Effects of the NIKalyMutation on NF-κB Activation by the Epstein-Barr Virus Latent Infection Membrane Protein, Lymphotoxin β Receptor, and CD40

Homozygosity for the aly point mutation in NF-κB-inducing kinase (NIK) results in alymphoplasia in mice, a phenotype similar to that of homozygosity for deletion of the lymphotoxin β receptor (LTβR). We now find that NF-κB activation by Epstein-Barr virus latent membrane protein 1 (LMP1) or by an LMP1 transmembrane domain chimera with the LTβR signaling domain in human embryonic kidney 293 cells is selectively inhibited by a wild type dominant negative NIK comprised of amino acids 624–947 (DN-NIK) and not by aly DN-NIK. In contrast, LMP1/CD40 is inhibited by both wild type (wt) and alyDN-NIK. LMP1, an LMP1 transmembrane domain chimera with the LTβR signaling domain, and LMP1/CD40 activate NF-κB in wt oraly murine embryo fibroblasts. Although wt andaly NIK do not differ in their in vitro binding to tumor necrosis factor receptor-associated factor 1, 2, 3, or 6 or in their in vivo association with tumor necrosis factor receptor-associated factor 2 and differ marginally in their very poor binding to IκB kinase β (IKKβ), only wt NIK is able to bind to IKKα. These data are compatible with a model in which activation of NF-κB by LMP1 and LTβR is mediated by an interaction of NIK or a NIK-like kinase with IKKα that is abrogated by the alymutation. On the other hand, CD40 mediates NF-κB activation through a kinase that interacts with a different component of the IKK complex.

NIK has an essential role in lymphoid organ development (12). The aly mutation results in a single amino acid change of glycine to arginine at mNIK codon 855 and can be rescued by transgenic expression of wild type NIK (12). Alymphoplasia (aly/aly) mice not only lack lymph nodes and Peyer's patches but also have abnormal spleen and thymus development, low serum Ig levels, and impaired B cell proliferation in response to LPS or CD40L (12,13). LT␤R Ϫ/Ϫ and aly/aly mice have similar developmental and immunological defects, and NIK has been implicated in LT␤R-mediated activation of NF-B (14,15). Indeed, LT␤R up-regulation of VCAM-1 is abnormal in aly/aly murine embryo fibroblasts (15). Also, CD40L-induced phosphorylation of IB␣ is abnormal in B lymphocytes from aly/aly mice, although phosphorylation of IB␣ in dendritic cells is normal (13).
Epstein-Barr virus (EBV) latent infection of human B lymphocytes causes B lymphocyte proliferation through expression of nuclear proteins and an integral membrane protein, LMP1, which mimics constitutively activated TNFRs (16). LMP1 has a short, arginine-rich, N-terminal cytoplasmic domain that is important for anchoring the first transmembrane domain, six hydrophobic transmembrane domains that mediate LMP1 aggregation in lipid rafts, and a 200-amino acid C-terminal cytoplasmic domain that has two sites that mediate EBV-induced B cell proliferation and NF-B activation (for review see Ref. 17). One site binds TRAF 3, 1, 2, and 5, whereas the second site binds TNFR-associated death domain protein (18,19). NF-B activation from either site is inhibited by overexpression of K429A/K430A kinase-negative DN-NIK (7). Thus, previous data are consistent with NIK having a significant role in LMP1 activation of NF-B. The experiments reported here further investigate the role of NIK and of the aly mutation in NF-B activation by LMP1, LT␤R, and CD40. by PCR with the BamHI and EcoRI primers and cloned into pGEX-2TK. The BamHI/EcoRI fragments were also used to make wt and aly DN-NIK for mammalian expression by subcloning into pCDNA3 (Invitrogen). Plasmids encoding TRAF1, TRAF2, TRAF3, IKK␣, IKK␤,  LMP1, CD40, and LMP1/CD40 have been described (7,18,20). The pCR-F-mTRAF6 plasmid was obtained from Dr. J. Inoue. The LMP1/ LT␤R construct was made by cloning the cytoplasmic domain from LT␤R into the previously described LMP1 pCDNA3 construct.
Cell Lines-293 and 293T cells were cultured as previously described (7). MEFs from aly/aly and wild type mice were obtained from Dr. T. Honjo. Primary MEFs were not amenable to transfection. Therefore, cells were infected with a human papilloma virus 16 E6/E7 retrovirus containing a neomycin resistance cassette (obtained from Dr-. P. Howley), and transformed cell lines were selected for G418 resistance and growth advantage. Cells were subsequently grown in Dulbecco's modified Eagle's medium with 20% fetal calf serum and antibiotics.
Transfections and Reporter Gene Assays-Transfections and reporter assays (7) were done with 350 ng per well of the 3xNF-B-luc reporter plasmid and 350 ng per well of pGK-␤-galactosidase as a transfection control. Measurements of luciferase and ␤-galactosidase activities were done with an Optocomp I luminometer (MGM Systems). Immune Precipitations-For transfections, 293T cells were seeded in 6-well plates at 5 ϫ 10 5 cells/well, and 24 h later 3 wells were transfected with 1 g of pCDNA3-F-TRAF2 or pRK5-myc-IKK␣ per well and 2, 3, or 4 g/well of wt or aly DN-NIK or pCDNA3 for each immune precipitation. Cells were lysed in Nonidet P-40 lysis buffer (1% Nonidet P-40, 150 mM NaCl, 0.5 mM EDTA, 20 mM Tris, pH 7.5) containing phenylmethylsulfonyl fluoride (1 mM) and aprotinin (22 g/ml) for 30 min on ice. Following lysis, cells were centrifuged at high speed for 10 min, precleared with protein G-Sepharose (Amersham Pharmacia Biotech) for 1 h at 4°C, normalized for total protein concentration, and 2 g of TRAF2 (C-20; Santa Cruz Biotechnology) antibody or 30 l of M2 beads (Amersham Pharmacia Biotech) were added. After rotation for 1 h at 4°C for C-20, protein G-Sepharose was added to the samples for an additional 1 h. Immune precipitates were washed 3-5 times with Nonidet P-40 lysis buffer. Samples were denatured in SDS-PAGE loading buffer and subjected to SDS-PAGE followed by immunoblotting (IB). IB antibodies were H-248 (NIK; Santa Cruz Biotechnology), M-110 (IKK␣; Santa Cruz Biotechnology), 9E10 (Myc; see Ref. 21), and C-20 (TRAF2; Santa Cruz Biotechnology).

Effects of wt or aly DN-NIK on EBV LMP1-mediated Activation of NF-B-LMP1 expression in 293T cells strongly acti-
vated NF-B as measured by a co-transfected reporter with three NF-B sites from the major histocompatibility complex I promoter upstream of luciferase and a pGK-␤-galactosidase control plasmid (see Fig. 1 and Refs. 7, 22, and 23). LMP1mediated NF-B activation was inhibited in a dose-dependent manner by co-expression of the wt DN-NIK fragment (aa 624 -947) (Fig. 1A). In contrast to the effect of wt DN-NIK, coexpression of the DN-NIK containing the aly point mutation G860R did not inhibit LMP1 activation of NF-B (Fig. 1B). wt and aly DN-NIK were expressed at equivalent levels, ␤-galactosidase levels did not vary more than 2-fold, and LMP1 expression was not affected by either wt or aly DN-NIK (Fig. 1C), excluding an artifactual basis for the differential effect. Thus, LMP1 activation of NF-B is resistant to aly but not to wt DN-NIK as had been previously noted for LT␤R (12). Similar results were obtained in multiple experiments in 293 and 293T cells.

Effects of wt or aly DN-NIK on LT␤R and CD40 Activation of NF-B-
To evaluate whether the differential effect of the wt and aly DN-NIK is specific for LMP1 and LT␤R signaling as opposed to other TNFRs, the effect of wt and aly DN-NIK on NF-B activation by LMP1, LT␤R, and CD40 was assayed in 293T and 293 cells. The LMP1 transmembrane domains were used to provide constitutive, ligand-independent receptor aggregation. Isogenic expression constructs were made in which the LMP1 C-terminal cytoplasmic domain was replaced with the LT␤R or CD40 C-terminal cytoplasmic domains. LMP1 and the LMP1/LT␤R chimera activated NF-B, and the activation was inhibited by wt but not by aly DN-NIK (see Fig. 1 and Fig.  2, A and B). wt and aly DN-NIK did not affect LMP1/LT␤R expression, and ␤-galactosidase levels did not vary more than 2-fold ( Fig. 2C and data not shown). These results are consistent with the previous observation that NF-B activation following LT␤R overexpression is inhibited by wt but not by aly DN-NIK (12). In contrast to the effects of wt but not aly DN-NIK on LMP1 or LMP1/LT␤R activation of NF-B, NF-B activation mediated by an LMP1/CD40 cytoplasmic domain chimera or by CD40 overexpression was inhibited by both wt and aly DN-NIK (Fig. 3, A and B and data not shown). CD40 and LMP1/CD40 expression were not affected by either wt or aly DN-NIK ( Fig. 3C and data not shown). These data support a model in which LMP1 and LT␤R activation of NF-B involves a downstream molecular interaction that can be inhibited by wt hNIK aa 624 -947 but not by the corresponding fragment with the glycine to arginine mutation at codon 860. NF-B activation by isogenic LMP1, LMP1/LT␤R, and LMP1/ CD40 was assessed in wt and aly/aly MEFs using the cotransfected NF-B-dependent luciferase reporter and control pGK-␤-galactosidase expression plasmids. Before initiating this series of experiments, the MEFs were first transformed with a human papilloma virus 16 E6 and E7-expressing retrovirus so as to enhance their growth and transfection efficiency (24). The surprising result was that LMP1, LMP1/LT␤R, and LMP1/CD40 activated NF-B similarly in wt and aly/aly MEFs indicating that NIK is not essential for NF-B activation by these receptors in fibroblasts (Fig. 4). Thus, the insensitivity of LMP1 and LMP1/LT␤R to aly DN-NIK inhibition of NF-B activation is not because of a specific and exclusive dependence of these receptors on wt NIK for NF-B activation.
aly NIK Interacts with TRAFs in Vitro and in Vivo-The aly mutation at codon 860 falls within both the TRAF binding domain (aa 624 -947) and the IKK binding domain (aa 735-947) of hNIK (1, 25) (Fig. 5A). The aly/aly phenotype is presumed to be because of the failure of aly NIK to interact with TRAFs (25). To more precisely determine the biochemical basis for the aly effect on NIK interactions, we compared the in vitro and in vivo binding of wt and aly NIK to TRAFs. hTRAFs 1, 2, and 3 and mTRAF6 were transcribed and translated with radiolabeled [ 35 S]methionine, and their ability to bind in vitro to equal amounts of wt or aly hNIK aa 624 -947 fused to GST (NIK 624 -947 -GST) was assessed. The aly mutation did not affect TRAFs binding to NIK 624 -947 -GST in vitro (Fig.  5B). However, TRAFs differed in their binding to NIK 624 -947 -GST; TRAF1 and -2 bound to NIK 624 -947 -GST at the 10 to 15% level, whereas less than 3% of TRAF3 or -6 bound to NIK 624 -947-GST (Fig. 5B).
Because TRAF2 is an important mediator of TNFR1-induced NF-B activation (26), TRAF2 associations with wt and aly NIK were evaluated in 293 cells that were co-transfected with a FLAG-tagged TRAF2 expression construct and either wt or aly DN-NIK (aa 624 -947). F-TRAF2 was immune-precipitated using M2-conjugated Sepharose beads, and the complexes were blotted for the presence of NIK. F-TRAF2 consistently brought down equivalent amounts of wt or aly DN-NIK (Fig. 5C). These data indicate that the aly/aly phenotype is unlikely to be because of the previously postulated inability of TRAFs to engage and associate with aly NIK.
Differential Association of wt and aly NIK with IKK␣ and IKK␤-hNIK aa 735-947 is not only part of the domain that binds to TRAFs but is also a sufficient domain for interaction with IKK␣ (3,27). hNIK aa 735-947 efficiently competes with wt NIK for binding to IKK␣ and is a dominant negative inhibitor of TNF-mediated NF-B activation (3). Surprisingly, only wt but not aly NIK 624 -947 -GST was able to pull down Myc-IKK␣ from lysates of 293T cells in which Myc-IKK␣ was expressed. About 10% of the Myc-tagged IKK␣ bound to wt NIK 624 -947 -GST, whereas binding to similar amounts of aly NIK 624 -947 -GST was below the limits of detection (Fig. 6A).
In similar experiments, the binding of FLAG-tagged IKK␤ expressed in 293T cells to wt or aly NIK 624 -947 -GST was assessed. As reported for full-length NIK (27), FLAG-tagged IKK␤ bound very weakly to wt NIK 624 -947 -GST, at a level of about 1% of input IKK␤ (Fig. 6A). Binding of F-IKK␤ to aly NIK 624 -947 -GST was about half the level of binding to wt and about twice as strong as to GST alone (Fig. 6A).
To evaluate the association of Myc-IKK␣ with wt or aly DN-NIK in vivo, 293T cells were co-transfected with Myc-IKK␣ and wt or aly DN-NIK expression vectors, lysed in non-ionic detergent, and immune-precipitated with anti-Myc antibody

FIG. 3. CD40-mediated NF-B activation is inhibited by both wt and aly DN-NIK.
A and B, this experiment was identical to that shown in Fig. 1, except that 1 g of CD40 was transfected instead of LMP1. The data shown are from one representative experiment of two with CD40 and three with LMP1/CD40 (not shown). C, IB of CD40 from cells transfected with CD40 alone (lane 1) or with CD40 plus 0.1 or 2 g of wt DN-NIK (lanes 2 and 3). Identical expression of CD40 was seen with CD40 plus aly DN-NIK (not shown). FIG. 4. LMP1, LMP1/LT␤R, and LMP1/CD40 activate NF-B at an equivalent level in wt and aly MEFs. LMP1 (100 ng), LMP1/ LT␤R (5 g), or LMP1/CD40 (100 ng) were transfected into either wt (f) or aly/aly ( ) MEFs, along with 3xB-luc and pGK-␤-galactosidase. Luciferase assays were performed as in Fig. 1. NF-B activation relative to control plasmid is shown for one representative of three independent experiments. followed by immune blotting for NIK. The efficiency of the Myc-IKK␣ IP was about 10% (data not shown). wt DN-NIK was readily detected in the immune precipitate, whereas aly DN-NIK was not detectable (Fig. 6B). Thus, the aly mutation results in virtually complete loss of interaction or association of DN-NIK with IKK␣.

DISCUSSION
The data presented here indicate that LMP1, LMP1/LT␤R, and LMP1/CD40 activate NF-B similarly in wild type and aly/aly murine embryo fibroblasts. Therefore, NIK either does not have a direct role in NF-B activation from any of these receptor cytoplasmic domains in MEFs, or another kinase substitutes for NIK in the transfected aly/aly fibroblasts and is unable to substitute under more physiologic conditions in vivo. Although this latter possibility seems a priori unlikely, similar substitution effects have been noted among MAP kinases in yeast, for example (28 -33). Further, the similarity between the aly/aly and LT␤R Ϫ/Ϫ phenotypes is most consistent with a physiologically significant role for the aly mutation in LT␤R signaling. Moreover, the inability of the aly DN-NIK to specifically block LMP1 and LMP1/LT␤R activation of NF-B in 293 cells supports the notion that these cytoplasmic domains signal through a pathway that is specifically blocked by wt and not by aly DN-NIK. Indeed, the biochemical studies further support the hypothesis that the effect is at the level of NIK per se, in indicating that aly hNIK is most abnormal in loss of interaction with IKK␣, is not abnormal in interaction with TRAF1, 2, 3, or 6, and is only minimally evident in diminished very weak interaction with IKK␤.
The simplest explanation of the inability of the aly DN-NIK to block LMP1 or LMP1/LT␤R activation of NF-B is that aly NIK is unable to bind to an essential mediator of that pathway. The aly mutation is within the NIK TRAF binding domain, and the failure of aly NIK to block some TNFRs has been attributed to a putative affect of the aly mutation on TRAF binding. Our data indicate that the aly mutation does not affect NIK binding to or association with TRAF1, 2, 3, or 6, making it less likely that the aly effect is at the level of TRAF interaction with NIK. Instead, we find aly NIK to be highly deficient in binding to IKK␣ and that both wt and aly mutant NIK bind poorly to IKK␤. Thus, the ability of wt NIK and the inability of aly NIK to block NF-B activation from LMP1 or LMP1/LT␤R are most consistent with a key role for a NIK-like kinase and IKK␣ in LMP1 and LT␤R activation of NF-B. Further, the blockade of CD40 and LMP1/CD40 activation of NF-B by both wt and aly DN-NIK are most compatible with the possibility that CD40 signaling through the IKK complex is mediated by a protein that can be blocked by either the wt or aly NIK C terminus.
During the preparation of this manuscript, two publications appeared that are relevant to these experiments. In one, NIK is found to associate with the p100 precursor to the NF-B subunit p52 and induce its phosphorylation and proteolytic processing (34). aly NIK, however, is unable to associate with p100 or induce p100 phosphorylation and processing (34). Consistent with these observations, p52 is not detected in aly/aly cells, despite the presence of p100 (34,35). However, although 2 M. L. and E. K., unpublished observations.

FIG. 5. wt and aly NIK bind to TRAFs in an equivalent manner.
A, a schematic diagram of NIK is shown. N-terminal regions were recently defined to play a role in regulating NIK activity by preventing the interaction between the C terminus of NIK and IKK␣ (39). The ATP binding pocket within the conserved MAP kinase domain is ablated by the K429A/K430A mutation (1). Thr 559 is the critical residue in the activation loop of the kinase domain that is phosphorylated by Tpl2 and is required for kinase activation (3). The region from aa 624 -947 has been defined as the necessary and sufficient TRAF binding domain (1), 2 whereas aa 735-947 within this domain is necessary and sufficient for IKK␣ binding (3). The alymphoplasia phenotype is caused by a single point mutation (Gly to Arg) occurring at amino acid 855 in mNIK (12) and is in a highly conserved region in hNIK at amino acid 860. BR, basic region; PRR, proline-rich region. B, [ 35 S]methionine-labeled, in vitrotranslated TRAFs 1, 2, and 3 or mTRAF6 were incubated with wt NIK 624 -947 -GST, aly NIK 624 -947 -GST, or GST alone. The 10% input lane indicates 10% of the total IVT reaction lysate used for each binding assay. Samples were resolved by SDS-PAGE, and binding was analyzed by phosphorimaging. C, 293T cells were co-transfected with F-TRAF2 and either wt or aly DN-NIK. F-TRAF2⅐NIK complexes were immunoprecipitated with anti-FLAG resin (M2; Sigma) followed by IB with anti-NIK antibody. 2% of whole cell lysates were probed with anti-TRAF2 and anti-NIK antibody.
FIG. 6. wt and aly NIK bind to IKK␤; however aly NIK does not bind to IKK␣. A, wt NIK 624 -947 -GST, aly NIK 624 -947 -GST, or GST were used to precipitate either Myc-IKK␣ or F-IKK␤ from transfected 293T cells. GST-bound IKKs were subjected to SDS-PAGE followed by IB for the appropriate tag. IB with IKK␣ antibody (M110; Santa Cruz Biotechnology) resulted in an identical result (data not shown). B, 293T cells co-transfected with the indicated amounts of wt or aly DN-NIK, along with a constant amount of Myc-IKK␣ expression vector, and were immunoprecipitated with anti-myc antibody followed by IB with anti-NIK antibody. Lysates contained equivalent amounts of IKK␣ and DN-NIK proteins (not shown). p52 Ϫ/Ϫ mice have major defects in germinal center formation and splenic architecture similar to aly/aly and LT␤R Ϫ/Ϫ mice, serum Ig levels and proliferation in response to LPS and CD40L appear relatively normal in p52 Ϫ/Ϫ mice but are abnormal in aly/aly mice (36,37). Therefore, the deficiency in p100 processing is likely to account for only part of the aly/aly phenotype. In a second very recent paper, B cells from IKK␣ Ϫ/Ϫ mice are found to be quite similar to aly/aly B cells in their response to LPS and CD40L (38). This report is consistent with our finding that aly NIK is unable to interact with IKK␣. Moreover, serum Ig levels are depressed in IKK␣ Ϫ/Ϫ bone marrow chimeras and in aly/aly mice but not in p52 Ϫ/Ϫ mice (38). Thus, the inability of aly NIK to process p100 into p52 and the inability of aly NIK to interact with IKK␣ may both contribute to the aly/aly phenotype.