Prostaglandin E2 (PGE2) production in immortalized, nontransformed cells derived from wild-type, cyclooxygenase 1–deficient (COX-1−/−) or cyclooxygenase 2–deficient (COX-2−/−) mice was examined after treatment with interleukin (IL)-1β, tumor necrosis factor α, acidic fibroblast growth factor, and phorbol ester (phorbol myristate acetate). Compared with their wild-type counterparts, COX-1−/− or COX-2−/− cells exhibited substantially enhanced expression of the remaining functional COX gene. Furthermore, both basal and IL-1–induced expression of cytosolic phospholipase A2 (cPLA2), a key enzyme-regulating substrate mobilization for PGE2 biosynthesis, was also more pronounced in both COX-1−/− and COX-2−/− cells. Thus, COX-1−/− and COX-2−/− cells have the ability to coordinate the upregulation of the alternate COX isozyme as well as cPLA2 genes to overcome defects in prostaglandin biosynthetic machinery. The potential for cells to alter and thereby compensate for defects in the expression of specific genes such as COX has significant clinical implications given the central role of COX in a variety of disease processes and the widespread use of COX inhibitors as therapeutic agents.
Prostaglandins, such as prostaglandin (PG)E2,1 are pivotal modulators of tissue homeostatsis and their aberrant regulation is known to cause serious pathophysiological consequences (1–4). PGE2 biosynthesis is regulated by successive metabolic steps involving the phospholipase A2– mediated release of arachidonic acid (AA) and its conversion to PGE2 by cyclooxygenase (COX), hydroperoxidase, and isomerase activities (1–4). Although cytosolic phospholipase A2 (cPLA2) is primarily responsible for agonist-induced AA release from membrane phospholipids (5, 6), secretory PLA2 may also be important in regulating AA availability via a transcellular mechanism (7). Conversion of AA to PGH2, the commited step in prostanoid biosynthesis, is mediated by cyclooxygenases, COX-1 and -2, which are encoded by two unique genes, located on different chromosomes (2). Generally, although COX-1 is constitutively expressed, the expression of COX-2 is highly inducible (2, 3). Based on their respective modes of expression, it is thought that COX-1 is primarily involved in cellular homeostasis, whereas COX-2 plays a major role in inflammation and mitogenesis. The COX isoenzymes are thought to be the primary target enzymes for nonsteroidal antiinflammatory drugs (NSAIDs), which act by inhibiting the COX activity of COX-1 and -2, thereby blocking their ability to convert AA to PGG2 (8, 9). In addition to the use of nonsteroidal antiinflammatory drugs as analgesics and for alleviation of acute and chronic inflammation, these agents have proven effective in decreasing the frequency of heart attacks and strokes (8, 10, 11), and in reducing the incidence of colon cancer (12, 13).
Since most cells invariably express both COX-1 and -2 under the appropriate conditions, it has been somewhat problematic to determine the exact contribution of the two COX isozymes towards basal and agonist-inducible PGE2 biosynthesis; the use of selective COX-1/COX-2 inhibitors to define relative contributions of the two isozymes has also resulted in limited success. The purpose of this study was to examine the effects of COX deficiency on the differential expression of the COX-1 and -2 isozymes, and compare the responses of wild-type, COX-1, and COX-2 knockout cells with respect to agonist-induced PGE2 biosynthesis. We demonstrate that the expression of COX, cPLA2, and PGE2 production are significantly increased in COX-deficient cells. Thus, COX deficiency, regardless of whether it is COX-1 or -2, results in the enhanced basal and inducible expression of the remaining COX isozyme as well as the elevated expression of cPLA2. We interpret these data to indicate that the elevated production of PGE2 in COX-1 or -2 isozyme-deficient cells is due to the compensatory expression of the remaining COX isozyme.
Materials And Methods
Isolation and Culture of COX-deficient Mouse Cells.
Lungs were collected from wild-type C57BL/6J (B6), COX-1–deficient (14), and COX-2–deficient (15) mice. The tissues were dissected into small pieces and grown underneath coverslips in 10-cm plates with MEM supplemented with PenStrep at a concentration of 300,000 U/liter penicillin G and 300 mg/liter streptomycin sulfate, nonessential amino acids (0.1 mM), Fungizone (1 mg/liter Amphotericin B), glutamine (292 mg/liter), ascorbic acid (50 mg/ liter), and 10% FCS in a humidified incubator with 5% CO2, and the media were changed three times per week. After 3 wk of culture under these conditions, only fibroblasts continued to grow. The PenStrep was reduced to 100,000 U/liter penicillin G and 100 mg/liter streptomycin sulfate and the media were replaced twice per week for another 3–5 wk. During subsequent passages, cells were maintained in DMEM containing high glucose and supplemented with PenStrep (100,000 U/liter penicillin G and 100 mg/liter streptomycin sulfate), nonessential amino acids (0.1 mM), Fungizone (1 mg/liter Amphotericin B), glutamine (292 mg/liter), ascorbic acid (50 mg/liter), and 10% FCS.
Transfection/Immortalization of COX-deficient Cells.
Subconfluent monolayers of COX-1−/− or COX-2−/− lung fibroblasts (passages 5–7) were cotransfected with pLE12S (containing adenovirus E1A gene; 8 μg/10-cm diam dish) and pREP4 (containing a hygromycin resistance gene; Invitrogen, Carlsbad, CA) plasmids (2 μg/10-cm diam dish) by the LipofectAMINE reagent (GIBCO BRL, Gaithersburg, MD). Plasmid pLE12S was a gift of Dr. Margaret Quinlan (University of Tennessee, Memphis, TN). Cells were maintained in the above media containing hygromycin (50 μg/ml) for 1 wk; hygromycin in the culture media was increased 50 μg/ml per week until the final concentration reached 250 μg/ml. Subsequent cell passage and subculture of cells used in all experiments was done in media containing 250 μg/ml hygromycin.
Treatment of COX-1−/− and COX-2−/− Cells with Cytokines and PMA.
Cells were seeded at 105 cells/ml in DMEM (high glucose) supplemented with PenStrep (100,000 U/liter penicillin G and 100 mg/liter streptomycin sulfate), nonessential amino acids (0.1 mM), Fungizone (1 mg/liter Amphotericin B), glutamine (292 mg/liter), ascorbic acid (50 mg/liter), 10% FCS, and 250 μg/ml hygromycin in 24-well (0.9 ml/well) flat-bottomed tissue culture plates (Costar, Cambridge, MA). Cells were incubated at 37°C in a humidified CO2 incubator (5% CO2) for 48 h until confluent. The medium was then replaced with fresh DMEM containing 0.5% FCS. Where indicated, cells were treated with IL-1 (0.25 ng/ml), TNF (5 ng/ml), acidic fibroblast growth factor (FGF; 10 ng/ml), or PMA (12.5 ng/ml) along with the appropriate vehicle controls; at these concentrations, neither cytokines nor PMA affected cell morphology or viability. In Western blot experiments, cells were treated as above except they were seeded in 6-well culture plates (2.7 × 105 cells/well).
Western Blot Analysis.
The medium from COX-deficient cells cultured in 6-well plates was aspirated and cell monolayers were washed with cold PBS and lysed in the Laemmli sample buffer. The samples were boiled for 3 min, and identical amounts of protein applied to SDS-PAGE (7.5%) and later transblotted to an Optitran membrane (Schleicher & Schuell, Keene, NH). The blot was blocked with 5% nonfat dry milk before incubation with either a rabbit antibody against murine COX-2 (Cayman Chemical, Ann Arbor, MI), a rabbit antibody against murine COX-1 (provided by Dr. D. DeWitt; reference 16), or mouse monoclonal anti-cPLA2 antiserum (Santa Cruz Biotechnology, Santa Cruz, CA). Proteins were detected using either enhanced chemiluminescence or enhanced chemifluorescence kits from Amersham Corp. (Arlington Heights, IL).
PGE2 in the media was measured by radioimmunoassay (RIA); this assay is based upon the competition by PGE2 in the test sample with labeled PGE2 for anti-PGE2 antibody binding sites. A 2–100-μl aliquot of culture medium was added to RIA assay buffer (0.1 mM phosphate buffer, pH 7.4, containing 0.9% sodium chloride, 0.1% sodium azide, and 0.1% gelatin), and mixed with the appropriate amount of labeled tracer and reconstituted antiserum. The mixture was incubated overnight at 4°C. Assay tubes were then placed in an ice bath, and 1 ml of cold charcoal-dextran suspension was added. After a 15-min incubation, the tubes were centrifuged at 2,200 g for 10 min at 4°C; the supernatants were decanted into scintillation vials, and radioactivity was determined by scintillation spectrometry. Percent binding was compared against a standard curve, and the amount of PGE2 in the sample was calculated. In each case the amount of PGE2 produced was normalized by cell number and all data is presented as picograms of PGE2 per 103 cells. To determine the potential effect of nonenzymatically produced, PGE2-immunoreactive products (e.g., isoprostanes) on RIA results, assays were performed in the presence of an effective COX-1 inhibitor (indomethacin, 1 mM) and NS-398 (1 μM), a COX-2 inhibitor, both of which block PGE2 biosynthesis, but should not block the production of nonenzymatically generated AA metabolites; products were further analyzed by radio-thin-layer chromatography (17).
Paired t test was used to determine the differences in the PGE2 levels between control samples of wild-type, COX-1−/− and COX-2−/− cells, and between control samples and samples from cytokine-treated cells. Differences were considered significant if P <0.05.
Results And Discussion
We examined the effects of IL-1 on PGE2 production in cells containing both COX isozymes (wild type) compared with cells that had only COX-1 (COX-2−/−) or COX-2 (COX-1−/−), respectively. As shown in Fig. 1,A both COX-1−/− or COX-2−/− cells synthesized 6–8-fold higher amounts of PGE2 compared with their wild-type counterparts. Interestingly, basal PGE2 production was higher in both COX-2−/− (66.71 ± 3.54 pg/103 cells; n = 6) and COX-1−/− (90.23 ± 3.29 pg/103 cells; n = 8) cells compared with wild-type (11.07 ± 0.62 pg/103 cells; n = 8). (All values are mean ± SE.) IL-1 treatment of wild-type and COX-1−/− cells further enhanced their PGE2 output. In contrast, IL-1 treatment of COX-2−/− cells did not significantly enhance PGE2 biosynthesis (Fig. 1 A).
These dramatic differences in basal PGE2 biosynthesis between wild-type and COX-deficient cells, prompted us to compare the expression of genes encoding three key enzymes (COX-1, COX-2, and cPLA2) regulating PGE2 biosynthesis in untreated and IL-1–treated wild-type, COX-1−/−, and COX-2−/− cells. A comparison of basal and IL-1– stimulated levels of COX-1 and COX-2 protein by immunoblot assay in wild-type and COX-1−/− cells is shown in Fig. 1, B and C. Consistent with numerous previous observations, the basal expression of COX-2 protein in wild-type cells was barely detectable (Fig. 1,B). Constitutive levels of COX-2 proteins were also significantly increased (2.4-fold) in untreated COX-1−/− cells (Fig. 1,B). The elevated level of COX-2 protein correlates well with the higher basal PGE2 levels in COX-1−/− cells compared with those in wild-type cells. When treated with IL-1, COX-2 protein levels increased moderately in wild-type cells (Fig. 1,B), but the increase in COX-2 protein was much more dramatic in COX-1−/− cells (41-fold). The overall pattern of COX-2 protein expression in wild-type and COX-1−/− cells correlated with increased PGE2 production seen in cells with unique COX phenotypes (see Fig. 1 A).
Next, we examined the basal and IL-1–stimulated levels of COX-1 protein in identically treated wild-type, COX-2−/−, and COX-1−/− cells, respectively (Fig. 1,C). In wild-type cell extracts, the level of COX-1 protein was barely detectable and IL-1 treatment was apparently inconsequential. This result was not unexpected since COX-1 expression is not known to be inducible under many conditions. We observed that basal expression of COX-1 protein in untreated COX-2−/− cells (Fig. 1,C) was much greater (14-fold) than that in wild-type cells. This overexpression of COX-1 protein corresponds with greater basal PGE2 levels in COX-2−/− cells, compared with the basal levels in wild-type cells. IL-1 had no stimulatory effect on COX-1 protein levels in COX-2−/− cells (Fig. 1 C) and as expected, COX-1−/− cells did not express detectable COX-1 protein. Another important enzyme in the prostaglandin biosynthetic pathway is PGE2 synthase, the isomerase that converts PGH2 to PGE2. Although PGE2 synthase has neither been sequenced nor cloned, making it difficult to study, available evidence does seem to indicate that this enzyme is not a rate-limiting reaction in PGE2 biosynthesis. However, based upon our findings, we cannot rule out the possibility that PGE2 synthase expression may also be altered in COX null cells.
To examine the possibility that iso-PGE2 or other isoprostanes (18) may be generated nonenzymatically from a buildup of endoperoxide intermediate that cross-reacts with the anti-PGE2 used in our RIA leading to erroneously high estimations of COX and/or PGE2 synthase activity, we performed two experiments. First, we treated wild-type and COX−/− cells with either indomethacin or NS-398, COX-1, and COX-2 selective inhibitors, respectively since these COX inhibitors should block PGE2 synthesis without affecting iso-PGE2 formation. We found that either indomethacin or NS-398 completely blocked both the basal and cytokine-induced formation of immunoreactive PGE2 in wild-type and COX−/− cells (data not shown). Second, radio-thin-layer chromatography was used to confirm that PGE2 was by far the predominate prostanoid product generated by wild-type and COX−/− cells and that no other AA metabolites in addition to PGE2 were generated in COX−/− cells (data not shown).
To compare the effects of IL-1 (see Fig. 1,A) to other inducers of PGE2 biosynthesis, we tested the effects of TNF, acidic FGF, and PMA on PGE2 production in wild-type, COX-2−/−, and COX-1−/− cells (Fig. 2). Compared to stimulated wild-type cells, there was significantly more PGE2 produced in either COX-1−/− or COX-2−/− cells, with the possible exception of TNF that induced comparable PGE2 biosynthesis in each cell type. In response to FGF, the amount of PGE2 that was produced by COX-2−/− cells was elevated and in COX-1−/− cells, PGE2 was even more dramatically elevated compared to wild type. Both COX-2−/− and COX-1−/− cells treated with PMA also produced much more PGE2 than wild type. Thus, in general, COX-isozyme deficiency results in increased PGE2 biosynthesis, but the relative contributions of COX-1 and COX-2 are clearly dependent upon the specific agonists involved.
Since constitutive COX-2 protein expression and PGE2 production in COX-1−/− cells was significantly enhanced, we were also curious about the status of cPLA2 gene expression in COX-deficient cells. We reasoned that cPLA2 activity could be involved in regulating levels of free AA for conversion to PGE2 and thereby could play a critical role in compensating for COX-isozyme deficiency. We were somewhat surprised to find that basal levels of cPLA2 protein in either COX-2−/− (Fig. 3,B) or COX-1−/− (Fig. 3,C) cells were significantly higher than levels of cPLA2 in wild-type cells (Fig. 3 A). It is conceivable, therefore, that enhanced expression of cPLA2 could directly contribute to higher PGE2 levels in both of the COX-deficient cells by generating greater AA substrate for PGE2 biosynthesis. Treatment of COX-1−/− or COX-2−/− cells with IL-1 resulted in a modest increase in the amount of cPLA2 protein (4-fold in COX-1−/− and 1.4-fold in COX-2−/−). This was in contrast to wild-type cells, which showed no change in the levels of cPLA2 protein after treatment with IL-1. As an important control, we examined the quantitative parameters of PGE2 production, and COX-1, COX-2, and cPLA2 gene expression in wild-type, COX-2−/−, and COX-1−/− cells from primary cell cultures and found essentially the same patterns in primary cells as those observed in the immortalized cells (data not shown). Therefore, the characteristic pattern of expression of COX-1, COX-2, and cPLA2 proteins in COX-2−/− and COX-1−/− cells is not elicited as a result of immortalization caused by the E1A adenovirus gene. Taken together these data indicate that COX-1−/− cells express enhanced levels of both basal and cytokine-stimulated COX-2 protein, and increased basal expression of cPLA2 protein. We postulate that the significantly increased levels of COX-2 and cPLA2 in COX-1−/− cells are likely to account for the increased rates of PGE2 biosynthesis; these data also implicate the existence of compensatory mechanisms for PGE2 production in COX-isozyme–deficient cells.
To distinguish between preferences of COX-1 and -2 for endogenous and/or exogenous AA for conversion to PGE2, and to verify that COX-1 was indeed expressed in COX-2−/− cells as judged by its ability to synthesize PGE2, we added exogenous AA to wild-type, COX-2−/−, or COX-1−/− cells that were either untreated or treated with IL-1, TNF, FGF, or PMA. The results shown in Fig. 4 demonstrate that cells expressing only COX-1 (COX-2−/−) synthesized similar amounts of PGE2 as wild-type or COX-1−/− cells supplemented with exogenous AA. Therefore, COX-1 is expressed and enzymatically active in COX-2−/− cells, but cytokines neither enhance COX-1 protein biosynthesis nor PGE2 biosynthesis. Thus, agonists that induce PGE2 biosynthesis in COX-2−/− cells in the absence of exogenous AA, do so by increasing endogenous substrate availability. Based on these data, we conclude that in COX-2−/− cells, substrate is likely to be limiting for constitutively expressed COX-1–mediated PGE2 biosynthesis. Fig. 4 also shows that COX-1−/− cells are able to use both exogenous and endogenous substrates (also see Fig. 1,A). However, IL-1, TNF, and FGF significantly enhanced the ability of COX-1−/− cells to produce PGE2, most likely by enhancing COX-2 expression as shown in Fig. 2. In addition, COX-1−/− cells treated with PMA did not produce elevated levels of PGE2, even when exogenous AA was provided. This indicates that PMA likely increased PGE2 production by increasing the availability of endogenous AA in COX-1−/− cells, whereas IL-1, TNF, and FGF likely affect AA mobilization and COX-2 expression. PMA affected the wild-type cells similarly. These results clearly raise the possibility that in COX-1 or -2 null cells, there is a coordinate upregulation of the expression and/or activities of COX-1, COX-2, and cPLA2, leading to increased PGE2 biosynthesis. These data also demonstrate that both COX-1−/− and COX-2−/− cells can effectively use AAs from either endogenous or exogenous sources.
Our data are consistent with the hypothesis that the long-term of COX-isozyme deficiency results in the altered expression of the remaining two enzymes that regulate mobilization and conversion of arachidonic acid to prostaglandins. Fig. 5 summarizes the patterns of COX and cPLA2 expression in COX null cells compared with normal cells in response to IL-1. The scheme shows the compensatory expression of the alternative COX isozyme and cPLA2 when one of the COX isozymes is absent. In cells lacking the housekeeping isozyme COX-1, overcompensation results in the overexpression of COX-2 and cPLA2, and in turn elevated PG biosynthesis. Similarly, cells lacking the inducible isozyme, COX-2, elicit the enhanced expression of COX-1 and cPLA2. Although we are uanble to comment as to the precise status of PGE2 synthase expression in wild-type, COX-1−/− or COX-2−/− cells, we have depicted its expression in each cell type; since PGE2 is the predominate prostanoid product, its expression would not appear to be rate-limiting given the great potential for PGE2 biosynthesis in the presence of exogenous AAs (see Fig. 4). Thus, our data clearly show that COX-deficient cells have the potential to overcome the lack of expression of one or the other COX isoenzymes by overexpressing the alternate COX isoform and increased cPLA2 expression. Such a potential mechanism for producing PGE2 by cells in vitro is not surprising since neither COX-1−/− (14) nor COX-2−/− mice (15) showed severe developmental arrest in utero or immediate postnatal mortality. However, in contrast to the results shown here using lung fibroblasts, Langenbach et al. (14) did not report any compensatory COX-2–mediated PGE2 production in glandular stomachs of COX-1–deficient mice, suggesting that tissue specificity may also be an important factor for further investigation. Together, these findings underscore the importance of elucidating the potential long-term effects of COX-1 or COX-2 inhibition with respect to alterations in the quantitiative and/or qualitative patterns of AA metabolism.
This work was supported by research funds from the Department of Veterans Affairs (DVA), The Arthritis Foundation, and grants AR39166 and AR26034 from the National Institutes of Health (National Institute of Arthritis and Musculoskeletal and Skin Diseases). R. Raghow is a Career Scientist of the DVA.
Address correspondence to Leslie R. Ballou, Department of Veterans Affairs Medical Center, Research Service (151), 1030 Jefferson Avenue, Memphis, TN 38104. Phone: 901-577-7283. Fax: 901-577-7273; E-mail: firstname.lastname@example.org
Abbreviations used in this paper: AA, arachidonic acid; COX, cyclooxygenase; cPLA, cytosolic phospholipase; FGF, fibroblast growth factor; PG, prostaglandin; RIA, radioimmunoassay.