Cytoplasmic Prostaglandin E 2 Synthase Is Dominantly Expressed in Cultured KAT-50 Thyrocytes, Cells That Express Constitutive Prostaglandin-endoperoxide H Synthase-2 BASIS FOR LOW PROSTAGLANDIN E 2 PRODUCTION*

The recent identification and cloning of two gluta-thione-dependent prostaglandin E 2 synthase (PGES) genes has yielded important insights into the terminal step of PGE 2 synthesis. These enzymes form efficient functional pairs with specific members of the prostag-landin-endoperoxide H synthase (PGHS) family. Microsomal PGES (mPGES) is inducible and works more efficiently with PGHS-2, the inflammatory cyclooxygenase, while the cytoplasmic isoform (cPGES) pairs functionally with PGHS-1, the cyclooxygenase that ordinarily exhibits constitutive expression. KAT-50, a well differentiated thyroid epithelial cell line, expresses high levels of PGHS-2 but surprisingly low levels of PGE 2 when compared with human orbital fibroblasts. Moreover, PGHS-1 protein cannot be detected in KAT-50. We report here that KAT-50 cells express high basal levels of cPGES but mPGES mRNA and protein are undetectable. Thus, KAT-50 cells express the inefficient PGHS-2/ cPGES pair, and this results in modest PGE 2 production. The high levels of cPGES and the absence of mPGES expression result from dramatic differences in the activities of their respective gene promoters. When mPGES is expressed in KAT-50 by transiently transfecting the cells, PGE 2 production is up-regulated

The field of prostaglandin biology has benefited enormously from the recent identification and cloning of two prostaglandinendoperoxide H synthase (EC 1.14.99.1, PGHS) 1 isoenzymes (1-6) and two forms of prostaglandin-E 2 synthases (EC 5.3.99.3, PGES) (7)(8)(9). The activities of PGHS involve the conversion of arachidonate, first into PGG 2 and subsequently into PGH 2 in two rate-limiting steps, catalyzed by discrete active sites on these enzymes (10). PGHS-1 is a constitutive enzyme involved in many "housekeeping" functions in most of the tissues and cell types thus far examined (1,2). In contrast, PGHS-2, the inflammatory cyclooxygenase, is expressed at extremely low levels under basal conditions in most cell types but is highly inducible by cytokines and growth factors (3)(4)(5)(6). The PGES enzymes are glutathione-dependent and involved in the conversion of PGH 2 to PGE 2 . cPGES localizes to the cytosol and is a constitutively expressed enzyme, identical to p23, a chaperone for the hsp90/glucocorticoid receptor complex (8). On the other hand, mPGES represents a microsomal protein, is inducible by cytokines, and glucocorticoid repressible (9). A number of different cell types have been examined and found to express both mPGES and cPGES (7)(8)(9).
A functional association linking PGHS-1 with cPGES and PGHS-2 with mPGES has been made on the basis of studies conducted in transfected cells overexpressing the enzymes (8,9). The authors of those earlier studies concluded that these associations yield efficient enzyme pairs. These initial findings have been extended subsequently to human synovial fibroblasts where treatment with IL-1␤ or tumor necrosis factor-␣ up-regulated mPGES mRNA and protein in a manner that is coordinated with an induction of PGHS-2 (11). More recently, this relationship between PGHS-2 and mPGES has also been demonstrated in orbital fibroblasts (12), cells that produce particularly high levels of PGE 2 when provoked by inflammatory cytokines such as IL-1␤, leukoregulin, and CD154 (13)(14)(15). In those cells, an overlap in the signal transduction pathways utilized by IL-1␤ to induce both enzymes was demonstrated (12). This signaling involved both p38 and ERK1/2 components of the mitogen-activated protein kinase pathways (12). Moreover, the induction by IL-1␤ of mPGES in orbital fibroblasts depends, at least in part, on the activity of PGHS-2. Glucocorticoids can block the up-regulation by IL-1␤ of both PGHS-2 and mPGES in synovial and orbital fibroblasts (11,12). Because these fibroblasts are believed to play important roles in the pathogenesis of rheumatoid arthritis and Graves' ophthalmopathy, respectively, it is tempting to implicate both PGHS-2 and mPGES in these disease processes.
KAT-50, an established line of human thyroid epithelial cells derived from non-neoplastic tissue, expresses high levels of constitutive PGHS-2 (16,17). Moreover, basal generation of PGE 2 in these cells can be substantially diminished by treating them with PGHS-2 selective inhibitors, such as SC58125. In fact, several types of mammalian cells exhibit high levels of unstimulated PGHS-2 expression, including hepatic stellate cells (18), bronchial epithelium (19), pancreatic islet (20), and granulosa cells (21). While the function of unprovoked PGHS-2 is uncertain, this enzyme may serve housekeeping roles under certain circumstances and therefore might replace the functions ordinarily ascribed to PGHS-1. In fact, PGHS-1 protein is undetectable in either un-stimulated or cytokine-activated KAT-50 cells. Moreover, most of the PGE 2 production detected in KAT-50 cells derived from the action of PGHS-2 (16). Thus KAT-50 cells represent a growing list of cell types, both well differentiated and neoplastic, where PGHS-2 represents the dominant, constitutive cyclooxygenase. Of particular interest is the lack of up-regulation of PGHS-2 expression in KAT-50 cells by factors such as serum and IL-1␤, agents that ordinarily enhance cyclooxygenase expression in most cell types. It was noted in the course of earlier studies that despite the high basal PGHS-2 levels, KAT-50 cells produce rather low levels of PGE 2 (16). Even when treated with exogenous arachidonate, KAT-50 cultures were found to generate PGE 2 levels substantially lower than those found in orbital fibroblasts when high PGHS-2 levels are induced. These observations suggested that some critical component(s) of the prostanoid synthetic machinery in KAT-50 might be suboptimally represented.
In this paper, we report that KAT-50 cells, despite expressing high levels of PGHS-2, produce very low levels of PGE 2 compared with orbital fibroblasts. The basis for this diminished prostanoid generation relates to an absence of detectable mPGES expression. Instead, KAT-50 cells express high levels of cPGES mRNA and protein. This disparity results from dramatically different activities of the respective PGES gene promoters. Thus, in these cells, PGHS-2 and cPGES are predominantly co-expressed, creating an inefficient enzyme pair. When mPGES expression was introduced by transiently transfecting KAT-50 cells, PGE 2 production was enhanced substantially. These observations suggest that a collaboration of PGHS-2 and cPGES can occur naturally in some cells. This pairing results in relatively low levels of PGE 2 generation.
Cell Culture-KAT-50 cells were a generous gift from Dr. K. Ain (University of Kentucky, Lexington, KY) (22). They were maintained in a humidified, 5% CO 2 incubator at 37°C covered with Eagle's medium (Invitrogen) supplemented with 10% fetal bovine serum (FBS) and antibiotics. These cells have been characterized as expressing, among other thyroid markers, thyroglobulin mRNA and the sodium/iodine symporter, and were supplied to us in January 1997 at passage 8 (22). Cells were serially passaged with gentle trypsin/EDTA treatment and were utilized between the 10 th and 35 th passage. Medium was changed every 3-4 days. Human orbital fibroblasts were obtained from explants of surgical waste emanating from individuals with severe thyroid-associated ophthalmopathy. Fibroblasts were cultivated as described previously (23) in Eagle's medium supplemented with 10% FBS, glutamine, and antibiotics. They were used between the 2 nd and 12 th passages.
Western Blot Analysis of Protein Expression-Relative levels of the cyclooxygenase and PGES proteins in fibroblasts and KAT-50 cells were determined by Western immunoblot analysis utilizing monoclonal an-tibodies generated against human PGHS-1 and PGHS-2, obtained from Cayman Chemical. Antisera against cPGES and mPGES were utilized as the primary antibodies for detecting the two terminal synthase enzyme proteins. Confluent cultures, usually cultivated in 60-mm-diameter plates, were shifted to serumless medium overnight. Monolayers were washed, harvested, and cellular protein was solubilized in an ice-cold buffer containing Nonidet P-40 (1%), 50 mM Tris-HCl (pH 7.5), and 200 M phenylmethylsulfonyl fluoride. Lysates were taken up in Laemmli buffer and subjected to SDS-polyacrylamide gel electrophoresis (SDS-PAGE), and the separated proteins were transferred to Immobilon-P membrane (Millipore, Medford, MA). The primary antibodies (cPGES, 1:500; mPGES, 1:1000; PGHS-1, 1:1000; PGHS-2, 1:1000) were incubated with the membranes at 4°C overnight, and these were then washed extensively and re-incubated with secondary, peroxidase-labeled antibodies for 1 h. Following washes, the ECL (Amersham Biosciences) detection system was used to generate the specific signals. Resulting bands were quantified densitometrically.
Northern Blot Analysis-Total cellular RNA was extracted using a method published by Chomczynski and Sacchi (24) from near-confluent 100-mm diameter plastic plates of KAT-50 cells and orbital fibroblasts that had been incubated without or with the test compounds indicated in the figure legends. Monolayers were covered with a solution containing guanidium isothiocyanate (ULTRASPEC, Biotecx Laboratories, Houston, TX), and RNA was precipitated from the aqueous phase by addition of isopropanol, washed with 75% ethanol, and solubilized in diethyl pyrocarbonate-treated water. Equal amounts of RNA (usually 10 -20 g) were electrophoresed in 1% agarose formaldehyde gels and transferred to Zeta-Probe (Bio-Rad) membrane. The integrity of the electrophoresed RNA was verified by UV inspection following ethidium bromide staining. [ 32 P]dCTP random-primed (Bio-Rad) PGHS and PGES probes were hybridized in ExpressHyb hybridization solution (CLONTECH, Palo Alto, CA) at 68°C for 1 h. Membranes were washed under high stringency conditions and exposed to X-Omat AR film (Eastman Kodak Co.) at Ϫ70°C with intensifier screens. To normalize the amounts of RNA transferred, membranes were stripped according to the manufacturer's instruction, or radioactivity was allowed to decay and the RNA was rehybridized with a radiolabeled human glyceraldehyde-3phosphate dehydrogenase cDNA probe. Radioactive DNA/RNA hybrids were quantified by subjecting autoradiographs to densitometric analysis.
Transfection of PGHS-2, mPGES, and cPGES Promoter Plasmid Constructs and mPGES cDNA into KAT-50 Cells-Cultured cells were allowed to proliferate to a state of 60 -80% confluence in six-well plates covered with medium containing 10% FBS, and then monolayers were washed with serum-free medium. Transfections were performed using LipofectAMINE PLUS (Invitrogen) following the manufacturer's instructions. Plamids utilized for these studies included Ϫ1800pGL2, containing Ϫ1840/ϩ123, and is thus 5 base pairs upstream from the ATG of the human PGHS-2 promoter (25,26). A 510-bp fragment of the human mPGES promoter spanning Ϫ538 to Ϫ28 was cloned with the Human GenomeWalker kit (CLONTECH) according to the instructions provided by the supplier. Two mPGES reverse primers used for the PCR reactions included 5Ј-CGCAGCTCAACTGTGGGTGTGATC-3Ј and 5Ј-GTGATCAGCTCGACAGAGGAGCAG-3Ј. An 1824-base fragment of the human cPGES promoter, spanning Ϫ1893 to Ϫ69, was PCR amplified using primers 5Ј-CAGTGCGCCAAGTTAATTGAGACC-3Ј and 5Ј-GTCGACTTCTCTCCGGTGGCGACT-3Ј. Amplified fragments were sequenced and cloned into pCR2.1-TOPO vector (Invitrogen). These were then cloned into the promoter-less luciferase vector, pGL2. Control pGL2 was used to determine the basal level of luciferase activity exhibited by these cells. To determine promoter activities, DNA was added to 100 l of serum-free medium and combined with 6 l of PLUS reagent and incubated at room temperature for 15 min. 4 l of Lipo-fectAMINE was added to 100 l of serum-free medium without antibiotics, and the suspension was mixed gently with the DNA/PLUS mixture. Complexes were allowed to form over 15 min at room temperature before being added to the washed monolayers. Cultures were incubated for 3 h at 37°C in 5% CO 2 atmosphere, and then the transfection mixture was removed and replaced with medium supplemented with 10% FBS. Cultures were incubated for 24 h, cell monolayers were washed in PBS, and 1ϫ passive lysis buffer (Promega) was added and cells scraped off the plates. The extract was vortexed and frozen at Ϫ80°C. Samples were thawed and assessed for luciferase activity in a Sirius luminometer (Berthold Detection System, Pforzheim, Germany). The luciferase activities were corrected for their respective Renilla luciferase levels and therefore should reflect the relative transfection efficiency.
For the mPGES expression studies, the open reading frame of the human mPGES cDNA was cloned into pcDNA3.1(Ϫ) (Invitrogen). Ei-ther empty vector or mPGES/pcDNA3.1(Ϫ) (0.6 g) was mixed with LipofectAMINE PLUS as described above. In control experiments conducted under identical conditions, cultures received the expression plasmid pIRES2-EGFP (CLONTECH) containing the coding region for green fluorescent protein (GFP) (0.6 g). Transfection efficiency was determined by trypsinization of the cells and manually counting those expressing GFP. The efficiency was used to normalize the data presented in Fig. 4A (inset) concerning PGE 2 production in cells before and after transfection with mPGES. PGE 2 Assay-PGE 2 levels were determined as described previously (12) utilizing an enzyme immunoassay (Amersham Biosciences). Briefly, KAT-50 cells were incubated in 24-well plates covered with medium supplemented with 10% FBS. One day prior to experimental manipulation, culture wells were shifted to medium without FBS, and the following day, the test compounds indicated in the figure legends were added. Thirty min prior to monolayer harvest, medium was removed and replaced with 150 l of PBS with the respective additives. Following the incubation, the PBS was removed quantitatively and subjected to the assay for PGE 2 following the manufacturer's instructions. These studies were conducted with three separate wells per treatment group. Data are expressed as the mean Ϯ S.D. of triplicate cultures from representative experiments.

KAT-50 Cells Generate Considerably Less PGE 2 than Do
Cytokine-activated Fibroblasts-KAT-50 cells express extraordinarily high levels of PGHS-2 under basal culture conditions. Despite high levels of enzyme, these cultures generate and release low levels of PGE 2 (Fig. 1). Moreover, IL-1␤ (10 ng/ml) treatment fails to enhance the production of this prostanoid. In contrast, orbital fibroblasts, maintained under identical conditions, exhibit a dramatic increase in PGE 2 synthesis in response to IL-1␤, resulting from an induction of PGHS-2. A substantial fraction of the low basal PGE 2 generation in KAT-50 cells can be attributed to constitutive PGHS-2 activity. This activity is susceptible to the PGHS-2-selective inhibitor, SC58125 (control, 123.8 Ϯ 3.2 pg/ml; SC58125, 45.2 Ϯ 2.4 pg/ml) (mean Ϯ S.D., n ϭ 3). The fractional inhibition is unchanged with IL-1␤ treatment. The extremely low levels of PGE 2 production, when expressed on a "per cell" basis, were dramatically below those observed in orbital fibroblasts (Fig. 1B).
In contrast, the PGE 2 production in untreated fibroblast cultures derives, in large part, from the activity of PGHS-1, which is expressed at relatively high levels in untreated and cytokine-exposed fibroblast cultures. SC58125 exerts modest effects on production of the prostanoid in fibroblasts under control conditions. In contrast, the vast majority of PGE 2 synthesis in IL-1␤-treated fibroblasts can be inhibited with PGHS-2-selective agents, including SC58125, consistent with earlier findings in these cells (12)(13)(14)(15).
KAT-50 Cells Express High Levels of PGHS-2 and cPGES under Basal Culture Conditions-We next began to investigate the basis for why KAT-50 cells generate low levels of PGE 2 , despite high constitutive PGHS-2 expression. Confluent cultures of KAT-50 cells incubated in medium supplemented with low concentrations of FBS, or in serum-free medium, express high levels of PGHS-2 protein ( Fig. 2A) and mRNA (Fig. 2B). PGHS-2 protein migrates as a 72-kDa band on Western analysis, while the transcript appears as a 5-kb band on Northern blot analysis. Basal levels of PGHS-2 are considerably higher in KAT-50 cells than those in orbital fibroblasts incubated under identical culture conditions. The impact of IL-1␤ on PGHS-2 expression also differs substantially in fibroblasts and KAT-50 cells. The cytokine fails to up-regulate PGHS-2 expression in KAT-50 cells. Levels of PGHS-2 are essentially unchanged with addition of IL-1␤ after 6 h. The same analysis reveals very different responses in fibroblasts. In those cells, IL-1␤ induces PGHS-2 mRNA and protein dramatically at 6 and 16 h, respectively, consistent with its actions in many other cell types (27).
Analysis of PGES isoform expression in KAT-50 cells reveals that cPGES is an abundant protein under basal culture conditions, and the levels of both mRNA and protein are invariant with regard to IL-1␤ treatment (Fig. 2, A and B). This was also the case in fibroblasts, although the levels of cPGES mRNA were considerably lower than those found in KAT-50. In contrast, mPGES mRNA and protein were undetectable in KAT-50 cells, regardless of whether the cultures had been treated with cytokine or were maintained under control conditions (Fig. 2). Fibroblast cultures, in contrast, exhibited a low level of basal expression, but the levels of mPGES mRNA and protein were enhanced substantially when cultures were treated with IL-1␤ for 6 and 16 h, respectively. Thus, the predominant synthase expressed in KAT-50 cells under control and IL-1␤-treated conditions is cPGES. Moreover, the levels of cPGES mRNA and protein are invariant in these cells with regard to cytokine

FIG. 1. Comparison of PGE 2 synthesis in KAT-50 cells and orbital fibroblasts.
A, experiment comparing PGE 2 production in fibroblasts and KAT-50 cells under control and IL-1␤ (10 ng/ml)-treated conditions without or with SC58125 (5 g/ml). B, PGE 2 production in fibroblasts and KAT-50 cells expressed on a per cell basis in medium without or with IL-1␤. Cells were allowed to proliferate to confluence in 24-well plates covered with growth medium supplemented with 10% fetal bovine serum. They were then shifted to serum-less medium with the additives indicated for 16 h. Thirty min before harvest, medium was replaced with PBS supplemented with the respective agents. The PBS was collected and subjected to assay for PGE 2 as described under "Experimental Procedures." Data are expressed as the mean Ϯ S.D. of three replicates from a representative experiment.
cPGES Is Dominantly Expressed in Cultured KAT-50 Thyrocytes treatment. Consistent with earlier studies, KAT-50 cells express high levels of PGHS-1 mRNA. PGHS-1 protein, in contrast, is undetectable under all culture conditions assessed. It would appear that the transcript for PGHS-1 is not translated in these cells. The molecular basis for this apparent blockade is not currently understood but KAT-50, unlike most cells, expresses a single functional cyclooxygenase isoform.

Absence of mPGES Promoter Activity in KAT-50 Cells under Basal and IL-1␤-treated Culture Conditions Accounts for Undetectable mPGES in
These Cells-We next determined whether differences in the respective promoter activities could account for the disparate levels of PGHS-2 and mPGES mRNA and protein expression observed in KAT-50 cells and orbital fibroblasts. Reporter genes linked to human promoter fragments for PGHS-2, cPGES, and mPGES were transiently transfected into both types of cells, and their activities were compared under basal and cytokine-treated conditions. With regard to PGHS-2 promoter activity, levels in untreated fibroblasts were at least 20-fold greater than the promoter-less reporter gene, and IL-1␤ (10 ng/ml) treatment for 3 h resulted in a 2-fold increase (Fig. 3). We have reported that the substantial impact of IL-1␤ on increasing the steady-state PGHS-2 mRNA levels in orbital fibroblasts is mediated primarily through the enhancement of transcript stability and not on PGHS-2 gene transcription (12). In untreated KAT-50 cells, the PGHS-2 promoter activity was twice that seen in control fibroblasts and IL-1␤ reduced the activity by 25%. With regard to mPGES promoter activity, fibroblasts not treated with the cytokine exhibited levels similar to those in cultures transfected with reporter genes lacking a promoter, and IL-1␤ treatment increased the activity levels by 3-fold. The mPGES promoter construct transfected in KAT-50 cells failed to exhibit any detectable activity, either in untreated cultures or in those treated with IL-1␤. With regard to the cPGES (p23) promoter, its activity was low in orbital fibroblasts but at least 30-fold higher in KAT-50 cells (Fig. 3). IL-1␤ treatment failed to influence cPGES promoter activity in either cell type, consistent with the findings concerning the relative steady-state mRNA levels observed on Northern blot analysis (Fig. 2B). Thus, the undetectable mPGES expression found in KAT-50 cells can be directly attributed to an absence of mPGES promoter activity in these cells. In contrast, the promoter for cPGES exhibited extremely high levels of activity in KAT-50 cells.
KAT-50 Cells Transfected with mPGES cDNA and Expressing the Terminal Synthase Generate Higher Levels of PGE 2 -We speculated that the limited capacity of KAT-50 cells to generate PGE 2 is a consequence of the "wrong" profile of prostanoid biosynthetic enzymes being expressed in those cells. We therefore next determined whether introducing the expression of mPGES to KAT-50 cells and thus creating the theoretically more effective PGHS/PGES pairing would enhance the production of PGE 2 in these cells. Cultures were transiently transfected with mPGES cDNA cloned into pcDNA3.1(Ϫ), and the resulting expression of that protein culminated in a substantial increase in PGE 2 synthesis in KAT-50 cells under basal culture conditions (Fig. 4A). The levels were up to 3-fold above those observed in cultures transfected with empty vector. Moreover, addition of graded concentration of exogenous arachidonate further enhanced the PGE 2 production. As Fig.  4B indicates, Western blots using anti-mPGES antibodies confirmed that the synthase protein was expressed in the transfected cultures. Potentially low transfection efficiency could result in a substantial underestimation of the impact mPGES expression might have on PGE 2 synthesis in KAT-50 cells. Therefore, we determined the efficiency by transfecting sister cultures with pcDNA GFP. Those studies revealed that the efficiency was 25.3% (range 19 -29%, n ϭ 3 experiments). Data presented in Fig. 4A (inset) represent the normalized PGE 2 generation in cultures transfected with mPGES expressed as a function of transfection efficiency. As the figure indicates, PGE 2 generation was enhanced more than 6.9-fold in cultures transfected with mPGES and treated with arachidonate (10 M). Thus, introducing mPGES expression in KAT-50 results in

FIG. 2. Western and Northern blot analyses of PGHS and PGES expression in KAT-50 cells and orbital fibroblasts.
A, analysis of protein expression in the absence or presence of IL-1␤. Confluent cultures of KAT-50 cells and orbital fibroblasts were incubated overnight in serum-free medium and then some received IL-1␤ (10 ng/ml) for 6 and 16 h, respectively. Monolayers were rinsed and proteins solubilized and subjected to Western blot analysis of the enzyme proteins using the procedures described under "Experimental Procedures." Proteins were subjected to electrophoresis and were then transferred to membranes. These were then analyzed using primary antibodies directed specifically against mPGES, cPGES, PGHS-2, and PGHS-1. The ECL detection system was used to generate specific signals. B, comparison of mRNA expression under control and IL-1␤-treated conditions. Cells were allowed to proliferate to confluence in 100-mm diameter plates covered with medium supplemented with 10% FBS. The cultures were shifted to serum-less medium overnight and then some received IL-1␤ (10 ng/ml) for 6 h. Monolayers were rinsed, and RNA was extracted as described under "Experimental Procedures." Samples were subjected to electrophoresis and then transferred to membranes and subjected to hybridization with the relevant cDNA probes. Radioactive DNA/RNA hybrids were detected by exposing the membranes to X-Omat film at Ϫ70°C. The signals were normalized by rehybridizing membranes with a cDNA probe for glyceraldehyde-3-phosphate dehydrogenase (GAPDH).
cPGES Is Dominantly Expressed in Cultured KAT-50 Thyrocytes a substantial increase in PGE 2 production, presumably by allowing the functional coupling of the synthase with PGHS-2. DISCUSSION Thyrocytes express high basal levels of PGHS-2 in culture and in situ (16,17). The cell line, KAT-50, retains the elevated PGHS-2 expression found in primary thyrocytes in vivo but the cells have lost their ability to express detectable PGHS-1 protein. These observations suggest that activities of PGHS-2 must, in some manner, suffice with regard to meeting the metabolic needs of the cell. Of note is the coincident expression in these cells of cPGES at relatively high levels but the absence of detectable mPGES mRNA or protein. This particular profile of enzyme expression results in considerably lower levels of PGE 2 generation in KAT-50 cells than those observed in orbital fibroblasts which express both PGHS-2 and mPGES following cytokine treatment (Fig. 2) (12). It is possible that the absence of an efficient pair of cyclooxygenase/PGES enzymes has resulted from adaptation to the high constitutive levels of PGHS-2, effectively protecting the cells from excess prostanoid concentrations. When the more efficient pair of enzymes is expressed, in this case by transfecting the cells with mPGES, PGE 2 synthesis is enhanced. This suggests strongly that the endogenously expressed PGHS-2 functions more efficiently when functionally coupled to mPGES. These results support the functional nature of the constitutively expressed PGHS-2 in KAT-50 cells. They indicate that abnormal compartmentalization of PGHS-2 protein has not rendered the enzyme inaccessible to critical pools of substrate. The current results indicate further that the disparity between mPGES and cPGES expression in KAT-50 cells is a consequence of vastly different levels of activity exhibited by the respective gene promoters. There exists no detectable activity of the mPGES promoter in these cells, in contrast to orbital fibroblasts where the promoter is active under both basal and IL-1␤-treated conditions.
The functional coupling between specific PGES and PGHS isoforms that leads to highly efficient PGE 2 production was first demonstrated in HEK293 cells that had initially undergone stable transfections with PGHS-1 or PGHS-2 and then were transiently transfected with one of the PGES isoforms (8,9). mPGES co-localized with PGHS-2 and exhibited a marked enzymatic preference for PGHS-2 when arachidonate was supplied from endogenous or exogenous sources (9). Likewise, cPGES and PGHS-1 formed a considerably more efficient pairing in these transfected cells (8). It would seem that KAT-50 cells represent a naturally occurring example of cells expressing a pair of enzymes not optimized for efficient PGE 2 production. Restoration of theoretically optimal fidelity between the cyclooxygenase and terminal prostaglandin synthase results in enhanced PGE 2 synthesis.
From the current studies, it would appear that IL-1␤ does not influence levels of cPGES expression in KAT-50 cells (Fig.  2). This finding is consistent with that in some cells types, such as HeLa, MKN45, HEK293, WI-38, CHO, and L929, but differs from the modest increase found in rat brain in vivo following treatment with lipopolysaccharide (8). In previous studies, we also failed to observe a change in cPGES levels in orbital fibroblasts treated with a variety of cytokines (12). mPGES was undetectable in KAT-50 by Western and Northern blot analysis, suggesting extremely low levels of expression of that enzyme. In contrast, the cells express high levels of cPGES. PGHS-1 constitutes the natural enzyme partner for cPGES (8). cPGES Is Dominantly Expressed in Cultured KAT-50 Thyrocytes mRNA. Further studies will be necessary to determine whether this transcript is competent to be translated.
An important role for PGE 2 in normal or pathological thyrocyte function has yet to be firmly established. High levels of PGHS-2 expression found in untreated KAT-50 cells and cultured primary thyrocytes suggest that this enzyme may predominate in prostanoid production found in thyroid. This possibility is supported by the finding of PGHS-2 protein in situ in thin-sectioned thyroid tissue from apparently healthy glands (16). The frequent involvement of the thyroid in inflammatory processes suggests that inherent properties of thyrocytes might underlie disease susceptibility. Likely participants in the tissue remodeling integral to inflammatory diseases of the thyroid include the arachidonate synthetic pathways. Limited numbers of observations have thus far been made concerning the capacity of thyrocytes to generate ecosanoids or the physiological implications of their synthesis in thyroid tissue. In FRTL-5 cells, thyrotropin can regulate arachidonate release from membrane phospholipids, the activity of PGHS, and influence the formation of prostaglandins (28). These actions were enhanced by insulin-like growth factor-1. Thyroid-stimulating immunoglobulins from patients with Graves' disease activate phospholipase A2 in FRTL-5 cultures, promote the release of free arachidonate, and enhance 1,4,5-trisphosphate generation (29,30). This action appears to be mediated through a different pathway from those effects that result in cAMP generation. Another study has shown that proliferation of FRTL-5 cells is accelerated by stimulating IgGs from patients with Graves' disease and that the inhibition of cyclooxygenase activity with indomethacin can block this effect of disease-specific immuno-globulins (31). The net impact of PGE 2 on many vascular beds often involves substantial vasodilatation (32). Because the thyroid is extremely vascular, local production of the prostanoid may represent a key step in the regulation of perfusion patterns in this organ. PGE 2 generated in thyroid tissue, regardless of the cell type generating it, can influence the immunity and condition the nature of inflammatory responses occurring in the gland. This prostanoid can bias the differentiation of naive T lymphocytes from T H 1 to the T H 2 phenotype (33). Moreover, PGE 2 can influence B lymphocyte and mast cell development (34 -36). Thus the profile of prostaglandin-generating enzymes expressed by thyrocytes can potentially influence the nature of immunocompetent cells infiltrating the gland in disease.
Our findings identify, for the first time, a cell type that expresses constitutive PGHS-2 as a single, dominant cyclooxygenase isoform and yet fails to express detectable mPGES under basal or IL-1␤-treated conditions. These results suggest that development of therapeutic agents selectively targeting mPGES might not interrupt the full spectrum of PGHS-2-dependent prostaglandin production. Coupling of PGHS-2 with cPGES may be more widespread than is currently appreciated. It will be of considerable interest to examine whether other cells expressing high constitutive levels of PGHS-2 exhibit similar patterns of PGES. Of particular potential importance is the high level of constitutive PGHS-2 found in the pancreatic islet (37). The inhibitory action by PGE 2 on glucose-provoked insulin release suggests that the functional relationship between PGHS-2 and the PGE 2 synthases could determine, at least in part, levels of glucose tolerance. FIG. 4. Impact of introducing mPGES expression in KAT-50 cells on PGE 2 production. A, cells were allowed to proliferate to 60 -80% confluence in medium supplemented with 10% FBS. They were then shifted to serum-free medium and were transiently transfected with plasmids containing mPGES cDNA or the empty vector (control) as described under "Experimental Procedures." Cells were then incubated in standard growth medium without or with arachidonate (0, 1, and 10 M). For the final 30 min of the incubation, medium was removed and replaced with PBS. Samples were collected and subjected to the PGE 2 assay. Data are expressed as the mean Ϯ S.D. of triplicate determinations. Inset, data were normalized to the efficiency of mPGES transfection, ascertained by transfecting KAT-50 with a plasmid containing GFP. Transfection efficiency was determined to be a mean of 25.4% in three experiments (range 19.2-29.5%). B, to confirm that cultures transfected with mPGES express the protein, KAT-50 cell layers were solubilized in harvest buffer and equivalent amounts of cellular protein subjected to Western blot analysis with anti-mPGES antibodies as described under "Experiment Procedures."

cPGES Is Dominantly Expressed in Cultured KAT-50 Thyrocytes
The current observations are potentially important in helping to define the spectrum of relationships that exists between enzymes in the PGE 2 biosynthetic pathway. They imply that, despite forming a more efficient pairing with mPGES, PGHS-2 can, under certain conditions, be co-expressed with cPGES. Whether such a "mismatch" represents a deliberate attempt by nature to limit the production of PGE 2 and therefore offers some adaptive advantage is uncertain. It is possible that in the future, cells with this and potentially other suboptimal pairings of PGHS and PGES enzymes, such as PGHS-1 co-expressed with mPGES, may be identified. Examination of neoplastic cells may prove especially interesting, since overexpression of PGHS-2 dominates the phenotype and apparently influences the clinical behavior of some tumors, including colorectal adenocarcinomas (38).