|Home | About | Journals | Submit | Contact Us | Français|
Anisomycin, a translational inhibitor secreted by Streptomyces spp., strongly activates the stress-activated mitogen-activated protein (MAP) kinases JNK/SAPK (c-Jun NH2-terminal kinase/stress-activated protein kinase) and p38/RK in mammalian cells, resulting in rapid induction of immediate-early (IE) genes in the nucleus. Here, we have characterized this response further with respect to homologous and heterologous desensitization of IE gene induction and stress kinase activation. We show that anisomycin acts exactly like a signalling agonist in eliciting highly specific and virtually complete homologous desensitization. Anisomycin desensitization of a panel of IE genes (c-fos, fosB, c-jun, junB, and junD), using epidermal growth factor (EGF), basic fibroblast growth factor, (bFGF), tumor necrosis factor alpha (TNF-α), anisomycin, tetradecanoyl phorbol acetate (TPA), and UV radiation as secondary stimuli, was found to be extremely specific both with respect to the secondary stimuli and at the level of individual genes. Further, we show that anisomycin-induced homologous desensitization is caused by the fact that anisomycin no longer activates the JNK/SAPK and p38/RK MAP kinase cascades in desensitized cells. In anisomycin-desensitized cells, activation of JNK/SAPKs by UV radiation and hyperosmolarity is almost completely lost, and that of the p38/RK cascade is reduced to about 50% of the normal response. However, all other stimuli produced normal or augmented activation of these two kinase cascades in anisomycin-desensitized cells. These data show that anisomycin behaves like a true signalling agonist and suggest that the anisomycin-desensitized signalling component(s) is not involved in JNK/SAPK or p38/RK activation by EGF, bFGF, TNF-α, or TPA but may play a significant role in UV- and hyperosmolarity-stimulated responses.
The bacterial compound anisomycin (54; reviewed in reference 30) (Fig. (Fig.1)1) inhibits translation by binding to 60S ribosomal subunits and blocking peptide bond formation, thereby preventing elongation and causing polysome stabilization (2, 30). More recently, the compound has been widely used as an extremely potent activator of kinase cascades in mammalian cells, especially the stress-activated mitogen-activated protein (MAP) kinase subtypes (8, 9, 11, 18, 20, 24, 25, 42, 45, 52, 61). Kyriakis et al. (34), studying a cycloheximide-activated kinase (p54 MAP kinase), identified a family of stress-activated protein kinases (SAPKs) encoded by three genes, each of whose transcripts may be alternatively processed. This subtype, which is more strongly activated by anisomycin, was independently identified as a UV radiation-activated kinase that binds to and phosphorylates the N terminus of c-Jun (c-Jun NH2-terminal kinase [JNK] [22, 28]). More recently, we showed, using in-gel kinase assays, that anisomycin strongly activates two kinases, p45 and p55, which we identified as MAP kinase-activated protein (MAPKAP) kinase 2 (MAPKAP K-2) (10). This finding implied that anisomycin must also activate its upstream kinases, and we subsequently showed that it potently activates the MKK6→p38/RK→MAPKAP K-2 cascade in these cells (25). Thus, anisomycin strongly activates two MAP kinase subtypes associated with the stress response. However, it is important to note that anisomycin equally strongly activates a third kinase, p70/85 S6 kinase (p70/85S6k), which phosphorylates ribosomal protein S6 (31). This response is sensitive to inhibition with rapamycin (31), indicating that it is mediated through the FRAP/TOR kinase (4, 5). Although we have shown that UV radiation also strongly stimulates S6 phosphorylation (9), p70/85S6k is not generally regarded as a stress kinase, whereas JNK/SAPKs and p38/RK are; all three are strongly activated by both UV radiation and anisomycin. As a result of its potent activation of MAP kinase subtypes which phosphorylate transcription factors such as c-Jun, ATF-2, and ternary complex factor in C3H 10T½ cells, anisomycin strongly induces transcription of several immediate-early (IE) genes (references 7, 9, 23, 25, and 26 and references therein).
The signalling and IE gene-inducing properties of anisomycin were originally thought to be secondary effects of translational arrest, arising either from loss of labile repressive proteins (reference 55 and references therein) or from the stress of translational arrest (34). However, the fact that anisomycin-stimulated signalling and gene induction responses are clearly demonstrable at concentrations below those required for inhibiting translation (subinhibitory concentrations [23, 39]), and conversely, that not all translational inhibitors activate signalling responses, invalidates this view. Puromycin and emetine have negligible signalling and gene-inducing effects, and although cycloheximide has some ability to activate these signalling responses, it is very much weaker than that of anisomycin, whereas it blocks translation equally well (23, 26a). These studies conclusively dissociate translational arrest from signalling and gene induction, but they do not exclude the possibility that anisomycin-induced signalling requires its interaction with ribosomes (see Discussion and references 29 and 39). Furthermore, it remains possible that the compound exerts unknown chemical toxicity in these cells, which may explain its ability to activate stress kinases and thereby the IE genes.
In this study, we examined whether anisomycin is capable of eliciting homologous desensitization, a characteristic of several true signalling agonists. Many signalling mechanisms undergo a transient refractory period wherein they do not respond to restimulation with the same agent (reviewed in reference 16). This phenomenon, called homologous desensitization, is caused by degradation of a receptor or signalling enzyme (16) or by negative-feedback mechanisms operating within signalling pathways (6) (see Discussion). In addition, other stimuli which utilize the same desensitized component(s) will also not elicit a response (heterologous desensitization), whereas agents which act either through distinct pathways or downstream of the desensitized component(s) continue to elicit normal responses (reviewed in reference 16). We have studied the desensitization of a panel of five IE genes to diverse agents and found that anisomycin elicits virtually complete homologous desensitization of all these genes. Further, we show that this desensitization arises because anisomycin loses its ability to activate the JNK/SAPK and p38/RK kinases in desensitized cells. Among several agents tested, only UV- and hyperosmolarity-induced kinase activation was compromised in anisomycin-pretreated cells. These studies show that anisomycin acts like a true signalling agonist in eliciting highly specific homologous desensitization and further that the anisomycin-desensitized component(s) is not required for the activation of JNK/SAPKs or p38/RK by growth factors or tumor necrosis factor alpha (TNF-α), whereas it may play a significant role in UV- and hyperosmolarity-induced activation of these kinases.
C3H 10T½ mouse fibroblasts were cultured in Dulbecco’s modified Eagle’s medium (DMEM) with 10% (vol/vol) fetal calf serum (FCS; Gibco). Confluent cultures were made quiescent by incubation for 12 to 18 h in DMEM containing 0.5% (vol/vol) FCS. Cells were either stimulated as indicated or, for desensitization analyses, exposed to the desensitizing agent for 3 h (pretreatment), and the second stimulus (restimulation) was then added directly to the culture medium. Stimuli used were human epidermal growth factor (EGF; 50 ng/ml; kindly provided by G. Panayotou, Ludwig Institute for Cancer Research, London, England), bovine basic fibroblast growth factor (bFGF; 20 ng/ml; Boehringer Mannheim), TNF-α (5 or 10 ng/ml, as indicated; R & D Systems), tetradecanoyl phorbol acetate (TPA; 100 nM; Sigma), and anisomycin (inhibitory [10 μg/ml] or subinhibitory [25 or 50 ng/ml]; Sigma). For UV irradiation, quiescent cells in 3 ml of DMEM–0.5% (vol/vol) FCS per 100-mm-diameter dish were exposed to 200 J of UV radiation (254 nm) per m2, delivered via a Spectrolinker XL-1000 (Spectronics Corp.). Medium was aspirated, and cells were harvested as described below at the indicated times after stimulation. The dose dependence of TNF-α-stimulated fos and jun induction has been analyzed at concentrations of 1 to 20 ng/ml (data not shown); at TNF-α concentrations of 5 to 10 ng/ml, there is no substantial difference between the strength of induction responses (see Fig. Fig.3).3). Where indicated, actinomycin D (20 ng/ml; Sigma) was added 5 min prior to pretreatment as detailed above.
C3H 10T½ cells were made quiescent and stimulated as described above. Total cellular RNA was isolated as described in reference 13. Aliquots containing 3 μg of RNA were resolved on formaldehyde-agarose gels essentially as described in reference 51 except that 0.41 M formaldehyde was used, as described in reference 12. RNA was transferred onto nylon membranes (Hybond-N+; Amersham) by capillary transfer, and hybridization performed as described in reference 15, using 32P-labelled fragments of the relevant genes. Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) mRNA was detected by using a 1-kb PstI fragment of murine cDNA in pBluescript KS− (Stratagene), kindly provided by D. R. Edwards. c-fos mRNA was detected by using a PstI/SalI fragment of v-fos derived from the 1-kb BglII/SalI fragment of v-fos (19) in pAT153. All other probes were derived from cDNA clones of fos and jun genes, which were generously provided by Rodrigo Bravo (Roche); a 1.7-kb EcoRV/HindIII fragment of fosB in pGEM-1 (Promega), a 0.75-kb EcoRI/SacII fragment derived from the 2.5-kb mouse c-jun clone pAH119 (49) in pUC19, a 1.8-kb EcoRI fragment of junB in pBluescript KS+ (Stratagene), and a 1.5-kb BamHI/HindIII fragment of junD in pBluescript KS+ (Stratagene). All Northern blots were sequentially hybridized to these six probes. For quantification of autoradiographs, densitometry was performed on a Molecular Dynamics instrument by using ImageQuant version 3.3., and all readings were corrected by reference to the corresponding GAPDH reading to compensate for slight variations in loading. A series of control and optimizing experiments designed to ensure that the methods used here produce an accurate reflection of the relative mRNA levels for these six genes is described in reference 24a.
Plasmids encoding glutathione S-transferase (GST)-cJun1-79 fusion proteins (28) were provided by M. Karin (University of California at San Diego). GST-fusion proteins were purified by affinity chromatography on glutathione (GSH)-agarose as described previously (9) and quantified by the bicinchoninic acid protein assay (Pierce). Kinase assays were performed by using a modification of the method described by Hibi et al. (28). C3H 10T½ cells in 60-mm-diameter dishes were harvested in 100 μl of buffer A (25 mM HEPES [pH 8.0], 0.3 M NaCl, 1.5 mM MgCl2, 0.2 mM EDTA, 0.1% Triton X-100 [TX-100], 20 mM sodium β-glycerophosphate, 0.1 mM sodium vanadate, protease inhibitors) (39, 40). Cell lysates were rotated at 4°C for 30 min, and the extract was cleared by centrifugation at 13,000 × g for 10 min. Cell extracts were diluted with 3 volumes of dilution buffer containing 20 mM HEPES (pH 8.0), 2.5 mM MgCl2, 0.1 mM EDTA, 0.05% TX-100, 0.5 mM dithiothreitol [DTT], 20 mM sodium β-glycerophosphate, 0.1 mM sodium orthovanadate, and protease inhibitors as described above. The diluted cell extract was rotated at 4°C for 30 min and centrifuged at 13,000 × g for 30 min. The supernatant was mixed with 10 to 20 μl of GSH-agarose suspension (Sigma) containing approximately 10 μg of GST-cJun1-79 fusion protein or GST alone (control for nonspecific binding). The mixture was rotated at 4°C overnight. After five washes in HEPES binding buffer (20 mM HEPES [pH 8.0], 2.5 mM MgCl2, 0.1 mM EDTA, 50 mM NaCl, 0.05% TX-100) and a final wash in kinase buffer (20 mM HEPES [pH 8.0], 20 mM MgCl2, 20 mM sodium β-glycerophosphate, 0.1 mM sodium vanadate, 2 mM DTT), beads were resuspended in 30 μl of kinase buffer containing 20 μM ATP and 3 μCi of [γ-32P]ATP. After 40 min at 30°C, the reaction was terminated by two washes with HEPES binding buffer. Phosphorylated proteins were boiled in 30 μl of 2× sodium dodecyl sulfate (SDS) sample buffer and resolved on SDS–10% polyacrylamide gels, which were subjected to autoradiography.
Quiescent C3H 10T½ cells treated as indicated were lysed in 75 μl of buffer containing 20 mM Tris-HCl (pH 8.0), 50 mM NaF, 5 mM MgCl2, 10 mM EGTA, 100 μM sodium vanadate, 1% TX-100, protease inhibitors as described previously (39, 40), and 1 μM microcystin-LR. Cell extracts were cleared by centrifugation at 6,000 × g for 10 min at 4°C and electrophoresed in SDS–14% polyacrylamide gels containing 200 μg of random copolymer l-glutamic acid and tyrosine (4:1; poly-Glu/Tyr; Sigma) per ml. After electrophoresis, SDS was removed by incubating gels in 20% isopropanol in 50 mM Tris-HCl (pH 8.0) (1 h, 250 ml), followed by 1 h in 50 mM Tris-HCl (pH 8.0)–1 mM DTT (250 ml). To denature proteins, gels were incubated for 1 h in 6 M guanidine-HCl (AnalaR; Sigma)–20 mM DTT–2 mM EDTA–50 mM Tris-HCl (pH 8.0) (50 to 100 ml). Proteins were renatured by incubation at 4°C, without agitation, in 250 ml of 1 mM DTT–2 mM EDTA–0.04% Tween 20–50 mM Tris-HCl (pH 8.0) for 12 to 18 h. For kinase assays, gels were equilibrated for 1 h in 10 ml of kinase buffer (40 mM HEPES [pH 8.0], 1 mM DTT, 0.1 mM EGTA, 20 mM MgCl2, 100 μM sodium vanadate), and the kinase reaction was carried out for 60 min in 10 ml of the same buffer containing 30 μM ATP and 10 μCi of [γ-32P]ATP (NEN) per ml. The gels were then washed extensively in 5% (wt/vol) trichloroacetic acid plus 1% sodium pyrophosphate (Sigma) until washes were free of radioactivity (usually four to five changes). Autoradiography of dried gels was performed by using RX-100 (Fuji) film with two intensifying screens.
To study desensitization, we had first to establish conditions where the original responses induced by the desensitizing stimulus had returned to basal levels prior to restimulation. After optimization using various stimuli, the following protocol was used for all desensitization experiments: C3H 10T½ cells were exposed to a desensitizing stimulus for 3 h (pretreatment), by which time both kinase activation and IE gene responses to the initial stimulus had largely diminished, and the second stimulus (restimulation) was then added directly to the medium. Note that although we have also observed anisomycin desensitization in other cell lines (CHO and HeLa), it is important to empirically determine the optimal desensitizing conditions for different cell lines, given that the period before the initial signals are lost will vary between cell lines and for different stimuli. For example, even in C3H 10T½ cells, which proved useful for studying desensitization of stress kinases and IE genes, it was not possible to analyze p70/85S6k desensitization because its activation by anisomycin persists well beyond the 3-h time point.
Because desensitization of IE gene induction is poorly described in the literature, we first characterized this phenomenon by using classical desensitizing stimuli and a panel of fos and jun genes (c-fos, fosB, c-jun, junB, and junD). With the exception of the junD mRNA level, fos and jun mRNA levels were generally low to undetectable after 3 h of stimulation with all agents used (see Fig. Fig.3A3A [lanes 3 and 8], Fig. Fig.4A4A [lane 3], and reference 26a). junD transcription is transient in these cells (23a); the transcripts persist because they are more stable than other jun mRNAs (48). The weakness and persistence of the junD response precluded quantitative analysis of desensitizing treatments on junD induction; quantitative data are presented for all other genes. Quantitative graphical representations of IE gene desensitization were obtained by densitometric scanning of autoradiographs, correcting for loading against GAPDH mRNA, and expressed as a percentage of the normal response seen in nondesensitized cells (see Materials and Methods).
Cells pretreated with EGF showed virtually total homologous desensitization of all five fos and jun genes to restimulation with EGF (Fig. (Fig.2A,2A, lanes 3 and 13; Fig. Fig.2B).2B). junD transcripts were not induced above the residual level remaining after the first stimulation (Fig. (Fig.2A,2A, lanes 3 and 13) (26a). When EGF-pretreated cells were exposed to other stimuli, the specificity of desensitization became apparent. After EGF pretreatment, bFGF-stimulated induction of c-fos and fosB was strongly inhibited, while junB induction was approximately 50% of the normal response (Fig. (Fig.2A,2A, lane 14; Fig. Fig.2B).2B). In contrast, bFGF-stimulated c-jun induction was relatively unaffected in EGF-desensitized cells. This result shows that bFGF receptors must still be functional at the cell surface and that the partial desensitization of other genes must occur within downstream signalling cascades or at the level of the gene itself (see Discussion). We commonly observe that when the desensitizing stimulus elicits strong ERK activation (EGF, bFGF, and TPA), the c-fos gene is consistently less responsive to restimulation with any other agent (Fig. (Fig.2A,2A, lanes 4, 5, 10, 14, and 17 to 19; Fig. Fig.2B)2B) (26a), which is probably due to postinduction autorepression that has been shown to occur at the level of the c-fos promoter (33, 38, 46, 53).
Pretreatment with EGF did not strongly desensitize IE gene induction in response to restimulation with UV radiation, with the notable exception of the c-fos gene (Fig. (Fig.2A,2A, lane 10; Fig. Fig.2B).2B). The inability of UV radiation to activate the c-fos gene and promoter in EGF-desensitized cells has been previously interpreted to support the case that UV radiation acts principally through the EGF receptor to activate this gene (50). Although we observe desensitization of c-fos, other IE genes remain clearly UV inducible (50 to 70% of the normal response [Fig. 2B]), supporting the view that c-fos induction is likely to be desensitized downstream of the EGF receptor and/or by autorepression of the c-fos gene. On balance, the partial desensitization seen here suggests that although some EGF receptor dimerization and activation may occur as a consequence of UV irradiation, it is likely that many other receptors will be similarly perturbed and contribute to gene activation (47, 50). Note that TPA induction of c-fos was also sensitive to EGF pretreatment (Fig. (Fig.2B),2B), although TPA does not utilize the EGF receptor (reference 27 and references therein). In contrast, TPA induction of c-jun was strongly augmented in EGF-desensitized cells, possibly due to synergy between protein kinase C- and EGF-stimulated signalling mechanisms (see below). Finally, after EGF pretreatment, all five IE genes remain clearly inducible by subinhibitory or inhibitory anisomycin, although the c-fos response is significantly reduced (Fig. (Fig.2A,2A, lanes 4 and 5; Fig. Fig.22B).
Pretreatment with bFGF resulted in virtually complete homologous desensitization of fos and jun induction in response to restimulation with bFGF (Fig. (Fig.2A,2A, lane 17; Fig. Fig.2B).2B). We also found that bFGF-desensitized cells responded very poorly to restimulation with EGF (Fig. (Fig.2A,2A, lane 16; Fig. Fig.2B).2B). This is an expected result, given that FGF causes transmodulation of EGF receptors, a phenomenon whereby activation of heterologous receptors, such as platelet-derived growth factor and bFGF receptors, results in phosphorylation and desensitization of the EGF receptor (3, 17, 21, 27). In bFGF-desensitized cells, we observed little, if any, effect on anisomycin-stimulated induction of fosB, c-jun, junB, and junD (Fig. (Fig.2A,2A, lanes 18 and 19; Fig. Fig.2B),2B), whereas the c-fos response was markedly reduced, again suggesting that the c-fos effects are gene specific and related to negative regulation at the promoter of this gene.
Having shown clear homologous and heterologous desensitization of IE gene induction by growth factors, we tested the effects of pretreating cells with agents which activate stress kinases, such as anisomycin and TNF-α (Fig. (Fig.3A,3A, lanes 3 to 6, 8 to 11, 16, and 22 to 24; Fig. Fig.3B).3B). Subinhibitory anisomycin was used for pretreatment to avoid complications due to translational arrest and because its signalling and gene-inducing responses at these concentrations are substantially over by 3 h (8, 9, 23). Nuclear run-on analyses as well as direct measurements of mRNA stability prove that subinhibitory anisomycin acts at the transcriptional level to induce these genes (23, 39).
Pretreatment with subinhibitory anisomycin produced very specific and virtually complete homologous desensitization of these cells to restimulation with either subinhibitory or inhibitory anisomycin (Fig. (Fig.3A,3A, lanes 5, 23, and 24; Fig. Fig.3B).3B). Note that the residual levels of c-jun and junB transcripts from the initial stimulation, which are maintained for up to 6 h after anisomycin pretreatment (26a), form the major contribution to the apparently less-than-complete desensitization seen for these two genes (Fig. (Fig.3A,3A, lane 3). These residual levels of c-jun and junB transcripts were reduced further if the anisomycin in the pretreatment medium was washed out prior to restimulation (see below), under which conditions we observe virtually complete homologous desensitization of both these genes (26a). junD transcripts remained at high levels for up to 6 h after pretreatment (Fig. (Fig.3A,3A, lane 3) (26a), and restimulation with subinhibitory or inhibitory anisomycin did not cause any further increase (Fig. (Fig.3A,3A, lanes 5, 23, and 24). Thus, anisomycin elicits virtually complete homologous desensitization of all five IE genes studied here.
In contrast, anisomycin-desensitized cells remain responsive to restimulation with EGF, bFGF, or TPA, showing that receptors and signalling mechanisms for all three ligands are still functional in anisomycin-desensitized cells. In fact, EGF- and TPA-stimulated c-fos, fosB, c-jun, and junB induction responses were generally enhanced in anisomycin-pretreated cells (Fig. (Fig.3A,3A, lane 4; Fig. Fig.3B).3B). bFGF-stimulated induction of c-jun was also enhanced in anisomycin-desensitized cells, whereas bFGF-induced c-fos, fosB, and junB transcripts were induced to approximately 50 to 70% of the normal response (Fig. (Fig.3A,3A, lane 22; Fig. Fig.3B).3B). Because the second stimulus is added directly to the pretreatment medium which contains anisomycin, the enhanced inductions seen here could arise from the known synergy between anisomycin and the restimulating agents. To verify this, we tested the effects of washing out the anisomycin with serum-free medium prior to restimulation. Under these conditions, the enhanced transcript levels in response to restimulation with EGF or bFGF were no longer observed (26a), although the cells remained completely homologously desensitized to restimulation with anisomycin. This finding suggests that the enhanced inductions are due to synergy between anisomycin in the pretreatment medium and the restimulating agent.
Because anisomycin is widely used as an agonist of stress responses, we then examined whether TNF-α- or UV radiation-induced responses were compromised in anisomycin-pretreated cells (Fig. (Fig.3A,3A, lane 6; Fig. Fig.3B).3B). Neither TNF-α nor UV radiation induces significant fosB transcription in C3H 10T½ cells, and compared to anisomycin, both agents are weak inducers of junD (Fig. (Fig.3A,3A, lanes 7 and 14; Fig. Fig.4A,4A, lanes 2, 9, and 12) (26a). Although c-fos was virtually completely desensitized to TNF-α restimulation and junB was inhibited by approximately 50% (Fig. (Fig.3A,3A, lane 6; Fig. Fig.3B),3B), TNF-α-stimulated c-jun induction appeared similar to that seen in nondesensitized cells (Fig. (Fig.3A,3A, lane 6), indicating that TNF-α receptors remain functional in anisomycin-desensitized cells. Restimulation of these cells with UV radiation produced a desensitization profile very similar to that seen with TNF-α, except that the c-fos gene remained significantly UV-inducible in anisomycin-desensitized cells (Fig. (Fig.33B).
We then tested the desensitizing effects of TNF-α, a physiological stimulus which, like anisomycin, activates the stress kinases relatively strongly (see below). Pretreatment with TNF-α resulted in virtually total homologous desensitization of all these genes (Fig. (Fig.3A,3A, lane 11; Fig. Fig.3B).3B). Although TNF-α does not induce fosB (Fig. (Fig.3A,3A, lane 7) (26, 26a), it appears to produce a repressive effect on this gene, as EGF-stimulated fosB induction was strongly inhibited (Fig. (Fig.3A,3A, lane 9; Fig. Fig.3B).3B). However, c-jun and junB remained fully EGF inducible, and c-fos induction was only partially inhibited in TNF-α-pretreated cells (Fig. (Fig.3B).3B). When TNF-α-pretreated cells were restimulated with anisomycin, differential effects were observed; c-fos, c-jun, and junD inductions were partially inhibited, while junB induction was markedly enhanced (Fig. (Fig.3A,3A, lane 10; Fig. Fig.3B).3B). Restimulation with UV radiation elicited a desensitization profile very similar to that seen with anisomycin, especially the enhanced induction of junB (Fig. (Fig.3A,3A, lane 16; Fig. Fig.3B).3B). These data suggest that TNF-α acts on its receptor to produce complete homologous desensitization and also elicits some downstream effects which cause partial and selective heterologous desensitization.
Finally, we examined whether a classical stress-inducing treatment, UV radiation, was capable of eliciting homologous or heterologous desensitization responses. Cells were exposed to 200 J of UV radiation per m2, left for 3 h, and then restimulated with various secondary stimuli (Fig. (Fig.4A).4A). This analysis showed that UV pretreatment resulted in widespread and virtually complete inhibition of induction of the c-fos, fosB, c-jun, and junB genes upon restimulation with all agents analyzed (Fig. (Fig.4A,4A, lanes 4 to 7, 13, 15, 17, and 19). Only UV-induced junD transcripts remained detectable after 3 h of pretreatment, and restimulation with bFGF and TPA resulted in a small increase in junD transcript levels (Fig. (Fig.4A,4A, lanes 17 and 19).
The general widespread inhibition of IE gene induction after UV pretreatment could be due to its well-known toxicity, such as its cross-linking effects. On the grounds that any such toxic inhibition would be apparent directly after UV treatment, we treated cells with UV radiation followed immediately by treatment with EGF, anisomycin, or a combination of the two (superinduction) (Fig. (Fig.4B,4B, EAn). As before, all fos and jun induction responses were strongly inhibited in cells exposed to UV radiation just before stimulation (Fig. (Fig.4B,4B, lanes 4, 6, and 8). Note that the order of costimulation is unimportant because these responses were also inhibited if EGF-anisomycin was added first and the cells were exposed immediately afterward to UV radiation (26a). Thus, UV radiation interferes with all fos and jun induction responses by a general toxic mechanism, in marked contrast to the highly specific and selective desensitizing effect elicited by anisomycin.
These data suggest that anisomycin acts not like a general stress or toxic stimulus but like a true signalling agonist in producing clear homologous desensitization in C3H 10T½ cells. We next investigated possible mechanisms which might account for this, particularly if it could be correlated with the loss of responsiveness of the kinases which mediate anisomycin-induced IE gene expression (25).
The activation characteristics of MAP kinase subtypes in C3H 10T½ cells by most agents used here have been reported previously (8, 9, 25). EGF and bFGF elicit very strong ERK activation but weakly and transiently activate JNK/SAPKs and p38/RK. TPA activates ERKs sustainedly but does not activate either JNK/SAPKs or p38/RK (8, 9, 25, 35a). Anisomycin does not activate ERKs, and UV radiation produces weak, barely detectable ERK activation; both of these agents very potently activate JNK/SAPKs and p38/RK (8, 9, 25). TNF-α also activates JNK/SAPKs and p38/RK (10a). It should be noted that no single MAP kinase subtype is indispensably required for fos and jun gene induction (9); anisomycin induces these genes without activating ERKs, and conversely, TPA induces them without JNK/SAPK or p38/RK activation.
JNK/SAPKs were assayed by exploiting their ability to bind to and phosphorylate the amino terminus of c-Jun (28) (see Materials and Methods). Subinhibitory anisomycin-stimulated activation of JNK/SAPKs peaks at 15 to 60 min (9) and then decreases to residual levels by 2 to 3 h (10b). After 3 h of pretreatment with subinhibitory anisomycin, a small amount of residual JNK/SAPK activity remained detectable (Fig. (Fig.5A,5A, lane 4). However, on restimulation with anisomycin at either an inhibitory (lane 5) or subinhibitory (lane 6) concentration, JNK/SAPK activation was no longer observed.
Activation of the p38/RK cascade was measured using its downstream kinase MAPKAP K-2 as an indicator. We have previously described an in-gel kinase assay specific for MAPKAP K-2 (10). Subinhibitory anisomycin elicits activation of MAPKAP K-2 which is maximal from 15 to 60 min (10, 25) and decreases to near basal levels by 2 to 3 h (10b). As with JNK/SAPKs, pretreatment with anisomycin leaves low residual MAPKAP K-2 activity detectable after 3 h (Fig. (Fig.5B,5B, lane 4). Furthermore, like the JNK/SAPKs, MAPKAP K-2 was no longer strongly activated by either inhibitory (Fig. (Fig.5B,5B, lane 5) or subinhibitory (Fig. (Fig.5B,5B, lane 6) anisomycin in anisomycin-desensitized cells.
Thus, anisomycin pretreatment produces a loss of responsiveness of both JNK/SAPKs and p38/RK to restimulation with anisomycin, which corresponds exactly with and probably explains the homologous desensitization of IE gene induction described above. The desensitization of these kinases to anisomycin is particularly remarkable given that the molecule is one of the most potent activators of JNK/SAPK and p38/RK cascades currently known.
We next examined whether, although no longer responsive to anisomycin, JNK/SAPKs and p38/RK remain present and responsive to other stimuli in anisomycin-desensitized cells. We therefore restimulated anisomycin-desensitized cells with heterologous stimuli and analyzed the state of JNK/SAPK and p38/RK activation. In control cells (Fig. (Fig.6A6A and B, lanes 4 and 5), EGF and bFGF elicited weak JNK/SAPK activation after 15 min (Fig. (Fig.6A)6A) and MAPKAP K-2 after 5 min (Fig. (Fig.6B).6B). In anisomycin-desensitized cells (Fig. (Fig.6C6C and D, lanes 16 and 17), JNK/SAPK and MAPKAP K-2 activation by these growth factors was not inhibited, and JNK/SAPK activation was in fact greater than in nondesensitized cells (Fig. (Fig.6C6C and D), reminiscent of the enhanced induction of specific IE genes under these conditions (Fig. (Fig.3A,3A, lanes 4 and 22; Fig. Fig.3B).3B). An unexpected result was that TPA, which was not able to stimulate either JNK/SAPK or MAPKAP K-2 (Fig. (Fig.6A6A and B, lanes 7) activation in control cells, was able to weakly activate these kinases, particularly MAPKAP K-2, in anisomycin-desensitized cells (Fig. (Fig.6C6C and D, lanes 19). These enhanced responses raise the possibility that the anisomycin-desensitized component(s) directly or indirectly acts negatively at some step(s) during the activation of JNK/SAPKs and p38/RK by EGF, bFGF, and TPA. However, as for the IE genes described above, the enhanced responses may also be due to the known synergy between these factors and anisomycin present in the pretreatment medium. The ability of TNF-α to activate JNK/SAPKs (Fig. (Fig.6A6A and C, lanes 6 and 18) and p38/RK (Fig. (Fig.6B6B and D) was completely unaffected in anisomycin-desensitized cells, suggesting that the anisomycin-desensitized signalling component(s) is not involved in TNF-α-stimulated activation of JNK/SAPKs and p38/RK (see Discussion).
It is worth noting that induction of certain genes by TNF-α and bFGF (TNF-α-induced c-fos and junB; bFGF-induced junB and, to a lesser extent, c-fos and fosB) is reduced in anisomycin-desensitized cells, in contrast to the enhanced or unaffected levels of JNK/SAPK or p38/RK activation these stimuli produce under these conditions; this must represent gene-specific influences brought about by anisomycin pretreatment and may be due either to negative cross talk within specific signalling pathways or to the loss or desensitization of other signalling mechanisms not studied here.
In contrast to the findings presented above, UV radiation, which strongly activated JNK/SAPKs and p38/RK (Fig. (Fig.6A6A and B, lanes 8) in control cells, elicited a substantially decreased response in anisomycin-desensitized cells (Fig. (Fig.6C6C and D, lanes 20). By densitometric quantitation, UV-stimulated JNK/SAPK activation was desensitized by approximately 90% and MAPKAP K-2 was desensitized by approximately 50% in anisomycin-pretreated cells. A very similar response was seen when anisomycin-desensitized cells were subjected to hyperosmotic shock. Hyperosmotic stress by either sorbitol or sodium chloride very strongly activated both JNK/SAPKs and p38/RK in these cells (Fig. (Fig.6A6A and B, lanes 11 and 12), whereas in anisomycin-pretreated cells, JNK/SAPK activation was very substantially desensitized to hyperosmotic shock and p38/RK was only partially desensitized (Fig. (Fig.6C6C and D, lanes 23 and 24).
Thus, the inability of anisomycin to activate JNK/SAPKs and p38/RK in desensitized cells is not due to the loss of these kinases or to any major aberration in upstream circuitry, as EGF, bFGF, and TNF-α are all able to activate JNK/SAPKs and p38/RK in anisomycin-desensitized cells. Only UV- and hyperosmolarity-induced activation of these kinases, especially JNK/SAPKs, is compromised by anisomycin pretreatment (see Discussion).
There are two possible causes for homologous desensitization in anisomycin-pretreated cells. One, modelled on conventional desensitization, is the loss or inactivation of signalling protein(s) crucial for the anisomycin response. However, because anisomycin is used here at subinhibitory concentrations, it is also possible that newly synthesized proteins translated from anisomycin-induced mRNA transcripts may be responsible for down-regulating anisomycin responsiveness in these cells. To distinguish between the two, we analyzed anisomycin-induced homologous desensitization of JNK/SAPK and p38/RK cascades in cells in which transcription was blocked with actinomycin D (Fig. (Fig.7B,7B, lanes 5 to 8). This showed that anisomycin desensitization of JNK/SAPKs (Fig. (Fig.7A,7A, upper panel) and MAPKAP K-2 (Fig. (Fig.7A,7A, lower panel, lanes 1 to 4) was still observed in the absence of transcription (Fig. (Fig.7B,7B, upper and lower panels). The same result was obtained in assays using 5,6-dichloro-1-β-d-ribofuranosyl-benzimidazole to inhibit transcription (35a). Note that anisomycin-stimulated activation of JNK/SAPKs and MAPKAP K-2 was stronger and the background levels were higher in the actinomycin D-treated cells (Fig. (Fig.7B).7B). By analogy to phosphatases for ERKs and JNK/SAPKs which are encoded by newly induced mRNAs (32, 37, 56), this may be a consequence of the inhibited transcription of the phosphatases that deactivate these kinases. We conclude that homologous desensitization in anisomycin-pretreated cells is not due to the fresh synthesis of inhibitory components but involves the more conventional mechanism of degradation or inactivation by negative feedback of some signalling component(s) crucial for the anisomycin response.
We have established conditions for analyzing homologous and heterologous desensitization in C3H 10T½ cells of a panel of IE genes upon pretreatment with diverse stimuli. This led to the surprising finding that anisomycin behaved like a true signalling agonist in eliciting highly specific and virtually total homologous desensitization of IE gene induction; these genes remain inducible by other stimuli in anisomycin-desensitized cells. We then investigated the mechanism by which this occurs and found that anisomycin no longer activates the JNK/SAPK and p38/RK cascades in desensitized cells, although these kinases remain present and activatable by other stimuli. However, in anisomycin-desensitized cells, JNK/SAPK activation by UV radiation and hyperosmotic shock was virtually completely ablated (<10% of the normal response), and MAPKAP K-2 activation by these treatments was inhibited by about 50%. The possible identity of the anisomycin-desensitized signalling components and the implications of these findings for upstream events controlling UV- and hyperosmotic shock-induced activation of the ERKs and JNK/SAPKs are discussed below.
The best-characterized mechanism of homologous desensitization involves internalization and recycling of surface receptors such as EGF receptors, resulting in a temporary loss of responsiveness of these cells to EGF (reference 27 and references therein). This is well established as a general phenomenon affecting many different receptors (reviewed in reference 16). There is also considerable evidence for postreceptor negative-feedback mechanisms which can prevent reactivation of the pathway. For example, within the ERK cascade, MEK, Sos, and Raf-1 have all been shown to be subject to feedback phosphorylation and/or desensitization when this cascade is activated (6, 11, 35, 57). Finally, desensitization may occur at the level of the gene itself; transcription factors and/or a regulatory element(s) acting upstream of IE genes may be specifically desensitized after induction, as exemplified by autorepression of the c-fos gene (33, 38, 46, 53).
All postreceptor desensitization mechanisms described above will obviously prevent responses to other stimuli which rely on the same down-regulated component. Heterologous desensitization can also occur at the level of receptors. For example, platelet-derived growth factor and bFGF activate serine/threonine kinases (ceramide-activated protein kinase  and protein kinase C [reference 27 and references therein]) that phosphorylate the EGF receptor, resulting in their conversion from high to low affinity (transmodulation [3, 17, 21]). This has been characterized in C3H 10T½ cells (27) and may contribute to partial desensitization of EGF-inducible IE genes in bFGF-pretreated cells. Cross-talk mechanisms, for example, the negative regulation of the ERK cascade in cyclic AMP-treated cells which is mediated by phosphorylation of Raf by protein kinase A, can also produce heterologous desensitization (58). Finally, degradation of specific signalling enzymes may cause loss of responsiveness to any stimuli that utilize that enzyme, the best-studied example of which is the loss of protein kinase C by proteolysis upon prolonged TPA pretreatment (reference 59 and references therein).
Although desensitization of signalling enzymes is well described in the literature, this study represents the first comprehensive characterization of this phenomenon on a panel of IE genes. We show here that pretreatment with every one of the physiological signalling ligands used here, which include EGF, bFGF, and TNF-α, elicited very clear and virtually complete homologous desensitization of all five fos and jun genes (Fig. (Fig.22 to to3).3). This may turn out to represent a general phenomenon elicited by diverse ligands which parallels the down-regulation of receptors at the cell surface. In this respect, we found that anisomycin behaves exactly like a true signalling ligand, eliciting highly specific homologous desensitization of all five IE genes.
Heterologous desensitization of IE gene induction, being the cumulative result of modulatory events occurring at all levels (receptors, signalling cascades, transcription factors, and upstream regulatory elements), appears much more complex. Nevertheless, some clear trends emerge. One of these is that the c-fos gene is uniquely sensitive to heterologous desensitization, much more so than any other gene studied here, and especially when the desensitizing stimulus strongly activates ERKs. Another unexpected finding is that pretreatment of cells can also strongly up-regulate induction of certain IE genes in response to restimulation with another ligand; this is particularly clear with TPA-stimulated c-jun induction in EGF-pretreated cells (Fig. (Fig.2B)2B) and with junB induction in TNF-α-pretreated cells (Fig. (Fig.3A3A and B). We also observe that although TNF-α does not activate fosB, it represses activation of fosB by any other stimulus. Finally, we show that UV pretreatment does not elicit selective desensitization but interferes nonspecifically, probably by its toxicity, to inhibit induction of all five IE genes studied here.
Assuming that these alterations in IE gene mRNA levels are reflected at the protein level, the up- and down-modulations described above might be expected to alter the composition of AP-1 complexes, which has obvious consequences for transcriptional regulation. The highly selective effects on IE gene induction reported here also imply that in the intact organism, where cells might be expected to be exposed to combinations of growth factors and cytokines which may occur in a particular sequence, the exact profile of IE gene expression would be under very complex regulation.
On investigating the mechanism of anisomycin-stimulated homologous desensitization of IE genes, we found that this correlates with and is probably caused by its inability to activate JNK/SAPKs and p38/RK in desensitized cells. This is a particularly striking result because anisomycin is among the most potent of activators of these two MAP kinase subtypes in C3H 10T½ cells (9, 10). Note that it is not possible to do similar desensitization assays with the third kinase, p70/85S6k, that anisomycin strongly activates, because even at subinhibitory concentrations, p70/85S6k remains active for up to 6 h (35a). Although nonresponsive to anisomycin, JNK/SAPK and p38/RK activation in anisomycin-desensitized cells is normal or augmented in response to EGF, bFGF, or TNF-α, indicating that the kinases are still present and the upstream circuitry linking them to the membrane receptors still intact. It is worth noting also that the augmented kinase activation correlates with correspondingly enhanced IE gene induction seen in these circumstances.
Of all agents tested, only UV radiation and hyperosmolarity elicited markedly reduced activation of JNK/SAPKs and p38/RK in anisomycin-desensitized cells. In both instances, JNK/SAPK activation was very strongly desensitized (<10% of the normal response) by anisomycin pretreatment, whereas p38/RK activation was desensitized to about 50% of the normal response. Using a specific inhibitor (SB 203580), we have recently shown that p38/RK is essential for the induction of all these genes by anisomycin or UV radiation (25, 26). The diminished activation of p38/RK by UV radiation in anisomycin-desensitized cells may explain the equivalent reduction in the IE gene induction elicited by this agent, in accord with the view that p38/RK is essential and rate limiting for UV-induced IE gene induction. Note that although we have attempted to analyze IE gene induction and desensitization in hyperosmotically shocked cells, it was not possible since the responses were extremely weak and variable.
The area within the intracellular signalling circuitry where anisomycin desensitization must occur is depicted schematically in Fig. Fig.8.8. It is important to stress that anisomycin desensitization of signalling and IE gene induction is clearly separate from its effects on translation; [35S]methionine-labelling experiments prove that the ability of anisomycin to bind ribosomes and block translation in these cells remains completely unaltered after anisomycin desensitization (60a). The desensitization cannot therefore be due to the loss of anisomycin-binding sites on the ribosome. Second, it is clear from experiments with transcriptional inhibitors that nascent products of anisomycin-induced transcripts are not required for anisomycin desensitization; it is therefore likely to be due to the loss or desensitization of specific anisomycin-activated signalling components. However, it is evident that desensitization is not due to the loss of JNK/SAPKs or p38/RK in these cells; the desensitizing event must therefore occur upstream of these two kinases. Third, it is also clear from this work that the desensitized components are not required for transmitting signals from transmembrane tyrosine kinases such as EGF or bFGF or TNF-α receptors to JNK/SAPKs and p38/RK. Thus, for the first time, it is possible to conclude that the signalling effects of anisomycin must originate outside the normal signalling pathways utilized by these receptors.
We also know that anisomycin is a strong activator of p70/85S6k, which is sensitive to ablation using rapamycin (14, 31); this finding indicates that anisomycin must act upstream of FRAP/TOR in these cells. Note that EGF utilizes both the FRAP/TOR→p70/85S6k and MKK6→p38/RK→MAPKAP K-2 pathways in these cells (25, 31); any anisomycin desensitization mediated via direct effects on these enzymes would produce heterologous desensitization to EGF as well, which was not seen. This suggests that the desensitizing event lies higher upstream of all these enzymes. Of a number of enzymes that lie upstream of JNK/SAPKs, it is clear that anisomycin strongly activates the immediately upstream kinase SEK1/JNKK/MKK4 (42, 52) and that this is essential for anisomycin-stimulated JNK/SAPK activation. However, it does not activate germinal center kinase (45) or require the small GTPases Rac and Cdc42 (18), which have been proposed to act via the p21-activated kinases (41) to activate JNK/SAPKs (1, 43, 44, 60).
These observations allow the hypothesis that anisomycin-stimulated signalling must originate further upstream of all the signalling events described here and utilizes enzymes which are not involved in receptor-mediated kinase activation and IE gene induction. Thus, the desensitizing event must involve either the anisomycin “receptor” itself or components close to and specific for the anisomycin “receptor” from which these signals originate. Current indications are that this component may also be involved in UV radiation- and hyperosmolarity-induced JNK/SAPK activation, although this requires further study. The down-regulation of receptors and signalling components is a common motif with true signalling agonists, and the observations reported here will help identify the putative signalling receptor for the anisomycin molecule.
C.A.H. and R.L.P. are funded by the Cancer Research Campaign, and E.C. is supported by a European Union fellowship. R.L.P. acknowledges the support of the Foundation Rene Touraine.
We thank N. Zhelev of this laboratory for help with the methionine-labelling experiments.