Identification of Signal-induced IκB-α Kinases in Human Endothelial Cells

Activation of the nuclear transcription factor-κB is an early event in endothelial activation. NF-κB activation is regulated by the inducible phosphorylation and subsequent degradation of the inhibitory subunit IκB-α. We identified two discrete kinases of approximately 36 and 41 kDa in the cytoplasm of human umbilical vein endothelial cells that specifically bind to and phosphorylate the IκB-α subunit. IκB-α kinase activity is transiently elevated following treatment with either tumor necrosis factor α, interleukin-1β, or bacterial lipopolysaccharides and precedes activation of either mitogen-activated kinase or Jun kinase. Furthermore, activation of the IκB-α kinases precedes both the appearance of hyperphosphorylated IκB-α and its subsequent degradation, as well as the translocation of NF-κB to the nucleus. Deletion mutagenesis of the IκB-α polypeptide revealed that these kinases bind in or around the ankyrin repeat domains and phosphorylate residues within the C terminus. These kinases, however, were not identical to casein kinase II and displayed a pharmacologic profile distinct from other known kinases. These kinases may represent components of a signal transduction pathway regulating IκB-α levels in vascular endothelium.

The vascular endothelium occupies a critically strategic interface between blood and body, where it plays a role in essentially all aspects of normal physiology (1,2). Injury or activation of the endothelium disrupts normal regulatory properties and results in altered endothelial cell function. Under certain inflammatory conditions such as reperfusion injury, bacterial and viral infections, and various autoimmune diseases, endothelial cells exhibit an altered phenotype, facilitating leukocyte adhesion and diapedesis. In addition, the endothelial surface becomes conducive to coagulation and thrombosis, and barrier function is compromised. These changes are cumulatively referred to as endothelial activation (3). A hallmark of endothelial activation is the induced expression of a range of proinflammatory genes, including endothelial cell adhesion molecules such as P-selectin, E-selectin, VCAM-1, and ICAM-1; chemotactic cytokines such as interleukin (IL) 1 -8 and monocyte chemoattractant protein-1; and prothrombotic molecules such as tissue factor and plasminogen activator inhibitor-1 (4 -6).
The nuclear transcription factor NF-B appears to play a central role in the process of endothelial activation (7,8). Many inducible genes involved in endothelial activation, including VCAM-1, ICAM-1, E-selectin, P-selectin, tissue factor, IL-6, IL-8, G-CSF, and c-myc, contain elements in their promoter regions that can be recognized by the nuclear factor NF-B/Rel family of transcription factors (reviewed in Ref. 4). NF-B is associated with rapid-response activation mechanisms (9,10). The NF-B subunits p50 (NFKB1) and p65 (RelA), belong to the Rel family of genes that also includes p52 (NFKB2), the v-Rel oncogene, c-Rel, and the Drosophila morphogen dorsal. In resting endothelium, NF-B is found in an inactive cytosolic form, complexed with an inhibitory subunit known as IB-␣ (11,13,14). IB-␣ binds preferentially to the p65 subunit of NF-B through interaction with sequences surrounding the nuclear localization signal of p65, thus preventing nuclear translocation of the NF-B complex (12). Stimulation of primary cultures of human umbilical vein endothelial cells (HU-VEC) with inflammatory mediators such as TNF␣ and bacterial lipopolysaccharide (LPS) results in the inducible degradation of IB-␣, the cytoplasmic inhibitor of NF-B, and the appearance of NF-B within the nucleus (13,14). In endothelial cells, activation of the NF-B pathway is rapid; IB-␣ degradation is complete within 5-10 min following cellular activation. The process of IB-␣ degradation has been the subject of intense research recently (15)(16)(17)(18)(19)(20)(21)(22)(23)(24). These studies demonstrated that IB-␣ is found as a phosphoprotein in resting cells and that agents that activate NF-B act to induce the hyperphosphorylation of IB-␣ (15,16). Hyperphosphorylation of IB-␣ is necessary, but not sufficient, for the activation of NF-B, with hyperphosphorylated IB-␣ remaining bound to the NF-B complex in the cytoplasm (15)(16)(17)(18)(19)(20)(21)(22). NF-B activation is achieved through the selective recognition and destruction of hyperphosphorylated IB-␣ by components of the ubiquitin-26 S proteasome system (21)(22)(23). Hyperphosphorylation of IB-␣ appears to serve as a signal for selective ubiquitination of the IB-␣ subunit (24).
We recently demonstrated that pharmacologic inhibition of IB-␣ hyperphosphorylation in human endothelial cells resulted in the inhibition of NF-B activation, and of E-selectin, VCAM-1 and ICAM-1 gene expression (13). The specific kinases that mediate inducible IB-␣ phosphorylation in endothelial cells have not been reported. However, several different kinases, including casein kinase II and protein kinase C, have been shown to phosphorylate IB-␣ either in vitro or in transformed cell lines (28,37,(41)(42)(43)(44)(45). In this study, we report the identification of two discrete kinases in the cytoplasm of human endothelial cells that can selectively bind and phosphorylate IB-␣. The activity of these IB-␣ kinases is transiently elevated following endothelial activation by pro-inflammatory stimuli, including TNF␣, interleukin-1␤, and LPS. Maximal activity precedes that of other known signal transducing kinases and precedes the appearance of hyperphosphorylated IB-␣, its destruction, and the translocation of NF-B to the nucleus. Deletion mutagenesis of the IB-␣ substrate reveals that the kinases bind within the ankyrin repeats and appear to phosphorylate residues in the C-terminal region. Our initial characterization of these enzymes suggests they are distinct from other known signal transducing kinases. We hypothesize that these IB-␣ kinases may play a role in the regulation of IB-␣ levels in vascular endothelium.

MATERIALS AND METHODS
Reagents-Human recombinant TNF␣ and IL-1␤ were purchased from Genzyme, Inc. (Boston, MA). LPS (Escherichia coli strain 0111:B4) was obtained from Sigma. Peptide-specific rabbit polyclonal antibodies against IB-␣, p50, and p65 were purchased from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). Myelin basic protein, calf intestinal alkaline phosphatase, lavendustin, and genestein were purchased from Life Technologies, Inc. Partially dephosphorylated bovine casein was obtained from Sigma. Anti-FLAG peptide monoclonal antibody conjugated to agarose beads was purchased from Eastman Kodak. Calphostin C, chelerythrine, KT5720, KT5823, and H89 were purchased from LC Laboratories (Woburn, MA). Human recombinant casein kinase II was purchased from Boehringer Mannheim. Peptide-specific rabbit polyclonal antisera against human casein kinase II were purchased from Upstate Biotechnology (Lake Placid, NY). All other reagents were purchased from Sigma.
Preparation of Cytoplasmic Extracts-To induce endothelial activation, cells were treated with either 300 units/ml (2000 units/ng) TNF␣, 0.5 ng/ml IL-1␤, or 0.1 g/ml LPS. The media were removed, and the cells were rinsed twice with ice-cold PBS (10 mM NaH 2 PO 4 , 150 mM NaCl, pH 7.4). Cells were trypsinized, pelleted, and resuspended in ice-cold lysis buffer (20 mM HEPES, pH 7.3, 50 mM NaCl, 10 mM MgCl, 1 mM EDTA, 1 mM EGTA, 1 mM sodium orthovanadate, 10 mM ␤-glycerol phosphate, 1 mM sodium fluoride, 1 mM phenylmethylsulfonyl fluoride, 1 mM dithiothreitol, 0.5% Nonidet P-40), 200 l/75-cm 2 flask. The total cell lysates were vortexed, incubated for 2 min at room temperature, and centrifuged 5 min at 5000 ϫ g at 4°C to remove nuclei and debris, and the resulting cytoplasmic extracts were transferred to tubes and stored at Ϫ80°C. An aliquot of each postnuclear fraction was taken for protein determination using the BCA protein assay (Pierce) with BSA as the standard.
Preparation of Recombinant Proteins-Recombinant IB-␣, expressed as a fusion protein with an N-terminal FLAG octapeptide (Kodak), was produced in E. coli INVFЈ and purified by immunoaffinity chromatography. Briefly, the IB-␣ cDNA (kind gift of Prof. P. A. Baeuerle, Freiburg, Germany) was subcloned into pFLAG-MAC expression vector (Kodak) by using the polymerase chain reaction (PCR) to introduce a HindIII site at the 5Ј end of the cDNA and a SmaI site at the 3Ј end (25). The sequences for the sense and antisense PCR oligonucleotides, respectively, were 5Ј-CAAGCTTCTCGTCCGCGCCATGT-TCCA-3Ј and 5Ј-TCCCGGGT TTGCACT CATAACGTCAGACG-3Ј. PCR product was digested with HindIII/SmaI restriction endonucleases (Life Technologies, Inc.), ligated into the HindIII/SmaI sites of the pFLAG-MAC vector, and transformed into E. coli strain INV␣FЈ. Midlog cultures were incubated in the presence of 0.1 mM isopropyl ␤-Dthiogalactopyranoside for 3 h, harvested, and disrupted by sonication in TEP buffer (10 mM Tris, pH 7.0, 1 mM EDTA, 1 mM phenylmethylsulfonyl fluoride, 1 mg/ml aprotinin). Insoluble material was removed by centrifugation. IB-␣ fusion protein was purified by affinity chromatog-raphy using M2 anti-FLAG affinity resin (Kodak). The affinity column was equilibrated with 3 ϫ 5-ml washes of 0.1 M glycine, pH 3.0, followed by 3 sequential washes with PBS. Cell extract was loaded in PBS, the column washed with 3 column volumes of PBS, and bound material was eluted with 6 ϫ 1-ml aliquots of 0.1 M glycine, pH 3.0. Eluant was immediately neutralized in 1 M Tris, pH 8.0, pooled, and dialyzed against 10 mM NaH 2 PO 4 , pH 7.4. An N-terminal deletion (⌬N) mutant missing amino acids 1-54 was generated by restriction digestion and religation at an in-frame XhoI site in the full-length IB-␣ pFLAG-MAC clone. A C-terminal deletion (⌬C) mutant missing amino acids 277-317 was generated by PCR as described above, except that an alternate antisense primer (5Ј-CGGATCCTCAGTTTTCTAGTGTCAGCCTGGC-CCAGCTG-3Ј) was utilized. Recombinant proteins were expressed and purified as described above. Recombinant human p50 protein was produced in E. coli as a fusion protein with an N-terminal FLAG octapeptide purification tag. Briefly, cDNA sequences encoding amino acids 1-399 of the p105 polypeptide were subcloned into pFLAG-MAC expression vector by using PCR to introduce a HindIII site at the 5Ј end of the cDNA and a SmaI site at the 3Ј end. The sequences for the PCR oligonucleotides were sense primer, 5Ј-GATCAAGCTTTTCAGAATG-GCAGAAGAT-3Ј; antisense primer, 5Ј-CGATCCCGGGTTCCAGTGC-CCCCTCCTCCA-3Ј. PCR product was digested with HindIII/SmaI restriction endonucleases (Life Technologies, Inc.), ligated into the HindIII/SmaI sites of the pFLAG-MAC vector, and transformed into E. coli strain INV␣FЈ. Recombinant protein was purified by M2 affinity chromatography as described above. Recombinant human p65 was expressed in sf9 insect cells. The cDNA encoding human p65 was subcloned into the baculovirus expression vector pVL1393 (PharminGen, San Diego, CA) utilizing PCR to introduce a BamHI site at the 5Ј end of the cDNA and an XbaI site at the 3Ј end. The sequences for the sense and antisense oligonucleotides, respectively, were 5Ј-GATCGGATCC-ATGGACTACAAGGACGACGATGACAAAGTCGACGAACTGTTCCCC-CTGAT-3Ј and 5Ј-GATCTCTAGATCACCCCCTTAGGAGCTGATCTG-ACTCAGCAGGGCTGAGAAGTCCATGTC-3Ј. The resulting PCR product was digested with BamHI and XbaI restriction endonucleases and ligated into the same sites in pVL1393. Insect sf9 cells were infected with recombinant baculovirus containing the pVL1393-p65 vector. Recombinant p65 protein was purified from extracellular fluid of infected cultures by immunoaffinity chromatography with anti-p65 antibodies (Santa Cruz Biotech, Santa Cruz, CA). Recombinant human c-JUN was expressed in E. coli as a glutathione S-transferase fusion protein. A c-Jun cDNA clone encoding the first 223 amino acids of c-Jun was amplified from human genomic DNA using PCR and the following sense and antisense primers: 5Ј-ATGACTGCAAAGATGGAAACGACC-3Ј and 5Ј-CGGACGTCGGCGCCCACGAC-3Ј. PCR product was cloned into the plasmid pCRII (Invitrogen, San Diego, CA). The c-Jun cDNA was subcloned into pGEX-4T-1 (Pharmacia Biotech) and transformed into E. coli strain DH5␣ (Life Technologies, Inc.). Synthesis of glutathione S-transferase-c-Jun fusion protein was induced by the addition of 100 M isopropyl ␤-D-thiogalactopyranoside to mid-log cultures. Cells were pelleted, resuspended in PBS, and lysed with 0.5 mg/ml lysozyme, followed by sonication. Triton X-100 (1% v/v) was added, and the lysed cells were incubated at room temperature with constant mixing for 30 min. Cell debris and nuclei were pelleted by centrifugation at 18,000 ϫ g for 10 min. The supernatant was passed through glutathione-Sepharose 4B and eluted with 15 mM reduced glutathione in 100 mM Tris, pH 8.0, and 100 mM NaCl. The eluate was dialyzed against 10 mM HEPES, pH 7.4, and concentrated using Centricon-30 microconcentrators (Amicon, Beverly, MA).
In-gel Kinase Assays-IB-␣, c-Jun, and MAPK isozymes were assayed for activity essentially as described earlier (26). Briefly, proteins from cytoplasmic extracts were separated by electrophoresis through an 8% SDS-PAGE gel in which either 30 g/ml recombinant ⌬N-, ⌬C-, or wild type-IB-␣, 75 g/ml c-Jun, or 100 g/ml myelin basic protein was included during gel polymerization. Following electrophoresis, the SDS was extracted with buffer 1 (50 mM Tris-Cl, pH 8, 5 mM ␤-mercaptoethanol) containing 20% (v/v) isopropyl alcohol. Proteins within the gel were denatured by incubation with buffer 1 containing 6 M guanidine-HCl for 45 min and then permitted to renature overnight at 4°C in buffer 1 containing 0.04% (v/v) Tween 40. Phosphorylation of the respective substrate was initiated by incubation of the gel for 60 min in buffer 2 (20 mM HEPES, pH 7.3, 10 mM MgCl 2 , 15 mM ␤-glycerophosphate. 0.5 mM sodium orthovanadate, 0.5 mM EGTA, 1 mM dithiothreitol) containing 150 Ci [␥-32 P]ATP. Following extensive washing to remove excess unincorporated 32 P, kinase activity was quantitated by autoradiography and scanning laser densitometry (Molecular Dynamics Model 300A, Sunnyvale, CA) utilizing ImageQuant 3.0 software. Statistical significance was determined using an unpaired Student's t test.
IKB-␣-agarose Bead Assay-A solid-phase kinase assay similar to that previously described for Jun N-terminal kinases was developed (27). The IB-␣ fusion protein (2 g) was incubated with 10 l of agarose beads conjugated with anti-FLAG M2 monoclonal antibody (Kodak) in 10 mM HEPES, pH 7.3. The IB-␣-agarose beads were washed twice with kinase reaction buffer (20 mM HEPES, pH 7.3, 10 mM MgCl 2 , 15 mM ␤-glycerophosphate, 0.5 mM sodium orthovanadate, and 0.5 mM EGTA), resuspended in 100 l of ice-cold reaction buffer containing 20 g HUVEC extract containing 1 mM phenylmethylsulfonyl fluoride, and incubated at 4°C for 2.5 h on a vertical turntable. The IB-␣-agarose bead complexes were recovered by centrifugation at 16,000 ϫ g for 1 min, washed three times with reaction buffer followed by centrifugation, and resuspended in 20 l of reaction buffer containing 200 M ATP and 5 Ci [␥-32 P]ATP. Following incubation for 10 min at 30°C, the reaction mix was denatured in SDS loading buffer at 100°C for 2 min and separated by electrophoresis on a 10 -20% gradient SDS-PAGE gel (Daiichi, Integrated Separation Systems, Natick, MA). The gel was stained with Coomassie dye, dried under vacuum, and exposed to Hyperfilm-MP autoradiography film (Amersham Corp.). Autoradiographs were quantitated using a model 300A computing densitometer and ImageQuant 3.0 software (Molecular Dynamics, Sunnyvale, CA). To detect the presence of dissociated kinase following ATP addition, the reaction mix was centrifuged to pellet the IB-␣-agarose beads, and 15 l of supernatant was removed to a fresh preparation of IB-␣-agarose beads and incubated at 30°C for 10 or 30 min in the presence of [␥ 32 P]ATP.
IB-␣ Kinase Plate Assay-Immulon-2 96-well microtiter plates (Dynatech, Chantilly, VA) were incubated with recombinant IB-␣ (500 ng/well in 0.1 M carbonate buffer, pH 9.2) overnight at 4°C. Following removal of this solution, plates were incubated overnight at 4°C in PBS containing 1% BSA and 0.02% sodium azide (200 l/well). This solution was removed from the wells and the plates washed four times with kinase buffer (20 mM HEPES, pH 7.3, 10 mM MgCl 2 , 15 mM ␤-glycerophosphate, 0.5 mM NaV 4 , 0.5 mM EGTA). For the kinase reaction, 1 g of cytoplasmic extract, 20 nM [␥ 32 P]ATP (10 mCi/ml, Amersham Corp.), and drug (when appropriate) were added into each well to a final volume of 50 l and incubated for 60 min at room temperature. Plates were washed 2-3 times with PBS and blot-dried, and 100 l/well Op-tiPhase "Polysafe" scintillant (Wallac, Gaithersburg, MD) was added. Transfer of [ 32 P] to immobilized IB-␣ was determined by liquid scintillation counting using a Wallac 1450 Microbeta counter with the appropriate cross-talk and normalization protocols.
Western Analysis-Cytoplasmic extracts (150 g) were subjected to electrophoresis on 12% SDS-PAGE gels, and the fractionated proteins electrophoretically transferred to Immobilon-P membranes (Millipore, Bedford, MA), using a Multiphor II semi-dry blotting device (Pharmacia). Membranes were incubated overnight in 5% non-fat milk powder, incubated with a polyclonal anti-IB-␣ antibody (0.1 g/ml) for 2 h in PBS containing 0.05% Tween-20 (PBS-T), and then washed three times with PBS-T. Membranes were then incubated with polyclonal donkey anti-rabbit IgG antibody (1:3000, v/v) conjugated with horseradish peroxidase (Amersham Int.). After extensive washing with PBS-T, chemiluminescent substrate was added (ECL system, Amersham Int.), and the membrane was subjected to autoradiography with Hyperfilm MP (Amersham Int.). In some experiments, the proteasome inhibitor calpain inhibitor I (N-Ac-Leu-Leu-norleucinal) (Calbiochem), suspended in dimethyl sulfoxide, was added to complete media to a final concentration of 250 M. Cells were treated with inhibitor for 60 min prior to addition of TNF␣. In experiments examining the effect of alkaline phosphatase treatment on IB-␣ phosphorylation, 150 units of calf intestinal alkaline phosphatase (Life Technologies, Inc.) were added to cytoplasmic extracts and incubated for 30 min at 37°C. Reactions were terminated by the addition of Laemmli loading buffer.
Electrophoretic Mobility Shift Assays-HUVEC were stimulated with TNF␣ as described above. Nuclear extracts were prepared as described previously (13). The concentration of protein in each extract was determined using the BCA protein assay (Pierce) using BSA as the standard. Double-stranded oligonucleotide containing a consensus NF-B recognition sequence (Promega, Madison, WI) was end-labeled with T4 polynucleotide kinase in the presence of [␥-32 P]ATP (Amersham Int.) (25). For each assay, 10 g of nuclear protein was incubated with 0.1 pmol of 32 P-labeled oligonucleotide in binding buffer (12 mM HEPES, 4 mM Tris-Cl, 60 mM KCl, 1 mM EDTA, 1 mM dithiothreitol, and 1 g of poly(dI-dC)⅐poly(dI-dC), pH 7.9, for 30 min at room temperature. Competition studies were performed by the addition of a 50-fold molar excess of unlabeled oligonucleotide to the binding reaction. Resultant protein-DNA complexes were resolved on native 5% polyacrylamide gels using a high ionic strength buffer (50 mM Tris-Cl, 380 mM glycine, 2 mM EDTA, pH 8.5). The gels were electrophoresed at 125 mA for 4 h, dried under a vacuum, and subjected to autoradiography using Hyperfilm MP.
Casein Kinase II Assay-Casein kinase II (CKII) activity was determined using a casein peptide-based assay system (Amersham Int.). Briefly, affinity-purified IB-␣ kinases or recombinant human CKII (Boehringer Mannheim) were incubated with casein peptide and [␥-32 P]ATP at 30°C for 30 min. Phosphorylation of this peptide was determined by scintillation counting of peptide captured on phosphocellulose membranes.
Q-Sepharose Chromatography-Cytoplasmic extract (100 ml) was thawed and concentrated to 23 ml with an Amicon Stirred Cell fitted with a YM10 membrane. The concentrate was clarified by centrifugation, and the supernatant was dialyzed (12 to 14,000 molecular weight cut-off membrane) against 4 liters of 20 mM HEPES buffer, pH 7.5 (buffer A). The dialyzed solution had a protein concentration of 0.51 mg/ml (12.7 mg total) and was added directly to a column (1.3 ϫ 3.0 cm) of Q-Sepharose equilibrated in buffer A. After addition of the protein sample, the column was washed with 180 ml of buffer A, and elution of protein was continued with 300 ml of a linear gradient of increasing NaCl concentration from 0 to 300 mM in buffer A (flow rate ϭ 0.5 ml/min). The final wash was with 50 ml of buffer A, containing 300 mM NaCl (buffer B). Fractions of 6 ml were collected and assayed for IB-␣ kinase activity using the plate assay described above.

RESULTS
An SDS-PAGE-based in-gel phosphorylation assay was employed to identify kinases present in HUVEC cytoplasmic extracts which were capable of directly phosphorylating IB-␣ protein. In initial assays, three such kinase activities were identified in extracts from primary cultured HUVEC (Fig. 1A,  lanes 1-5). The apparent molecular masses of these kinases were 36, 41, and 43 kDa. IB-␣ kinase activity was consistently associated with the 36-and 41-kDa species, with the 43-kDa species being apparent in only some assays. In resting cells, the majority of kinase activity was associated with the 36-kDa molecular species. In extracts from cells stimulated for 5 min with the pro-inflammatory cytokine TNF␣, the activities of the 36-and 41-kDa kinases were consistently elevated, on average 2.6-and 8-fold (Fig. 1B). The activity of these kinases returned close to that of unstimulated cells by 10 min after TNF␣ addition. In similar assays performed in the absence of recombinant IB-␣ (Fig. 1A, lanes 19 -23), no kinase activity was detectable at either 36 or 41 kDa. Such a finding demonstrated that the activity of these kinases was dependent upon the presence of IB-␣ substrate and was not due to autophosphorylation. In assays using extracts from HUVEC stimulated with either IL-1␤ or LPS, known inducers of NF-B activation, the 36-and 41-kDa IB-␣ kinase activities were also elevated with temporal kinetics similar to that seen with TNF␣ (data not shown).
To examine whether these kinases might represent members of known cytokine signal transduction pathways, in-gel kinase assays were performed utilizing either recombinant c-Jun or myelin basic protein (MBP) polymerized in the gel in place of recombinant IB-␣. These proteins represent substrates for Jun N-terminal kinases (JNK) and members of the mitogenactivated protein kinase (MAPK) family, respectively (26,37). With these assays it was shown that the 55-and 46-kDa isoforms of JNK were present in endothelial cells, and the activity of these kinases was elevated 15 min following TNF␣ stimulation (Fig. 1A, lanes 6 -10). JNK activation was also detectable in cells treated for 30 min with LPS (lane 11). JNK activity in HUVEC extracts was consistently lower than that seen for the IB-␣ kinases. No activity capable of phosphorylating c-Jun was detectable at either 36 or 41 kDa, demonstrating that the IB-␣ kinase activities were distinct from the JNK activities. Activity of the 42-and 44-kDa isoforms of MAP kinase (ERK1 and 2) was undetectable in unstimulated endothelial cells; weak MAPK activity was, instead, associated with several spe-cies of differing molecular weight (Fig. 1A, lanes 12-16). ERK1 and ERK2 activities were not dramatically increased in lysates from TNF␣-stimulated HUVEC. ERK1 and ERK2 activation was detectable using this assay, however, as evidenced following insulin treatment of differentiated NIH 3T3/L1 adipocytes (lanes 17 and 18). Analysis of MAPK activity utilizing a peptide-based MAPK assay (Amersham Int.) confirmed that TNF␣ treatment of HUVEC resulted in little if any MAPK activation (data not shown). The IB-␣ kinases identified with these assays, therefore, appeared distinct from JNK or MAP kinases, both in molecular mass and kinetics of activation following TNF␣ treatment.
The temporal relationship between IB-␣ kinase activation and the appearance of NF-B within the nucleus of TNF␣activated HUVEC was investigated (Fig. 2). Consistent with earlier observations, activity of the 41-and 36-kDa IB-␣ kinases was elevated by 3 min following TNF␣ addition ( Fig. 2A). IB-␣ protein was detectable in the cytoplasm of unstimulated HUVEC, and by 5 min following TNF␣ addition, a slowermigrating form of IB-␣ was visible on Western blots (Fig. 2B).
To confirm that this immunoreactive species represented hyperphosphorylated IB-␣, we examined IB-␣ levels in HUVEC stimulated in the presence of calpain inhibitor I, an inhibitor of the 26 S proteasome responsible for IB-␣ degradation. As reported earlier (19 -23), the presence of calpain inhibitor I prevented degradation of IB-␣ and resulted in the accumulation of the slower-migrating species (Fig. 2D, lane 3). Treatment of this extract with alkaline phosphatase resulted in the complete ablation of this slower-migrating species (lane 4), demonstrating that this represented a hyperphosphorylated form of IB-␣. Levels of cytoplasmic IB-␣ diminished dramatically by 10 min (Fig. 2B), and reduced cytoplasmic levels of IB-␣ were associated with the appearance of NF-B DNA binding activity within the nucleus of these cells (Fig. 2C). Activation of the IB-␣ kinases, therefore, precedes the appearance and destruction of hyperphosphorylated IB-␣ and the translocation of NF-B to the nucleus.
Using an IB-␣-agarose bead assay, we determined that the IB-␣ kinases were capable of binding IB-␣ prior to the phosphorylation event. Recombinant IB-␣, immobilized on agarose beads, was incubated with TNF␣-activated HUVEC extracts and the agarose bead complexes recovered by centrifugation and extensive washing to remove unassociated factors. Addition of radiolabeled ATP to these bead complexes resulted in phosphorylation of the IB-␣ bound to the agarose, demonstrating that a kinase activity was associated with the IB-␣-agarose beads (Fig. 3, lane 1). This binding of kinase activity was dependent upon the presence of divalent cation but not on the presence of ATP (data not shown). No kinase activity was detectable when extracts were incubated with agarose beads in the absence of IB-␣ protein, even when IB-␣ protein was added to the kinase reaction after washing and centrifugation. To determine if the IB-␣ kinase activity was released from the IB-␣-agarose complex after phosphorylation, we recovered su- Cells were harvested and cytoplasmic extracts prepared as described under "Materials and Methods." Equal amounts of protein were subjected to SDS-PAGE in-gel kinase analysis. Activity of the 36-(Ⅺ) and 41-kDa (f) kinase activities were quantified by scanning laser densitometry and expressed as the fold increase over that seen in resting endothelium. Results were collated from three separate experiments and are presented as mean Ϯ S.E. *, p Ͻ 0.05.

FIG. 2. Relationship between IB-␣ kinase activation, IB-␣ degradation, and NF-B nuclear translocation in TNF␣-stimulated HUVEC.
HUVEC monolayers were stimulated with TNF␣ (300 units/ml) for 1-20 min. Cells were harvested and cytoplasmic and nuclear extracts were prepared as described under "Materials and Methods." A, in-gel assay for IB-␣ kinase activity. B, IB-␣ levels in cytoplasmic extracts were detected by immunoblotting using anti-IB-␣ polyclonal antibodies as described under "Materials and Methods." C, NF-B DNA binding activity in nuclear extracts was determined by electrophoretic mobility shift assays as described under "Materials and Methods." D, the slower migrating species present in HUVEC extracts is hyperphosphorylated. Extracts were immunoblotted using anti-IB-␣ polyclonal antibodies. Lanes contain HUVEC extract from 1) untreated cells, 2) cells treated with TNF␣ for 20 min, 3) cells treated with 250 M calpain inhibitor I for 30 min prior to addition of TNF␣ for 20 min, and 4) the extract then incubated with alkaline phosphatase. pernatants from the kinase assay following centrifugation to remove the IB-␣-agarose beads and examined the presence of kinase activity by addition of IB-␣ in solution. IB-␣ added to cleared supernatants was efficiently phosphorylated (lanes 2 and 3), suggesting that the IB-␣ kinase activity was released following phosphorylation of the IB-␣ bound to the beads. Kinase activity was dependent upon the addition of fresh IB-␣ (lane 4) and was therefore not due to release of radiolabeled, phosphorylated IB-␣ protein from the agarose beads.
Recent studies demonstrate that Ser-32 and Ser-36 in the N-terminal region of the IB-␣ polypeptide are targets of inducible phosphorylation in vivo (17,24). In addition, consensus sequences for CKII and protein kinase C, and a Ser/Thr-rich PEST sequence are present in the C-terminal region of IB-␣ (28,29,37). We utilized the in vitro kinase assay and deletion mutants of the IB-␣ polypeptide to identify those regions required for kinase binding and phosphorylation. Cellular extracts were incubated in the presence of agarose beads containing either IB-␣, in which residues 1-54 in the amino terminus were deleted (⌬N), or IB-␣, in which the carboxyl-terminal 40 residues were deleted (⌬C). After unassociated factors were washed away, an in vitro kinase assay was performed, with the IB-␣-agarose now serving as the substrate. Similar to the wild type, the IB-␣ ⌬N protein was capable of binding a kinase activity and serving as an efficient substrate for phosphorylation (Fig. 4A, lane 5). This kinase activity was released from the agarose bead complex following phosphorylation and was detectable in the supernatant (lane 11). No kinase activity was detectable, however, when HUVEC extracts were incubated with the IB-␣ ⌬C-agarose beads (lane 6). This was not due to an inability of the kinase activity to bind to the IB-␣ protein, as kinase activity capable of phosphorylating wild type IB-␣ was detectable in the cleared supernatant following addition of ATP (lane 12). From these studies (summarized in Fig. 4B), it appears that the IB-␣ kinase activity bound physically to the IB-␣ protein within the ankyrin repeat domains and phosphorylated residues present in the C terminus of IB-␣.
To determine whether one or both of the 36-and 41-kDa kinases identified by the in-gel assay (Fig. 1) were those responsible for the kinase activity which bound to IB-␣-agarose (Figs. 3 and 4), we utilized IB-␣-agarose affinity chromatography to purify kinase activity from HUVEC extracts and then examined the bound material using the in-gel IB-␣ kinase assay. Using this approach, we determined that both the 36and 41-kDa kinases bound to the IB-␣-agarose (Fig. 5, lanes 1  and 2). To determine whether these kinases phosphorylate distinct regions of IB-␣, we examined their ability to phospho-rylate the ⌬N and ⌬C mutants using the in-gel assay. Both kinases were capable of phosphorylating IB-␣ ⌬N but neither was capable of phosphorylating IB-␣ ⌬C protein (lanes 3-6). Such a finding is consistent with the results of the in vitro kinase assays and suggests that both kinases bind to IB-␣ and phosphorylate residues contained in the carboxyl terminus of IB-␣.
To better understand the substrate specificity of the affinitypurified IB-␣ kinases, we examined their ability to phosphorylate a diverse set of substrates. These substrates included IB-␣, the p50 and p65 components of NF-B, the transcription factor c-Jun, and the MAPK substrate myelin basic protein (Fig. 6). Affinity-purified IB-␣ kinases were capable of phosphorylating IB-␣ but were not capable of phosphorylating either recombinant p50 or p65 (Fig. 6, lanes 3-5, lower panel). In addition, the IB-␣ kinases did not phosphorylate the c-Jun transcription factor (lane 7). The IB-␣ kinases were capable of phosphorylating the MAPK substrate myelin basic protein (lane 6), although this activity was approximately 100-fold less  1, 3, 5, and 7) and affinitypurified IB-␣ kinase (lanes 2, 4, 6, and 8) were analyzed in a series of in-gel kinase assays with equal amounts (30 g/ml) of wild type, ⌬N, or ⌬C IB-␣ substrate embedded in the gel as described under "Materials and Methods." than that observed using IB-␣ as the substrate. The IB-␣ kinases, therefore, appear to be specific for IB-␣ as a substrate.
We examined the effects of a panel of inhibitors of known kinases on IB-␣ kinase activity using an in vitro plate-based assay. HUVEC extracts were added to microtiter plate wells in which IB-␣ protein had been immobilized to the plate surface. Following an incubation, the plates were extensively washed to remove unassociated factors and a kinase reaction performed. Inhibitors of tyrosine kinases (lavendustin, genestein) (29,30), protein kinase C (chelerythrine, calphostin C) (31, 32), cAMPregulated kinases (KT5720, H89) (33,34), and cGMP-regulated kinases (KT5823) (34) were included at varying concentrations to test for their ability to inhibit the kinase reaction (Table I). Of these agents, only H89 and genestein had any significant inhibitory effect on IB-␣ kinase activity in these assays. The IC 50 for inhibition of IB-␣ kinase activity by these agents was determined to be 8.1 and 12 M, respectively.
CKII has been demonstrated to associate physically with IB-␣ in Jurkat and U937 cells and to be capable of phosphorylating residues within the C terminus of IB-␣ (28,37). It was possible, therefore, that the IB-␣ kinases we identified in endothelial cells represented CKII. To explore this possibility, we examined the affinity-purified endothelial IB-␣ kinases for CKII activity using a peptide-based CKII assay. Surprisingly, the purified IB-␣ kinases displayed little CKII activity, in marked contrast to purified CKII itself (Fig. 7). The IB-␣ kinases demonstrated greater than 10-fold selectivity toward IB-␣ as compared with casein as a substrate. Purified CKII was also capable of phosphorylating both IB-␣ and the casein peptide. However, in contrast to the purified IB-␣ kinases, CKII demonstrated a clear preference for the casein peptide, rather than IB-␣. These data suggested the possibility that the purified IB-␣ kinases are distinct from CKII.
To explore the possible relatedness of the endothelial IB-␣ kinases to CKII, we subjected HUVEC extracts to anion exchange chromatography on Q-Sepharose and examined the relative positions of CKII and the IB-␣ kinases in the elution profile. Using an IB-␣ plate assay, we determined that multiple peaks of IB-␣ kinase activity could be resolved using this procedure (Fig. 8A). Examination of these peaks by the IB-␣ in-gel kinase assay revealed that the 36-and 41-kDa IB-␣ kinases identified previously were bound to and eluted from the Q-Sepharose column between 100 and 200 mM NaCl (Fig.  8B, lanes 4 and 5). Western blots of equal amounts of protein from each of the peaks of kinase activity in Fig. 8A were examined for CKII immunoreactivity. CKII was found to be present in HUVEC extracts (Fig. 8C, lane 1) but, under the conditions employed, was not retained by the Q-Sepharose column and was detected in the unbound fraction (lane 2). No CKII immunoreactivity was present in those fractions that contained the 36-and 41-kDa IB-␣ kinase activities (lanes 4 and 5), and no CKII activity could be identified in these fractions using the casein peptide assay (data not shown). Consistent with the earlier findings, the IB-␣ kinases present in endothelial cells appear to be immunologically and biochemically distinct from CKII. DISCUSSION We recently demonstrated that NF-B is rapidly activated in multiple organs, including the lung, liver, and myocardium during acute inflammation in vivo (45,46). In particular, rapid activation of NF-B was observed in vascular endothelial cells isolated from the liver (46). Activated NF-B is also detectable in the lung vasculature following aerosol antigen challenge (58). In both situations, the rapid activation of this transcription factor precedes the induction of endothelial cell adhesion molecule and chemokine gene expression, leukocyte recruitment, and leukocyte-mediated tissue injury. Rapid activation of NF-B has also been detected in vascular endothelium lining the wound edge following balloon injury in the rat aorta (49), and activated NF-B is present in atherosclerotic lesions in man (59) and in spinal cord lesions in mice with active experimental allergic encephalomyelitis (60). These studies define a key role for NF-B activation as an early signaling event in  activated vascular endothelium in vivo and demonstrate a correlation between the activation kinetics of NF-B in endothelium in culture and in vivo. Because NF-B is a key nuclear transcription factor regulating the expression of a wide range of inflammatory genes, elucidation of components of the system regulating IB-␣ degradation and NF-B activation in vascular endothelium may lead to the identification of novel targets for anti-inflammatory drug development.
In this report we describe the identification of two novel cytoplasmic kinases that, we hypothesize, form components of an IB-␣ regulatory system in human endothelial cells. These kinases have been identified based upon their ability to bind to and phosphorylate IB-␣ protein, and this interaction served as the basis for their partial purification by affinity chromatography. We focused upon examining the relationship of these kinases to known signal-transducing enzymes and examining the domains of IB-␣ important for kinase recognition and phosphorylation. By multiple criteria, the kinases appear distinct from other known kinases, and their mode of interaction with IB-␣ is novel. TNF␣, IL-1␤, or LPS treatment of HUVEC resulted in a transient activation of these kinases, and this activation preceded the appearance of hyperphosphorylated IB-␣ protein in the cytoplasm. These may play a role in the NF-B activation pathway through their ability to mediate phosphorylation of IB-␣, the cytoplasmic inhibitor of NF-B. To our knowledge, this represents the first report of kinases present in primary cultures of human endothelial cells with such activities.
Activation of the NF-B pathway in endothelial cells occurs in response to a wide range of cellular stimuli, including cytokines such as TNF␣ and IL-1␤, bacterial cell-wall products, UV irradiation, and fluid shear stress (reviewed in Ref. 4). Translocation of NF-B from the cytoplasm to the nucleus of these cells is achieved through a rapid decrease in the cytoplasmic levels of IB-␣ (13,14). Hyperphosphorylated IB-␣ was detectable within 5 min following endothelial activation with TNF␣, and IB-␣ levels fell dramatically by 10 min, suggesting the signal transduction mechanisms regulating IB-␣ phosphorylation and destruction must be activated rapidly in these cells. The 36-and 41-kDa IB-␣ kinase activities were routinely elevated by 3-5 min following TNF␣ addition, consistent with a role for these kinases as part of such a rapidly activated signal transduction mechanism. The 41-kDa kinase was activated to a greater extent than the 36-kDa kinase, raising the possibility that these species are differentially regulated in endothelium. TNF␣ also rapidly activates the JNK signal transduction pathways in a variety of cells, leading to the phosphorylation-dependent activation of the c-Jun subunit of the transcription factor AP-1 (reviewed in Ref. 38). JNK activation was detectable in TNF␣-activated endothelial cells; however, this was not apparent until 15-20 min following TNF␣ addition. Consistent with reports in other cell types, MAPK activity is not considerably activated by TNF␣ treatment of HUVEC (39,40). The IB-␣ kinases present in endothelium are distinct in both molecular size and kinetics of activation, therefore, from either the p42 or p44 species of MAPK or the 55-or 46-kDa species of JNKs.
The IB-␣ kinases display a high degree of substrate specificity for IB-␣ in vitro, being unable to phosphorylate either the p50 or p65 subunits of the cytoplasmic NF-B complex, or the cytokine-inducible c-Jun transcription factor. In addition, from a panel of inhibitors of known kinases, only H89 and genestein displayed any inhibitory activity in an in vitro IB-␣ kinase assay, and then only at concentrations at least 200-fold , and 23 ml (total 12.7 mg of protein) of a concentrated HUVEC cytoplasmic proteins was added by gravity flow. The column was then washed with 180 ml of buffer A, and proteins were eluted with 300 ml of a linear gradient of increasing NaCl concentration from 0 to 300 mM in buffer A at a flow rate of 0.5 ml/min. Fractions of 6 ml were collected for further analysis. Results are representative of at least three independent experiments. A, identification of multiple peaks of IB-␣ kinase activity. Protein elution was monitored by absorbance at 280 nm (solid line) and IB-␣ kinase activity as measured by the plate assay (dashed line). Four peaks of kinase activity were identified (labeled a-d) B, IB-␣ in-gel kinase analysis of peaks a-d (lanes 2-5) from Q-Sepharose column chromatography. HUVEC cytoplasmic extract was included as a control (lane 1). C, Western blot analysis for immunoreactive CKII. Equal amounts of protein from HUVEC cytoplasmic extract (lane 1) or peaks a-d (lanes 2-5) were subjected to SDS-PAGE and transferred to nylon membranes as described under "Materials and Methods." The presence of immunoreactive CKII in each fraction was analyzed using an anti-CKII antibody.
above those required to inhibit cAMP-regulated and tyrosine kinases, respectively (30,34). Both H89 and genestein act as competitive inhibitors with respect to ATP, and therefore, it is not surprising that at high concentrations they may affect IB-␣ kinase activity. Therefore, in addition to displaying distinct biochemical characteristics and substrate specificities, the IB-␣ kinases also appear to be pharmacologically distinct from other known kinases. Such an activity has not been reported before.
The IB-␣ kinases were capable of recognizing and binding IB-␣ in the absence of ATP. Deletion of either the N-terminal 54 or C-terminal 40 residues of the IB-␣ polypeptide did not abolish kinase binding, suggesting that the kinases bound to IB-␣ within the region encompassing the ankyrin repeats. Binding to a region distinct from the site of phosphorylation is a property shared with other cytoplasmic kinases that phosphorylate transcription factors, such as JNK (38). The exact role of the ankyrins within IB-␣ in the binding of the kinases is at this time unclear. A number of proteins in addition to IB-␣ have been found to contain variable numbers of ankyrin repeats similar to the tandemly linked 33-residue motifs first described in erythrocyte ankyrin proteins (50). Ankyrin repeats have been demonstrated to play an important role in proteinprotein interactions, and it is possible, therefore, that the ankyrin repeats in IB-␣ may not only play a role in interaction with subunits of NF-B (reviewed in Ref. 9) but that they may also play a role in kinase binding.
In transformed cell lines such as Jurkat, HeLa, and U937, IB-␣ degradation appears to require the phosphorylation of multiple residues on the IB-␣ polypeptide, with phosphorylation serving as a signal for the selective recognition and degradation of IB-␣ by components of the ubiquitin-proteasome system (17-24, 28, 29, 37). In vitro, IB-␣ can be phosphorylated by many different kinases, including protein kinases C and A (41), a protein kinase C--associated kinase (42), hemeregulated eukaryotic initiation factor-2␣ kinase (43), Raf-1 (44), and CKII (28,37). Both the 36-and 41-kDa IB-␣ kinases were able to bind, but not phosphorylate, a truncated form of IB-␣ lacking the C-terminal 40 residues. This suggests that both kinases phosphorylate residues within the C-terminal 40 residues of IB-␣. Phosphopeptide maps from IB-␣ phosphorylated in vitro with cellular extracts supported this finding, with greater than 90% of the label associated with peptides derived from the C-terminal 40 residues (data not shown). This region of IB-␣ is rich in PEST residues, a peptide motif frequently associated with rapid protein turnover (39). Phosphorylation of residues within this region occurs in vivo (28,29,37), and recent studies using either murine pre-B cell lines or Jurkat T cells have demonstrated a role for CKII in this phosphorylation (28,37). Based upon the substrate specificity, molecular size, kinetics of activation, and pharmacologic profile, it appears the kinases identified in this study are distinct from those reported to phosphorylate IB-␣ in vitro. Surprisingly, the IB-␣ kinases we identified do not appear to represent CKII. This conclusion is based upon the finding that the IB-␣ kinases in HUVEC cytoplasmic extracts could be purified separately from CKII by chromatographic techniques. Affinitypurified fractions containing the 36-and 41-kDa IB-␣ kinases contained little CKII activity, and no immunoreactive CKII protein could be detected in these fractions. CKII is known to phosphorylate a variety of nuclear factors including c-Jun (51). However, the affinity-purified endothelial IB-␣ kinases were unable to phosphorylate c-Jun in vitro and were instead highly specific for IB-␣ as a substrate. Based upon these findings, we conclude that the IB-␣ kinases we identified in human endothelial cells are distinct from CKII. These findings at first appear to be in conflict with those using transformed cell lines of murine and human origin. The studies we performed utilized primary cultures of human vascular endothelial cells, and this is the first report we are aware of where IB-␣ kinases have been identified in primary cultures of cells of human origin. It is possible, therefore, that some differences may exist in the nature or relative abundance of components of the IB-␣ regulatory system in primary, as compared with transformed, cell lines. Indeed, CKII expression is aberrantly elevated in many transformed cell lines (52)(53)(54), including many of those in which studies of the IB-␣ regulatory system have been undertaken. It will be important to examine the presence and identity of IB-␣ kinases in other human primary cells and to examine the presence of these kinases in various cell lines. In this respect, we have identified the presence of IB-␣ kinases similar to those in HUVEC in Hela and NIH3T3 cell lines and are currently characterizing these activities. 2 Recent evidence reveals that the C-terminal 40 residues of IB-␣ (residues 277-317), which include the phosphorylation sites for the kinases identified in this study, are entirely dispensable for TNF␣-induced degradation of IB-␣ and instead are important for constitutive turnover of IB-␣ in resting cells (28,56). A number of studies have demonstrated that ubiquitindependent degradation of IB-␣ is regulated by signal-induced phosphorylation at two specific residues, serine residues 32 and 36 within the N terminus of the IB-␣ polypeptide (15)(16)(17)(18)(19)(20)(21)(22)(23)(24)55). Indeed, during the conclusion of our studies, a large, multisubunit kinase that phosphorylates IB-␣ at Ser-32 and Ser-36 has been reported to be present in HeLa cells (57). Remarkably, this kinase requires the ubiquitin-activating enzyme (E1), a specific ubiquitin carrier (E2) of the Ubc4/Ubc5 family, and ubiquitin for activity. Because of their different patterns of phosphorylation of IB-␣, the IB-␣ kinases we identified might be predicted to play a role in the constitutive turnover of IB-␣ in resting endothelial cells through phosphorylation of residues in the PEST domain. However, we observed that the activity of the IB-␣ kinases was transiently elevated by TNF␣ treatment, suggesting that signal-induced phosphorylation of the C terminus of IB-␣ may occur. This process could enhance the rapid degradation of IB-␣ in activated endothelium. Elucidation of the molecular identity of these kinases and experimental manipulation of their activities in endothelial cells will be required to clearly define their potential role in IB-␣ degradation and regulation of NF-B.