Gadd45 mediates the NF-B suppression of JNK signalling by targeting MKK7/JNKK2

NF-B/Rel transcription factors control apoptosis, also known as programmed cell death. This control is crucial for oncogenesis, cancer chemo-resistance and for antagonizing tumour necrosis factor (TNF)-induced killing^{1, 2}. With regard to TNF, the anti-apoptotic activity of NF-B involves suppression of the c-Jun N-terminal kinase (JNK) cascade^{3, 4, 5}. Using an unbiased screen, we have previously identified Gadd45/Myd118, a member of the Gadd45 family of inducible factors⁶, as a pivotal mediator of this suppressive activity of NF-B³. However, the mechanisms by which Gadd45 inhibits JNK signalling are not understood. Here, we identify MKK7/JNKK2 — a specific and essential activator of JNK^{7, 8} — as a target of Gadd45, and in fact, of NF-B itself. Gadd45 binds to MKK7 directly and blocks its catalytic activity, thereby providing a molecular link between the NF-B and JNK pathways. Importantly, Gadd45 is required to antagonize TNF-induced cytotoxicity, and peptides disrupting the Gadd45/MKK7 interaction hinder the ability of Gadd45, as well as of NF-B, to suppress this cytotoxicity. These findings establish a basis for the NF-B control of JNK activation and identify MKK7 as a potential target for anti-inflammatory and anti-cancer therapy.

JNK1/2/3 are the downstream component of one of the major mitogen-activated protein kinase (MAPK) cascades, also comprising the extracellular-signal-regulated kinases 1 and 2 (ERK1/2) and p38(///) cascades^{8, 9, 10}. Generally, activation of JNK (and p38) is associated with induction of apoptosis, whereas activation of ERK is linked to cell growth and survival^{8, 9, 10}. MAPKs are activated by phosphorylation by MAPK kinases (MAPKKs), which in turn are activated through phosphorylation by MAPKK kinases (MAPKKKs). To understand the basis for the Gadd45 control of JNK signalling, we examined whether Gadd45 physically interacted with kinases in these cascades^{8, 9, 10}. Haemagglutinin (HA)-tagged kinases were transiently expressed in 293 cells, alone or in combination with Flag-tagged Gadd45, and associations were assessed by combined immunoprecipitation and western blot assays. Gadd45 bound to ASK1, but not to other MAPKKKs capable of interacting with TRAF2 (ref. 8; Fig. 1a, left), a factor required for JNK activation by TNF¹¹. As reported previously¹², Gadd45 also associated with MEKK4/MTK1, a MAPKKK that instead is not induced by TNF^{8, 13}. Notably, Gadd45 interacted strongly with MKK7/JNKK2, but not with the other JNK kinase, MKK4/JNKK1, the p38-specific activators MKK3b and MKK6, the ERK kinase, MEK-1, or MAPKs^{8, 9, 10} (Fig. 1a, middle and right, and Fig. 1b). Gadd45 interactions were confirmed in vitro. Glutathione S-transferase (GST)−Gadd45, but not GST alone, precipitated a large fraction of the MKK7 input (Fig. 1c), whereas it absorbed only a small fraction of ASK1 or MEKK4. Hence, Gadd45 interacts with JNK-inducing kinases and most avidly with MKK7.

Figure 1. Gadd45 physically interacts with kinases in the JNK pathway. (a, b) Western blots with anti-Flag immunoprecipitates (top) or total lysates (middle and bottom) from 293 cells, showing association of Gadd45 with ASK1, MEKK4 and MKK7. (c) GST pull-down assays using GST- or GST−Gadd45-coated beads and 35S-labelled in-vitro-translated proteins. 40% of the inputs are shown. Full Figure and legend (54K)

Next, we examined whether the association of Gadd45 with these kinases had functional consequences in vivo. Remarkably, whereas in IBM-Hygro 3DO control clones³ TNF activated MKK7 strongly, this activation was abolished in clones expressing Gadd45 (Fig. 2a). Gadd45 also blocked induction of MKK7 by stress stimuli (see Supplementary Information, Figs. S1a, b). Inhibition was specific, as Gadd45 had no effect on induction of other MAPKKs (that is, MKK4, MKK3/6, and MEK1/2) by either treatment with TNF, or 12-O-tetradecanoylphorbol-13-acetate (TPA) and ionomycin (T/I; Figs. 2b and 2c, respectively). ASK1 and MEKK1 were activated weakly by TNF, as shown previously^{14, 15}, and similarly, this activation was unaffected by Gadd45 (Fig. 2b). Thus, Gadd45 selectively blocked induction of MKK7 phosphorylation/activity by TNF.

Figure 2. Gadd45 and NF-B specifically inhibit MKK7 in vivo. Western blots with antibodies against phosphorylated (P) or total kinases, and kinase assays (K.A.) showing MAPKK and MAPKKK activation by TNF or T/I in IBM-Hygro and IBM-Gadd45 clones (a−c) and in Neo and IBM 3DO clones (d, e). In a and d, MKK7 phosphorylation (P-MKK7) was monitored by combined immunoprecipitation (anti-P-MKK7 antibodies) and western blotting (anti-total MKK7 antibodies). Non-specific (n.s.) bands are indicated. In b, Kinase-dead (KD)-MKK7 is the MKK7 mutant, K149M. Full Figure and legend (94K)

Previously, we have shown that Gadd45 mediates the suppression of JNK signalling by NF-B³. Indeed, MKK7 was also inhibited by NF-B (Fig. 2d). Whereas in control 3DO clones (Neo), MKK7 activation by TNF returned to basal levels by 40 min (mirroring the JNK response³), this activation remained sustained in NF-B null clones (IBM). MKK7 down-regulation correlated with Gadd45 induction by NF-B (data not shown). Furthermore, NF-B did not inhibit MKK4, MKK3/6 or MEK1/2 (Figs. 2d and 2e), thereby recapitulating the effects of Gadd45 on MAPK cascades. Indeed, Gadd45 was up-regulated by T/I in Neo, but not in IBM, clones (see Supplementary Information Fig. S1c; see also Fig. 3a, bottom right).

Figure 3. Gadd45 is a direct inhibitor of MKK7. (a) Immunoprecipitations followed by western blots, showing physical association between endogenous Gadd45 and MKK7 (top) in 3DO cells treated with T/I (2 h) or left untreated (Un). Protein levels are shown (bottom). (b, g) Coomassie brilliant blue staining, showing purity of the proteins used in c−e. Size markers (S.M.) are indicated. (c) In vitro pull-down assays with purified proteins, showing direct physical interaction between His6/T7−Gadd45 and GST−MKK7. Precipitated GST proteins and bound His6/T7-tagged proteins were visualized with Coomassie (CS) and by western blotting with anti-T7 antibodies, respectively. Inputs of His6/T7-tagged proteins are indicated. The fraction of His6/T7−Gadd45 and His6/T7−JIP1 binding to GST−MKK7 was calculated relative to a standard curve generated from known protein concentrations17. (d, e) Kinase assays showing specific inhibition of active MKK7 by purified GST−Gadd45 and His6−Gadd45 in vitro. Flag-tagged kinases were immunoprecipitated from 293 cells treated with TNF (10 min) or left untreated and pre-incubated with the indicated concentrations of Gadd45 polypeptides. (f) Western blots showing exogenous kinase levels in 293 cells. Full Figure and legend (87K)

Importantly, an interaction between endogenous Gadd45 and MKK7 was detected readily (Fig. 3a). Anti-Gadd45 antibodies¹⁶ co-immunoprecipitated MKK7 from T/I-treated 3DO cells exhibiting strong Gadd45 expression (Fig. 3a, bottom right), but not from untreated cells lacking detectable Gadd45. MKK7 was present at comparable levels in stimulated and unstimulated cells (Fig. 3a, bottom left) and was not co-precipitated by isotype-matched control antibodies (data not shown). The interaction was confirmed by using anti-MKK7 antibodies for immunoprecipitation and the anti-Gadd45 antibody for western blots (Fig. 3a, top right). Anti-MEKK1 antibodies failed to co-precipitate Gadd45, further demonstrating the specificity of the MKK7−Gadd45 association. To determine whether Gadd45 binds to MKK7 directly, we used purified recombinant proteins (Fig 3b). Purified GST−MKK7 or GST were incubated in vitro with increasing amounts of purified His₆−Gadd45 or control His₆−JIP1 (ref. 17), and the fraction of His₆-tagged polypeptides binding to GST proteins was visualized by western blotting. His₆−Gadd45 associated specifically with GST−MKK7 (Fig. 3c), and this association was tighter than that of the physiological MKK7 regulator, JIP1 (refs 8, 10), with half-maximal binding (HMB) values of 390 nM for Gadd45 and >650 nM for JIP1 (Fig. 3c, left; note that JIP1 was used under non-saturating conditions). We conclude that endogenous Gadd45 and MKK7 associate through direct, high-affinity contact.

Thus, we examined whether Gadd45 inhibited active MKK7 in vitro. Flag−MKK7 was immunoprecipitated from TNF-treated or untreated 293 cells, and kinase assays were performed in the presence of purified His₆−Gadd45, GST−Gadd45 or control proteins (Fig. 3d; see also Fig. 3g). Both Gadd45 polypeptides, but neither GST nor His₆−EF3 (ref. 18), blocked phosphorylation of GST−JNK1 by MKK7 in a dose-dependent manner (Fig. 3d). Of note, MKK7 auto-phosphorylation was also blocked (data not shown). Consistent with the in vivo findings (Fig. 2), the inhibitory activity of Gadd45 was specific. In fact, even at high concentrations, this factor did not hamper MKK4, MKK3b or ASK1 (despite its ability to bind to ASK1 in overexpression studies; Fig. 3e; see also Fig. 3f, total levels). Hence, Gadd45 is a potent and specific inhibitor of MKK7. Indeed, the effects of Gadd45 on TNF-induced MKK7 phosphorylation (see Fig. 2a, middle) may be caused by inhibition of the ability of MKK7 to autophoshorylate (data not shown) and/or to function as a substrate for upstream kinases. Nevertheless, this effect seems to be independent of ASK1, as Gadd45 does not interfere with MKK7 phosphorylation by this MAPKKK, either in vivo or in vitro (Figs. 2b and 3e, respectively; ASK1). Altogether, the findings identify MKK7 as a target of Gadd45, and in fact, of NF-B. Of interest, MKK7 is a selective activator of JNK^{8, 10}. Furthermore, its ablation abolishes JNK activation by TNF⁷ and rescues RelA null cells from TNF-induced killing¹⁹. Thus, blockade of MKK7 is alone sufficient to explain the effects of Gadd45 on JNK signalling; that is, its specific and near-complete suppression of this signalling³.

Gadd45 shares no homology with phosphatases and is not known to have enzymatic activity. Furthermore, we have found no evidence of such activity (data not shown). Thus, to understand mechanisms of kinase inactivation, we mapped the Gadd45-binding region(s) of MKK7 using sets of amino- and carboxy-terminally truncated MKK7 polypeptides (see Supplementary Information Fig. S2a, c, respectively). MKK7 (63−401), MKK7 (91−401) and MKK7 (132−401) bound to GST−Gadd45 specifically and with affinity comparable to that of full-length MKK7, whereas mutants starting between amino acids 157 and 213 interacted weakly with GST−Gadd45 (see Supplementary Information, Fig. S2b). Ablation of a region extending to or beyond residue 232 abolished binding. Analysis of C-terminal truncations confirmed the presence of a Gadd45-interaction domain between residues 141 and 161 (see Supplementary Information, Fig. S2d; compare MKK7(1−140) and MKK7(1-161)), but failed to identify the C-terminal binding region identified above, suggesting that Gadd45 interacts with this latter region more weakly. Hence, MKK7 contacts Gadd45 through two distinct regions located within residues 132−161 and 213−231 (hereafter referred to as regions A and B, respectively).

To define more precisely the interaction regions and determine whether they are sufficient for binding, we tested the association of Gadd45 with overlapping peptides spanning these regions (Fig. 4a). As shown in Fig. 4b, both region A and B bound to GST−Gadd45, even when isolated from the context of MKK7, and peptides 132−156 and 220−234 (that is, peptides 1 and 7, respectively; see Fig. 4c) were sufficient to recapitulate this binding. Interestingly, both peptides are found within the MKK7 kinase domain⁸, and peptide 1 spans the ATP-binding site, Lys 149, required for catalytic function^{8, 20}, suggesting that Gadd45 inactivates MKK7 by masking critical residues. To clarify this issue, we generated alanine-scanning mutations across peptide 1 (Fig. 4c). As shown in Fig. 4d, residues 133−140, and to a lesser extent 145−152, seemed crucial for the interaction with Gadd45 (peptide 1 and Ala-I, Ala-II, Ala-IV and Ala-V mutants; compare inputs and bound fractions). Furthermore, the association of Gadd45 with peptide 1 and with MKK7 itself, but not with peptide 7, was inhibited by ATP in a dose-dependent manner (Fig. 4e), suggesting that Gadd45 and ATP binding to MKK7 are sterically incompatible. This is reminiscent of the mechanism by which p27^KIP1 inhibits cyclin-dependent kinase 2 (Cdk2; ref. 21).

Figure 4. Gadd45 blocks MKK7 by contacting a peptidic region in its catalytic domain. (a) Schematic representation of the MKK7 peptides used for binding assays. Interaction regions are in grey. (b, d, e) GST pull-downs showing binding of GST−Gadd45 to the indicated 35S-labelled, in-vitro-translated MKK7 products. 40% of the inputs are shown in b and d. In e, ATP was used as indicated. (c) Amino-acid sequence of Gadd45-interacting peptides 1 and 7, and the peptide 1 mutants used in d. Lys 149 is marked by an asterisk. Amino acids involved in binding to Gadd45 are in grey; darkness of shading correlates with their apparent relevance to this binding. (f) Kinase assay (K.A.), showing that binding to peptidic region 1 is required for MKK7 inactivation by Gadd45. Flag−MKK7 was immunoprecipitated from 293 cells treated with TNF for 10 min. Full Figure and legend (94K)

Thus, we examined whether MKK7, Gadd45-binding peptides interfered with the Gadd45 inhibitory activity. Indeed, peptide 1 prevented MKK7 inactivation by Gadd45, whereas peptide 7 or control peptides did not (Fig. 4f). Consistently, only peptide 1 was capable of disrupting the MKK7−Gadd45 interaction in vitro (see Supplementary Information, Fig. S2e). Hence, in the context of full-length MKK7, peptide 7 is seemingly insufficient or unavailable for contacting Gadd45. We conclude that kinase inactivation by Gadd45 requires contact with region A, but not with region B, and most probably involves blockade of the MKK7 access to ATP.

These data predict that preventing MKK7 inhibition by Gadd45 in vivo should sensitize cells to TNF-induced apoptosis. To test this hypothesis, MKK7-mimicking peptides were fused to a cell-permeable HIV-TAT peptide²² and transduced into cells. As shown in Supplementary Information Figs S3a−d, peptides entered cells with equivalent efficiency. Remarkably, peptide 1 markedly increased susceptibility of IBM-Gadd45 cells to TNF-induced killing, whereas dimethyl sulphoxide (DMSO)-treated cells were resistant to this killing, as expected³ (Fig. 5a, left; see also Supplementary Information Fig. S4a). Other peptides, including peptide 7, had no effect on the apoptotic response to TNF. Importantly, peptide 1 exhibited marginal basal toxicity (see Supplementary Information, Fig. S4a, left) indicating that its effect was specific for cytokine stimulation. Further linking the in vivo effect of peptide 1 to Gadd45, the pro-apoptotic activity of Ala mutant peptides correlated with their apparent binding affinity for Gadd45 in vitro (Figs 4d and 5a, right; see also Supplementary Information Fig. S3c−e).

Figure 5. Gadd45-mediated suppression of MKK7 is required to block TNF-induced apoptosis. (a, b) Apoptosis assays, showing that peptide 1 effectively promotes cytotoxicity by TNF in IBM-Gadd45 and Neo 3DO clones, respectively. Amino-acid sequence of the mutant peptides can be found in Supplementary Information, Fig. S3e. (c, d) Apoptosis assays, showing that both peptide 1 and peptide 2 can facilitate TNF-induced killing in wild-type MEFs, and that only peptide 2 can promote this killing in Gadd45 null MEFs, respectively. MEFs were from twin embryos and were used at passage 4 (p4). In a−d, values (expressed as arbitrary units) were obtained by subtracting background values of untreated cells from values of TNF-treated cells, and represent the mean ( standard deviation) of three experiments. Full Figure and legend (54K)

Consistent with the notion that MKK7 is a target of NF-B (Figs. 2d and 2e), peptide 1 promoted TNF-induced killing in NF-B-proficient cells (Neo; Fig. 5b; see also Supplementary Information Fig. S4b), which are normally refractory to this killing^{2, 3}. As observed with Gadd45-expressing clones, this peptide exhibited minimal toxicity in untreated cells (Supplementary Information Fig. S4b), and mutation of residues required for interaction with Gadd45 abolished its effects on TNF cytotoxicity (Fig. 5b, right). Together, the findings demonstrate that Gadd45 halts the JNK cascade by inhibiting MKK7 and causally links the Gadd45 protective activity to this inhibition. Furthermore, blockade of MKK7 is crucial to the suppression of apoptosis by NF-B, and this blockade is mediated, at least in part, by induction of Gadd45.

A recent report suggested that ablation of Gadd45 does not affect TNF-induced programmed cell death in mouse embryonic fibroblasts (MEFs)²³, as might have been expected from our data. Thus, we tested the effects of MKK7-derived peptides in these cells. Surprisingly, cytokine-induced toxicity was enhanced by both peptide 1 and peptide 2 in wild-type fibroblasts, whereas other peptides had no effect on this toxicity (Fig. 5c and data not shown; see also Supplementary Information, Fig. S4c). This contrasts with our observations in 3DO lymphoid cells, where only peptide 1 promoted killing by TNF (Fig. 5b). As peptide 2 does not bind to Gadd45 (Figs. 4b and 4d), its pro-apoptotic activity may be caused by displacement of other inhibitory factor(s) from MKK7.

Consistently, the activity of peptide 2 was retained (and, in fact, enhanced) in gadd45^-/- MEFs (Fig 5d; see also Supplementary Information, Fig. S4d). Remarkably, however, Gadd45 ablation rendered these cells completely insensitive to the cytotoxic effects of peptide 1, indicating that in wild-type fibroblasts these effects were caused by inactivation of Gadd45. Together, these findings demonstrate that the MKK7 inhibitory mechanism activated in response to TNF is tissue-specific (shown by the distinct effects of MKK7 peptides in 3DO cells and fibroblasts; Figs 5b-d), and that, at least in MEFs, this mechanism is functionally redundant. These findings also provide compelling evidence that Gadd45 is required to antagonize TNF-induced killing (Fig. 5c). Indeed, the apparent lack of apoptotic phenotype previously reported in gadd45^-/- fibroblasts²³ seems to be caused by activation of compensatory mechanisms in these cells, mechanisms that are not mounted during acute Gadd45 inactivation by peptide 1.

Here we have identified a mechanism for the control of JNK signalling by Gadd45. Gadd45 associates tightly with MKK7 and inhibits its enzymatic activity by contacting critical residues in the catalytic domain; this inhibition is crucial for the suppression of TNF-induced apoptosis. Interactions with other kinases do not seem relevant to Gadd45-mediated control of JNK activation and programmed cell death by TNF, as MEKK4 is not involved in TNF receptor signalling^{8, 13}, and ASK1 is seemingly unaffected by Gadd45 (Figs. 2b and 3e). Indeed, peptides that interfere with binding of Gadd45 to MKK7 inhibit the Gadd45 protective activity against TNF (Figs 4f, 5a, 5c, and 5d; also see Supplementary Information, Fig. S2e). Importantly, the targeting of MKK7 is critical for suppression of apoptosis by NF-B itself. NF-B-deficient cells fail to down-modulate MKK7 induction by TNF (Fig. 2d), and MKK7-mimicking peptides hinder the ability of NF-B to block cytokine-induced killing (Figs. 5b and 5c). Our data are consistent with a model whereby NF-B activation induces expression of Gadd45, which in turn inhibits MKK7, resulting in the suppression of JNK signalling and ultimately in apoptosis triggered by TNF. These findings identify a crucial molecular link between the NF-B and JNK pathways. Indeed, the relevance of this link is underscored by knockout studies showing that Gadd45 is essential for blocking TNF-induced apoptosis (Figs 5c, d; also F.Z., S.P. & G.F., unpublished observations). In some tissues, however, other NF-B-inducible factors might contribute to suppress induction of MKK7 by TNF (Figs. 5c and 5d).

Chronic inflammatory conditions such as rheumatoid arthritis and inflammatory bowel disease are driven by a positive feedback loop created by mutual activation of TNF and NF-B^{24, 25, 26}. Furthermore, several malignancies depend on NF-B for their survival¹, a process that might involve suppression of JNK signalling⁵. Our data suggest that blockade of the NF-B ability to shut down MKK7 may promote apoptosis of self-reactive/pro-inflammatory cells and, perhaps, of cancer cells, thereby identifying the MKK7−Gadd45 interaction as a potential therapeutic target. Interestingly, pharmacological compounds that disrupt binding of Gadd45 to MKK7 might uncouple anti-apoptotic and pro-inflammatory functions of NF-B, and so circumvent the potent immunosuppressive side-effects observed with global NF-B blockers^{1, 5, 24, 26}, currently used to treat these illnesses. The pro-apoptotic activity of MKK7 peptides in NF-B-proficient cells implies that NF-B-inducible factors target MKK7 through, or in proximity to, its Gadd45-binding surface, thereby proving in principle the validity of this therapeutic approach.

pcDNA-HA-GCKR, pCEP-HA-MEKK1, pcDNA-HA-ASK1, pCMV5-HA-MEKK3, pCMV5-HA-MEKK4, pcDNA-HA-MEK1, pMT3-HA-MKK4, pSR-HA-JNK1, pMT2T-HA-JNK3, pcDNA-HA-ERK1, pSR-HA-ERK2, pcDNA-Flag-p38, pcDNA-Flag-p38, pcDNA-Flag-p38 and pcDNA-Flag-p38 were provided by A. Leonardi, H. Ichijo, J. Landry, R. Vaillancourt, P. Vito, T.H. Wang, J. Wimalasena, and H. Gram. pcDNA-HA-Gadd45, pGEX-JNK1, pET28-His₆/T7-JIP1 (expressing the MKK7-binding domain of JIP1b), and pProEx-1.His₆-EF3 (expressing edema factor 3) were described previously^{16, 17, 18, 27}. All other Flag- or HA-coding constructs were generated using pcDNA (Invitrogen, Carlsbad, CA). For bacterial expression, sub-clonings were in the following vectors: His₆/T7−Gadd45 (Figs. 3b and 3c) in pET-28 (Novagen, Madison, WI); His₆−Gadd45 (Fig. 3d, e and g) in pProEx-1.H₆ (ref. 18); GST−p38, GST−MKK7, GST−KD−MKK7 (containing the K149M mutation; Fig. 2b), and GST−Gadd45 in pGEX (Amersham, Piscataway, NJ). To prime in vitro transcription/translations, we generated: pBluescript(BS)-MEKK4, pBS-ASK1 and pBS-MKK7 (Fig. 1c); pBS-based plasmids expressing N-terminal truncations or polypeptidic fragments of human MKK7 (see Supplementary Information, Fig. S2a and Fig. 4a, respectively), or peptide 1 point mutants (Fig. 4c). To enhance radiolabelling, short peptidic fragments were expressed fused to enhanced green fluorescent protein (eGFP; Clontech, Palo Alto,CA). ASK1¹⁻⁷⁵⁷ (encoding amino acids 1−757 of ASK1; Fig. 1c) and C-terminal MKK7 truncations (see Supplementary Information, Fig. S2c) were obtained by linearizing pBS-ASK1 and pBS-MKK7, respectively, with appropriate restriction enzymes. Detailed cloning information will be provided upon request. All clonings were confirmed by sequencing and appropriate restriction digestion.

3DO clones were described previously³. Treatments were as follows: murine TNF (Peprotech, Rock Hill, NJ), 1,000 U ml^-1 (Figs. 2a, b, d and 5b; see also Supplementary Information, Fig. S4b), 10 U ml^-1 (Fig. 5a and see Supplementary Information, Fig. S4a), or 1,000 U ml^-1 plus 0.3 g ml^-1 cycloheximide (CHX; Fig. 5c, d; also see Supplementary Information, Fig. S4c, d); human TNF (Peprotech), 2,000 U ml^-1 (Figs. 3d−f and 4f); TPA plus ionomycin (Sigma, St Louis, MO), 100 ng ml^-1 and 1 M, respectively (Figs 2c, e, 3a; also see Supplementary Information, Fig. S1c). Treatments with H₂O₂ and sorbitol were as described previously³ (see Supplementary Information Fig. S1a, b). In Fig. 5a, b, and in Supplementary Information Fig. S4a, b, pre-treatment with HIV-TAT peptides (5 M) or DMSO was for 30 min and incubation with TNF was for an additional 4 and 3.5 h, respectively. In Fig. 5c, d, and in Supplementary Information Fig. S4c, d, peptides were used at 10 M and incubation with TNF was for 4 h. Apoptosis was measured by using the Cell Death Detection ELISA^PLUS kit (Roche, Basel Switzerland). To assess intracellular incorporation, peptides were labeled with FITC either at the N terminus during synthesis (see Supplementary Information, Fig. S3a, b) or after HPLC purification by using the FluoReporter FITC protein labelling kit (Molecular Probes, Eugene, OR; see Supplementary Information, Fig. S3c, d). Cells were incubated with 5 M FITC-labelled peptides for 20 min, subjected to trypsinization, washed three times with PBS and examined by flow cytometry (FCM) or confocal microscopy.

Generation of gadd45^-/- fibroblasts.

Gadd45 null mice were generated with the help of the Transgenic and Knockout facility at the University of Chicago by using embryonic stem cells and standard homologous recombination-based technology. MEFs were isolated from mouse embryos at day 14 post-coitum. Generation and characterization of gadd45^-/- mice and fibroblast lines will be described elsewhere (F.Z., S.P. & G.F., unpublished observations).

GST precipitations with in-vitro-translated proteins (Figs. 1c, 4b, 4d, and 4e, and Supplementary Information Fig. S2b, d and e) or purified proteins (Fig. 3c), and kinase assays, were performed as described previously^{6, 17, 27, 28, 29, 30}. In Fig. 4e, ATP pre-incubation was for 20 min. His₆/T7−Gadd45, His₆/T7−JIP1, His₆−Gadd45, His₆−EF3 and GST proteins were purified from bacterial lysates as detailed elsewhere^{17, 18, 28}, and dialysed against buffer A¹⁷ (Fig. 3b, c) or 5 mM sodium phosphate buffer (pH 7.6; Figs. 3d, 3e, 3g and 4f). Kinase pre-incubation with recombinant proteins was for 10 min (Figs. 3d, 3e and 4f), and GST−Gadd45 pre-incubation with synthetic peptides or DMSO (-) was for an additional 20 min (Fig. 4f and Supplementary Information Fig. S2e). MKK7 phosphorylation was monitored by immunoprecipitation with anti-P-MKK7 antibodies (developed at Cell Signalling, Beverly, MA) before western blotting with anti-total MKK7 antibodies. For co-immunoprecipitations, extracts were prepared in immunoprecipitation buffer¹².

The anti-MKK7 antibodies were: Fig. 2a and 2d and Supplementary Information Fig. S1a and S1b, kinase assays (goat; Santa Cruz Biotechnology, Santa Cruz, CA); Figs 2a, 2d, and 3a, bottom left, and Supplementary Information Figs. S1a and S1b, western blots, and Fig. 3a, top right, immunoprecipitations (rabbit; Santa Cruz); Fig. 3a, top left, western blot (mouse monoclonal; BD Pharmingen, San Diego, CA). Other antibodies were: anti-Flag from Sigma; anti-P-MKK4, anti-P-MKK3/6, anti-P-MEK1/2, anti-total MKK3 and anti-total MEK1/2 from Cell Signalling; anti-T7 from Novagen; anti-HA, anti-total MKK4, anti-total ASK1 (kinase assays and western blots), and anti-total MEKK1 (kinase assays, western blots, and co-immunoprecipitations) from Santa Cruz. The anti-Gadd45 monoclonal antibody (5D2.2) was described elsewhere¹⁶.

	Top

1. Orlowski, R.Z. & Baldwin, A.S. Jr. NF-B as a therapeutic target in cancer. Trends Mol. Med. 8, 385−389 (2002). | Article | PubMed  | ISI | ChemPort |
2. Karin, M. & Li, A. NF-B at the crossroads of life and death. Nature Immunol. 3, 221−227 (2002). | Article | PubMed  | ISI | ChemPort |
3. De Smaele, E. et al. Induction of gadd45 by NF-B downregulates pro-apoptotic JNK signaling. Nature 414, 308−313 (2001). | Article | PubMed  | ISI | ChemPort |
4. Javelaud, D. & Besancon, F. NF-B activation results in rapid inactivation of JNK in TNF-treated Ewing sarcoma cells: a mechanism for the anti-apoptotic effect of NF-B. Oncogene 20, 4365−4372 (2001). | Article | PubMed  | ISI | ChemPort |
5. Franzoso, G., Zazzeroni, F. & Papa, S. JNK: a killer on a transcriptional leash. Cell Death Diff. 10, 13−15 (2003). | Article | ISI | ChemPort |
6. Sheikh, M.S., Hollander, M.C. & Fornace, A.J. Jr. Role of Gadd45 in apoptosis. Biochem. Pharmacol. 59, 43−45 (2000). | Article | PubMed  | ISI | ChemPort |
7. Tournier, C. et al. MKK7 is an essential component of the JNK signal transduction pathway activated by proinflammatory cytokines. Genes Dev. 15, 1419−1426 (2001). | Article | PubMed  | ISI | ChemPort |
8. Davis, R.J. Signal transduction by the JNK group of MAP kinases. Cell 103, 239−252 (2000). | Article | PubMed  | ISI | ChemPort |
9. Johnson, G.L. & Lapadat, R. Mitogen-activated protein kinase pathways mediated by ERK, JNK and p38 protein kinases. Science 298, 1911−1912 (2002). | Article | PubMed  | ISI | ChemPort |
10. Chang, L. & Karin, M. Mammalian MAP kinase signaling cascades. Nature 410, 37−40 (2001). | Article | PubMed  | ISI | ChemPort |
11. Yeh, W.C. et al. Early lethality, functional NF-B activation, and increased sensitivity to TNF-induced cell death in TRAF2-deficient mice. Immunity 7, 715−725 (1997). | Article | PubMed  | ISI | ChemPort |
12. Takekawa, M. & Saito, H. A family of stress-inducible GADD45-like proteins mediate activation of the stress-responsive MTK1/MEKK4 MAPKKK. Cell 95, 521−530 (1998). | Article | PubMed  | ISI | ChemPort |
13. Takekawa, M., Posas, F. & Saito, H. A human homolog of the yeast Ssk2/Ssk22 MAP kinase kinase kinases, MTK1, mediates stress-induced activation of the p38 and JNK pathways. EMBO J. 16, 4973−4982 (1997). | Article | PubMed  | ISI | ChemPort |
14. Kim, J.W. et al. Glycogen synthase kinase 3 is a natural activator of mitogen-activated protein kinase/extracellular signal-regulated kinase kinase kinase 1 (MEKK1). J. Biol. Chem. 278, 13995−14001 (2003). | Article | PubMed  | ISI | ChemPort |
15. Saitoh, M. et al. Mammalian thioredoxin is a direct inhibitor of apoptosis signal-regulating kinase (ASK) 1. EMBO J. 17, 2596−2606 (1998). | Article | PubMed  | ISI | ChemPort |
16. Zazzeroni, F. et al. Gadd45 mediates the protective effects of CD40 co-stimulation against Fas-induced apoptosis. Blood 102, 3270−3279 (2003). | Article | PubMed  | ISI | ChemPort |
17. Matsuda, S. et al. C-Jun N-terminal kinase (JNK)-interacting protein 1b/Islet-brain-1 scaffolds Alzheimer's amyloid precursor protein with JNK. J. Neurosci. 21, 6597−6607 (2001). | PubMed  | ISI | ChemPort |
18. Drum, C.L. et al. An extended conformation of calmodulin induces interactions between the structural domains of adenylyl cyclase from Bacillus anthracis to promote catalysis. J. Biol. Chem. 275, 36334−36340 (2000). | Article | PubMed  | ISI | ChemPort |
19. Deng, Y., Ren, X., Yang, L., Lin, Y. & Wu, X. A JNK-dependent pathway is required for TNF-induced apoptosis. Cell 115, 61−70 (2003). | Article | PubMed  | ISI | ChemPort |
20. Moriguchi, T. et al. A novel SAPK/JNK kinase, MKK7, stimulated by TNF and cellular stresses. EMBO J. 16, 7045−7053 (1997). | Article | PubMed  | ISI | ChemPort |
21. Russo, A.A. et al. Crystal structure of the p27Kip1 cyclin-dependent-kinase inhibitor bound to the cyclin A−Cdk2 complex. Nature 382, 325−331 (1996). | Article | PubMed  | ISI | ChemPort |
22. Vives, E., Brodin, P. & Lebleu, B. A truncated HIV-1 Tat protein basic domain rapidly translocates through the plasma membrane and accumulates in the cell nucleus. J. Biol. Chem. 272, 16010−16017 (1997). | Article | PubMed  | ISI | ChemPort |
23. Amanullah, A. et al. Cell survival and Gadd45-factor deficiency. Nature 424, 741−742 (2003). | Article | PubMed  | ISI | ChemPort |
24. Tak, P.P. & Firestein, G.S. NF-B: a key role in inflammatory diseases. J. Clin. Invest. 107, 7−11 (2001). | PubMed  | ISI | ChemPort |
25. Romas, E., Gillespie, M.T. & Martin, T.J. Involvement of receptor activator of NF-B ligand and tumor necrosis factor- in bone destruction in rheumatoid arthritis. Bone 30, 340−346 (2002). | Article | PubMed  | ISI | ChemPort |
26. Yamamoto, Y. & Gaynor, R.B. Therapeutic potential of inhibition of the NF-B pathway in the treatment of inflammation and cancer. J. Clin. Invest. 107, 135−142 (2001). | PubMed  | ISI | ChemPort |
27. Lu, X., Nemoto, S. & Lin, A. Identification of c-Jun NH₂-terminal protein kinase (JNK)-activating kinase 2 as an activator of JNK but not p38. J. Biol. Chem. 272, 24751−24754 (1997). | Article | PubMed  | ISI | ChemPort |
28. Franzoso, G. et al. Activation of the serum response factor by p65/NF-B. EMBO J. 15, 3403−3412 (1996). | PubMed  | ISI | ChemPort |
29. Dorion, S., Lambert, H. & Landry, J. Activation of the p38 signaling pathway by heat shock involves the dissociation of glutathione S-transferase Mu from ASK1. J. Biol. Chem. 277, 30792−30797 (2002). | Article | PubMed  | ISI | ChemPort |
30. Chadee, D.N., Yuasa, T. & Kyriakis, J.M. Direct activation of mitogen-activated protein kinase kinase kinase MEKK1 by the Ste20p homologue GCK and the adapter protein TRAF2. Mol. Cell. Biol. 22, 737−749 (2002). | PubMed  | ISI | ChemPort |

	Top

We thank M. Peter, H. Singh, C.-R. Wang, and P. Ashton-Rickardt for critical comments on the manuscript. We are grateful to H. Ichijo, J. Landry, A. Leonardi, P. Vito, H. Gram, R. Vaillancourt, T.H. Wang, and J. Wimalasena for plasmids, and G. Taroli for help with manuscript preparation. We also thank L. Degelstein of the Transgenic and Knockout facility and C. McShan of the Monoclonal Antibody facility, both at the University of Chicago, for help with the generation of Gadd45 knockout mice and the anti-Gadd45 antibody, 5D2.2, respectively. This work was supported in part by the Damon Runyon Scholar Award of the Cancer Research Fund and National Institutes of Health grants R01-CA84040 and R01-CA098583.

Competing interests statement:  The authors declare that they have no competing financial interests.