Characterization of IκB Kinases

The NF-κB transcription factor is activated by a wide variety of stimuli, including phorbol esters such as 12-O-tetradecanoylphorbol-13-acetate. In its inactive state, NF-κB is sequestered in the cytoplasm tethered to an inhibitor protein, IκB. Activation comprises the rapid phosphorylation of IκB-α at N-terminal sites, which presumably marks IκB-α for proteolytic degradation and leads to release of NF-κB into the nucleus. In addition, IκB-α is constitutively phosphorylated at the C terminus, which may be a prerequisite for proper IκB function. Protein kinase C (PKC) is activated by 12-O-tetradecanoylphorbol-13-acetate and has been previously reported to phosphorylate IκB-α in vitro. As PKC has turned out to constitute a multigene family encoding isozymes with different biological functions, we have reinvestigated IκB-α phosphorylation by PKC using recombinant PKC isozymes expressed in insect cells. While crude PKC preparations were efficient IκB-α kinases, highly purified PKC isozymes completely failed to phosphorylate IκB-α. Biochemical separation of porcine spleen yielded at least two fractions with IκB-α kinase activity, both of which were devoid of detectable PKC isozymes. One peak contained both Raf-1 and casein kinase II (CKII). Purified Raf-1 does not phosphorylate IκB-α directly, but associates with CKII, which efficiently phosphorylates the C terminus of IκB-α. Two-dimensional phosphopeptide mapping and high pressure liquid chromatography-mass spectroscopy analysis showed that all IκB-α kinases induced phosphorylation at the same prominent sites in the C terminus. Our results clearly indicate that PKC isozymes α, β, γ, δ, ε, η, and ζ as well as Raf-1 are not IκB-α kinases. They furthermore demonstrate that IκB-α is targeted by several kinases, one of which appears to be CKII.

NF-B is composed of a dimer of related proteins belonging to the Rel superfamily (reviewed in Refs. 1 and 2). The induction of NF-B activity is an immediate-early event, when cells are exposed to inflammatory cytokines, such as tumor necrosis factor-␣ or interleukin-1, phorbol esters, e.g. TPA, 1 UV radiation, or hydrogen peroxide (reviewed in Ref. 3). The activation of NF-B is a unique paradigm for the regulation of a transcription factor by subcellular compartmentalization. In the inactive state, NF-B is complexed with cytosolic proteins collectively designated as inhibitors of NF-B, IB, which retain NF-B in the cytoplasm. The prototypic and best studied IB is IB-␣. Activation induces the phosphorylation of IB-␣, ubiquitination, and its subsequent proteolytic degradation (4 -6). As a consequence, NF-B can translocate to the nucleus and activate the transcription of target genes. The role of IB-␣ phosphorylation is not entirely defined, but recent work suggests that IB-␣ is phosphorylated on multiple sites located at the C and N termini (4, 6 -8). Phosphorylation of the C-terminal sites is constitutive (7,8), while phosphorylation of the N terminus has been suggested to be the target of inducible phosphorylation (4,6). Phosphorylation of these sites does not suffice to release NF-B, but rather seems to mark IB-␣ for degradation (5).
The central role of phosphorylation in the NF-B activation pathway has evoked an intense interest in the identification of kinases that phosphorylate IB-␣. A prime candidate for such kinases has been PKC. Phorbol esters, which activate PKC, are efficient inducers of NF-B, and PKC has been reported to phosphorylate IB in vitro (9). These studies were carried out at a time, however, when PKC was characterized mainly as a biochemical entity that could be activated by phorbol esters, phospholipids, and calcium. Molecular cloning of PKC cDNAs has shown that PKC is a multigene family comprising at least 10 genes, whose protein products differ in structure and biological effects (reviewed in Ref. 10). The classical PKC isozymes ␣, ␤I, ␤II, and ␥ feature all the properties of the initial biochemical description. The novel PKCs (PKC-␦, -⑀, -, -, and -) lack the calcium-binding domain and hence are calcium-independent. The atypical PKCs (PKCand PKC-) do not bind and respond to phorbol ester or calcium, but instead, at least PKCmay be activated by ceramide (11). In the cell, ceramide is generated in response to tumor necrosis factor-␣ and can mediate NF-B induction (12). Moreover, overexpression of individual PKC isozymes in NIH 3T3 fibroblasts and 32D promyelocytes has revealed very diverse biological effects. PKC-⑀ and PKC-can transform NIH cells, whereas PKC-␦ inhibits proliferation (13). 2 In 32D cells, PKC-␣ and PKC-␦, but not other isozymes, make these cells susceptible to differentiation into macrophages upon TPA treatment (14).
These observations prompted us to reinvestigate the phosphorylation of IB-␣ using recombinant PKC isozymes produced in the baculovirus/Sf9 cell expression system. Unexpectedly, highly purified PKC isozymes failed to phosphorylate IB-␣, while crude PKC preparations were active as IB-␣ kinases. The IB-␣ kinase activity could be separated from PKC by further purification. In concordance with these results, we could not detect PKC isozymes in IB-␣ kinase preparations from chromatographically fractionated porcine spleen extracts. These separations yielded two different peaks with IB-␣ kinase activity, one of which copurified with Raf-1 and CKII, which both have previously been shown to function as IB-␣   Fractions from a Mono-Q column (indicated at the top) were used in IB kinase assays and for Western blot analysis. Samples were resolved by SDS-PAGE and transferred to nitrocellulose membranes. The blots were stained with an antiserum recognizing Raf-1 (crafVI) or CKII. kinases (7,8). Purified CKII as well as the other IB-␣ kinase fractions phosphorylated the same sites in IB-␣ located at the C terminus.

MATERIALS AND METHODS
Expression and Purification of PKC Isozymes-PKC isozymes were purified from Sf9 cells infected with baculoviruses expressing the different PKCs (15). Purification of PKC isozymes was carried out essentially as described (16). In short, cell pellets of 3 ϫ 10 9 baculovirusinfected Sf9 cells were lysed in 100 ml of Tris buffer (pH 7.5) containing 1 mM EDTA, 2 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride, and 1% Triton X-100. The homogenate was centrifuged at 100,000 ϫ g for 60 min, and the clear supernatant was applied to a 20 ϫ 100-mm Q-Sepharose fast flow column. The column was eluted with a 500-ml linear gradient of 0 -500 mM NaCl. Fractions containing high PKC activity (as measured by histone phosphorylation) were pooled and applied to a 10 ϫ 100-mm hydroxylapatite column. The column was eluted with a 200-ml linear gradient of 0 -400 mM KH 2 PO 4 (pH 7.5) containing 2 mM dithiothreitol and 1 mM EDTA. Active fractions were pooled again, adjusted to 1 M KCl, and loaded onto a 5 ϫ 50-mm phenyl-Superose column. The column was eluted with a 50-ml linear gradient of 1 to 0 M NaCl, and the purity of the active fractions was assayed by SDS-PAGE. Fractions containing PKC as the major protein were pooled; dialyzed overnight against 20 mM Tris buffer (pH 7.5) containing 1 mM EDTA, 1 mM dithiothreitol, and 50% glycerol; and subsequently stored at Ϫ80°C.
Preparation of Raf-1 and IB-␣-Raf-1 was expressed in Sf9 cells as a GST fusion protein and purified as described (17). GST-IB-␣ was a kind gift from Dr. Ulli Siebenlist. GST-IB-␣ was prepared as described (17), cleaved with thrombin to remove the GST portion, and stored at Ϫ70°C.
Purification of IB-␣ Kinases from Spleen-5 g of frozen porcine spleen were homogenized in a mortar under liquid nitrogen. Purification of the IB-␣ kinases was carried out following a procedure similar to that described above for PKC, using a Mono-Q or Q-Sepharose fast flow column followed by a hydroxylapatite column.
Two-dimensional Peptide Maps-1 g of IB-␣ was phosphorylated as described above, and bands of phosphorylated IB-␣ were cut out from an SDS gel and processed for two-dimensional tryptic peptide mapping as described using pH 8.9 buffer for the first dimension and 1-butanol/pyridine/acetic acid/water (15:10:3:12) for the second dimension (17,18).
HPLC-Mass Spectroscopy Analysis-Analysis was performed on an ABI Model 172 micropurification system connected to a Perkin-Elmer API 100 quadrupol mass spectrometer and a Berthold radioactivity detector. IB-␣ was phosphorylated and digested as described above. Tryptic peptides from 5 g of IB-␣ were loaded onto a Pharmacia Biotech reversed-phase column. A gradient of 0 -35% acetonitrile in 100 min and subsequently of 35-70% in 10 min was run at 40 l/min.  Masses were determined in 0.1-atomic mass unit steps over an m/z range from 500 to 3000 with an orifice voltage of 20 V.
In-gel Kinase Assays-Protein was resolved on a 10% SDS-polyacrylamide gel containing 0.2 mg/ml IB-␣. After electrophoresis, the gel was washed twice for 10 min with 250 ml of 50 mM HEPES (pH 7.4), 5 mM 2-mercaptoethanol, and 20% isopropyl alcohol. After equilibration for 1 h at room temperature with 250 ml of 50 mM HEPES (pH 7.4) and 5 mM 2-mercaptoethanol, the gel was denatured twice for 30 min at room temperature with 6 M guanidinium Cl in 50 mM HEPES (pH 7.4) and 5 mM 2-mercaptoethanol. The gel was subsequently renatured overnight at 4°C in 500 ml of 50 mM HEPES (pH 7.4) and 5 mM 2-mercaptoethanol containing 0.04% Tween. The gel was equilibrated for 30 min at 30°C in kinase buffer (25 mM HEPES (pH 7.4), 5 mM 2-mercaptoethanol, 10 mM MgCl 2 , and 90 M Na 3 VO 4 ) and incubated in kinase buffer containing 250 Ci of [␥-32 P]ATP (specific activity of 3000 Ci/mmol) for 1 h at 30°C. The gel was washed extensively four times for 4 h in 500 ml of 5% trichloroacetic acid and 10 mM sodium pyrophosphate and subsequently dried and exposed to x-ray film.
Western Blot Analysis-Proteins were resolved on a 10% SDS-polyacrylamide gel and subsequently electroblotted onto nitrocellulose filters. Nonspecific binding to nitrocellulose was blocked with 5% nonfat dry milk for 30 min, and the blots were subsequently incubated with primary antibody and, after extensive washing with Tris-buffered saline and 1% Tween 20, with secondary antibody for 1 h each. The primary antibodies used were either a monoclonal antibody against PKC-␣ (Upstate Biotechnology, Inc.) or antisera against classical PKCs (201, directed against the pseudosubstrate domain of the PKCs), PKC-␦ (19), c-Raf (17), or CKII (Upstate Biotechnology, Inc.). Secondary antibodies coupled to horseradish peroxidase were purchased from Dianova. Immunoreactive bands were visualized using the ECL system (Amersham Corp.).

RESULTS
Different PKC isozymes were produced in the baculovirus/ Sf9 cell system and purified by successive chromatography on Q-Sepharose fast flow, hydroxylapatite, and phenyl-Superose columns. Column fractions were monitored for PKC activity employing histone as substrate. Active fractions were pooled and further purified. The presence of PKC was confirmed by Western blotting with appropriate PKC isozyme-specific antibodies. In parallel, the purity was assessed by staining duplicate gels with Coomassie Blue. A routine purification of PKC-␣ is shown in Fig. 1. PKC-␣ preparations of Ͻ10% purity (Fig.  1A) efficiently phosphorylated recombinant IB-␣ purified from Escherichia coli (Fig. 1B). The IB-␣ kinase activity, however, could be separated from PKC-␣ by further purification, suggesting that it copurified with kinases that were distinct from PKC-␣. Additional evidence was obtained with the specific PKC inhibitor GF109203X, which did not affect the IB-␣ kinase activity, but reduced the histone kinase activity by Ͼ95% (data not shown).
Similar results were obtained with PKC-␤II, -␥, -␦, -⑀, -, and -. The IB-␣ kinase activity of these PKC preparations could be removed by enrichment of PKC above 50% (by phenyl-Superose chromatography, step 3 in the purification protocol). A representative assay is shown in Fig. 2. Purified preparations of PKC-␣, -␦, and -efficiently phosphorylated histone or myelin basic protein (MBP), but completely failed to phosphorylate 38-kDa IB-␣. To exclude the possibility that the IB-␣ preparations contained an unspecific PKC inhibitor, a mixture of histone, MBP, and IB-␣ was used as substrate. IB-␣ was not phosphorylated and did not reduce the phosphorylation of histone or MBP by PKC.
To corroborate these findings, we attempted to purify IB-␣ kinases from porcine spleen. Spleen lysates were separated by chromatography on a HiLoad-Q-Sepharose column (Fig. 3A). Individual fractions were assayed for kinase activity using IB-␣ as substrate. Two peaks of IB-␣ kinase activity were recovered, pooled, and further fractionated on a hydroxylapatite column (Fig. 3B). Their elution profiles differed from those of PKC-␣, -␤, and -␥, which were determined in parallel experiments.
The peaks containing IB-␣ kinases were examined for the presence of PKC by kinase assays using histone as substrate and by Western blotting with anti-PKC antibodies (data not shown). Both methods failed to detect appreciable amounts of PKC in the IB-␣ kinase fractions, confirming the results obtained in the course of purification of PKC isozymes expressed in insect cells. As Raf-1 and CKII have been reported to phosphorylate IB-␣, the Western blots were also tested for the presence of these two kinases. As shown in Fig. 4, pool II contained readily detectable amounts of Raf-1 and CKII. While Raf-1 was also present in fractions that were devoid of appreciable IB-␣ kinase activity, CKII was present only in the fractions of the second peak containing IB-␣ kinase activity.
To assess the question of which of these two enzymes does in fact phosphorylate IB-␣, we used purified CKII (Upstate Biotechnology, Inc.) and purified Raf-1 (17). As shown in Fig. 5, both enzyme preparations contained IB-␣ kinase activity. The addition of heparin, a CKII inhibitor (20), completely abolished the IB-␣ kinase activity of both the Raf-1 and CKII preparations. As shown in Fig. 5, the Raf-1 kinase activity against Mek was not inhibited at all with heparin, while CKII activity using casein as substrate was abolished in the presence of heparin. Furthermore, the Raf-1 preparations, although pure as judged by Coomassie Blue staining (Ref. 17 and data not shown), did contain detectable amounts of CKII, as can be seen in the Western blot analysis shown in Fig. 5. These results suggest that CKII associates with Raf-1 under the conditions used and that the associating CKII, but not Raf-1, phosphorylates IB-␣.
To eliminate association of CKII with Raf-1, the immobilized GST-Raf fusion protein was washed with radioimmune precipitation assay buffer (Tris-buffered saline containing 1% Triton X-100, 0.5% deoxycholate, and 0.1% SDS). As shown in Fig. 6, these additional washes resulted in the essentially complete absence of both CKII as well as IB-␣ kinase activity, while the Raf-1 kinase activity (using Mek as substrate) was unchanged. To further substantiate the finding that the CKII associating with Raf-1, and not Raf-1 itself, phosphorylates IB-␣, we assayed the Mek and IB-␣ kinase activity of both wild-type and kinase-negative Raf-1. Both wild-type and kinase-negative preparations of Raf-1 contained roughly equal amounts of IB-␣ kinase activity, while the Mek kinase activity was present only in the wild-type Raf-1 preparation, but was completely absent in the kinase-negative mutant (Fig. 7).
As several protein kinases can at least be partially renatured after SDS electrophoresis and their presence detected in in-gel kinase assays, we attempted to visualize IB-␣ kinase activity in in-gel kinase assays using SDS gels that were polymerized in the presence or absence of recombinant IB-␣. The results of these experiments are shown in Fig. 8. While the presence of CKII was readily detectable using this approach in fractions of pool II of our IB-␣ kinase preparations as well as in Raf-1 immunoprecipitates, no activity could be detected in pool I. These experiments further supported that CKII efficiently phosphorylates IB-␣, but unfortunately, they failed to help identify the IB-␣ kinase present in pool I.
These results demonstrate that IB-␣ serves as substrate for at least two different kinases. Therefore, the pattern of phosphorylation sites induced by these kinases was compared by two-dimensional phosphopeptide mapping (Fig. 9). IB-␣ was phosphorylated by crude PKC-␣ preparations, purified Raf-1, CKII, or the IB-␣ kinases purified from spleen; isolated by SDS-PAGE; and digested with proteases. Phosphopeptides were resolved on thin-layer cellulose plates by electrophoresis at pH 8.9 in the first dimension followed by ascending chromatography in the second dimension and autoradiographed. The standard digestion protocol with trypsin proved unsatisfactory, resulting in smeary maps (data not shown), probably due to the large size of the tryptic phosphopeptide. Therefore, a combination of trypsin and Asp-N endoproteases was used for subsequent experiments, which allowed the resolution of two major phosphopeptides. All IB-␣ kinases ( Fig. 9) yielded identical phosphopeptide maps, indicating that a common set of phosphorylation sites is targeted by different kinases.
To identify the phosphorylation sites in IB-␣, we performed HPLC-mass spectroscopy analysis of the tryptic digests. Digestion with trypsin yielded one radioactive peak that eluted at ϳ35% acetonitrile (Fig. 10) and comigrated with a peptide with a mass of 5917 Da. This corresponds to the mass of the Cterminal tryptic peptide IQQQ . . . (amino acids 265-314, 5906 Da). Digestion with trypsin and Asp-N yielded two radioactive peaks (Fig. 10). The major mass detected in the first peak was 2935 Da, with a second peak at 3015 Da, which was absent in the unphosphorylated preparation. This corresponds exactly to the peptide IQQQLGQLTLENLQMLPESEDEESY (amino acids 265-289) and its singly phosphorylated form, respectively. The masses found in the second peak, however (2015 and 2175 Da, the latter only present in phosphorylated IB-␣), could not be correlated with any of the predicted masses from the IB-␣ sequence. Since both peptides resulting from the additional Asp-N digest must be derived from the only radiolabeled peak in the trypsin digest, they must lie within the large tryptic peptide at amino acids 265-314. To investigate this assumption, we generated a GST-IB-␣ mutant protein with a stop codon at amino acid 236. This protein in fact was not phosphorylated by any of the IB-␣ kinases tested here (data not shown). Hence, the phosphorylation sites of all IB-␣ kinases described here must lie within amino acids 242-307, and no phosphorylation of Ser-32 or Ser-36 could be detected.
To prove the identity of the phosphorylated peptide (amino acids 265-288), we synthesized the identical peptide. This synthetic peptide was readily phosphorylated by CKII and comigrated exactly with the first peptide resulting from the trypsin and Asp-N digest of IB-␣ (Fig. 11). DISCUSSION PKC isozymes expressed in insect cells were purified and assayed as IB-␣ kinases. Enriched PKC preparations (ϳ10% pure) efficiently phosphorylated IB-␣, whereas highly purified PKCs, including PKC-, failed. This result was not due to a loss of PKC activity or to the presence of a PKC inhibitor in the IB-␣ preparations used. Furthermore, in crude preparations of PKC that still contained IB-␣ kinase activity, we could inhibit the histone and MBP kinase activity with a highly specific PKC inhibitor, while the activity of the IB-␣ kinase remained unchanged. In a complementary experiment, we partially purified IB-␣ kinases from porcine spleen. Fractions active as IB-␣ kinases were devoid of PKC, further proving that PKC is not an IB-␣ kinase. The latter experiments demonstrated that at least two IB-␣ kinase activities could be separated. One of the IB-␣ kinase peaks contained both Raf-1 and CKII; the latter was in fact identified as IB-␣ kinase. A Raf-1 mutant, BXB, rendered constitutively activated by deletion of the regulatory domain, has been shown to phosphorylate IB-␣ in vitro and to activate NF-B transcription in cells (17,21). While the in vitro phosphorylation of IB-␣ is certainly due to the presence of contaminating CKII, it still remains unclear how Raf-1 can initiate the down-regulation of IB and hence NF-kB activation if it does not phosphorylate IB.
From our studies, it is clear that IB-␣ can be targeted by at least two distinct kinases. This finding is not unexpected, given the pleiotropic modes of NF-B induction. These different signaling pathways seem to converge on the level of IB-␣ phosphorylation, as distinct kinases phosphorylate the same set of sites resolved on two-dimensional phosphopeptide maps of trypsin/Asp-N-digested IB-␣. Unfortunately, the kinase activity present in pool I could not be renatured after SDS electrophoresis under various different conditions. Hence, we are unable to even speculate on the nature of this second IB-␣ kinase.
We could not detect any phosphorylation of IB-␣ serines 32 and 36, which have been indirectly implicated as IB-␣ phosphorylation sites (4,6). Since kinases that phosphorylate these sites might only be active in, for example, TPA-stimulated cells, we also purified IB-␣ kinases from TPA-stimulated WEHI 3Z cells (data not shown). We again only obtained the two pools of IB-␣ kinases that were described above, but no activity that phosphorylated IB-␣ on other sites, including serines 32 and 36, could be detected. Although IB-␣ point mutants at serines 32 and 36 have been reported to be resistant to tumor necrosis factor-␣-induced degradation, direct phosphorylation of these sites has not yet been shown.
Recently, two laboratories reported that IB-␣ is constitutively phosphorylated by CKII (6,8), which is in good agreement with our data. The phosphorylation sites described for CKII match the sites we identified here. Although McElhinny et al. (8) show phosphorylation of IB-␣ by PKC, they also state that the efficiency is ϳ1000-fold less than with CKII, suggesting that the phosphorylation actually results from impurities in the PKC preparation. The authors were, as we were, unable to detect phosphorylation of IB-␣ at the N terminus, including serines 32 and 36. It is tempting to speculate that the phosphorylation of IB-␣ at the C terminus is required for an additional phosphorylation at serines 32 and 36. Since the IB-␣ protein used to detect IB-␣ kinases was unphosphorylated, the "inducible" phosphorylation at the N terminus would not be detected. In conclusion, our data indicate that PKC isozymes and Raf-1 are not IB-␣ kinases; however, Raf-1 associates with an IB-␣ kinase, CKII.