between JNK/p38 kinase activation and translational arrest is important for the induction of rapid apoptosis by selected inhibitors of protein synthesis. Consequently, the relative ability of individual trichothecenes to trigger translational arrest and JNK/p38 kinase activation are likely to determine their efficacy as antineoplastic drugs. Importantly, the one trichothecene that has been tested for antineoplastic activity in clinical trials (1-5) is a relatively weak activator of JNK/p38 kinases (see below). Classification of natural or synthetic trichothecenes using the functional parameters defined in this study might facilitate selection of the most promising compounds for testing in clinical trials.

Materials
Trichothecenes and other protein synthesis and protease inhibitors were obtained from Sigma unless indicated otherwise. Stock solutions for all compounds were prepared in Me 2 SO at 3.3 mM, except for emetine, which was dissolved in water at 10 mg/ml.

Cell Treatments
The Jurkat human T-lymphoid cell line was grown in RPMI 1640 medium supplemented with 10% heat-inactivated fetal bovine serum, 500 units/ml penicillin and 500 g/ml streptomycin. For various treatments, cells were collected at 1.0 -1.5 ϫ 10 6 /ml and resuspended at 1.0 ϫ 10 7 /ml in the fresh growth medium. Anisomycin (3.8 M), trichothecenes (10 M), or other protein synthesis inhibitors (or equivalent volumes of solvents for control samples) were added in a volume not exceeding 1% of the total culture volume and incubated at 37°C for the indicated times. For treatments with two reagents, cells were incubated with the first reagent (or solvents for control samples) for 30 min at 37°C before addition of the second reagent (or solvents for control samples) and continued culture at 37°C for indicated periods of time. Cells were collected by centrifugation at 2,000 ϫ g for 1 min at 4°C, washed twice with ice-cold phosphate-buffered saline and then frozen in liquid nitrogen for storage at Ϫ80°C until further analysis.

Apoptosis Induction Assays
DNA Fragmentation Assay-After 3 h of treatment with various agents, 5 ϫ 10 6 cells per treatment were lysed in 0.5 ml of 10 mM Tris (pH7.5), 1% Triton X-100, 5 mM EDTA, incubated on ice for 10 min, vortexed for 5 s, and lysates were clarified for 5 min at 4°C in an Eppendorf microcentrifuge at the top speed. 0.45 ml of the supernatants were extracted once with an equal volume of phenol/chloroform (1:1) and aqueous phases were adjusted to 0.5 M NaCl and precipitated with equal volumes of isopropanol, followed by overnight incubation at Ϫ20°C. Precipitates were collected by centrifugation (10 min) at 4°C in an Eppendorf microcentrifuge at the top speed, pellets were washed with 70% ethanol, air dried and resuspended in 40 l of 10 mM Tris (pH 7.5), 1 mM EDTA, 50 g/ml RNase A. Following a 30 min incubation at 37°C, 10 l aliquots were separated on 1.2% agarose gels in TAE buffer as described (31).
Caspase-3 and Poly(ADP)ribose Polymerase (PARP) Cleavage-The cell lysates used for enzymatic assay of caspase-3 (see above) were also subjected to Western blotting analysis with caspase-3 (CPP32)-specific antibodies (PharMingen, San Diego, CA) according to the manufacturer's instructions and with PARP-specific monoclonal antibodies C-2-10 as described (33).

Activation of Stress-activated Kinase p38
Activation of kinase p38 was assayed as described previously (36) by Western blotting with antibody 9211 (New England Biolabs, Beverly, MA), recognizing the activated (phosphorylated) form of p38 kinase. Duplicate filters were probed with antibody 9212, recognizing both phosphorylated and unphosphorylated forms of p38 to verify equal loading.

Protein Synthesis Inhibition Assays
Protein synthesis was assayed by measuring the incorporation of labeled amino acids into cellular proteins, essentially as described by Ausbel et al. (37) with the following modifications: Jurkat cells were grown and collected as described above, washed once with Hanks' balanced salt solution and resuspended at 2 ϫ 10 6 /ml in cysteine-and methionine-free RPMI 1640 medium supplemented with 10% dialyzed heat-inactivated fetal bovine serum, incubated for 15 min at 37°C and treated in triplicate with protein synthesis inhibitors (or with corresponding solvents for control samples) for 20 min at 37°C before the addition of 50 Ci/ml of 35 S-labeled methionine/cysteine mixture (NEN Life Science Products) and incubation for an additional 20 min at 37°C. Cells were then centrifuged at 2,000 ϫ g for 1 min at 4°C, washed twice with ice-cold phosphate-buffered saline and solubilized in lysis buffer (200 l/10 6 cells; 10 mM Tris, pH 7.2, 0.1% SDS, 1% Triton X-100, 1% sodium deoxycholate, 150 mM NaCl, 20 g/ml chymostatin, 3 g/ml leupeptin, 14 g/ml pepstatin A, 1.7 mg/ml benzamidine, and 10 g/ml aprotinin). After a 10-min incubation on ice, cell lysates were vortexed for 10 s and clarified by centrifugation at 15,000 ϫ g for 10 min. 50 l of the supernatants were mixed with 500 l of 100 g/ml bovine serum albumin, and proteins were precipitated by the addition of 500 l of 20% trichloroacetic acid. After 20 min on ice, precipitated proteins were collected by filtration through glass microfiber filters (GF/C, Whatman), washed with 10 ml of 10% trichloroacetic acid and 5 ml of ethanol, and air dried. Incorporation of radiolabeled amino acids into cellular proteins was quantitated by liquid scintillation counting in a Packard 1600 TR counter. , we identified trichothecenes that induce strong (e.g. nivalenol, scirpentriol, and T-2 triol), intermediate (e.g. acetyldeoxynivalenol, acetoxyscirpenol, and HT-2), or weak (e.g. verrucarin, T-2 toxin) activation of JNK/p38 kinases. The different activity of these compounds cannot be explained by differential cell permeability, as several trichothecenes (e.g. deoxynivalenol and 3-acetyldeoxynivalenol; T-2 triol and acetyl-T-2 toxin) that differ dramatically in their ability to activate JNK/p38 kinases (Fig. 1), are similarly potent inhibitors of protein synthesis (see Table I, and below). The strong correlation between the ability of individual trichothe-cenes to activate JNKs and p38 (compare Fig. 1, B and C), suggests that both of these MAP kinases are activated via the same mechanism, the ribotoxic stress response. Therefore, our data indicate that structural differences between individual trichothecenes can influence their ability to trigger the ribotoxic stress response.

Activation of JNK and p38 MAP Kinases by Trichothecenes-We
Induction of Apoptosis by Trichothecenes-During our analysis of JNK activation by various trichothecenes, we noticed that many trichothecenes induce what appears to be a typical apoptotic cell death in Jurkat cells. The relative ability of individual trichothecenes to induce various manifestations of apoptosis was assessed by monitoring internucleosomal DNA fragmentation ( Fig. 2A), processing of pro-caspase-3 (Fig. 2B), activation of DEVD-specific caspases (Fig. 2C), and cleavage of one of the major caspase-3 substrates, PARP (Fig. 2D). This analysis identified trichothecenes within each structural subfamily that are strong (e.g. deoxynivalenol, scirpentriol, and T-2 triol), intermediate (e.g. nivalenol, diacetoxyscirpentriol, HT-2), and weak (e.g. 3-acetyldeoxynivalenol, varrucarin, T-2) inducers of apoptosis. Comparison of results presented in Figs. 1 and 2 reveals that activation of JNK/p38 kinases is not sufficient for the induction of apoptosis (see also Table I). Thus trichothecenes that similarly activate JNK/p38 kinases (e.g. T-2 triol and T-2 tetraol, Fig. 1) can differ significantly in their ability to induce apoptosis as measured by caspase-3 activation (Fig. 2, B and C). Nevertheless, the most potent apoptotic trichothecenes strongly activate JNK/p38 kinases, suggesting that kinase activation might contribute to the efficient induction of rapid apoptosis.
The sequential activation of stress-induced MAP kinases and caspases differs in different experimental systems (25, 28, 38 -40). Because of the strong correlation between JNK activation and trichothecene-induced apoptosis (compare Figs. 1 and 2; summarized in Table I), we determined the temporal order of JNK and caspase-3 activation by several trichothecenes and anisomycin (Fig. 3). The kinetics of JNK activation by anisomycin and T-2 triol are similar, with each drug producing maximal activation within 15 min (Fig. 3A) followed by detectable caspase activation at 1-2 h (Fig. 3, B and C). This sequential order of JNK and caspase-3 activation is even better illustrated by T-2 tetraol, which has a slower rate of cellular uptake than T-2 toxin (10, 41) and therefore requires 2 h for maximal activation of JNK (Fig. 3A), and 3 h for detectable caspase activation (Fig. 3B). Therefore, in response to trichothecenes and anisomycin, JNK activation precedes caspase-3 activation, distinguishing this process from a similarly rapid Fas-induced apoptosis in which JNK/p38 kinases are activated after caspase-3 during the later stages of cell death (38 -40). A more direct test of a role for JNKs in the activation of caspases would be provided by the demonstration that dominant negative JNK inhibits trichothecene-induced caspase activation. Unfortunately, the low transfection efficiency of Jurkat cells prevented us from obtaining useful information from this experiment.   (42). Inhibitors of protein synthesis can promote the induction of apoptosis in response to inflammatory cytokines that activate JNK/p38 kinases (e.g. Fas-ligand, tumor necrosis factor-␣), suggesting that the survival pathway, but not the death pathway, requires new protein synthesis (43,44). The possibility that differential inhibition of protein synthesis might determine the functional response to trichothecene-induced JNK/ p38 kinase activation prompted us to compare the ability of (10 M) as in Fig. 2B, and DEVD-specific caspase activity in 50 g of lysates was measured as described under "Experimental Procedures." D, trichothecenes induce PARP cleavage. Cells were treated with the indicated trichothecenes (10 M) as in Fig. 2B, lysed as described under "Experimental Procedures" (enzymatic assay for caspase-3 activity), and 25 g of each lysate was analyzed by Western blotting using monoclonal antibodies to PARP. Arrows indicate PARP, 116-kDa native PARP polypeptide; PARP*, 89-kDa fragment generated by caspase-3 digestion (56, 57). individual trichothecenes to inhibit protein synthesis in Jurkat cells. Table I compares the ability of individual trichothecenes to inhibit protein synthesis, activate caspase-3, and activate JNK. Although there is no obvious correlation between any two trichothecene-induced effects (e.g. protein synthesis inhibition versus JNK activation; protein synthesis inhibition versus caspase activation; JNK activation versus caspase activation), the tendency for apoptotic trichothecenes to strongly inhibit protein synthesis and strongly activate JNKs (Table I) suggests that these two effects might cooperate in the induction of apoptosis. To test this possibility, we measured the dose-dependent inhibition of protein synthesis produced by trichothecenes that strongly activate JNK kinases (Ͼ9-fold activation; Table I) and derived IC 50 values for this functional response (Fig. 4A). We then compared the activation of caspase-3 at a trichothecene concentration (10 M) at which each of these compounds maximally activate JNK1 (determined in dose-response experiments using the assay described in Fig. 1B). Under these conditions, caspase activation is a linear function of IC 50 (Fig.  4B), suggesting that inhibition of protein synthesis and activation of JNK/p38 kinases both contribute to the activation of caspase-3. Although the assay for JNK activation is not sufficiently quantitative to allow a similar analysis of trichothecenes that strongly inhibit protein synthesis, we compared the ability of these trichothecenes to activate caspase-3 at a concentration (10 M) that inhibits protein synthesis by Ͼ95% (see Table I). Under these conditions, caspase activation is a linear function of JNK activation (Fig. 4C), again suggesting that trichothecene-induced translational arrest and JNK/p38 kinase activation cooperate in the induction of apoptosis.

Deacetylanisomycin, T-2 Toxin, and Verrucarin Block the Activation of JNK and Caspase-3 by Both Anisomycin and
Apoptotic Trichothecenes-The observation that translational arrest and JNK/p38 kinase activation are independently triggered by individual trichothecenes led us to question whether binding to a common ribosomal site is required for trichothecene-and anisomycin-induced activation of JNK/p38 and caspase-3. If activation of JNK/p38 kinases requires ribosome binding, inactive anisomycin derivatives or trichothecenes that inhibit translation without activating JNK/p38 kinases (e.g. T-2 toxin, verrucarin, see Table I) might inhibit the function (i.e. JNK/p38 kinase and caspase-3 activation) of apoptotic trichothecenes and/or anisomycin. Deacetylanisomycin (DA) is an anisomycin analog that enters cells, binds to ribosomes, and inhibits protein synthesis (albeit with 10,000-fold lower potency than anisomycin; data not shown). When used at a concentration that inhibits protein synthesis by 65% (300 g/ml), it fails to activate JNKs on its own, and inhibits activation of JNKs by T-2 triol (10 M), T-2 tetraol (10 M), and anisomycin (3.8 M) (Fig. 5A). At similar concentrations, DA also inhibits anisomycin-induced translational arrest in Jurkat cells, as well as in rabbit reticulocyte lysates, suggesting that its functional effects are a consequence of ribosome binding (data not shown). T-2 toxin (10 M) (Fig.  5A) and verrucarin (10 M) (data not shown) similarly inhibit the activation of JNKs by these compounds. Pretreatment with either DA (300 g/ml), T-2 toxin (10 M), or verrucarin (10 M) also prevents caspase-3 activation in Jurkat cells cultured with apoptotic trichothecenes (T-2 triol, diacetylverrucarol, and deoxynivalenol, Fig. 5B). The concentrations of anisomycin, T2toxin and verrucarin used in these experiments is at or above the IC 50 for binding to free ribosomes (approximately 10 M, 0.6 M and 10 M, respectively) (10), consistent with a role for competitive displacement at the level of ribosome binding. The suggestion that DA and nonapoptotic trichothecenes can occupy to the peptidyltransferase site and inhibit the function of anisomycin and apoptotic trichothecenes is consistent with a role for ribosome binding in the activation of JNK/p38 kinases and, subsequently, caspase-3.
Activation of the Ribotoxic Stress Response by Anisomycin Does not Require Active Translation-Activation of JNK/p38 kinases in response to ribotoxic stress has been proposed to require on-going protein translation (11). If this is true, the profound translational arrest induced by T-2 toxin and verrucarin could account for the inhibition of JNK/p38 kinase activation observed in Fig. 5A. Translational arrest could not, however, account for the inhibition of JNK/p38 kinase activation induced by DA, a compound that only partially inhibits protein synthesis under the conditions employed. We therefore compared the ability of anisomycin to activate JNKs in Jurkat cells pre-incubated in the absence or presence of T-2 toxin (10 M, 30 min, conditions that reduced protein synthesis to Ͻ98% of control levels). The ability of T-2 toxin to prevent anisomycin-induced JNK activation could be overcome at high concentrations of anisomycin (Fig. 5C), suggesting that displacement of T-2 toxin from the peptidyltransferase site might allow anisomycin to activate JNK/p38 kinases in the absence of protein synthesis.
Further evidence that ribosome binding is required for peptidyltransferase inhibitors to activate JNK/p38 kinases and induce apoptosis was provided by analyzing the effects of structurally unrelated compounds that compete with trichothecenes for ribosome binding. Emetine was previously reported to block JNK activation by anisomycin (11). We found that emetine blocks anisomycin-and T-2 tetraol-induced JNK activation (Fig. 6A, lanes 5-8), as well as JNK activation induced by other trichothecenes (nivalenol, fusarenon, and trichothecin, data not shown), suggesting that emetine may block the access of anisomycin or trichothecenes to a common ribosomal binding site. Interestingly, emetine does not block T-2 triol-induced JNK activation (Fig. 6A, lane 6). Although this reinforces our conclusion that active translation is not required for trichothecene-induced JNK activation, it is surprising because T-2 triol differs from T-2 tetraol only in the addition of a 3-methylbutyryloxy group (R4 in Fig. 1A) at the C8 position.
Harringtonine, a plant alkaloid that is structurally unrelated to either anisomycin or trichothecenes, competes with these compounds for binding to the ribosomal peptidyltransferase site (45,46). As shown in Fig. 6A, harringtonine weakly activates JNKs on its own, but efficiently blocks both anisomycin-and trichothecene-induced JNK activation (Fig. 6A, lanes  9 -12). Both emetine and harringtonine also inhibit caspase activation by apoptotic trichothecenes (Fig. 6B). (Here, the inability of emetine to prevent JNK activation by T-2 triol is consistent with in its relative inability to block caspase activation by this trichothecene.) Because extended treatment with many inhibitors of protein synthesis can induce apoptosis (19,20,47), it is not surprising that treatment with emetine or harringtonine alone induced a low level of caspase-3 activation (Fig. 6B); however, the ability of both emetine and harringtonine to block anisomycin-or trichothecene-induced caspase-3 activation suggests that the mechanisms of apoptosis induction by anisomycin/trichothecenes on one hand, and emetine or harringtonine, on the other, are different. DISCUSSION Activation of JNK1 by protein synthesis inhibitors that bind to or alter the structure of 28 S ribosomal RNA (e.g. blasticidin S, gougerotin, anisomycin, ricin toxin, sarcin toxin) led Iordanov et al. (11) to propose the existence of a ribotoxic stress response in eukaryotic cells. The ability of ribosomes to sense cellular stress and activate signaling pathways that alter cellular function has been well characterized in prokaryotes. In response to amino acid starvation, prokaryotic ribosomes produce guanosine 3Ј,5Ј-bispyrophosphate (ppGpp), a nucleoside analogue that arrests transcription of genes encoding translafrom Fig. 4A. Caspase activation produced by 10 M T-2 triol, scirpentriol, diacetylverrucarol, nivalenol, or T-2 tetraol as a function of IC 50 was analyzed by a best fit simple polynomial, and the regression coefficient was calculated using the Cricket graph program (Cricket Software, Malvern, PA). C, the relative ability of trichothecenes (scirpentriol, T-2 triol, diacetoxyscirpentol, acetyldeoxynivalenol, HT-2, acetyl T-2, T-2 toxin, verrucarin, all at 10 M) that strongly inhibit protein synthesis (Ͼ95% inhibition, Table I) to activate JNKs and caspase-3 were compared. The best fit simple polynomial and regression coefficient were calculated using the Cricket graph program.  1-4) or T-2 toxin (10 M) (lanes [5][6][7][8], before the addition of Me 2 SO (lanes 1 and 5) or the indicated concentrations of anisomycin. After a further 2-h incubation, cell lysates were prepared for quantitation of JNKs activity as described under "Experimental Procedures." tion factors, a response that promotes survival under starvation conditions (48). By promoting the activation of JNK and p38 MAP kinases, eukaryotic ribosomes might similarly induce the transcription of genes that regulate the cellular response to stress. Our results provide support for this concept by showing that selected trichothecenes, compounds that target the ribosomal peptidyltransferase site (6,7,9,10,49), also activate JNK and p38 MAP kinases. The rich structural diversity of the trichothecenes allowed us to carry out a limited structure: function analysis of compounds that target this functional domain on the large ribosomal subunit. By comparing the ability of individual trichothecenes to inhibit protein synthesis, activate JNK/p38 kinases, and induce apoptosis, we have grouped these compounds into distinct functional classes. Most trichothecenes strongly inhibit protein synthesis (Table I). Among these, induction of apoptosis is linearly correlated with the ability to activate JNK and p38 MAP kinases (Fig. 4B). Trichothecenes that inhibit protein synthesis without activating JNKs (e.g. acetyl T-2, T-2 toxin, and verrucarin), induce a 3-4-fold increase in caspase activation, indicating that JNK/ p38 kinases may not be necessary for low level caspase activation. Among trichothecenes that strongly activate JNK/p38 kinases, induction of apoptosis is linearly correlated with the ability to inhibit protein synthesis. Taken together, these results reveal that the ability of individual trichothecenes to induce rapid apoptosis may require both translational arrest and JNK/p38 kinase activation.
The ability of emetine to inhibit anisomycin-, palytoxin (a modulator of Na/K ATPase function), and UV-induced JNK1 activation led Iordanov et al. (11,50,51) to conclude that active translation is required to trigger the ribotoxic stress response.
Our results showing that high concentrations of anisomycin can overcome the T-2 toxin-induced inhibition of JNK activation (Fig. 5C) require a re-examination of this conclusion. T-2 toxin inhibits protein synthesis by Ͼ98% under these conditions, indicating that anisomycin can activate JNKs in the absence of active translation. The further observation that T-2 triol (but not T-2 tetraol or anisomycin) activates JNKs in cells rendered translationally incompetent by pretreatment with emetine (producing Ͼ98% inhibition of protein synthesis) is also not consistent with a requirement for active ribosomes in this process (see Fig. 6A). Although the binding site of emetine is located on the small ribosomal subunit, its close proximity to the trichothecene binding site on the large ribosomal subunit allows it to compete with T-2 toxin for ribosome binding (49,52,53). The ability of emetine to block T-2 tetraol, but not T-2 triol-induced JNK activation suggests that emetine can interfere with the binding of the tetraol, but not the triol to the peptidyltransferase site on the large ribosomal subunit. The ability of emetine to also inhibit palytoxin and UV-induced JNK activation further suggests that emetine either stabilizes ribosomal structure, or prevents the binding of a natural ribosomal ligand involved in the ribotoxic stress response (alternatively, the activation of JNK/p38 kinases by palytoxin and UV irradiation may be dependent upon protein synthesis). Pactomycin, another compound that inhibits the ribotoxic stress response (11), also binds to the small ribosomal subunit. Although it is not known to interfere with the binding of anisomycin or trichothecenes, photoaffinity labeling experiments indicate that it also interacts with the large ribosomal subunit (54,55), suggesting that it could block access to the anisomycinand trichothecene-binding site. The identification of trichothecene derivatives (e.g. T-2 tetraol and, to a lesser degree, 3-acetyldiacetoxyscirpenol, Table  I) that activate JNK and p38 kinases without significantly affecting protein synthesis suggests that different parts of the trichothecene molecule are responsible for translational arrest and JNK/p38 kinase activation. The functional effects of these compounds could result from the displacement of ribosomebinding molecules (e.g. elongation factors, charged tRNAs, mRNAs, etc.), the induction of a conformational change in the ribosome itself, or both. The ability of plant toxins such as ricin and sarcin to activate JNK1 following the catalytic depurination or cleavage of 28 S RNA (11) suggests that a conformational change in the ribosome can result in JNK/p38 kinase activation. How this conformational change leads to JNK1 activation remains to be determined. The ability of some trichothecenes, but not others, to activate JNK/p38 kinases suggests that the presence or absence of side groups that interact with ribosome target sites or displace the binding of ribosomeassociated molecules might determine whether JNK/p38 kinases are activated. The effect that various modifications of trichothecene structure have on JNK activation seems to be a direct one, because these modifications do not affect the ability of most trichothecenes to enter cells and rapidly inhibit translation (Table I). Although the variable R1-R4 trichothecene side groups (Fig. 1A) must determine their relative ability to inhibit protein synthesis and activate JNK/p38 kinases, there is no obvious correlation between the structure of these groups and trichothecene function.
In conclusion, our results suggest that: (i) the functional effects of trichothecenes are initiated by ribosome binding; (ii) trichothecene-induced translational arrest and JNK/p38 kinase activation are independent dissociable events; and (iii) translational arrest and JNK/p38 kinase activation appear to cooperate in the induction of apoptosis. Consequently, classification of trichothecenes based on their relative ability to inhibit protein synthesis, activate JNK/p38 kinases, and induce apoptosis might facilitate the selection of natural and synthetic compounds for clinical trials in human cancers.