- Email: [email protected]
The catalytic subunits of IκB kinase, IKK-1 and IKK-2, contain non-equivalent active sites when expressed as homodimers
BBRC Biochemical and Biophysical Research Communications 293 (2002) 1508–1513 www.academicpress.com
The catalytic subunits of IjB kinase, IKK-1 and IKK-2, contain non-equivalent active sites when expressed as homodimersq James R. Burke* and Joann Strnad Drug Discovery and Exploratory Development, Bristol-Myers Squibb Pharmaceutical Research Institute, P.O. Box 4000, Princeton, NJ 08543, USA Received 23 April 2002
Abstract The signal-inducible phosphorylation of serines 32 and 36 of IjBa is the key step in regulating the subsequent ubiquitination and proteolysis of IjBa which then releases NF-jB to promote gene transcription. The multisubunit IjB kinase responsible for this phosphorylation contains two catalytic subunits, termed IKK-1 and IKK-2. It has been shown that both subunits catalyze the phosphorylation of IjBa as well as an autophosphorylation at a C-terminal cluster of serines. With baculovirus/insect cell-expressed homodimeric IKK-1 or IKK-2, inhibitors such as ADP or a peptide inhibitor (corresponding to amino acid residues 26–42 of IjBa with Ser-32 and Ser-36 changed to aspartates) inhibited autophosphorylation and IjBa phosphorylation reactions with different potencies. ADP was more potent against IjBa phosphorylation as compared to autophosphorylation, while the peptide inhibitor showed the opposite effect. Pseudo-Dixon plots of the inhibition with ADP were linear while non-linear plots were obtained with the peptide inhibitor, suggesting a cooperative effect in the case of the latter. Using different concentrations of IKK-1, autophosphorylation was shown to be intramolecular. These results indicated that there were non-equivalent active sites present within the preparations of recombinant homodimers of IKK-1 and IKK-2. The peptide inhibitor showed equivalent inhibitory effects with wild-type IKK-1 and the S176E/S180E mutant. In contrast, ADP showed equipotent inhibition against the S176E/S180E mutantcatalyzed autophosphorylation and IjBa phosphorylation reactions. A model is proposed in which the phosphorylation state of the activation loop of IKK-1 or IKK-2 affects the active site conformation of the enzyme such that the two forms catalyze the autophosphorylation and IjBa phosphorylation reactions with different affinities. In addition, the two active sites within the dimer appear to act in a cooperative fashion so that binding of peptide inhibitor at one active site affects the conformation of the other active site. Ó 2002 Elsevier Science (USA). All rights reserved.
The transcriptional activator NF-jB normally resides in the cytoplasm in unstimulated cells as an inactive complex with a member of the IjB inhibitory protein family. This class of protein includes IjB-a, IjB-b, and IjB-e, all of which contain ankyrin repeats necessary for complexation with NF-jB (for a review, see [1]). In the case of IjB-a, the most carefully studied member of the class, stimulation of cells with agents such as TNFa or LPS results in the phosphorylation of IjB-a at Ser-32 and Ser-36 [2]. This phosphorylation serves as a critical signal for subsequent ubiquitination and proteolysis of q Abbreviations: ADP, adenosine 50 -diphosphate; ATP, adenosine 50 -triphosphate; GST-IjB-a, IjB-a fusion protein with glutathione S-transferase tag; IKK, IjB kinase; msIKK, multisubunit IjB kinase; [33P]ATP, adenosine 50 –½c–33 P]triphosphate. * Corresponding author. Fax: +1-609-252-6058. E-mail address: [email protected] (J.R. Burke).
IjB-a leaving NF-jB free to translocate to the nucleus and promote gene transcription [3–5]. A mutant in which both Ser-32 and Ser-36 are changed to alanine prevents signal-induced activation of NF-j B and results in an IjB-a which is neither phosphorylated, ubiquitinated nor proteolytically digested [5]. Analogous serines have been identified in both IjB-b and IjB-e and phosphorylation at these residues appears to regulate the proteolytic degradation of these proteins by a similar mechanism [6,7]. Because the expression of many pro-inflammatory genes such as cytokines and adhesion molecules is regulated by NF-jB (for reviews, see [8,9]), the inhibition of this signal-inducible phosphorylation of IjB would be an important target for novel antiinflammatory agents. The IjB kinase (IKK) which catalyzes this phosphorylation is a high molecular weight (500–900 kDa) multisubunit complex comprised of two catalytic
0006-291X/02/$ - see front matter Ó 2002 Elsevier Science (USA). All rights reserved. PII: S 0 0 0 6 - 2 9 1 X ( 0 2 ) 0 0 4 1 7 - 5
J.R. Burke, J. Strnad / Biochemical and Biophysical Research Communications 293 (2002) 1508–1513
subunits (termed IKK-1 and IKK-2) and a regulatory subunit termed IKK-c/NEMO (for a review, see [10]). The IKK-1 and IKK-2 subunits are highly homologous proteins (52% identity, >70% homology) which have N-terminal kinase domains followed by leucine zipper and helix–loop–helix domains [10]. The catalytic subunits appear to dimerize, either as homo- or heterodimers, through interactions between leucine zipper domains. The helix–loop–helix domain appears to play a role in mediating the activation of the kinase domain through signals mediated through the IKK-c/NEMO regulatory subunit [11]. Similar to the regulation in a wide range of other kinases, both IKK-1 and IKK-2 contain an activation loop within their kinase domain which can be phosphorylated by upstream serine kinases, leading to greatly enhanced enzymatic activity [12]. Phosphorylation at the two serines within this activation loop (S176 and S180 in IKK-1 and S177 and S181 in IKK-2) can be mimicked by mutating these residues to glutamates [13]. In addition, both IKK-1 and IKK-2 catalyze an autophosphorylation which results not in phosphorylation at the activation loop, but at a cluster of serines in the C-terminal helix–loop–helix domain [14]. This autophosphorylation appeared to diminish enzymatic activity and has been suggested to prevent prolonged activation of IKK. In this paper, we show that homodimers of recombinantly expressed IKK-1 or IKK-2 have non-equivalent binding sites as evidenced by the kinetic characteristics of inhibitors, both those that bind to the ATP binding site as well as an inhibitor which binds at the IjB-a binding site. The results demonstrate that these non-equivalent active sites differ in their affinities toward autophosphorylation versus IjB-a phosphorylation. While the phosphorylation state of the activation loop of the enzyme appears to play a role in the non-equivalent nature of the active sites, other mechanisms are involved as well. Experimental procedures Materials. GST-IjBa was purchased from Santa Cruz Biotechnology; [33 P]ATP (1000 Ci/mmol) was purchased from Amersham and ADP and staurosporine were from Sigma–Aldrich. The peptide inhibitor which corresponds to amino acid residues 26–42 of IjBa with Ser-32 and Ser-36 changed to aspartates (LDDRHDD GLDDMKDEEY, N-terminal acetylated and C-terminal amidated, see [15]) was synthesized by Research Genetics (Huntsville, Alabama). GST-tagged IKK-1 and IKK-2 were expressed in High Five cells and purified as previously reported [15]. A GST-tagged IKK-1 mutant in which S176 and S180 were mutated to glutamates (see [13]) was also expressed and purified in this manner. Enzyme assays. Assays were performed by adding enzyme (IKK2, IKK-1, or mutated IKK-1, typically to a final concentration of 0:5 lg/mL unless noted) at 30 °C to solutions of 100 lg/mL GST-IjBa and 5 lM [33 P]ATP in 40 mM Tris–HCl, pH 7.5, containing 4 mM MgCl2 , 34 mM sodium phosphate, 3 mM NaCl, 0.6 mM potassium phosphate, 1 mM KCl, 1 mM dithiothreitol, 3% (w/v) glycerol, and
1509
250 lg/mL bovine serum albumin. The specific activity of [33 P]ATP used in the assay ranged from 15 to 150 Ci/mmol. After 5 min, the kinase reactions were stopped by the addition of 2 Laemmli’s sample buffer and heat treated at 90 °C for 1 min. The samples were then loaded onto NuPAGE 10% Bis–Tris gels (Novex, San Diego, CA). After completion of SDS–PAGE, gels were dried on a slab gel dryer. The bands were then detected using a 445Si PhosphorImager (Molecular Dynamics) and the radioactivity was quantified using the ImageQuant software while employing a mean background correction factor for each 33 P-IjB-a band. It should be noted that the amount of radioactivity measured in this way is in arbitrary units and absolute values from one experiment cannot necessarily be compared to those from another experiment. Under these conditions, the degree of phosphorylation of GST-IjBa was linear with time and concentration of enzyme.
Results Using [33 P]ATP as the phosphate source in an IKK assay, it is possible to measure the amount of phosphorylated IjBa and autophosphorylation products simultaneously by employing SDS–PAGE and phosphorimager analysis of the reaction mixture, since the molecular weights of IjBa and IKK proteins differ. Under conditions of the assays used in the present study, roughly equivalent amounts of radioactivity were associated with autophosphorylation and phosphorylated IjBa bands in the absence of inhibitors. The amount of radioactivity associated with each band was linear with respect to time for at least 5 min (results not shown). As might be expected, given that there is only one active site within each molecule of enzyme, the broad-spectrum kinase inhibitor staurosporine inhibited IKK-1-catalyzed autophosphorylation and IjBa phosphorylation to the same degree as shown in Fig. 1A. In contrast to this result, the analogous experiment with IKK-2 showed that staurosporine inhibited the catalyzed phosphorylation of IjBa somewhat more potently than autophosphorylation (see Fig. 1B). The results are represented as pseudo-Dixon plots (i.e., ratio of the rate of product formed in the absence to that in the presence of inhibitor versus concentration of inhibitor) to more easily gauge the relative inhibition of each phosphorylation product. The use of ADP, which has been shown to inhibit IKK-1 and IKK-2 competitively with respect to ATP [16], showed an even more pronounced effect, with the enzyme-catalyzed IjBa phosphorylation inhibited considerably more potently than autophosphorylation with both IKK-1 and IKK-2 (see Fig. 2). That these pseudoDixon plots are linear indicates that Michaelis–Menten kinetics are still in effect (competitive, non-competitive, and uncompetitive inhibitors would all be expected to give linear Dixon plots, see [17]). Quite different results were obtained when a peptide corresponding to amino acid residues 26–42 of IjBa (with Ser-32 and Ser-36 mutated to aspartates) was used as an inhibitor. Since the serines have been removed, this
1510
J.R. Burke, J. Strnad / Biochemical and Biophysical Research Communications 293 (2002) 1508–1513
Fig. 1. Pseudo-Dixon plot of the inhibition by staurosporine of autophosphorylation and phosphorylation of IjBa catalyzed by (A) IKK-1 and (B) IKK-2. The amount of phosphorylated products was analyzed by SDS–PAGE and phosphorimager quantitation of the respective bands of 5-min assays. The ratio of the amount of product formed in the absence to that formed in the presence of inhibitor is represented as ðv0 ÞO =ðv0 ÞI . Closed circles, phosphorylation of IjBa; open circles, autophosphorylation. Plots are representative examples of multiple experiments.
peptide cannot act as a substrate and, instead, is an inhibitor [15]. As shown in Fig. 3, this inhibitor was more potent against the IKK-1 and IKK-2-catalyzed autophosphorylation than against the phosphorylation of fulllength IjBa. In addition, these pseudo-Dixon plots were curved rather than linear, with complete inhibition observed at concentrations of peptide >3 mg/mL. There are three possible mechanisms by which ADP and the peptide would inhibit the enzyme-catalyzed autophosphorylation and phosphorylation of IjBa with different potencies. The first possibility is that autophosphorylation and IjBa phosphorylation are catalyzed by the same active site, but autophosphorylation diminishes the catalytic activity of the enzyme toward IjBa phosphorylation. Inhibition of autophosphoryla-
Fig. 2. Pseudo-Dixon plot of the inhibition by ADP of autophosphorylation and phosphorylation of IjBa catalyzed by (A) IKK-1 and (B) IKK-2. The amount of phosphorylated products was analyzed by SDS–PAGE and phosphorimager quantitation of the respective bands of 5-min assays. The ratio of the amount of product formed in the absence to that formed in the presence of inhibitor is represented as ðv0 ÞO =ðv0 ÞI . Closed circles, phosphorylation of IjBa; open circles, autophosphorylation. Plots are representative examples of multiple experiments.
tion by the peptide inhibitor, for example, would appear to be more potent than inhibition of IjBa phosphorylation since the enzyme is in a more active nonphosphorylated state for longer period of time. This situation would result in IjBa phosphorylation apparently enhanced from that expected based on the observed autophosphorylation inhibition. There is evidence in the literature that autophosphorylation diminishes enzymatic activity [14]. Evidence against this possibility comes from the observation that under conditions of the assays used in the present work, the rates of product formation (both autophosphorylation and IjBa phosphorylation) were linear over the 5-min assay time in both the absence and presence of inhibitors (results not shown). Moreover, the more potent inhibi-
J.R. Burke, J. Strnad / Biochemical and Biophysical Research Communications 293 (2002) 1508–1513
Fig. 3. Pseudo-Dixon plot of the inhibition by peptide inhibitor of autophosphorylation and phosphorylation of IjBa catalyzed by (A) IKK-1 and (B) IKK-2. The amount of phosphorylated products was analyzed by SDS–PAGE and phosphorimager quantitation of the respective bands of 5 min assays. The ratio of the amount of product formed in the absence to that formed in the presence of inhibitor is represented as ðv0 ÞO=ðv0 ÞI . Closed circles, phosphorylation of IjBa; open circles, autophosphorylation. Plots are representative examples of multiple experiments.
tion against IjBa phosphorylation as compared to autophosphorylation observed with ADP cannot be explained in this manner. A second possibility is that the autophosphorylation is intermolecular and having a bound inhibitor changes the conformation of IKK such that the protein binds with altered affinity to the active site of a second enzyme molecule. For example, IKK with peptide inhibitor bound may have a larger dissociation constant from another IKK molecule catalyzing autophosphorylation than the dissociation constant of unbound IKK from another IKK molecule. This would result in more potent inhibition against autophosphorylation than against phosphorylation of IjBa in the assay: That is, inhibition of autophosphorylation would reflect occupation of active sites with inhibitor as well as an increase in the dissociation constant of inhibitor-bound autophosph-
1511
orylation ‘‘substrate.’’ To explain the effect observed with ADP, the conformation of ADP-bound protein would have to be recognized more effectively (i.e., smaller dissociation constant of ADP-bound IKK from the active site of another IKK enzyme). Therefore the inhibition of autophosphorylation due to occupying active sites with ADP would have to be somewhat offset by an increased affinity of ADP-bound IKK in an autophosphorylation reaction. To test whether the autophosphorylation is intermolecular, the inhibition of IKK-1 by ADP at two enzyme concentrations was measured. As shown in Fig. 4, the inhibition of autophosphorylation was independent of enzyme concentration. To reconcile this observation with an intermolecular autophosphorylation requires that the concentration of substrate (enzyme in the case of autophosphorylation) be much lower than the KM for autophosphorylation. If the concentration of substrate was near or greater than the KM , the potency of inhibition would have been much more sensitive to changes in substrate concentration. It would also be expected that if the autophosphorylation was intermolecular, 10-times lower enzyme concentration would have resulted in 100-times lower autophosphorylation product (10-times less catalyst multiplied by 10-times lower substrate concentration operating below the KM ). However, the amount of autophosphorylation was found to be only 9.9-times less at 0:14 lg=mL enzyme as compared to 1:4 lg/mL. These results therefore are consistent only with intramolecular autophosphorylation. Since IKK-1 and IKK-2 are isolated as dimers (in this case homodimers), it is important to note that intramolecular autophosphorylation does not discriminate between true intramolecular autophosphorylation (catalyzing phosphorylation on the same polypeptide
Fig. 4. Dependence of IKK-1 concentration on the inhibition of autophosphorylation by ADP. The amount of autophosphorylation product was analyzed by SDS–PAGE and phosphorimager quantitation of 5-min assays using IKK-1 at 1:4 lg/mL (closed circles) or 0:14 lg/mL (open circles). The ratio of the amount of product formed in the absence to that formed in the presence of inhibitor is represented as ðv0 ÞO=ðv0 ÞI . Plot is a representative example of multiple experiments.
1512
J.R. Burke, J. Strnad / Biochemical and Biophysical Research Communications 293 (2002) 1508–1513
chain as the active site) or phosphorylation of one subunit by the other subunit within a dimer. The third possible explanation for the differential potency of inhibitors against autophosphorylation as compared to phosphorylation of IjBa is that there are two non-equivalent forms of the enzyme present in the preparations. In this case, one form would contribute more to the autophosphorylation reaction while the other form of the enzyme plays a larger role in catalyzing IjBa phosphorylation. With ADP being more potent against IjBa phosphorylation, ADP would have to bind with greater affinity to the form of the enzyme contributing more to this reaction as compared to the form contributing more to autophosphorylation. The opposite would be true for the peptide inhibitor. This raises the interesting possibility that the two different forms of enzyme within the preparations simply represent two populations of enzyme which differ by the phosphorylation state at the MAP kinase consensus site within the activation loop of the N-terminal catalytic domain (autophosphorylation occurs in the C-terminal helix–loop–helix domain). Phosphorylation of the activation loop greatly affects enzymatic activity. While it has not been determined with IKK-1 or IKK-2, other proteins expressed from the baculovirus/insect cell expression system have been shown to be phosphorylated, sometimes to a high degree, presumably by endogeneous insect MAP kinases during expression [18]. To test this, the S176E/S180E mutant of IKK-1 was expressed and purified. It has been shown that this mutant mimics phosphorylation at these sites and gives constitutively activated enzyme [13]. As compared to the results with the wild-type IKK-1 shown in Fig 2A, ADP was equipotent against both autophosphorylation and IjBa phosphorylation when the mutant enzyme was used (see Fig. 5). This provides evidence, there are two forms of the wild-type enzyme (differentiated by the phosphorylation state of the enzyme) which bind inhibitors such as ADP with different affinities and have non-equivalent activities against autophosphorylation and IjBa phosphorylation. Interestingly, the inhibitor was more potent against both mutant IKK-1-catalyzed autophosphorylation and IjBa phosphorylation than was observed against either activity in wild-type enzyme. These results are in contrast to those observed with the peptide inhibitor which behaved in approximately the same manner toward mutant enzyme to that observed with the wild-type enzyme, with non-linear pseudo-Dixon plots and more potent inhibition against autophosphorylation than against phosphorylation of IjBa (compare Fig. 6 to Fig. 3A). This indicates that the effect of peptide is not dependent on two forms of the enzyme, differentiated by the phosphorylation states, contributing differentially to autophosporylation versus phosphorylation of IjBa.
Fig. 5. Pseudo-Dixon plot of the inhibition by ADP of autophosphorylation and phosphorylation of IjBa catalyzed by S176E/S180E mutant IKK-1. The amount of phosphorylated products was analyzed by SDS–PAGE and phosphorimager quantitation of the respective bands of 5-min assays. The ratio of the amount of product formed in the absence to that formed in the presence of inhibitor is represented as ðv0 ÞO =ðv0 ÞI . Closed circles, phosphorylation of IjBa; open circles, autophosphorylation. Plot is a representative example of multiple experiments.
Fig. 6. Pseudo-Dixon plot of the inhibition by peptide inhibitor of autophosphorylation and phosphorylation of IjBa catalyzed by S176E/S180E mutant IKK-1. The amount of phosphorylated products was analyzed by SDS–PAGE and phosphorimager quantitation of the respective bands of 5-min assays. The ratio of the amount of product formed in the absence to that formed in the presence of inhibitor is represented as ðv0 ÞO=ðv0 ÞI . Closed circles, phosphorylation of IjBa; open circles, autophosphorylation. Plot is a representative example of multiple experiments.
Discussion The present work demonstrates that recombinantly expressed homodimers of IKK-1 or IKK-2 contain two (or more) non-equivalent forms of the enzyme which catalyze the autophosphorylation and IkBa phosphorylation reactions with different affinities. Inhibitors such as ADP and an IjBa peptide inhibitor bind with nonequivalent affinities to these two forms of the enzyme such that more potent inhibition was observed against one of
J.R. Burke, J. Strnad / Biochemical and Biophysical Research Communications 293 (2002) 1508–1513
the phosphorylated products than against the other. That there are non-equivalent binding sites may be explained by a mixture of forms isolated from the baculovirus/insect cell expression system which differ in the degree of phosphorylation at the activation loop of the enzyme. Phosphorylation has been shown to greatly affect the activity of the IKKs [13,19]. Mutation of serines 176 and 180 of IKK-1 to phosphomimicking glutamates yields constitutively active (and homogeneous) enzyme which obliterated the differences in potencies ADP showed against autophosphorylation and IjBa phosphorylation. Interestingly, the differential effect of the peptide inhibitor on autophosphorylation and IjBa phosphorylation was insensitive to these mutations, suggesting that the active sites within preparations of this mutated enzyme remain non-equivalent. A possible explanation, especially relevant since both IKK-1 and IKK-2 are isolated as homodimers, is that the non-equivalent active sites are within the same dimer such that binding of a peptide substrate to one active site affects the second active site within the dimer (i.e., cooperativity or half-of-site reactivity, [20]). Unlike staurosporine or ADP which binds at the ATP binding site of the enzyme, the peptide inhibitor shows non-linear pseudo-Dixon plots. This is consistent with a cooperative effect between the two active sites since simple Michaelis–Menten kinetics are no longer in effect. Instead, the binding of peptide inhibitor at one site may enhance the binding affinity for the peptide at the second active site yielding the cooperative effect seen in Fig. 3. A cooperative effect between active sites within a dimer is not completely surprising, since it has been shown that the active site(s) of the multisubunit complex (containing heterodimeric IKK1 and IKK-2 along with the regulatory subunit IKK-c) binds IjBa substrate with considerably higher affinity than homodimers lacking IKK-c [15]. This indicates that the presence of other subunits affects the conformation of the active site(s). The model proposed here therefore is one in which phosphorylation state of the activation loop of IKK affects the active site conformation of the enzyme such that the two forms catalyze autophosphorylation and IjBa phosphorylation reactions with different affinities. This change in active site conformation would reflect changes both at the peptide binding site (to explain different efficiencies toward autophosphorylation and IjBa phosphorylation) and at the ATP binding site (reflecting the change in affinity of ADP and, in the case of IKK-2, staurosporine). In addition, the two active sites within the dimer may show cooperative interactions so that binding of peptide inhibitor at one active site affects the conformation of the other active site. While the present work used only homodimers of IKK-1 or IKK-2, it is uncertain how this model relates to heterodimers of IKK-1/IKK-2. It is interesting to speculate that this model may explain some aspects concerning the receptor-mediated activation of IKK in cells. For
1513
instance, it may be that receptor-mediated activation of kinase signal transduction cascades results in phosphorylation of the activation loop of one of the catalytic subunits within the complex (IKK-1, for instance). This phosphorylation then alters the protein conformation so that the catalytic efficiency is optimized toward phosphorylation of the activation loop of its binding partner in the complex (IKK-2?). We are continuing to investigate these questions to ascertain the role of phosphorylation as well as cooperativity between active sites in the regulation of this highly complex and poorly understood enzyme. Acknowledgments The authors thank Dr. Rolf-Peter Ryseck for providing the clones for IKK-1, IKK-2, and mutated IKK-1 used in the expression of these proteins.
References [1] S.T. Whiteside, A. Israel, Semin. Cancer Biol. 8 (1997) 75–82. [2] K. Brown, S. Gertsberger, L. Carlson, G. Franzoso, U. Siebenlist, Science 267 (1995) 1485–1488. [3] T.S. Finco, A.A. Beg, A.S. Baldwin Jr., Proc. Natl. Acad. Sci. USA 91 (1994) 11884–11888. [4] L. Baldi, K. Brown, G. Franzoso, U. Siebenlist, J. Biol. Chem. 271 (1996) 376–379. [5] M. Roff, J. Thompson, M.S. Rodriquez, J.-M. Jacque, F. Baleux, F. Arenzana-Seisdedos, R.T. Hay, J. Biol. Chem. 271 (1996) 7844–7850. [6] R. Weil, C. Laurent-Winter, A. Israel, J. Biol. Chem. 272 (1997) 9942–9949. [7] S.T. Whiteside, J.-C. Epinat, N.R. Rice, A. Israel, EMBO J. 16 (1997) 1413–1426. [8] U. Siebenlist, G. Franzoso, K. Brown, Annu. Rev. Cell Biol. 10 (1994) 405–455. [9] P.A. Bauerle, D. Baltimore, Cell 87 (1997) 13–20. [10] M. Karin, Y. Ben-Neriah, Ann. Rev. Immunol. 18 (2000) 621–663. [11] M.J. May, F. D’Acquisto, L.A. Madge, J. Glockner, J.S. Pober, S. Ghosh, Science 289 (2000) 1550–1554. [12] H. Nakano, M. Shindo, S. Sakon, S. Hishinaka, M. Mihara, H. Yagita, K. Okumura, Proc. Natl. Acad. Sci. USA 95 (1998) 3537– 3542. [13] F. Mercurio, H. Zhu, B.W. Murray, A. Shevchenko, B.L. Bennett, J.W. Li, D.B. Young, M. Barbosa, M. Mann, A. Manning, A. Rao, Science 278 (1997) 860–866. [14] M. Delhase, M. Hayakawa, Y. Chen, M. Karin, Science 284 (1999) 309–313. [15] J.R. Burke, M.K. Wood, R.-P. Ryseck, S. Walther, C.A. Meyers, J. Biol. Chem. 274 (1999) 36146–36152. [16] Q.K. Huynh, H. Boddupalli, S.A. Rouw, C.M. Koboldt, T. Hall, C. Sommers, S.D. Hauser, J.L. Pierce, R.G. Combs, B.A. Reitz, J.A. Diaz-Collier, R.A. Weinberg, B.L. Hood, B.F. Kilpatrick, C.S. Tripp, J. Biol. Chem. 275 (2000) 25883–25891. [17] I.H. Segel, in: Enzyme Kinetics, Wiley, New York, 1975, pp. 110– 144. [18] M.G. de Carvalho, A.L. McCormack, E. Olson, F. Ghomashchi, M.H. Gelb, J.R. Yates, C.C. Leslie, J. Biol. Chem. 271 (1996) 6987–6997. [19] J.A. DiDonato, M. Hayakawa, D.M. Rothwarf, E. Zandi, M. Karin, Nature 388 (1997) 548–554. [20] A. Fersht, in: Enzyme Structure and Mechanism, W.H. Freeman, New York, 1985, pp. 263–272.
The catalytic subunits of IjB kinase, IKK-1 and IKK-2, contain non-equivalent active sites when expressed as homodimersq James R. Burke* and Joann Strnad Drug Discovery and Exploratory Development, Bristol-Myers Squibb Pharmaceutical Research Institute, P.O. Box 4000, Princeton, NJ 08543, USA Received 23 April 2002
Abstract The signal-inducible phosphorylation of serines 32 and 36 of IjBa is the key step in regulating the subsequent ubiquitination and proteolysis of IjBa which then releases NF-jB to promote gene transcription. The multisubunit IjB kinase responsible for this phosphorylation contains two catalytic subunits, termed IKK-1 and IKK-2. It has been shown that both subunits catalyze the phosphorylation of IjBa as well as an autophosphorylation at a C-terminal cluster of serines. With baculovirus/insect cell-expressed homodimeric IKK-1 or IKK-2, inhibitors such as ADP or a peptide inhibitor (corresponding to amino acid residues 26–42 of IjBa with Ser-32 and Ser-36 changed to aspartates) inhibited autophosphorylation and IjBa phosphorylation reactions with different potencies. ADP was more potent against IjBa phosphorylation as compared to autophosphorylation, while the peptide inhibitor showed the opposite effect. Pseudo-Dixon plots of the inhibition with ADP were linear while non-linear plots were obtained with the peptide inhibitor, suggesting a cooperative effect in the case of the latter. Using different concentrations of IKK-1, autophosphorylation was shown to be intramolecular. These results indicated that there were non-equivalent active sites present within the preparations of recombinant homodimers of IKK-1 and IKK-2. The peptide inhibitor showed equivalent inhibitory effects with wild-type IKK-1 and the S176E/S180E mutant. In contrast, ADP showed equipotent inhibition against the S176E/S180E mutantcatalyzed autophosphorylation and IjBa phosphorylation reactions. A model is proposed in which the phosphorylation state of the activation loop of IKK-1 or IKK-2 affects the active site conformation of the enzyme such that the two forms catalyze the autophosphorylation and IjBa phosphorylation reactions with different affinities. In addition, the two active sites within the dimer appear to act in a cooperative fashion so that binding of peptide inhibitor at one active site affects the conformation of the other active site. Ó 2002 Elsevier Science (USA). All rights reserved.
The transcriptional activator NF-jB normally resides in the cytoplasm in unstimulated cells as an inactive complex with a member of the IjB inhibitory protein family. This class of protein includes IjB-a, IjB-b, and IjB-e, all of which contain ankyrin repeats necessary for complexation with NF-jB (for a review, see [1]). In the case of IjB-a, the most carefully studied member of the class, stimulation of cells with agents such as TNFa or LPS results in the phosphorylation of IjB-a at Ser-32 and Ser-36 [2]. This phosphorylation serves as a critical signal for subsequent ubiquitination and proteolysis of q Abbreviations: ADP, adenosine 50 -diphosphate; ATP, adenosine 50 -triphosphate; GST-IjB-a, IjB-a fusion protein with glutathione S-transferase tag; IKK, IjB kinase; msIKK, multisubunit IjB kinase; [33P]ATP, adenosine 50 –½c–33 P]triphosphate. * Corresponding author. Fax: +1-609-252-6058. E-mail address: [email protected] (J.R. Burke).
IjB-a leaving NF-jB free to translocate to the nucleus and promote gene transcription [3–5]. A mutant in which both Ser-32 and Ser-36 are changed to alanine prevents signal-induced activation of NF-j B and results in an IjB-a which is neither phosphorylated, ubiquitinated nor proteolytically digested [5]. Analogous serines have been identified in both IjB-b and IjB-e and phosphorylation at these residues appears to regulate the proteolytic degradation of these proteins by a similar mechanism [6,7]. Because the expression of many pro-inflammatory genes such as cytokines and adhesion molecules is regulated by NF-jB (for reviews, see [8,9]), the inhibition of this signal-inducible phosphorylation of IjB would be an important target for novel antiinflammatory agents. The IjB kinase (IKK) which catalyzes this phosphorylation is a high molecular weight (500–900 kDa) multisubunit complex comprised of two catalytic
0006-291X/02/$ - see front matter Ó 2002 Elsevier Science (USA). All rights reserved. PII: S 0 0 0 6 - 2 9 1 X ( 0 2 ) 0 0 4 1 7 - 5
J.R. Burke, J. Strnad / Biochemical and Biophysical Research Communications 293 (2002) 1508–1513
subunits (termed IKK-1 and IKK-2) and a regulatory subunit termed IKK-c/NEMO (for a review, see [10]). The IKK-1 and IKK-2 subunits are highly homologous proteins (52% identity, >70% homology) which have N-terminal kinase domains followed by leucine zipper and helix–loop–helix domains [10]. The catalytic subunits appear to dimerize, either as homo- or heterodimers, through interactions between leucine zipper domains. The helix–loop–helix domain appears to play a role in mediating the activation of the kinase domain through signals mediated through the IKK-c/NEMO regulatory subunit [11]. Similar to the regulation in a wide range of other kinases, both IKK-1 and IKK-2 contain an activation loop within their kinase domain which can be phosphorylated by upstream serine kinases, leading to greatly enhanced enzymatic activity [12]. Phosphorylation at the two serines within this activation loop (S176 and S180 in IKK-1 and S177 and S181 in IKK-2) can be mimicked by mutating these residues to glutamates [13]. In addition, both IKK-1 and IKK-2 catalyze an autophosphorylation which results not in phosphorylation at the activation loop, but at a cluster of serines in the C-terminal helix–loop–helix domain [14]. This autophosphorylation appeared to diminish enzymatic activity and has been suggested to prevent prolonged activation of IKK. In this paper, we show that homodimers of recombinantly expressed IKK-1 or IKK-2 have non-equivalent binding sites as evidenced by the kinetic characteristics of inhibitors, both those that bind to the ATP binding site as well as an inhibitor which binds at the IjB-a binding site. The results demonstrate that these non-equivalent active sites differ in their affinities toward autophosphorylation versus IjB-a phosphorylation. While the phosphorylation state of the activation loop of the enzyme appears to play a role in the non-equivalent nature of the active sites, other mechanisms are involved as well. Experimental procedures Materials. GST-IjBa was purchased from Santa Cruz Biotechnology; [33 P]ATP (1000 Ci/mmol) was purchased from Amersham and ADP and staurosporine were from Sigma–Aldrich. The peptide inhibitor which corresponds to amino acid residues 26–42 of IjBa with Ser-32 and Ser-36 changed to aspartates (LDDRHDD GLDDMKDEEY, N-terminal acetylated and C-terminal amidated, see [15]) was synthesized by Research Genetics (Huntsville, Alabama). GST-tagged IKK-1 and IKK-2 were expressed in High Five cells and purified as previously reported [15]. A GST-tagged IKK-1 mutant in which S176 and S180 were mutated to glutamates (see [13]) was also expressed and purified in this manner. Enzyme assays. Assays were performed by adding enzyme (IKK2, IKK-1, or mutated IKK-1, typically to a final concentration of 0:5 lg/mL unless noted) at 30 °C to solutions of 100 lg/mL GST-IjBa and 5 lM [33 P]ATP in 40 mM Tris–HCl, pH 7.5, containing 4 mM MgCl2 , 34 mM sodium phosphate, 3 mM NaCl, 0.6 mM potassium phosphate, 1 mM KCl, 1 mM dithiothreitol, 3% (w/v) glycerol, and
1509
250 lg/mL bovine serum albumin. The specific activity of [33 P]ATP used in the assay ranged from 15 to 150 Ci/mmol. After 5 min, the kinase reactions were stopped by the addition of 2 Laemmli’s sample buffer and heat treated at 90 °C for 1 min. The samples were then loaded onto NuPAGE 10% Bis–Tris gels (Novex, San Diego, CA). After completion of SDS–PAGE, gels were dried on a slab gel dryer. The bands were then detected using a 445Si PhosphorImager (Molecular Dynamics) and the radioactivity was quantified using the ImageQuant software while employing a mean background correction factor for each 33 P-IjB-a band. It should be noted that the amount of radioactivity measured in this way is in arbitrary units and absolute values from one experiment cannot necessarily be compared to those from another experiment. Under these conditions, the degree of phosphorylation of GST-IjBa was linear with time and concentration of enzyme.
Results Using [33 P]ATP as the phosphate source in an IKK assay, it is possible to measure the amount of phosphorylated IjBa and autophosphorylation products simultaneously by employing SDS–PAGE and phosphorimager analysis of the reaction mixture, since the molecular weights of IjBa and IKK proteins differ. Under conditions of the assays used in the present study, roughly equivalent amounts of radioactivity were associated with autophosphorylation and phosphorylated IjBa bands in the absence of inhibitors. The amount of radioactivity associated with each band was linear with respect to time for at least 5 min (results not shown). As might be expected, given that there is only one active site within each molecule of enzyme, the broad-spectrum kinase inhibitor staurosporine inhibited IKK-1-catalyzed autophosphorylation and IjBa phosphorylation to the same degree as shown in Fig. 1A. In contrast to this result, the analogous experiment with IKK-2 showed that staurosporine inhibited the catalyzed phosphorylation of IjBa somewhat more potently than autophosphorylation (see Fig. 1B). The results are represented as pseudo-Dixon plots (i.e., ratio of the rate of product formed in the absence to that in the presence of inhibitor versus concentration of inhibitor) to more easily gauge the relative inhibition of each phosphorylation product. The use of ADP, which has been shown to inhibit IKK-1 and IKK-2 competitively with respect to ATP [16], showed an even more pronounced effect, with the enzyme-catalyzed IjBa phosphorylation inhibited considerably more potently than autophosphorylation with both IKK-1 and IKK-2 (see Fig. 2). That these pseudoDixon plots are linear indicates that Michaelis–Menten kinetics are still in effect (competitive, non-competitive, and uncompetitive inhibitors would all be expected to give linear Dixon plots, see [17]). Quite different results were obtained when a peptide corresponding to amino acid residues 26–42 of IjBa (with Ser-32 and Ser-36 mutated to aspartates) was used as an inhibitor. Since the serines have been removed, this
1510
J.R. Burke, J. Strnad / Biochemical and Biophysical Research Communications 293 (2002) 1508–1513
Fig. 1. Pseudo-Dixon plot of the inhibition by staurosporine of autophosphorylation and phosphorylation of IjBa catalyzed by (A) IKK-1 and (B) IKK-2. The amount of phosphorylated products was analyzed by SDS–PAGE and phosphorimager quantitation of the respective bands of 5-min assays. The ratio of the amount of product formed in the absence to that formed in the presence of inhibitor is represented as ðv0 ÞO =ðv0 ÞI . Closed circles, phosphorylation of IjBa; open circles, autophosphorylation. Plots are representative examples of multiple experiments.
peptide cannot act as a substrate and, instead, is an inhibitor [15]. As shown in Fig. 3, this inhibitor was more potent against the IKK-1 and IKK-2-catalyzed autophosphorylation than against the phosphorylation of fulllength IjBa. In addition, these pseudo-Dixon plots were curved rather than linear, with complete inhibition observed at concentrations of peptide >3 mg/mL. There are three possible mechanisms by which ADP and the peptide would inhibit the enzyme-catalyzed autophosphorylation and phosphorylation of IjBa with different potencies. The first possibility is that autophosphorylation and IjBa phosphorylation are catalyzed by the same active site, but autophosphorylation diminishes the catalytic activity of the enzyme toward IjBa phosphorylation. Inhibition of autophosphoryla-
Fig. 2. Pseudo-Dixon plot of the inhibition by ADP of autophosphorylation and phosphorylation of IjBa catalyzed by (A) IKK-1 and (B) IKK-2. The amount of phosphorylated products was analyzed by SDS–PAGE and phosphorimager quantitation of the respective bands of 5-min assays. The ratio of the amount of product formed in the absence to that formed in the presence of inhibitor is represented as ðv0 ÞO =ðv0 ÞI . Closed circles, phosphorylation of IjBa; open circles, autophosphorylation. Plots are representative examples of multiple experiments.
tion by the peptide inhibitor, for example, would appear to be more potent than inhibition of IjBa phosphorylation since the enzyme is in a more active nonphosphorylated state for longer period of time. This situation would result in IjBa phosphorylation apparently enhanced from that expected based on the observed autophosphorylation inhibition. There is evidence in the literature that autophosphorylation diminishes enzymatic activity [14]. Evidence against this possibility comes from the observation that under conditions of the assays used in the present work, the rates of product formation (both autophosphorylation and IjBa phosphorylation) were linear over the 5-min assay time in both the absence and presence of inhibitors (results not shown). Moreover, the more potent inhibi-
J.R. Burke, J. Strnad / Biochemical and Biophysical Research Communications 293 (2002) 1508–1513
Fig. 3. Pseudo-Dixon plot of the inhibition by peptide inhibitor of autophosphorylation and phosphorylation of IjBa catalyzed by (A) IKK-1 and (B) IKK-2. The amount of phosphorylated products was analyzed by SDS–PAGE and phosphorimager quantitation of the respective bands of 5 min assays. The ratio of the amount of product formed in the absence to that formed in the presence of inhibitor is represented as ðv0 ÞO=ðv0 ÞI . Closed circles, phosphorylation of IjBa; open circles, autophosphorylation. Plots are representative examples of multiple experiments.
tion against IjBa phosphorylation as compared to autophosphorylation observed with ADP cannot be explained in this manner. A second possibility is that the autophosphorylation is intermolecular and having a bound inhibitor changes the conformation of IKK such that the protein binds with altered affinity to the active site of a second enzyme molecule. For example, IKK with peptide inhibitor bound may have a larger dissociation constant from another IKK molecule catalyzing autophosphorylation than the dissociation constant of unbound IKK from another IKK molecule. This would result in more potent inhibition against autophosphorylation than against phosphorylation of IjBa in the assay: That is, inhibition of autophosphorylation would reflect occupation of active sites with inhibitor as well as an increase in the dissociation constant of inhibitor-bound autophosph-
1511
orylation ‘‘substrate.’’ To explain the effect observed with ADP, the conformation of ADP-bound protein would have to be recognized more effectively (i.e., smaller dissociation constant of ADP-bound IKK from the active site of another IKK enzyme). Therefore the inhibition of autophosphorylation due to occupying active sites with ADP would have to be somewhat offset by an increased affinity of ADP-bound IKK in an autophosphorylation reaction. To test whether the autophosphorylation is intermolecular, the inhibition of IKK-1 by ADP at two enzyme concentrations was measured. As shown in Fig. 4, the inhibition of autophosphorylation was independent of enzyme concentration. To reconcile this observation with an intermolecular autophosphorylation requires that the concentration of substrate (enzyme in the case of autophosphorylation) be much lower than the KM for autophosphorylation. If the concentration of substrate was near or greater than the KM , the potency of inhibition would have been much more sensitive to changes in substrate concentration. It would also be expected that if the autophosphorylation was intermolecular, 10-times lower enzyme concentration would have resulted in 100-times lower autophosphorylation product (10-times less catalyst multiplied by 10-times lower substrate concentration operating below the KM ). However, the amount of autophosphorylation was found to be only 9.9-times less at 0:14 lg=mL enzyme as compared to 1:4 lg/mL. These results therefore are consistent only with intramolecular autophosphorylation. Since IKK-1 and IKK-2 are isolated as dimers (in this case homodimers), it is important to note that intramolecular autophosphorylation does not discriminate between true intramolecular autophosphorylation (catalyzing phosphorylation on the same polypeptide
Fig. 4. Dependence of IKK-1 concentration on the inhibition of autophosphorylation by ADP. The amount of autophosphorylation product was analyzed by SDS–PAGE and phosphorimager quantitation of 5-min assays using IKK-1 at 1:4 lg/mL (closed circles) or 0:14 lg/mL (open circles). The ratio of the amount of product formed in the absence to that formed in the presence of inhibitor is represented as ðv0 ÞO=ðv0 ÞI . Plot is a representative example of multiple experiments.
1512
J.R. Burke, J. Strnad / Biochemical and Biophysical Research Communications 293 (2002) 1508–1513
chain as the active site) or phosphorylation of one subunit by the other subunit within a dimer. The third possible explanation for the differential potency of inhibitors against autophosphorylation as compared to phosphorylation of IjBa is that there are two non-equivalent forms of the enzyme present in the preparations. In this case, one form would contribute more to the autophosphorylation reaction while the other form of the enzyme plays a larger role in catalyzing IjBa phosphorylation. With ADP being more potent against IjBa phosphorylation, ADP would have to bind with greater affinity to the form of the enzyme contributing more to this reaction as compared to the form contributing more to autophosphorylation. The opposite would be true for the peptide inhibitor. This raises the interesting possibility that the two different forms of enzyme within the preparations simply represent two populations of enzyme which differ by the phosphorylation state at the MAP kinase consensus site within the activation loop of the N-terminal catalytic domain (autophosphorylation occurs in the C-terminal helix–loop–helix domain). Phosphorylation of the activation loop greatly affects enzymatic activity. While it has not been determined with IKK-1 or IKK-2, other proteins expressed from the baculovirus/insect cell expression system have been shown to be phosphorylated, sometimes to a high degree, presumably by endogeneous insect MAP kinases during expression [18]. To test this, the S176E/S180E mutant of IKK-1 was expressed and purified. It has been shown that this mutant mimics phosphorylation at these sites and gives constitutively activated enzyme [13]. As compared to the results with the wild-type IKK-1 shown in Fig 2A, ADP was equipotent against both autophosphorylation and IjBa phosphorylation when the mutant enzyme was used (see Fig. 5). This provides evidence, there are two forms of the wild-type enzyme (differentiated by the phosphorylation state of the enzyme) which bind inhibitors such as ADP with different affinities and have non-equivalent activities against autophosphorylation and IjBa phosphorylation. Interestingly, the inhibitor was more potent against both mutant IKK-1-catalyzed autophosphorylation and IjBa phosphorylation than was observed against either activity in wild-type enzyme. These results are in contrast to those observed with the peptide inhibitor which behaved in approximately the same manner toward mutant enzyme to that observed with the wild-type enzyme, with non-linear pseudo-Dixon plots and more potent inhibition against autophosphorylation than against phosphorylation of IjBa (compare Fig. 6 to Fig. 3A). This indicates that the effect of peptide is not dependent on two forms of the enzyme, differentiated by the phosphorylation states, contributing differentially to autophosporylation versus phosphorylation of IjBa.
Fig. 5. Pseudo-Dixon plot of the inhibition by ADP of autophosphorylation and phosphorylation of IjBa catalyzed by S176E/S180E mutant IKK-1. The amount of phosphorylated products was analyzed by SDS–PAGE and phosphorimager quantitation of the respective bands of 5-min assays. The ratio of the amount of product formed in the absence to that formed in the presence of inhibitor is represented as ðv0 ÞO =ðv0 ÞI . Closed circles, phosphorylation of IjBa; open circles, autophosphorylation. Plot is a representative example of multiple experiments.
Fig. 6. Pseudo-Dixon plot of the inhibition by peptide inhibitor of autophosphorylation and phosphorylation of IjBa catalyzed by S176E/S180E mutant IKK-1. The amount of phosphorylated products was analyzed by SDS–PAGE and phosphorimager quantitation of the respective bands of 5-min assays. The ratio of the amount of product formed in the absence to that formed in the presence of inhibitor is represented as ðv0 ÞO=ðv0 ÞI . Closed circles, phosphorylation of IjBa; open circles, autophosphorylation. Plot is a representative example of multiple experiments.
Discussion The present work demonstrates that recombinantly expressed homodimers of IKK-1 or IKK-2 contain two (or more) non-equivalent forms of the enzyme which catalyze the autophosphorylation and IkBa phosphorylation reactions with different affinities. Inhibitors such as ADP and an IjBa peptide inhibitor bind with nonequivalent affinities to these two forms of the enzyme such that more potent inhibition was observed against one of
J.R. Burke, J. Strnad / Biochemical and Biophysical Research Communications 293 (2002) 1508–1513
the phosphorylated products than against the other. That there are non-equivalent binding sites may be explained by a mixture of forms isolated from the baculovirus/insect cell expression system which differ in the degree of phosphorylation at the activation loop of the enzyme. Phosphorylation has been shown to greatly affect the activity of the IKKs [13,19]. Mutation of serines 176 and 180 of IKK-1 to phosphomimicking glutamates yields constitutively active (and homogeneous) enzyme which obliterated the differences in potencies ADP showed against autophosphorylation and IjBa phosphorylation. Interestingly, the differential effect of the peptide inhibitor on autophosphorylation and IjBa phosphorylation was insensitive to these mutations, suggesting that the active sites within preparations of this mutated enzyme remain non-equivalent. A possible explanation, especially relevant since both IKK-1 and IKK-2 are isolated as homodimers, is that the non-equivalent active sites are within the same dimer such that binding of a peptide substrate to one active site affects the second active site within the dimer (i.e., cooperativity or half-of-site reactivity, [20]). Unlike staurosporine or ADP which binds at the ATP binding site of the enzyme, the peptide inhibitor shows non-linear pseudo-Dixon plots. This is consistent with a cooperative effect between the two active sites since simple Michaelis–Menten kinetics are no longer in effect. Instead, the binding of peptide inhibitor at one site may enhance the binding affinity for the peptide at the second active site yielding the cooperative effect seen in Fig. 3. A cooperative effect between active sites within a dimer is not completely surprising, since it has been shown that the active site(s) of the multisubunit complex (containing heterodimeric IKK1 and IKK-2 along with the regulatory subunit IKK-c) binds IjBa substrate with considerably higher affinity than homodimers lacking IKK-c [15]. This indicates that the presence of other subunits affects the conformation of the active site(s). The model proposed here therefore is one in which phosphorylation state of the activation loop of IKK affects the active site conformation of the enzyme such that the two forms catalyze autophosphorylation and IjBa phosphorylation reactions with different affinities. This change in active site conformation would reflect changes both at the peptide binding site (to explain different efficiencies toward autophosphorylation and IjBa phosphorylation) and at the ATP binding site (reflecting the change in affinity of ADP and, in the case of IKK-2, staurosporine). In addition, the two active sites within the dimer may show cooperative interactions so that binding of peptide inhibitor at one active site affects the conformation of the other active site. While the present work used only homodimers of IKK-1 or IKK-2, it is uncertain how this model relates to heterodimers of IKK-1/IKK-2. It is interesting to speculate that this model may explain some aspects concerning the receptor-mediated activation of IKK in cells. For
1513
instance, it may be that receptor-mediated activation of kinase signal transduction cascades results in phosphorylation of the activation loop of one of the catalytic subunits within the complex (IKK-1, for instance). This phosphorylation then alters the protein conformation so that the catalytic efficiency is optimized toward phosphorylation of the activation loop of its binding partner in the complex (IKK-2?). We are continuing to investigate these questions to ascertain the role of phosphorylation as well as cooperativity between active sites in the regulation of this highly complex and poorly understood enzyme. Acknowledgments The authors thank Dr. Rolf-Peter Ryseck for providing the clones for IKK-1, IKK-2, and mutated IKK-1 used in the expression of these proteins.
References [1] S.T. Whiteside, A. Israel, Semin. Cancer Biol. 8 (1997) 75–82. [2] K. Brown, S. Gertsberger, L. Carlson, G. Franzoso, U. Siebenlist, Science 267 (1995) 1485–1488. [3] T.S. Finco, A.A. Beg, A.S. Baldwin Jr., Proc. Natl. Acad. Sci. USA 91 (1994) 11884–11888. [4] L. Baldi, K. Brown, G. Franzoso, U. Siebenlist, J. Biol. Chem. 271 (1996) 376–379. [5] M. Roff, J. Thompson, M.S. Rodriquez, J.-M. Jacque, F. Baleux, F. Arenzana-Seisdedos, R.T. Hay, J. Biol. Chem. 271 (1996) 7844–7850. [6] R. Weil, C. Laurent-Winter, A. Israel, J. Biol. Chem. 272 (1997) 9942–9949. [7] S.T. Whiteside, J.-C. Epinat, N.R. Rice, A. Israel, EMBO J. 16 (1997) 1413–1426. [8] U. Siebenlist, G. Franzoso, K. Brown, Annu. Rev. Cell Biol. 10 (1994) 405–455. [9] P.A. Bauerle, D. Baltimore, Cell 87 (1997) 13–20. [10] M. Karin, Y. Ben-Neriah, Ann. Rev. Immunol. 18 (2000) 621–663. [11] M.J. May, F. D’Acquisto, L.A. Madge, J. Glockner, J.S. Pober, S. Ghosh, Science 289 (2000) 1550–1554. [12] H. Nakano, M. Shindo, S. Sakon, S. Hishinaka, M. Mihara, H. Yagita, K. Okumura, Proc. Natl. Acad. Sci. USA 95 (1998) 3537– 3542. [13] F. Mercurio, H. Zhu, B.W. Murray, A. Shevchenko, B.L. Bennett, J.W. Li, D.B. Young, M. Barbosa, M. Mann, A. Manning, A. Rao, Science 278 (1997) 860–866. [14] M. Delhase, M. Hayakawa, Y. Chen, M. Karin, Science 284 (1999) 309–313. [15] J.R. Burke, M.K. Wood, R.-P. Ryseck, S. Walther, C.A. Meyers, J. Biol. Chem. 274 (1999) 36146–36152. [16] Q.K. Huynh, H. Boddupalli, S.A. Rouw, C.M. Koboldt, T. Hall, C. Sommers, S.D. Hauser, J.L. Pierce, R.G. Combs, B.A. Reitz, J.A. Diaz-Collier, R.A. Weinberg, B.L. Hood, B.F. Kilpatrick, C.S. Tripp, J. Biol. Chem. 275 (2000) 25883–25891. [17] I.H. Segel, in: Enzyme Kinetics, Wiley, New York, 1975, pp. 110– 144. [18] M.G. de Carvalho, A.L. McCormack, E. Olson, F. Ghomashchi, M.H. Gelb, J.R. Yates, C.C. Leslie, J. Biol. Chem. 271 (1996) 6987–6997. [19] J.A. DiDonato, M. Hayakawa, D.M. Rothwarf, E. Zandi, M. Karin, Nature 388 (1997) 548–554. [20] A. Fersht, in: Enzyme Structure and Mechanism, W.H. Freeman, New York, 1985, pp. 263–272.