Regulation of cyclooxygenases by protein kinase C. Evidence against the importance of direct enzyme phosphorylation.

Cyclooxygenases (COXs) are key prostaglandin biosynthetic enzymes. While COX-1 expression is largely constitutive, COX-2 is highly regulated by cytokines, growth factors, and tumor promoters, such as the protein kinase C (PKC) activator, phorbol 12-myristate 13-acetate (PMA). While phosphorylation of transcription factors may regulate COX transcription, the existence of PKC consensus sequences suggests that direct enzyme phosphorylation might also regulate differential expression of the enzymes. Nevertheless, phosphorylation of both human recombinant COX-1 and COX-2 by rat brain PKC in vitro was minimal, as was phosphorylation of peptides based on PKC consensus sequences in COX-1 (less than 4% of the phosphorylation of the PKC-α pseudosubstrate peptide). Similarly, phosphorylation of the corresponding COX-2 peptides was not observed using either the phosphocellulose paper absorption method or electrospray mass spectrometry. MEG-01 and NIH 3T3 cells were labeled with [32P]orthophosphate to investigate COX phosphorylation in vivo. COX-2 synthesis was induced by PMA (100 nM) or serum stimulation in NIH 3T3 cells. COX-1 was expressed constitutively in MEG-01 cells. Specific polyclonal antibodies raised against sequences of human COX-1 (Ala24-Cys35) and COX-2 (Asn580-Lys598) were used for immunoprecipitation. Neither COX-1 nor COX-2 was phosphorylated in vivo, irrespective of the presence of a phosphatase inhibitor (1 μM okadaic acid). Although COX-1 and COX-2 are differentially regulated, no differences were observed in terms of susceptibility to phosphorylation by PKC either in vitro or in vivo. Despite regulated expression of COX-2 by PMA and the existence of consensus sequences for PKC phosphorylation, it appears that it is an unfavorable substrate for this enzyme.

The unstable metabolites of arachidonic acid (prostaglandins, leukotrienes, and epoxyeicosatrienoic acids) are thought to be of importance in the modulation of complex biological responses, such as vascular homeostasis, ion transport, and inflammatory reactions in vivo (1). The major rate-limiting enzymes in prostaglandin formation are the cyclooxygenases (COXs). 1 Two COX genes have been cloned. COX-1 is generally expressed constitutively; however, its expression may be regulated by sex hormones and certain cytokines (2). COX-2, by contrast, is rarely expressed constitutively but is highly regulated by tumor promoters, growth factors, and cytokines (1). Given this discrepancy between the enzymes, COX-2 is thought to be the form that predominates in inflammatory states (3)(4)(5) and, perhaps, in cancer (6,7). Consequently, considerable effort has been invested in the development of selective inhibitors of this isoform (3,5,8).
The two human COX isoforms exhibit substantial (61%) amino acid identity. Much of the sequence disparity resides in the amino and carboxyl-terminal ends of the enzymes (9,10). Among the residues conserved are those critical to both peroxidase and cyclooxygenase activities and the serine target for aspirin acetylation (11). Resolution of the crystal structures of both enzymes (12,13) confirms a more accommodating active site in COX-2, which may explain a differential substrate affinity and a different profile of products formed by the enzymes following acetylation by aspirin (14,15).
The primary sequence of the COX cDNAs suggests several sites for posttranslational modification. Both enzymes contain multiple glycosylation sites (9,16,17). They may also be subjected to phosphorylation. Phosphorylation/dephosphorylation reactions are employed widely to regulate protein function (18). For example, the cytosolic phospholipase A 2 is phosphorylated in a mitogen-activated protein kinase-dependent manner (19), although the functional consequences of this modification are unclear (20). The COX enzymes, particularly COX-2, are also potential targets for this form of regulation. The tumor promoter, phorbol 12-myristate 13-acetate (PMA), is a potent activator of protein kinase C (PKC) (21) and induces COX-2 expression in many cell types (1). Potential serine and threonine targets for PKC exist in both COX enzymes; however, the 18-amino acid carboxyl-terminal extension of COX-2 contains several such target residues, including a PKC consensus sequence (22). Consistent with the possibility that PKC might regulate COX-2 expression, specific inhibition of the kinase or its depletion (23) prevents COX-2 induction by PMA or interleukin-1␣ in endothelial cells.
Given these observations, we wished to address the hypothesis that COX-2 was directly phosphorylated by PKC. Despite its regulation by PMA and the existence of potential target sequences for PKC, COX-2 is an unfavorable substrate for PKC, even during induction of the enzyme by PMA in vivo.

Materials
Aprotinin, leupeptin, pefabloc, soybean trypsin inhibitor, and the protein kinase A catalytic subunit were obtained from Boehringer * Supported by National Institutes of Health Grant HL54500 and a grant from the Southeastern Pennsylvania Affiliate of the American Heart Association (to R. V.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. ‡

In Vitro Phosphorylation Assays
In vitro phosphorylation of COX-1 and COX-2 was carried out as described previously (24)  Reactions were carried out at 37°C for the indicated times and terminated by trichloroacetic acid precipitation with 50 l of ice-cold 10% trichloroacetic acid. After incubation on ice for 30 min, the samples were centrifuged at 12,000 ϫ g for 10 min, and the pellets were dissolved in Laemmli sample buffer (26). Proteins were analyzed by SDS-PAGE (8% acrylamide). The gels were stained with Coomassie Blue, and the bands corresponding to COX were analyzed by densitometry. The gels were then dried and autoradiographed. The bands corresponding to COX were excised from the dried gel and counted for radioactivity, correcting for the amount of COX loaded in each lane. Accurate quantification of H1 phosphorylation is not possible in these experimental conditions, since the trichloroacetic acid precipitation of H1 is not complete; for this reason, the supernatants of H1 samples were also loaded in the gel. Quantitation of H1 phosphorylation was accomplished by summing the radioactivity of the precipitated sample with that of its supernatant.
Phosphorylation reactions using these peptides or the PKC-␣ pseudosubstrate peptide (RFARKGSLRQKNV), as a control, were performed using a phosphocellulose paper absorption method (27,28). Briefly, reactions were carried out in a 50-l volume at 30°C for 10 min in the presence of 250 g/ml bovine serum albumin (BSA), 50 mM Tris-HCl, pH 7.5, 100 M CaCl 2 , 7.5 mM magnesium acetate, 25 M ATP, 1 M PMA, 100 g/ml phosphatidylcholine/phosphatidylserine (4:1 ratio), 1 Ci of [␥-32 P]ATP (6000 Ci/mmol, 10 mCi/ml), and 0.3-0.5 l of rat brain PKC. Twenty-five l of reaction mixture was spotted onto a 1-inch square of phosphocellulose paper (P81, Whatman, Maidstone, UK) at the end of the incubation period. The paper was washed three times with 0.45% phosphoric acid, soaked in acetone, and counted for radioactivity. Samples containing no substrates were included in every experiment to evaluate the autophosphorylation of PKC. The radioactivity counted in these samples (basal) was subtracted from the counts obtained in the other samples.
When human recombinant COX phosphorylation was assessed with the phosphocellulose paper absorption method, BSA was omitted (due to the high phosphorylation background observed when it was included in the reaction mixture), while DTT 5 mM was added; 1 l of PKC/ sample was used. Reactions were carried out at 30°C for different times up to 60 min, and at the end of the incubation period, half of the reaction mixture was spotted onto P81 paper, while the other half was trichloroacetic acid-precipitated and analyzed by SDS-PAGE.
Human recombinant COX-1 and COX-2 were also used as substrates for protein kinase A. The reactions were performed as we have previously described (29). Briefly, phosphorylation was carried out in 50-l reactions containing 12.5 mM HEPES, pH 7.3, 5 mM MgCl 2 , 5 mM EGTA, 20 M ATP, 2 Ci of [␥-32 P]ATP, and 0.5 l of protein kinase A catalytic subunit. The reactions were allowed to proceed at 37°C for different times (0, 15, 30, 45, and 60 min) and terminated by trichloroacetic acid precipitation, as described above.

Reverse Phase HPLC/Electrospray Mass Spectrometry
Phosphorylation of the peptides was carried out as described above; at the end of the incubation time, samples were put on ice and then frozen and stored at Ϫ20°C until the analysis. A 20-l aliquot of a 30 M solution was separated using a 1.0 mm ϫ 25-cm Phenomenex Primesphere C18 column (Phenomenex, Torrance, CA). The HPLC system (Applied Biosystems 140B dual syringe pump; Applied Biosystems, Foster City, CA) was operated at a flow rate of 50 l/min starting with a mobile phase composition of 90% solvent A (0.05% trifluoroacetic acid in water) and 10% solvent B (0.05% trifluoroacetic acid, 90% acetonitrile in water), which was increased linearly to 40% solvent B in 30 min. The separation was monitored by a UV detector (Applied Biosystems 785A UV detector; Applied Biosystems) set at 214 nm and a VG Quattro II mass spectrometer (Fison Instruments, Beverly, MA) equipped with a coaxial electrospray probe and triple quadrupole analyzer further downstream. The sampling cone voltage was set to 30 V, the capillary voltage to 3.5 kV, and the source temperature to 75°C. The mass spectrometer was scanned continuously from m/z 300 to 1000 with a scan duration of 5 s.
The relative amount of the peptide phosphorylation was calculated from the selected ion chromatography peak area obtained in the liquid chromatography/mass spectrometry experiments.

Measurement of COX-1 and COX-2 Activity in Vitro
PGF 2␣ production was measured after incubation of COX-2 with arachidonic acid to determine if PKC phosphorylation could affect enzymatic activity. Two g of recombinant COX-2 were incubated with PKC as described above, with the difference that no [␥-32 P]-ATP was added. DTT was also omitted from the reaction mixture, because it impairs COX activity. The absence of DTT only slightly decreased phosphorylation, as detected by autoradiography, in a selected experiment in which [␥-32 P]ATP was added to the reaction mixture. Phosphorylation reactions were stopped after a 1-h incubation at 37°C; half of the sample (25 l) was mixed with 225 l of a buffer containing 50 mM Tris-HCl, pH 8, 1 mM phenol, 1 M hematin, and 1 mg/ml BSA. Peroxide-free arachidonic acid (100 M) was added, and incubations were carried out for 1 min at 37°C. Reactions were stopped by the addition of 150 l of glacial acetic acid. PGF 2␣ was extracted with 3 ml of ether containing 1 mg of triphenylphosphine, and derivatization was carried out as we have described previously (30). PGF 2␣ was analyzed by gas chromatography mass spectrometry using a derivatized 18 O 2 -labeled PGF 2␣ internal standard.

In Vivo Phosphorylation Experiments
COX-2-NIH 3T3 fibroblasts were grown in 60-mm culture dishes in Dulbecco's modified Eagle's medium (DMEM) containing 2 mM glutamine, 100 units/ml penicillin, 100 g/ml streptomycin, and 10% fetal bovine serum at 70% confluence and then shifted to DMEM with 1% serum for 48 h. Cells were washed and incubated for 3 h at 37°C with 32 P as orthophosphate (aqueous solution, carrier-free, specific activity 10 mCi/ml), 0.3 mCi/ml in DMEM without sodium phosphate. COX-2 synthesis was induced either by PMA (100 nM) stimulation of serumstarved cells or by stimulation of the cells with 20% serum in the presence or absence of 100 nM PMA or of the phosphatase inhibitor, okadaic acid (1 M). Me 2 SO was added at the same concentration (0.1%), irrespective of whether the cells were treated with PMA. Cells were washed in phosphate-buffered saline and lysed with 150 l of ice-cold lysis buffer (50 mM Tris-HCl, pH 8, 1 mM EDTA, 1% Nonidet P-40, 0.5% SDS, 0.5 mM DTT, 10 g/ml soybean trypsin inhibitor, 10 g/ml aprotinin, 1 g/ml leupeptin, 0.5 mM 4-(2-aminoethyl)-benzensulfonyl fluoride hydrochloride (pefabloc), 1 mM NaF, 1 mM sodium orthovanadate) at the end of the incubation period. Cells were scraped, and the DNA was sheared by passing the cell lysates through a 21gauge needle. The lysates were then centrifuged for 15 min at 12,000 ϫ g at 4°C, and the protein concentration in the supernatant was determined using a microbicinchoninic acid assay (Pierce) with BSA as a standard. Immunoprecipitation was carried out as we have previously described (17), using a specific polyclonal antibody, kindly donated by Dr. Jacques Maclouf. Ten mg of protein A-Sepharose CL-4B was incubated with 25 l of a specific rabbit polyclonal antibody raised against the C-terminal sequence of human COX-2 (Asn 580 -Lys 598 ) in 500 l of buffer (150 mM NaCl, 50 mM phosphate buffer, pH 7.4). After an overnight incubation at 4°C, protein A-Sepharose was washed five times with radioimmune precipitation buffer buffer (50 mM Tris-HCl, pH 8, 150 mM NaCl, 1% Nonidet P-40, 0.5% deoxycholic acid, 0.1% SDS) and incubated for 3 h at 4°C with cell lysates (600 g of the proteins) in 500 l. After five washings in radioimmune precipitation buffer, protein A-Sepharose was mixed with Laemmli buffer (26) under reducing conditions, heated for 5 min at 95°C, and subjected to 10% acrylamide SDS-PAGE. Thirty l of the supernatants of the immunoprecipitation reactions were also run in the gel. The gel was transferred onto a nitrocellulose membrane (Schleicher & Schuell, Göttingen, Germany) with a semidry transfer unit (Hoefer Scientific Instruments, San Francisco, CA). Transfer was performed in 25 mM Tris, 192 mM glycine buffer, pH 8.3, containing 0.01% SDS and 20% methanol for 1.5 h at 200 mA. Blots were stained with 0.4% Ponceau Red in 0.3% trichloroacetic acid and exposed to Kodak X-Omat film at Ϫ80°C. Immunoblotting was performed to assess the effectiveness of COX-2 immunoprecipitation. Western blots were saturated overnight at 4°C with 5% nonfat dry milk in Tris-buffered saline (50 mM Tris-HCl, pH 7.4, 250 mM NaCl, 0.1% Tween 20) and incubated with a mouse monoclonal antibody (1:2000) in Tris-buffered saline containing 0.1% Tween 20 and 1% milk for 1 h at room temperature. Blots were washed in the same buffer without milk and incubated with a peroxidase-conjugated donkey anti-mouse IgG diluted 1:5000 in the buffer containing milk for 1 h. After washing to eliminate the excess of antibody, enhanced chemiluminescence substrates were used to reveal positive bands according to the manufacturer's instructions, and bands were visualized after exposure to Kodak BIO-MAX films.
COX-1-COX-1 phosphorylation was assessed in MEG-01 cells rather than NIH 3T3 fibroblasts for two reasons. First, we found that MEG-01 cells express a higher amount of COX-1 compared with NIH 3T3 cells in preliminary immunoblotting experiments. Second, immunoprecipitation of NIH 3T3 cell lysates with a specific rabbit polyclonal antibody raised against the N-terminal sequence of human COX-1 (Ala 24 -Cys 35 ), was suboptimal, possibly due to a low affinity of the antibody for the native murine protein.
MEG-01 cells were grown in RPMI 1640 containing 2 mM glutamine, 100 units/ml penicillin, 100 g/ml streptomycin, and 10% fetal bovine serum. Seven million cells/sample were washed and labeled in 3 ml of DMEM without sodium phosphate with 0.3 mCi/ml 32 P as orthophosphate as described above, in the presence or in the absence of 100 nM PMA and 1 M okadaic acid. 0.1% Me 2 SO was added to the control sample.
Cells were incubated 3 h at 37°C, washed, and lysed in Nonidet P-40 buffer (50 mM Tris-HCl, pH 8, 150 mM NaCl, 1% Nonidet P-40) containing protease and phosphatase inhibitors as described above. Nonidet P-40 buffer was chosen for its low detergent concentration to retain the antibody binding sites and to avoid solubilization of background proteins. After lysis of the cells, double immunoprecipitation was carried out using a rabbit polyclonal antibody raised against the N-terminal sequence of human COX-1 (Ala 24 -Cys 35 ), kindly donated by Dr. Jacques Maclouf. The immunoprecipitation steps were essentially those described above, with the exception that protein A-Sepharose was washed three times in Nonidet P-40 buffer instead of in radioimmune precipitation buffer; the first immunoprecipitation was performed for about 7 h, while the second was overnight. Samples were analyzed by SDS-PAGE, and the gel was blotted as described above. The membrane was autoradiographed and then subjected to immunoblotting to visualize the COX-1 protein. Immunoblotting was performed as described above, except that the antibodies were diluted in Tris-buffered saline containing 0.1% Tween 20 but no milk. A monoclonal anti COX-1 antibody was used at 5 g/ml, and a peroxidase-conjugated donkey anti-mouse IgG was used at a 1:5000 dilution. Bands were visualized by enhanced chemiluminescence, as described above.

RESULTS
In Vitro Phosphorylation Experiments-Human recombinant COX-1 and COX-2 were used as substrates for purified rat brain PKC in in vitro assays. Autoradiography of the gels revealed an apparent time-dependent phosphorylation of both COX-1 and COX-2 (Fig. 1). However, the radioactivity associated with the bands corresponding to COX-1 and COX-2 was low and accounted for less than 30% of that found in phosphorylated histone after a 60-min incubation (Table I). COX-1 and COX-2 did not appear to be phosphorylated by protein kinase A (data not shown).
COX-1 and COX-2 phosphorylation by PKC was also assessed by the phosphocellulose paper absorption method. No radioactivity above basal level was observed for either COX-1 or COX-2 by PKC in concentrations of 0.3-10 g/50 l when the incubation period was prolonged for up to 30 min. Incubation of COX-2 with PKC did not modify enzyme activity as assessed by the formation of PGF 2␣ (data not shown).
To assess phosphorylation of the enzymes in more detail, we synthesized peptides based on the putative consensus sequences for PKC-dependent phosphorylation (22). Phosphorylation of the peptides was first evaluated by phosphocellulose paper absorption. The PKC-␣ pseudosubstrate peptide was highly phosphorylated by PKC, with a K m of 6.5 Ϯ 1.5 M (n ϭ 3). The COX-2 peptides (3-100 M), by contrast, were not phosphorylated by PKC. Similarly, phosphorylation of two of the four COX-1 peptides (HQFFKTSGKMG and VIRES-REMRL) was not observed. The other two COX-1 peptides that we studied (TWLRNSLRPSP and IVKTATLKKLV) were slightly phosphorylated by PKC. However, the radioactivity associated with these peptides was very low; it accounted for less than 4% of the radioactivity associated with the PKC-␣ pseudosubstrate peptide used at the same concentration (Table  II). Given this observation and that we failed to reach a plateau of phosphorylation at concentrations of peptides up to 1.3 mM, we could not calculate an accurate K m . The phosphorylation of peptide TWLRNSLRPSP (10 M) remained low (0.4% of that observed in the positive control), even when the period for phosphorylation was extended to 30 min.
Peptides were also analyzed by electrospray mass spectrometry. The phosphorylation of the PKC-␣ pseudosubstrate peptide could be readily observed with this method. When the peptide was preincubated with PKC, a shift of the molecular mass (from 1559.6 to 1639.8 daltons) consistent with the addition of one phosphate group was observed in 85% of the molecules. By contrast, phosphorylation of the two COX-2 peptides or of two of the COX-1 peptides was not detected.
Consistent with the phosphocellulose paper absorption method, phosphorylation of two COX-1 peptides was also detected by electrospray mass spectrometry (Fig. 2). Thus, the molecular mass of TWLRNSLRPSP increased from 1325.2 to 1405.2 daltons in the presence of PKC, and the molecular mass of IVKTATLKKLV increased from 1212.4 to 1292.8 daltons.  However, only 12 and 1.9% of the molecules were phosphorylated, respectively. In Vivo Phosphorylation Experiments-We also studied the possible phosphorylation of COX-1 and COX-2 in vivo. To determine whether COX-1 and COX-2 were phosphorylated, we immunoprecipitated these enzymes from 32 P-labeled MEG-01 and NIH 3T3 cells, respectively.
No phosphorylated bands were detected in the immunoprecipitated NIH 3T3 cell lysates, either when COX-2 synthesis was induced by PMA (100 nM) in serum-starved cells (Fig. 3A, upper part) or by serum stimulation, even in the presence of PMA (100 nM), okadaic acid (1 M), or a combination of both (Fig. 3B, upper part). Okadaic acid, a serine/threonine phosphatase inhibitor, was added to exclude a rapid dephosphorylation reaction, which could theoretically account for a lack of detectable phosphorylation. To verify that COX-2 was actually immunoprecipitated, immunoblotting was performed, demonstrating the presence of COX-2 in the immunoprecipitated samples (Fig. 3, A and B, lower parts). The efficiency of cell labeling was demonstrated by the high protein phosphorylation observed in the supernatants of the immunoprecipitated samples (Fig. 3, A and B, upper parts).
Similar results were obtained when COX-1 was immunoprecipitated with a specific antibody from 32 P-labeled MEG-01 cell lysates; no radioactivity associated with COX-1 was observed, even in cells treated with PMA (100 nM) with or without okadaic acid (1 M) (Fig. 4, upper panel). The effectiveness of the immunoprecipitation was again verified by immunoblotting (Fig. 4, lower panel). DISCUSSION COX-2 regulation is thought to be of importance in the elaboration of prostanoids in inflammation and, perhaps, in cancer. It is likely that both the expression and function of such an enzyme is highly controlled. This may include phosphorylation of transcription factors, such as the nuclear factor-B (NF-B), c-Jun, c-Fos, and CREB (31). Although promoter analysis of the COX-2 gene has been relatively limited, recognition motifs for NF-B, the nuclear factor for interleukin-6 expression (NF-IL6), PMA, and cyclic AMP have been described in the human COX-2 gene (32). In addition to a role for phosphorylation in transcriptional activation of the COX-2 gene, it is possible that the enzyme would be subject to posttranslational modification by direct phosphorylation. Thus, precedent has been established for the biological significance of such modifications, most notably modulation of the interaction of heptahelical receptors with G proteins during receptor desensitization (33,34). Additionally, enzyme function or localization may be modified by  phosphorylation. Thus, the cytosolic phospholipase A 2 , a form of the enzyme with a high affinity for phospholipids containing arachidonic acid, acting at a step proximal to that of the COXs, is phosphorylated by mitogen-activated protein kinase (19). Similarly, direct phosphorylation of adenylate cyclase by PKC (35) or protein kinase A (36) may mediate cross-talk among G protein-coupled receptors.
COX-2 is induced by diverse growth factors, tumor promoters, and cytokines (1), albeit with different kinetics (17). The molecular events that mediate this response are poorly understood. Recently, Blanco et al. (23) have demonstrated that COX-2 induction by PMA and interleukin-1␣ in endothelial cells appears to be dependent upon both PKC and tyrosine phosphorylation. Thus, putative inhibitors of PKC and of tyrosine kinases prevented COX-2 induction by PMA and interleukin-1␣. Depletion of PKC by prolonged incubation with PMA had a similar effect. The potent induction of COX-2 expression by short term incubation with PMA, known to result in PKC activation (4,9,10,17) also suggests a role for PKC-dependent phosphorylation in COX-2 regulation. The possibility of direct phosphorylation of COXs was suggested by the existence of some conventional PKC consensus sequences in the enzymes, together with additional potential target serines and threonines outside the boundaries of these motifs. Intriguingly, five serines and four threonines are contained within the 18-amino acid carboxyl-terminal tail extension that is peculiar to COX-2. It also includes a conventional PKC consensus sequence.
Several methodological approaches were used in our study to document COX phosphorylation. The first of these was conventional: incubation of the COX enzymes with PKC in the presence of radioactive ATP, electrophoresis, and autoradiography. Apparent phosphorylation of both enzymes was detected. However, this was modest when compared with a positive control (histone H1) substrate. Furthermore, when using the phosphocellulose paper absorption method, we failed to detect phosphorylation of either enzyme. These experiments indicated that quantitative phosphorylation of the whole enzymes by PKC was, at best, minimal and suggest that misleading results could be obtained for such in vitro assays. Experiments with protein kinase A also suggested that COXs were unfavorable substrates for this kinase.
To address the potential for phosphorylation in more detail, we synthesized peptides based on the putative consensus sequences for PKC in both enzymes. We chose peptides in which a serine or a threonine is surrounded by positively charged residues, such as arginine or leucine (37). Using the phosphocellulose paper method, we failed to detect phosphorylation in all but two of the COX-1-based peptides (TWLRNSLRPSP and IVKTATLKKLV). These were slightly phosphorylated. Again, the degree of phosphorylation was trivial by comparison with that of a PKC-␣ pseudosubstrate peptide. Internally consistent results were obtained with electrospray mass spectrometry, which confirmed trivial phosphorylation of only the two COX-1 peptides by PKC.
Although these experiments suggested that both COX enzymes were unfavorable substrates for PKC-dependent phosphorylation in vitro, we wished to address the issue in vivo. Particularly, we were interested in investigating this possibility during COX-2 induction by PMA, a known stimulator of PKC (21). We did this in NIH 3T3 cells in the presence and absence of okadaic acid, a phosphatase inhibitor. Rapid activation of phosphatases (38) could theoretically confound the interpretation of such experiments. COX-2 was induced, either by PMA in serum-starved cells or by incubation with serum. However, immunoprecipitated COX-2 was not phosphorylated in either case, despite presumptive activation of endogenous PKC by PMA (39,40). Similar results were obtained in MEG-01 cells for COX-1. Induction is not necessary in these cells, since COX-1 is expressed constitutively. However, immunoprecipitation of the enzyme indicates that it is not phosphorylated.
Despite the existence of PKC consensus sequences, enzyme induction by PKC activators, and prevention of induction by PKC inhibition and depletion, our results indicate that COX-2 is an unlikely substrate for direct phosphorylation by this enzyme in vitro or in vivo. Should phosphorylation play a role in COX expression, it is likely to involve other kinases and/or act at the level of transcription.