Mitogen-activated Protein Kinase/ERK Kinase Kinases 2 and 3 Activate Nuclear Factor-κB through IκB Kinase-α and IκB Kinase-β*

Recent evidence indicates that nuclear factor-κB (NF-κB), a transcription factor critically important for immune and inflammatory responses, is activated by a protein kinase cascade. The essential features of this cascade are that a mitogen-activated protein kinase kinase kinase (MAP3K) activates an IκB kinase (IKK) that site-specifically phosphorylates IκB. The IκB protein, which ordinarily sequesters NF-κB in the cytoplasm, is subsequently degraded by the ubiquitin-proteasome pathway, thereby allowing the nuclear translocation of NF-κB. Thus far, only two MAP3Ks, NIK and MEKK1, have been identified that can activate this pathway. We now show that MEKK2 and MEKK3 can in vivoactivate IKK-α and IKK-β, induce site-specific IκBα phosphorylation, and, relatively modestly, activate an NF-κB reporter gene. In addition, dominant negative versions of either IKK-α or IKK-β abolish NF-κB activation induced by MEKK2 or MEKK3, thereby providing evidence that these IKKs mediate the NF-κB-inducing activities of these MEKKs. In contrast, other MAP3Ks, including MEKK4, ASK1, and MLK3, fail to show evidence of activation of the NF-κB pathway. We conclude that a distinct subset of MAP3Ks can activate NF-κB.

Upon exposure to a wide variety of agents, including the proinflammatory cytokine TNF-␣, lipopolysaccharide, oxidative stress, and the HTLV-I Tax protein, the IB protein is phosphorylated at its N terminus. In the case of IB␣, the most extensively studied IB isoform, this phosphorylation occurs at Ser-32 and Ser-36 (3,4). This phosphorylation event targets IB for degradation by the ubiquitin-proteasome pathway (5), allowing the subsequent nuclear translocation of NF-B.
An IB kinase (IKK) complex with a native molecular mass of 700 kDa was originally identified in cytoplasmic extracts of HeLa cells and shown to perform the site-specific phosphorylation of IB␣ (6,7). A significant advance was the subsequent cloning of the cDNAs for the catalytic, protein kinase subunits of this complex, IKK-␣ and IKK-␤ (8 -12). Several lines of evidence now indicate that IKK-␣ and IKK-␤ can be regulated by phosphorylation. The initial indications were that the IKK complex can be activated in vitro by the MAP3K MEKK1 (MAPK/ERK kinase kinase 1) (7) and that the complex, activated either in vitro by MEKK1 or in vivo by exposure of cells to TNF-␣ can be inactivated by phosphatase treatment (7,8). Additional work then demonstrated that (i) mutation of potential phosphoacceptor residues to alanine in the activation loop of IKK-␣ or IKK-␤ abrogated activity (12), (ii) mutation of these same residues in IKK-␤ to the phosphoresidue mimetic glutamic acid results in its constitutive activation (12), (iii) both IKK-␣ and IKK-␤ can be activated in vivo when overexpressed with MEKK1 or the related MAP3K NF-B-inducing kinase (NIK) (10,11,(13)(14)(15), and (iv) immunoprecipitated NIK can phosphorylate immunoprecipitated IKK-␣ (16). Therefore, an important conceptual advance in our understanding of NF-B regulation is that it can be activated by protein kinase cascade, the core elements of this cascade being a MAP3K and an IKK (7,10,17).
These findings have already begun to provide a framework for understanding how NF-B can be activated by diverse stimuli. For example, compelling evidence has been presented to show that NIK mediates the NF-B-inducing activity of TNF-␣ (17,18), whereas MEKK1 mediates the NF-B-inducing activity of Tax (15). Thus, different stimuli can activate NF-B by targeting different MAP3Ks.
These findings moreover raise the possibility that yet other MAP3Ks might activate NF-B. MAP3Ks were originally identified as components of signaling cascades in which a MAP3K phosphorylates and activates a MAP2K, which in turn phosphorylates and activates a MAPK; the latter include the mitogen-activated ERK and the stress-activated c-Jun N-terminal kinase (JNK, also known as stress-activated protein kinase) and p38 families (19). Here we show that MEKK2 and MEKK3, but not certain other MAP3Ks, can activate NF-B, and show that this activation occurs by their activation of IKK-␣ and IKK-␤.
Tissue Culture and Transfection-HeLa cells were maintained as described (7). Transfections performed in 3.5-cm-diameter wells were conducted by calcium phosphate precipitation (25) or by using Fugene 6 according to the manufacturer's instructions (Boehringer Mannheim). CAT and protein measurements were performed as described (7,26).
Immunoprecipitations-Cells were washed once with Dulbecco's phosphate-buffered saline containing 1 mM EDTA and then lysed by the addition of 1 ml of buffer B (14) containing 10 g/ml leupeptin and 1 mM phenylmethylsulfonyl fluoride. After centrifugation of the whole cell lysate at 16,000 ϫ g for 10 min at 4°C, the supernatant was incubated with 10 l of M2-agarose with end over end rotation for 1 h at 4°C. The resin was then washed three times with buffer B and eluted by the addition of 20 l of 2ϫ SDS-PAGE loading buffer.
Protein Kinase Assays-Immunocomplex kinase assays for IKK and JNK were performed essentially as described (14), except that 10 Ci of [␥-32 P]ATP was employed per assay, 1 g of GST-IB␣ (5-55) was used in the IKK assays, and the ATP concentration employed to initiate the IKK reactions was 50 M instead of 200 M. Kinase activities were quantitated using a Molecular Dynamics Storm 860 PhosphorImager.

MEKK2 and MEKK3 Induce NF-B Activity and Site-specific
Phosphorylation of IB␣-To examine the possibility that MAP3Ks other than MEKK1 or NIK might activate NF-B HeLa cells were cotransfected with a reporter gene that contains two NF-B binding sites and expression constructs for a series of MAP3Ks, including MEKK2 (20), MEKK3 (20), the catalytic domain of MEKK4 (⌬MEKK4) (21), apoptosis signalregulating kinase 1 (ASK1) (22), and mixed-lineage kinase 3 (MLK3, also known as protein-tyrosine kinase 1 or SH3 domain-containing proline-rich kinase (23,27,28). All can activate the JNK pathway (20 -22, 28). In addition, MEKK3, MEKK4, and ASK1 can activate the p38 pathway (22,29,30), whereas MEKK2 and MEKK3 can activate the ERK pathway (20). As shown in Fig. 1A, under conditions where overexpression of the positive controls MEKK1 and NIK induces activation of the NF-B reporter gene, overexpression of MEKK3 (as reported previously; Ref. 31) and, to a lesser extent, MEKK2 induces activation as well. ⌬MEKK4, ASK1, and MLK3 did not induce activation in either these cells (Fig. 1A) or the murine fibroblast cell line L929 (data not shown) but as expected did induce robust activation of coexpressed JNK1 in HeLa cells (data not shown).
NIK and MEKK1 activate NF-B by inducing the site-specific phosphorylation of IB. In the case of IB␣, this phosphorylation, which occurs at Ser-32 and Ser-36, is manifested by slower mobility when IB␣ is examined by SDS-PAGE (3,4). To examine whether MEKK2 and MEKK3 might act through the same mechanism, HeLa cells were cotransfected with expression vectors for Flag-tagged wild-type or phosphorylation-defective (S32A/S36A) IB␣ and expression vectors for MEKK2, MEKK3, or empty expression vector. The Flag-tagged IB␣ was then immunoprecipitated with anti-Flag antibodies and examined by Western blotting with anti-IB␣ antibodies. As shown in Fig. 1B, MEKK2 and MEKK3 both induce the appearance of a more slowly migrating IB␣ species (top panel, lanes 3 and 5, upper bands) that is abolished when an S32A/ S36A IB␣ mutant is examined (lanes 4 and 6), consistent with this species being N-terminally phosphorylated IB␣. ⌬MEKK4, ASK1, and MLK3 did not induce the appearance of this more slowly migrating species (data not shown). This IB␣ species was examined further by reprobing this blot with antibodies specific for phospho-Ser-32 IB␣. As shown in Fig. 1B  (bottom panel, lanes 3 and 5), the slower migrating IB␣ species induced by MEKK2 or MEKK3 is immunoreactive with these antibodies. We conclude that MEKK2 and MEKK3 can induce site-specific, N-terminal phosphorylation of IB␣ in vivo.
MEKK2 and MEKK3 Activate Both IKK-␣ and IKK-␤-Both MEKK1 and NIK induce site-specific phosphorylation of IB by activating IKK-␣ and IKK-␤. To examine whether MEKK2 or MEKK3 acts by the same mechanism, HeLa cells were cotransfected with expression constructs for Flag-tagged IKK-␣, IKK-␤, or JNK1 and expression constructs for MEKK1, MEKK2, MEKK3, NIK, or MLK3. The IKK or JNK was then immunoprecipitated with anti-Flag antibodies, and the kinase activities of the immunoprecipitated proteins were measured by using as substrates GST fused to the N terminii of IB␣ (residues 5-55) or c-Jun (residues 1-79), respectively, in the presence of [␥-32 P]ATP.
The potencies of MEKK1, MEKK2, MEKK3, and NIK in activating IKK-␣, IKK-␤, or an NF-B reporter gene were analyzed in more detail (Fig. 3). All four MAP3Ks induce dosedependent increases in the activities of coexpressed IKK-␣ or IKK-␤. In the case of coexpressed IKK-␣, the dose response curves are roughly comparable (Fig. 3A; see also Fig. 2A). In the case of coexpressed IKK-␤, NIK is a somewhat less potent activator than the other MAP3Ks (Fig. 3B). In contrast, NIK is a substantially more potent activator of an NF-B reporter gene than the other three MAP3Ks (Fig. 3C). For example, the NF-B reporter gene activity induced by 40 ng of NIK expression vector is comparable or even greater than that induced by 4000 ng of expression vector for either MEKK1, MEKK2, or MEKK3.
Dominant Negative IKK-␣ and Dominant Negative IKK-␤ Inhibit MEKK2-and MEKK3-induced NF-B Activation-The experiments described above indicate that MEKK2 and MEKK3 can activate both IKK-␣ and IKK-␤ in vivo. To examine whether this activation is functionally significant, HeLa cells were cotransfected with expression constructs for MEKK1, MEKK2, MEKK3, or empty expression vector, expression constructs for dominant negative, catalytically inactive IKK-␣ (K44A), IKK-␤ (K44A), or empty expression vector, and an NF-B reporter gene. As shown in Fig. 4, under conditions where activation of the NF-B reporter gene induced by MEKK1 is almost completely inhibited by dominant negative IKK-␣ or dominant negative IKK-␤ (13,14,32), that induced by either MEKK2 and MEKK3 is completely abolished. This therefore provides evidence that IKK-␣ and IKK-␤ mediate the NF-B inducing activity of MEKK2 and MEKK3. DISCUSSION Here we identify two additional, previously cloned MAP3Ks: MEKK2 and MEKK3, which now join NIK and MEKK1 as activators of IKK and NF-B, thereby enlarging our framework for understanding NF-B activation. Such knowledge is essen- Titration experiments reveal that MEKK2 and MEKK3 are comparable in potency to NIK in activating coexpressed IKK-␣ or IKK-␤ but are substantially less potent than NIK in activating an NF-B reporter gene (Fig. 3). Possible factors that might contribute to this apparent discrepancy include the following: (i) MEKK2 and MEKK3 might activate intracellular pathways that inhibit the NF-B pathway and therefore could be relatively less effective in activating an NF-B reporter gene; (ii) NIK might activate other components of the NF-B pathway besides IKK and thus could be relatively more potent in activating an NF-B reporter gene; and (iii) overexpressed IKK might respond less sensitively than endogenous IKK to coexpressed MAP3K and therefore might not accurately reflect activation of the NF-B pathway (33). In any case, relative potencies in transient overexpression assays cannot be used as the sole criterion for assessing physiologic significance. A particularly pertinent example is provided by the fact the HTLV-I protein Tax activates NF-B through MEKK1 (15) despite the fact that this MAP3K, like MEKK2 or MEKK3, is substantially less potent than NIK in activating an NF-B reporter gene (14,18).
MEKK2 and MEKK3, like other MAP3Ks, contain both catalytic and noncatalytic domains. The catalytic domains of MEKK2 and MEKK3 share 96% homology, consistent with the fact that both can activate NF-B, whereas their noncatalytic domains are 65% homologous (20). MEKK2 is activated by treatment of cells by epidermal growth factor (34). In addition, MEKK2 and MEKK3 bind 14-3-3 proteins, an interaction that is mediated at least in part through their catalytic domains but that does not modulate their JNK-inducing activities (35).
Further experimentation will be required to determine the detailed mechanism by which MEKK2 and MEKK3 activate IKK-␣ and IKK-␤. NIK activates IKK-␣ by inducing phosphorylation of Ser-176 in the activation loop of the latter (16). Phosphorylation of Ser-177 and/or Ser-181 in IKK-␤ is essential for its activity, because a double S176A/S181A mutation abolishes activity (12). Therefore, MEKK2 and MEKK3 might directly phosphorylate these IKK residues, particularly because these residues are components of canonical MAP2K activation loop motifs (SXXXS) (12) that might be predicted to be substrates for MAP3Ks. It is worth noting, however, that definitive experimental evidence that either NIK or MEKK1 directly induce IKK-␣ or IKK-␤ catalytic activity has yet to be reported.
The fact that many MAP3Ks have the capacity to activate distinct pathways now raises the problem of how specificity in signaling pathways is achieved. One example of this is provided by the observation that Tax activates MEKK1 and induces potent NF-B activity (15) but only modest JNK activity (15,36), despite the fact that MEKK1 overexpression coordinately activates both (7). Thus, the relative capacities of a MAP3K to activate distinct signaling pathways may be modulated in a manner that is stimulus-specific.