IKK-i and TBK-1 are enzymatically distinct from the homologous enzyme IKK-2: comparative analysis of recombinant human IKK-i, TBK-1, and IKK-2.

NF-kappaB is sequestered in the cytoplasm by the inhibitory IkappaB proteins. Stimulation of cells by agonists leads to the rapid phosphorylation of IkappaBs leading to their degradation that results in NF-kappaB activation. IKK-1 and IKK-2 are two direct IkappaB kinases. Two recently identified novel IKKs are IKK-i and TBK-1. We have cloned, expressed, and purified to homogeneity recombinant human (rh)IKK-i and rhTBK-1 and compared their enzymatic properties with those of rhIKK-2. We show that rhIKK-i and rhTBK-1 are enzymatically similar to each other. We demonstrate by phosphopeptide mapping and site-specific mutagenesis that rhIKK-i and rhTBK-1 are phosphorylated on serine 172 in the mitogen-activated protein kinase kinase activation loop and that this phosphorylation is necessary for kinase activity. Also, rhIKK-i and rhTBK-1 have differential peptide substrate specificities compared with rhIKK-2, the mitogen-activated protein kinase kinase activation loop of IKK-2 being a more favorable substrate than the IkappaBalpha peptide. Finally, using analogs of ATP, we demonstrate unique differences in the ATP-binding sites of rhIKK-i, rhTBK-1, and rhIKK-2. Thus, although these IKKs are structurally similar, their enzymatic properties may provide insights into their unique functions.

NF-B 1 is a ubiquitous transcription factor that plays an important role in the regulation of a wide variety of genes involved in immune, inflammatory, and stress responses (1)(2)(3)(4). In resting cells, NF-B is sequestered in the cytoplasm in an inactive state by association with members of the IB family of inhibitory proteins (IB␣, IB␤, or IB⑀), the best characterized being IB␣ (5)(6)(7)(8). Stimulation of cells with an agonist results in phosphorylation, ubiquitination, and degradation of IBs, thus releasing NF-B for nuclear translocation and acti-vation of gene transcription (9 -11). Two IB kinases (IKK-1 or IKK␣ and IKK-2 or IKK␤), which specifically phosphorylate the critical serines in IBs, have been cloned and characterized by several laboratories (12)(13)(14)(15)(16)(17)(18)(19)(20)(21)(22). A third adapter protein, NEMO (NF-B essential modulator, also called IKK␥) is necessary for IKK phosphorylation and activation by upstream kinases (23)(24)(25). IKK-1 and IKK-2 are ubiquitously expressed in most human tissues and appear to be the converging point in the activation of NF-B by a diverse array of agonists (15,19).
IKK-1 is an 85-kDa, 745-amino acid protein that contains an N-terminal serine/threonine kinase catalytic domain, a leucine zipper-like amphipathic helix, and a C-terminal helix-loop-helix domain (13,26). IKK-2 is an 87-kDa, 756-amino acid protein with the same overall structure as IKK-1. IKK-1 and IKK-2 are 52% identical overall, with 65% identity in the kinase domain and 44% identity in the C-terminal region, which contains the leucine zipper and helix-loop-helix domains. The kinase activities of IKK-1 and IKK-2 are regulated by phosphorylation. Both enzymes contain canonical mitogen-activated protein kinase kinase (MAPKK) activation loops, which are the targets for phosphorylation and hence activation by upstream kinases (14,28,29). Several experimental approaches from different laboratories have indicated that IKK-2, rather than IKK-1, is essential for NF-B activation in response to a wide range of inflammatory and stress stimuli including tumor necrosis factor ␣ and interleukin-1␤ (14,21,29). Additionally, IKK-2 also demonstrates a significantly more potent kinase activity using IB␣ or IB␤ as substrates.
Recently, two homologs of IKK-1 and IKK-2 have been described, called IKK-i (also known as IKK-⑀) and TBK-1 (also known as T2K or NAK), and activation of either of these kinases results in NF-B activation (30 -34). IKK-i and TBK-1 show 30% amino acid identity to IKK-2 in the N-terminal kinase domain and an overall similar topology in the C terminus including a leucine zipper-like domain as well as a helixloop-helix region (30 -34). Additionally, the canonical activation loop motif of IKK-i and TBK-1 differ from IKK-2 with a glutamic acid being present instead of a serine (EXXXS) (30 -32). IKK-i and TBK-1 also differ from IKK-2 in several other respects. First, they both specifically phosphorylate serine 36 in IB␣ and not serine 32. Second, IKK-i is predominantly expressed in immune cells and tissues, including peripheral blood leukocytes, spleen, and thymus, respectively (30). TBK-1 is constitutively and ubiquitously expressed, with the highest level of expression in the testis (32). Third, mRNA for IKK-i, but not IKK-2, can be induced in multiple cell lines in response to several agonists including cytokines and lipopolysaccharide (30). Finally, IKK-i and TBK-1 show significant kinase activity when overexpressed in and isolated from nonstimulated mammalian cells, and both associate with TRAF/TANK proteins within the cell (35,36). In contrast, IKK-2 and IKK-1 demonstrate significant kinase activity only when isolated from mammalian cells that have been stimulated with agonists (14,21). Taken together, it is clear that the IKK isoforms activate NF-B via distinct mechanisms, have differential tissue distribution, associate within distinct signaling complexes in cells, and differ in their enzymatic properties as well as their substrate recognition.
Both homodimers and heterodimers of rhIKK-1 and rhIKK-2 have been purified to homogeneity and characterized enzymatically by a number of laboratories (14 -22). However, neither rhIKK-i nor rhTBK-1 has been purified and characterized. In this report, we have cloned, expressed in a baculovirus expression system, and purified rhIKK-i and rhTBK-1 to compare and contrast their enzymatic properties to each other and rhIKK-1 and rhIKK-2 homodimers. We show that rhIKK-i and rhTBK-1 exhibit a significantly higher catalytic activity compared with rhIKK-1 or rhIKK-2 using IB␣ as the substrate. However, the K m for IB␣ peptide with respect to IKK-i and TBK-1 is Ͼ40fold higher than the IB␣ K m for IKK-1 and Ͼ200-fold higher than that for IKK-2. In contrast, the K m for the MAPKK activation loop peptide of rhIKK-2 is significantly lower for both rhTBK-1 and rhIKK-i. We demonstrate that rhIKK-i and rhTBK-1, like rhIKK-1 and rhIKK-2, are phosphorylated during their expression and that this phosphorylation is necessary for kinase activity. We demonstrate by peptide mapping that both rhIKK-i and rhTBK-1 are phosphorylated on serine 172 in the MAPKK activation loop and that substitution of this serine to either alanine or glutamic acid results in decreased kinase activity. Additionally, we show that, as previously reported for rhIKK-1 and rhIKK-2 (22), ADP is a competitive inhibitor of both rhIKK-i and rhTBK-1 but with unique IC 50 values of inhibition. Using a panel of ATP and ADP analogs, we further demonstrate that the ATP-binding sites for rhIKK-i and rhTBK-1 are distinct from each other and from either rhIKK-1 or rhIKK-2. These observations suggest the possibility of developing selective inhibitors of each kinase to better understand their roles in NF-B activation. Thus, the characterization of rhIKK-i and rhTBK-1 defines unique biochemical properties that provide insight into their roles within the NF-B activation pathway.

Materials
All reagents used were of the highest grade commercially available. Anti-FLAG M2 affinity resin, FLAG peptide, antibodies specific for the FLAG peptide, ATP and ADP and their analogs, bovine serum albumin, Nonidet P-40, protease inhibitors, and dithiothreitol (DTT) were obtained from Sigma-Aldrich. Anti-TBK-1 antibody (M-375) was from Santa Cruz Biotechnologies. Recombinant protein phosphatase was from New England Biolabs. Peptides were either purchased from American Peptide Co. or made in the peptide synthesis laboratory at Pharmacia. [␥-33 P]ATP (2500 Ci/mmol) was purchased from Amersham Pharmacia Biotech. cDNA used to obtain full-length clones of IKK-i and TBK-1 and Advantage cDNA polymerase mix for PCRs were from CLONTECH (Palo Alto, CA).

Cloning and Expression
Cloning hIKK-i cDNA-Human T cell Jurkat cDNA was used as a template for PCR (Advantage cDNA polymerase mix) and primers (Sigma-Genosys, Houston, TX) homologous to the areas containing the initiation codon or the termination codon according to the published sequence of IKK-i (GenBank TM accession number AB016590). Those primers also contained a BamHI site at the 5Ј end of the forward oligonucleotide or a EcoRI site at the 5Ј end of the reverse oligonucleotide. The sequences of the oligonucleotides of IKK-i were: forward, 5Ј-ACGTACGGATCCATGCAGAGCACAGCCAATTACCTGTGGCACA-CAGATGA, and reverse, 5Ј-ACGTACGAATTCTTAGACATCAGGAGG-TGCTGGGACTCTATTTAGCCGTT.
The PCR was run on a 1% agarose gel, and a band of ϳ2 kb was gel-purified. The purified DNA was cut with BamHI and EcoRI and cloned into pFASTBAC (Invitrogen) containing an N-terminal FLAGcoding region (pMON46007) to yield pMON48029.
Cloning of hTBK-1 cDNA-Human placenta cDNA was used as a template for a PCR using elongase enzyme (Invitrogen) and primers (Midland Certified Reagent Company, Midland, TX) homologous to the areas containing the initiator methionine or the terminal stop codon according to the published sequence of TBK-1 (GenBank TM accession number AF191838.1). Those primers also contained a BamHI site at the 5Ј end of the forward oligonucleotide or a NotI site at the 5Ј end of the reverse oligonucleotide. The sequences of the oligonucleotides of TBK-1 were: forward, 5Ј-GATCGGATCCATGCAGAGCACTTCTAATCATCTG, and reverse, 5Ј-GATCGCGGCCGCTTAAAGACAGTCAACGTTGCGA-AGG.
The PCR was run on a 1% agarose gel, and a band of ϳ2 kb was gel-purified. The purified DNA was cut with BamHI and NotI and cloned into pFASTBAC (Invitrogen) containing an N-terminal FLAG coding region (pMON46007) in frame with the initiator methionine. An isolate was obtained, fully sequenced, and identified as pMON48028.
Site-directed Mutagenesis-The serine residues at positions 172 in IKK-i and TBK-1 were mutated to alanine or glutamic acid residues by two-step PCR-mediated mutagenesis. Briefly, the forward cloning oligonucleotide (described in the cloning sections) for each cDNA was used to prime a PCR along with the reverse mutant oligonucleotide containing the desired codon change. Simultaneously, the forward mutant oligonucleotide was used to prime a second PCR with the reverse cloning oligonucleotide as described in the cloning sections. Those products were gel-purified, and a fraction of each was used as a template in fresh PCRs containing the original cloning oligonucleotides. The products were gel-purified, restriction-digested with the appropriate enzymes, and ligated into the vectors described in the cloning sections. All oligonucleotides were from the Midland Certified Reagent Company, and the PCR reagents were from Invitrogen. The mutant oligonucleotides are shown below, where the letter following the letter S indicates the resulting amino acid residue at position 172. Entire cDNAs were sequenced to verify the desired mutations and to confirm that no other mutations had occurred. The mutant oligonucleotides are: IKK-i SEF, 5Ј-TGATGAGAAGTTCGTCGAAGTCTATGGGACTGAGGAG-3Ј; IKK-i SER, 5Ј-CTCCTCAGTCCCATAGACTTCGACGAACTTCTCATCA-3Ј; IKK-i SAF, 5Ј-TGATGAGAAGTTCGTCGCGGTCTATGGGACTGAGG-AG-3Ј; IKK-i SAR, 5Ј-CTCCTCAGTCCCATAGACCGCGACGAACTTC-TCATCA-3Ј; TBK-1 SEF, 5Ј-TGATGAGCAGTTTGTTGAACTGTATGG-CACAGAAG-3Ј; TBK-1 SER, 5Ј-CTTCTGTGCCATACAGTTCAACAA-ACTGCTCATCA-3Ј; TBK-1 SAF, 5Ј-TGATGAGCAGTTTGTTGCTCTG-TATGGCACAGAAG-3Ј; and TBK-1 SAR, 5Ј-CTTCTGTGCCATACAG-AGCAACAAACTGCTCATCA-3Ј.
Insect Cell Expression-TBK-1 and IKK-i were expressed in Sf9 insect cells using the commercially available Bac-to-Bac TM baculovirus expression system (Invitrogen). Following the manufacturer's suggested protocols, donor plasmids pMON48028 and pMON48029 were used to generate recombinant baculovirus Bacmid DNAs containing human TBK-1 and IKK-i, respectively, via transposition in Escherichia coli. Purified bacmid miniprep DNAs were checked by restriction digestion for the presence of the correct inserts and then used to transfect Sf9 insect cells for production of recombinant viruses. Transfection was accomplished using Cell Fectin TM reagent (Invitrogen) following the standard protocol for Sf9 cells. Briefly, a transfection mixture containing 5 l of the appropriate miniprep DNA and 5 l of Cell Fectin in 1 ml of SF-900 serum-free medium was added to cells that had been seeded in a 6-well tissue culture plates at 9 ϫ 10 5 cells/well. After 5 h, the mixture was removed, and the cells were fed 3 ml of complete IPL-41 medium (Invitrogen). After 3 days, the medium containing recombinant virus was cleared of cell debris and stored at 4°C. Infecting fresh Sf9 cells with a portion of each transfection medium made larger virus stocks, designated P1. The larger stocks were titered by plaque assay and used for production of recombinant proteins. To generate material for purification, fresh Sf9 cells at ϳ1 ϫ 10 6 cells/ml were infected with virus at a multiplicity of infection of 0.1. The infections were allowed to proceed for 72 h, then the cells were harvested, and the resulting pellets were stored at Ϫ80°C until needed.
Isolation of rhIKK-i-Cells from a 20-liter fermentation of baculovirus-infected cells expressing N-terminal FLAG-tagged IKK-i were microfluidized and centrifuged at 26,000 ϫ g for 1 h. The supernatant was collected, and the pH was adjusted to 7.6 with sodium hydroxide. 25 ml of anti-FLAG M2 affinity gel, pre-equilibrated in buffer A, was added to the supernatant pool and allowed to mix overnight. The resin was batch-washed using four resin volumes of buffer A/wash, poured into a 26/20 column, and washed with 15 resin volumes of buffer B. The FLAG-tagged IKK-i was eluted using FLAG peptide in buffer A. Dithiothreitol was added to the pool to a final concentration of 5 mM, followed by concentrating in a stirred cell using an Amicon YM 30 membrane. Bovine serum albumin was added to the concentrated pool to a final concentration of 0.1%, which was then frozen at Ϫ80°C in aliquots. Recombinant hIKK-i (Ser 3 Glu and Ser 3 Ala) mutants from a 2-liter fermentation were isolated following procedures described above.
Isolation of rhTBK-1-Cells from a 20-liter fermentation of baculovirus-infected cells expressing FLAG-tagged TBK-1 were microfluidized, adjusted to pH 7.6, and centrifuged as described for IKK-i. 50 ml of anti-FLAG M2 antibody resin was added to the pool and allowed to mix overnight. The resin was batch-washed using 4 bed volumes of buffer A/wash, then poured into a 26/20 column, and washed with 15 volumes of buffer B. The TBK-1 protein was eluted with FLAG peptide in buffer A. Dithiothreitol was added to the pool to a final concentration of 5 mM followed by concentration with an Amicon YM 30 membrane. Bovine serum albumin was added to the concentrated pool to a final concentration of 0.1%, and the sample was aliquoted and stored at Ϫ80°C until use. Recombinant hTBK-1 (Ser 3 Glu and Ser 3 Ala) mutants from a 2-liter fermentation were isolated following similar procedures.

SDS-PAGE and Western Blotting
Purified samples of rhIKK-i and rhTBK-1 were analyzed by SDS-PAGE (4 -12% Bis Tris NuPAGE gel) run in MES buffer. The proteins were detected by Coomassie stain. For Western blot analyses, the proteins were transferred to nitrocellulose membranes (Novex) and detected by chemiluminescence (SuperSignal) using anti-FLAG antibody or anti-TBK-1 antibody for IKK-i and TBK-1, respectively.

Phosphatase Treatment
Phosphatase treatment was as previously described for rhIKK-1 and rhIKK-2 (22). Briefly, 4 -8 g of rhIKK-i or rhTBK-1 was immunoprecipitated with 40 g of FLAG antibody followed by coupling to protein G-Sepharose beads. Immunoprecipitated rhIKK enzymes were washed and resuspended in 50 mM Tris-HCl, pH 7.6, containing 0.1 mM EDTA and 2 mM MnCl 2 . Following an incubation with recombinant protein phosphatase (800 units) at room temperature for 30 min, cold lysis buffer (20 mM HEPES, pH 7.6, containing 0.5% Nonidet P-40, 50 mM NaCl, 1 mM EDTA, 1 mM EGTA, 5 mM DTT, 1 mM sodium orthovanadate, and 10 mM ␤-glycerophosphate) was added to the samples to stop the reaction. After several washes, 10% of the beads were removed for Western blot analysis, and the remaining material was pelleted and resuspended in 100 l of the kinase buffer used for in vitro enzyme assay.

Phosphopeptide Analysis
IKK-i was concentrated and dialyzed into 6 M guanidine, 0.5 M Tris, pH 8.5, 5 mM dithiothreitol for 1 h. The free cysteines were alkylated with 15 mM iodoacetic acid on ice in the dark for 30 min as described (37). Alkylated protein was dialyzed against 2 M urea in 0.01 M Tris, pH 8.0. Protein was digested with trypsin (Promega, Madison, WI) at a 60:1 substrate to enzyme ratio. The resulting peptides were separated by reversed-phase HPLC using a 130 A microbore HPLC system (Applied Biosystems, Foster City, CA). 1 ϫ 150 mm Vydac (Hesperia, CA) C-18 300, a reversed-phase column was used for the separation, employing a linear gradient of acetonitrile/water with 0.1% trifluoroacetic acid. The fractions were collected and analyzed by matrix-assisted laser desorption/ionization (MALDI) mass spectrometry using a Voyager DE-RP time-of-flight mass spectrometer (PerSeptive Biosystems, Farmingham, MA). A 0.25-l aliquot of the collected fraction was mixed at a 1:4 ratio with matrix (50 mM solution of ␣-cyano-4-hydroxycinnamic acid) and allowed to dry. The MALDI mass spectrometry data were collected in the reflectron mode, using an average of 100 scans. The measured masses of the tryptic peptides were compared with the expected masses based on the known amino acid sequence. Potentially phosphorylated peptides were identified by the increase of 80 Da or multiples of 80 Da over the expected mass of a given tryptic peptide. To confirm the identity of the phosphopeptides and to determine the site of phosphorylation, the phosphopeptide-containing fractions were subjected to tandem mass spectrometry using a Q-T mass spectrometer (Micromass, Inc., Beverly, MA) fitted with a nano-electrospray ionization source (38,39). 1-3 l of the sample was transferred into a PicoTip glass nanospray emitter with a 2 m-inner diameter tip (New Objective, Woburn, MA). Up to 20 min of fragmentation data were collected and averaged for each sample.

Enzyme Assay
Kinase activity was measured essentially as described previously for IKK-2 (22) using a 23-amino acid residue peptide derived from IB␣ (Gly-Leu-Lys-Lys-Glu-Arg-Leu-Leu-Asp-Asp-Arg-His-Asp-Ser 32 -Gly-Leu-Asp-Ser 36  Following incubation at 25°C for 30 min, 33 P-IB␣ was separated from unincorporated [ 33 P]ATP by the transfer of a suitable aliquot of the reaction to a SAM 96 biotin capture plate, followed by sequential washes with 2 M NaCl and 2 M NaCl containing 1% H 3 PO 4 essentially as described (22). The plates were dried and counted in a Top-Count NXT after the addition of 25 l of Microscint 20. Alternatively, the reaction was stopped by the addition of 150 l of AG1XB resin in 900 mM sodium formate buffer, pH 3 (the resin is in a slurry of 1 volume resin to 2 volume of sodium formate buffer). The resin was allowed to settle, and 50 l of supernatant was transferred to a top count plate followed by the addition of 150 l of Microscint 40, mixed well, and counted. Both assays gave results similar to those reported earlier (22), although the background was lower with the biotin capture method. The resin capture assay was used in experiments where nonbiotinylated peptide substrates were used. Biotinylated IB␣ peptide was used in both assays and yielded similar values. One unit of enzyme activity is defined as the amount of enzyme required to catalyze the transfer of 1 nmol of phosphate from ATP to the peptide substrate/min. Specific activity is expressed as units/mg of protein.
For K m determination of purified enzymes, various concentrations of ATP or IB␣ were used in the assay at a fixed concentration of the second substrate. For ATP K m , the assays were carried out with 15 or 20 ng of IKK-i or TBK-1, respectively, with IB␣ at 500 and 750 M, respectively, and ATP was varied from 0.31 to 10 M. For IB␣ K m , ATP was fixed at 10 M for IKK-i and 15 M for TBK-1, and IKB␣ concentration was varied from 31 to 1000 M. Experiments involving IKK-2 were essentially as described earlier (22).

RESULTS AND DISCUSSION
IKK-i and TBK-1 have recently been described as isoforms of IKK-1 and IKK-2 based on similarities in their overall structure and sequence. Although rhIKK-1 and rhIKK-2 have been purified to homogeneity and enzymatically characterized by many investigators, the properties of rhIKK-i and rhTBK-1 have not been described thus far. To better understand the enzymatic similarities and differences between these four IKK isoforms, we have cloned, expressed, and purified rhIKK-i and rhTBK-1 and compared their enzymatic properties to those of rhIKK-1 and rhIKK-2.
Both rhIKK-i and rhTBK-1 containing a FLAG epitope tag at the N terminus were expressed in a baculovirus system and purified to homogeneity. Single bands at the expected molecular masses between 80 -85 kDa were confirmed by SDS-PAGE followed by Coomassie staining, and the identities of rhIKK-i and rhTBK-1 were verified by Western blot analysis (Fig. 1A). Both rhIKK-i and rhTBK-1 exhibited a pH optimum between 7 and 8, similar to all other kinases. Purified enzymes were stable for at least 9 months at Ϫ80°C when stored in buffer containing 0.1% bovine serum albumin, 0.1% Nonidet P-40, 10% glycerol, 5 mM DTT, and protease inhibitors (data not shown).
Like IKK-1 and IKK-2, TBK-1 and IKK-i are IB kinases, because they are able to phosphorylate IB␣ as a substrate. We determined the kinetic parameters of rhIKK-i and rhTBK-1 using a 23-amino acid peptide encompassing residues 19 -41 of IB␣ as the phosphoacceptor peptide and compared them with those of rhIKK-1, rhIKK-2, and rhIKK-1/IKK-2 heterodimer ( Fig. 1B and Table I). Both rhIKK-i and rhTBK-1 had a dramatically higher enzymatic activity compared with rhIKK-2 under comparable assay conditions (Fig. 1B). Table I compares the kinetic parameters of rhIKK-i and rhTBK-1 to our previously published findings on rhIKK-2 homodimer, rhIKK-1 homodimer, and rhIKK-1/IKK-2 heterodimer. The values for kinetic parameters shown in Table I depict data generated concomitantly with those of IKK-i and TBK-1 to compare values under identical assay conditions. Note that the K cat and specific activity values reported here for IKK-1 and IKK-2 and IKK-1/IKK-2 heterodimer are 2-3-fold higher than our previously published values generated with early enzyme preparations. Further optimization of storage conditions have resulted in increased stability of all IKK isoforms. Table I shows several key findings. First, the K cat for rhIKK-i and rhTBK-1 are Ͼ40 -50-fold higher than either rhIKK-2 homodimer or rhIKK-2/IKK-1 heterodimer and Ͼ300-fold higher than rhIKK-1 homodimer. Second, IB␣ is a significantly better substrate for rhIKK-2 homodimer and rhIKK-1/ IKK-2 heterodimer with more than a 100-fold lower K m compared with rhIKK-i and rhTBK-1. Comparable results were obtained when IB␣-GST fusion protein was used as a substrate (data not shown). Thus, the catalytic efficiency (K cat /K m ) using IB␣ peptide as the substrate is 8 -10-fold higher for rhIKK-2 and rhIKK-1/IKK-2 heterodimer compared with rhIKK-i and rhTBK-1. This is an intriguing result and may indicate that IB␣ is not a preferred physiological substrate for either rhIKK-i or rhTBK-1. The data in Table I also confirm that the catalytic efficiency of rhIKK-1 is the lowest of the four isoforms by virtue of its low K cat and its 5-10-fold higher K m for the IB␣ peptide compared with rhIKK-2. Interestingly, Zandi et al. (40) have shown that when full-length IB␣ is used as a substrate, the K m values for IKK-1 and IKK-2 are comparable. However, the catalytic efficiency of both rhIKK-2 and rhIKK-1 is reported to be significantly higher when IB␣-NF-B complex is used as a substrate instead of free IB␣. Thus, IB␣ may be a significantly better substrate for rhIKK-i and rhTBK-1 when it is associated with other cellular proteins such as NF-B.
The MAPKK activation loops of IKK-1 and IKK-2 are similar, each containing the SXXXS motif with Ser 176 and Ser 180 and with Ser 177 and Ser 181 being the phosphoacceptor serines, respectively. This sequence differs from that in either IKK-i or TBK-1, which contain an EXXXS motif, where the phosphoacceptor serine is Ser 172 (Fig. 2A). Thus, we wanted to evaluate the role of phosphorylation of the MAPKK activation loop in the regulation of IKK-i and TBK-1 kinase activity (Fig. 2). Both rhTBK-1 and rhIKK-i had phosphatase-sensitive kinase activity (Fig. 2B). These data are similar to that of rhIKK-2 shown here and previously reported for rhIKK-1, rhIKK-2, and rhIKK-1/IKK-2 heterodimer (22). Note that the decrease in kinase activity after phosphatase treatment was due to dephosphorylation because heat treatment of the phosphatase abolished its ability to inactivate rhIKK-i and rhTBK-1. Likewise, the constitutively active rhIKK-2 (S177E,S181E) did not demonstrate a loss of activity with phosphatase treatment, indicating a specific effect of phosphatase on the susceptible phosphorylated IKK isoforms. Similar results were obtained with the rhIKK-i (S172E) mutant enzyme, in that the kinase activity was not sensitive to phosphatase treatment (data not shown). The Western blot analysis confirms that the amount of kinase remained constant during the phosphatase treatment and the subsequent kinase assay. To identify the specific phosphorylated residue, we subjected rhIKK-i and rhTBK-1 to tryptic digestion followed by MALDI mass spectrometry to identify the phosphopeptide. Tandem mass spectrometry of the phosphopeptide-containing fractions identified Ser 172 as the phosphorylated residue in rhIKK-i (Fig. 2C) and rhTBK-1 (data not shown). Thus, like rhIKK-1 and rhIKK-2, rhTBK-1 and rhIKK-i are activated by phosphorylation of their MAPKK activation loops during expression.
To confirm the role of phosphorylation of IKK-i and TBK-1 in the regulation of kinase activity, we constructed mutants of IKK-i and TBK-1 in which serine 172 was replaced with either alanine or glutamic acid. Each mutant including rhIKK-i (S172A), rhIKK-i (S172E), rhTBK-1 (S172A), and rhTBK-1 (S172E) was cloned, expressed, and purified to homogeneity. The enzymatic properties of the purified mutant enzymes were compared with the respective wild type enzymes under comparable assay conditions (Fig. 3). Neither rhIKK-i (S172A) nor rhTBK-1 (S172A) had kinase activity (data not shown). These results are in agreement with the results reported by Shimada et al. (30) and Peters et al. (31), who showed that IKK-i (S172A), when overexpressed and immunoprecipitated from a mammalian system, had no kinase activity. Our results further extend these observations to TBK-1 (S172A), which has not been reported thus far. In contrast to the S172A constructs, S172E mutants of rhIKK-i and rhTBK-1 had kinase activity, although the specific activity was reduced Ͼ100-fold compared with wild type rhIKK-i (Fig. 3A) or wild type rhTBK-1 (Fig. 3B). However, the K m values for ATP and IB␣ of the rhIKK-i (S172E) and rhTBK-1 (S172E) mutants were comparable with the respective wild type enzymes (Table II). Thus, the reduced specific activity of the mutant (S172E) enzymes is mainly due to the rate of catalysis rather than altered substrate binding. These results are similar to our previous finding with rhIKK-2 (S177E,S181E) reported earlier and shown in Table II  of rhIKK-i (S172E), these results may still be in agreement with those reported by Shimada et al. (30), who showed a loss of kinase activity of the S712E mutant, and Peters et al. (31), who showed a highly reduced kinase activity of the IKK-i (S172E) mutant enzyme. Note that purified rhIKK-i (S172E) and rhTBK-1 (S172E) mutants have 68-and 226-fold lower activity, respectively, than their respective wild type kinases. Thus, this decreased level of kinase activity may not have been detectable using the immunoprecipitated kinase in an SDS-PAGE assay reported previously (30,31). Interestingly, the specific activity of both rhIKK-i and rhTBK-1 with the (S172E) mutation have higher kinase activities than the comparable mutation of IKK-2 (S177E, S181E). However, all of the Ser 3 Glu mutants have decreased activities compared with their respective wild type kinases.
Both rhIKK-i and rhTBK-1 overexpressed in mammalian cells have been shown to specifically phosphorylate serine 36 of IB␣, whereas IKK-1 and IKK-2 phosphorylate both serine 32 and serine 36 (15, 30 -34). This suggests that the peptide substrate requirements for the IKK isoforms may be different. Table III shows several unique differences in the peptide substrate specificity of rhIKK-i and rhTBK-1 compared with rhIKK-2. We did not evaluate these substrates against IKK-1 because of its low specific activity and thus low signal to noise ratio in the resin assay. First, rhIKK-i and rhTBK-1 specifically phosphorylate serine 36 of IB␣ with K m and K cat values comparable with the parent IB␣ peptide (Table III, peptide  series A). Thus, peptide variants in which serine 36 was substituted with alanine (Ser 32 -Ala 36 ) or with a phosphoserine (Ser 32 -Ser(P) 36 ) were not substrates for rhIKK-i and rhTBK-1 but were comparable substrates to the parent (Ser 32 -Ser 36 ) peptide with respect to rhIKK-2. In contrast, serine 32 peptide variants containing alanine (Ala 32 -Ser 36 ) or phosphoserine (Ser(P) 32 -Ser 36 ) were efficient substrates for rhIKK-i, rhTBK-1, and rhIKK-2 with catalytic efficiency comparable with that of the wild type (Ser 32 -Ser 36 ) peptide. Interestingly, peptide variants containing a phosphoserine at either position 32 or 36 had comparable catalytic efficiency with respect to rhIKK-2, because the K cat /K m ratio were 94 h Ϫ1 M Ϫ1 and 62 h Ϫ1 M Ϫ1 , respectively. These data suggest that the phospho-  2. Phosphorylation of rhIKK-i and rhTBK-1 is required for kinase activity; phosphorylation occurs on serine 172 in the activation loop. A, MAPKK activation loop sequence comparison between IKK-1, IKK-2, IKK-i, and TBK-1. B, FLAG-tagged rhIKK-i, rhTBK-1, and rhIKK-2 and mutated rhIKK-2 (rhIKK-2(SS-EE)) were immunoprecipitated using anti-FLAG antibody. WT, wild type. The immunoprecipitated proteins were treated with buffer (open bars), heat-inactivated recombinant protein phosphatase (hatched bars), or active recombinant protein phosphatase (filled bars) for 30 min. Kinase activity of phosphatase-treated enzyme was expressed as a percentage of kinase activity of the respective buffer-treated control. Western blot analyses post-immunoprecipitation are also shown. C, MALDI mass spectrometric scan of the principal phosphopeptide identified from tryptic digest of rhIKK-i. The HPLC fraction was mixed at 1:4 ratio of the matrix, and MALDI mass spectrometry data were collected in the reflectron mode, using an average of 100 scans. The measured mass (shown in italics) of the tryptic peptide (peak A) was compared with the calculated mass (shown in bold) based on the known amino acid sequence. Phosphorylated peptides (peak B) were identified by a 80-Da mass increase over the expected mass of the tryptic peptide. The site of phosphorylation was confirmed by subjecting the phosphopeptide-containing fraction to tandem mass spectrometry. The confirmed sequence of the phosphopeptide is shown, and the residue with the asterisk is identified to be phosphorylated. rylation of serine 32 and serine 36 occurs in random order. If phosphorylation occurred sequentially, one of the phosphorylated peptides would be expected to have a significantly higher catalytic efficiency. Likewise, substitution of serine 32 with either alanine or phosphoserine had minimal effect on the catalytic efficiency with respect to the specific phosphorylation of serine 36 by rhIKK-i and rhTBK-1. Thus, serine 32 in the 23-amino acid IB␣ peptide affects neither binding nor catalysis for rhIKK-i and rhTBK-1.
The second unique difference in the peptide substrate specificity of the three IKK isoforms is that short, truncated (7 residues) peptides derived from the minimal phosphorylation motif of IB␣ were efficient substrates for rhIKK-i and rhTBK-1 but were not substrates for rhIKK-2, up to a concentration of 1 mM (Table III, peptide series B). The K cat /K m ratio for rhIKK-i with respect to the short peptide represented by the sequence, DSGLDSM was 8.7 h Ϫ1 M Ϫ1 and thus comparable with the K cat /K m ratio for the IB␣ parent peptide, which was 9.7 h Ϫ1 M Ϫ1. Likewise, the K cat /K m ratio of the truncated peptide for rhTBK-1 was 2.6 h Ϫ1 M Ϫ1 compared with 6.4 h Ϫ1 M Ϫ1 for the parent IB␣ peptide. Additionally, both rhIKK-i and rhTBK-1 maintained their specificity for serine 36, even in the truncated peptide. However, unlike peptide series A, when phosphoserine was substituted for serine 32 in peptide series B, it was no longer a substrate for rhIKK-i or rhTBK-1. This may be because the additional negative charge contributed by the phosphoserine residue in the truncated peptide may interfere with binding or phosphotransfer to the acceptor serine 36. In contrast to the 7-amino acid truncated IB␣ peptide, the larger 23-amino acid IB␣ peptide has several positively charged amino acids such as lysines and arginine that could compensate for a negatively charged phosphoserine residue.
To further define peptide substrate specificity, we tested a dozen diverse peptides containing serines that were substrates for other kinases (e.g. c-Jun peptide, HSP-27, CREB-tide, Fos peptide, etc.), and none of these peptides were substrates for either rhIKK-i, rhTBK-1, or rhIKK-2 (data not shown). Finally, we examined a 17-amino acid peptide (residues 170 -187) derived from the activation loop of IKK-2. This peptide was indeed a more efficient substrate than IB␣ for both rhIKK-i and rhTBK-1 (Table III,    We have previously demonstrated that ADP is a competitive inhibitor of rhIKK-2 at the ATP site with an IC 50 value in the range of 1-2 M. ADP also inhibits both rhIKK-i and rhTBK-1, but with different IC 50 values (Fig. 4A). The Ͼ10-fold difference in the IC 50 of ADP inhibition against rhIKK-i and rhTBK-1 is interesting given the fact that the K m for ATP is comparable for the two enzymes. Like IKK-2, ADP is competitive at the ATP site for both rhIKK-i as well as rhTBK-1 (Fig. 4, B and C). To further extend these observations, we determined the IC 50 values for various ATP and ADP analogs for rhIKK-i and rhTBK-1 and compared them to previously reported results for rhIKK-1 and rhIKK-2 (Table IV). Note that the IC 50 values were comparable between rhIKK-2 and rhIKK-1. Interestingly, selective inhibition of rhIKK-i, rhTBK-1, and rhIKK-2 was noted with minor differences in the ATP analog structure. For example, ␣,␤-methylene-adenosine 5Ј-triphosphate and 2Ј-deoxyadenosine 5Ј-triphosphate showed comparable IC 50 values against rhIKK-2, rhIKK-i, and rhTBK-1, whereas 2,3dideoxyadenosine 5Ј-triphosphate was Ͼ10-fold more potent against rhIKK-i and rhTBK-1 compared with rhIKK-2. In contrast, adenylyl-imidodiphosphate was Ͼ10-fold more potent against rhIKK-2 than either rhIKK-i or rhTBK-1. Interestingly, ␤,␥-methylene adenosine triphosphate, which only differs from the close analog ␣,␤-methylene adenosine triphosphate with respect to the position of the methylene bridge, gave selective inhibition of rhTBK-1. Additionally, ADP analogs such as adenosine 5Ј-O-(2-thiodiphosphate), like ADP, were 10-fold more potent against rhIKK-2 compared with rhIKK-i and 100-fold more potent compared with rhTBK-1. These observations demonstrate that the ATP-binding sites of rhIKK-2, rhIKK-i, and rhTBK-1 are indeed unique and amenable for the design of selective inhibitors.
In summary, in the present report, we have cloned, expressed, and purified rhIKK-i and rhTBK-1 and compared FIG. 4. Differential inhibition of rhIKK-i, rhTBK-1, and rh1KK-2 by ADP. A, kinase activity of each enzyme was determined following incubation with increasing concentration of ADP using [␥-33 P]ATP and IB␣ peptide concentrations that were 2.5-3 times their respective K m values for each enzyme. Closed squares, IKK-2; closed triangles, IKK-i; closed circles, TBK-1. B, competitive inhibition of rhIKK-i by ADP with respect to the ATP site. The concentrations of ADP used were 6.25,12.5, and 25 M. C, competitive inhibition of rhTBK-1 by ADP with respect to the ATP site. The concentrations of ADP used were 50, 100, and 200 M. their enzymatic properties with those of rhIKK-2 and rhIKK-1.
Despite the overall structural similarity to IKK-1 and IKK-2, IKK-i and TBK-1 are enzymatically distinct. rhIKK-i and rhTBK-1 share several similarities. The catalytic rate of rhIKK-i and rhTBK-1 are 50 -100-fold higher than that of rhIKK-2. They clearly differ from rhIKK-2 with respect to their peptide substrate specificity. The fact that the best substrate identified for rhIKK-i and rhTBK-1 is the activation loop peptide of IKK-2 confirms several reports indicating that both of these kinases biochemically map upstream of IKK-2 in the NF-B activation pathway. Thus, both TBK-1 and IKK-i may act as IKK-2 activating kinases. Additionally, both IKK-i and TBK-1 interact with complexes of TANK and TRAF proteins (35,36). Baud et al. (41) previously demonstrated that the oligomerization of TRAF proteins was sufficient to activate the IKK signalosome complex. The ATP analog studies show distinct differences in the ATP sites of the three isoforms. This is indeed intriguing because the ATP K m values do not reflect these active site differences. Differential inhibition of rhIKK-2, rhIKK-i, and rhTBK-1 is particularly interesting, because ADP is a product inhibitor. The unique IC 50 values may reflect differences in enzyme mechanism, with potential differences in the product inhibition patterns. IKK-1 and IKK-2 have been demonstrated to exhibit a classical random sequential mechanism (17,18), but IKK-i and TBK-1 deviate from the classical random sequential mechanism. 2 Finally, the selective inhibition of the IKK isoforms with analogs of ATP opens up the possibility of using selective inhibitors of IKK-i and TBK-1 to understand their roles in NF-B activation.