Effects of Hypoxia on Monocyte Inflammatory Mediator Production

Blood-derived monocytes are found at sites of inflammation as well as in solid tumors and atherosclerotic arteries. They are an abundant source of inflammatory eicosanoids such as prostaglandin E2 (PGE2) and thromboxane A2, which are formed via arachidonic acid (AA) metabolism by cyclooxygenase-1/2 (COX-1/2). In vitro studies of inflammatory mediator production are conducted invariably in room air, which does not reflect the oxygen tensions found in monocyte-containing lesions, which are frequently hypoxic. In this work we examined the effects of hypoxia at levels reported in these lesions, on monocyte COX-2 expression, the related events that lead to eicosanoid synthesis, and relationships with tumor necrosis factor (TNF)-α synthesis. In fresh human monocytes exposed to hypoxia (1% O2), there was an increase in COX-2 protein compared with cells in normoxia, and this was attributable to increased transcription and mRNA stability. However, the synthesis of PGE2 and thromboxane A2 was reduced in hypoxia and did not reflect the increased level of COX-2. Monocytes prelabeled with [3H]AA followed by lipopolysaccharide stimulation in the presence of hypoxia showed a reduced release of AA compared with cells in normoxia. In addition, hypoxia resulted in decreased phosphorylation of the p44/42 mitogen-activated protein kinase and of cytosolic phospholipase A2. Hypoxia also increased TNF-α synthesis, which appeared to play a role in COX-2 expression, and the observed increase TNF-α synthesis appeared to result from reduced PGE2 synthesis. Overall, the results suggest the existence of an autocrine loop of regulation between monocyte eicosanoid and TNF-α production, which is dysregulated in hypoxia and establishes hypoxia as being an important environmental determinant of inflammatory mediator production.

sorter analysis. Contaminant cells were essentially all lymphocytes. For the maintenance of minimal LPS contamination, mononuclear cell isolation procedure was performed under sterile conditions, and elutriator tubing was treated with E-Toxa-Clean, 70% ethanol, and Milli-Q water before each elutriation.
Cell Stimulation-Elutriated monocytes were resuspended (2 ϫ 10 6 / ml) in RPMI 1640 medium supplemented with 10% (v/v) heat-inactivated low LPS FCS, L-glutamine, Hepes, 100 units/ml penicillin, and 100 g/ml gentamycin. One-ml aliquots were incubated in Minisorp tubes at 37°C in 5% CO 2 as indicated with either 10 M arachidonic acid (AA) or 1 M calcium ionophore A23187 for 15 min or 200 ng/ml LPS for 0 -18 h. All cell stimulations with AA or A23187 were performed in RPMI (FCS-free). Incubations with LPS were performed in complete medium (10% FCS) at 37°C. Cell suspensions were centrifuged, and cell-free supernatants were stored at Ϫ20°C until eicosanoid determination.
Hypoxia-Ambient oxygen concentrations of 1% were maintained using a controlled incubator with CO 2 /O 2 monitoring and CO 2 /N 2 gas sources (Edwards Instrument Co., Wilmington, MA). CO 2 was maintained at 5%. Culture medium was preequilibrated overnight before cell exposure and maintained at a pH of 7.3. The pO 2 of the medium was 33 mm Hg in hypoxia and 154 mm Hg in normoxia. The appearance of cells in hypoxia was indistinguishable from those maintained in normoxic conditions by light microscopy.
Eicosanoid Measurement-TXA 2 has a t1 ⁄2 of ϳ30 s under physiological conditions and is hydrolyzed to the stable metabolite TXB 2 , which was measured. PGE 2 and TXB 2 levels were determined by RIAs. The TXB 2 assay used rabbit antiserum raised against thyroglobulin-conjugated TXB 2 (18).
TNF-␣ Measurement by ELISA-Nunc plates were coated with mouse monoclonal coating antibody against human TNF-␣ (5 g/ml in 0.2 mol/liter Na 2 CO 3 , pH 9.4) overnight at 4°C. The plate was then blocked by the addition of 200 l of 0.5% bovine serum albumin for 1 h at 37°C. Serial dilutions of human recombinant TNF-␣ (ranging from 20 to 0.312 ng/ml) or 50-l samples diluted 1:3 were added together with 50 l of mouse monoclonal (matched pair antibody) against 0.05 g/ml human TNF-␣ for 2 h at room temperature. Plates were washed between steps with phosphate-buffered saline containing 0.05% Tween 20. 100 l of Extravidin® peroxidase (1:4,000 dilution in 0.5% bovine serum albumin) was added for 15 min at 37°C. Finally, 100 l of the peroxidase substrate, TMB (Sigma), in 0.5 M phosphate citrate buffer (according to manufacturer's protocol) was added. The reaction was stopped by the addition of 100 l of 2 M H 2 SO 4 . Absorbance was measured at 450 nM in a microplate reader (model 450, Bio-Rad).
Northern Blot-Total RNA was isolated using TriZol (Invitrogen) according to manufacturer's protocol. Total RNA (10 g/lane) was heated at 68°C for 10 min, electrophoresed on a 1% agarose-formaldehyde gel, transferred to a positively charged nylon membrane (Hybond Nϩ, Amersham Biosciences), and UV cross-linked. Membranes were prehybridized for 3 h at 55°C and subsequently hybridized overnight at 43°C with random primer [ 32 P]dCTP-labeled human COX-2 cDNA or glyceraldehyde-3-phosphate dehydrogenase probe using a GIGAprime DNA Labeling Kit (Bresatec, Adelaide, Australia). The COX-2 cDNA probe was prepared by reverse transcription-PCR as described (19). Equal RNA loading efficiency was determined by visualization of 28 S and 18 S bands over UV light or glyceraldehyde-3-phosphate dehydrogenase.
COX-2 Promoter-Reporter Construct-A vector containing a 7-kb fragment (GenBank accession number AF044206) of the COX-2 promoter region was the gift of Dr. Steven Prescott of the Huntsman Cancer Institute (University of Utah) (20). This was used as the template for amplification by PCR of a fragment containing bases Ϫ531 through ϩ65 relative to the COX-2 transcriptional start site. A reporter construct driven by this segment was responsive to hypoxia in endothelial cells (21). Briefly, the conditions for the PCR were 0.2 unit/reaction AmpliTaq Gold® (Applied Biosystems), 1.5 mM MgCl 2 , 1ϫ (10ϫ) buffer (100 mM Tris-HCl, pH 8.3, 500 nM KCl, 15 mM MgCl 2 , 0.01% (w/v) glycerin), 0.2 mM dNTPs (New England Biolabs), 100 ng/reaction each primer and high performance liquid chromatography grade water (Sigma) to a total volume of 20 l. Conditions were 95°C for 10 min, 30 cycles of (94°C for 30 s, 50°C for 30 s, 72°C for 1 min), and 72°C for 10 min. The primers used had specific restriction sites built into the 5Ј-most ends to facilitate ligation in a specific orientation into pGL3-Basic (Promega). This vector contains the gene coding for the firefly luciferase gene but no promoter. The specific primers were, with the restriction sites in bold italics, fproϪ531COX-2 (5Ј-GCGGTACCGTT-ACTCGCCCCAGTCTGTC-3Ј) and rproϩ65COX-2 (5Ј-GGCTCG-AGCGAGGCGCTGCTGAGGAG-3Ј).
The PCR product was purified using the MinElute™ PCR Purifica-tion Kit (Qiagen), restricted in separate reactions with KpnI and XhoI (New England Biolabs) according to manufacturer's instructions, and then purified further for transformation using the MinElute™ Reaction Cleanup Kit (Qiagen). The isolated restricted PCR product was then ligated at the KpnI and XhoI sites located in the multiple cloning site of the pGL3-Basic vector using T4 DNA Ligase (Promega). This vector pGL3-COX-2Ϫ531 was then transformed into MAX Efficiency® DH5␣™-competent cells (Invitrogen) according to the manufacturer's instructions. Sequencing (ABI Prism® model 3700) confirmed the orientation and sequence of COX-2Ϫ531. Overnight cultures were then grown and plasmid isolated using the Endofree® Plasmid Maxi Kit (Qiagen) to ensure minimal LPS contamination. Transient Transfection-U937 monocytic cells were plated in 12-well plates (2 ϫ 10 6 cells/2 ml) in RPMI with 10% FCS and 50 ng/ml phorbol 12-myristate 13-acetate, which promotes differentiation after 3-5 days of treatment (22). After differentiation, cells were transfected using Jet PEI (PolyTransfection), according to the manufacturer's instructions. Briefly, 4 g of the pGL3-COX-2Ϫ531 construct was suspended in 75 l of 150 mM sterile NaCl solution. Also 4 l of Jet PEI solution was suspended in 75 l of 150 mM sterile NaCl solution. The Jet PEI/NaCl solution was then added to the DNA/NaCl solution and incubated at room temperature for 30 min. The medium in the wells was then changed to fresh medium, and 150 l of the DNA/Jet PEI was added to each well. The transfection was allowed to proceed for 5 h, and the medium replaced again with either hypoxic or normoxic medium. The cells were then stimulated with 100 ng/ml serum-treated zymosan for specified times. Following the transfection period, the medium was removed and discarded and the cells lysed with Passive Lysis Buffer supplied in the Dual Luciferase™ Reporter Assay Kit. The lysate was then assayed for luciferase activity.
Statistical Analysis-Results are expressed as the mean Ϯ S.E. of triplicate incubations. Analysis of variance followed by the Newman-Keuls multiple comparisons test was used to identify the statistically significant differences between treatments using WINKS (Texasoft, Cedar Hill, TX). human monocytes in a time-dependent manner over 18 h. The up-regulation of COX-2 mRNA and protein was greatly potentiated by hypoxia (1% O 2 ) ( Fig. 1). This augmentation of COX-2 expression by hypoxia was observed with a variety of costimuli ( Fig. 2).

Effect of Hypoxia on
It was reported that hypoxia can increase transcription of COX-2 in endothelial cells (21), and therefore this mode of regulation in monocytes was examined. Many attempts to transfect fresh human monocytes transiently with a COX-2 promoter/luciferase reporter construct were unsuccessful. However, the human monocytic cell line U937 was transfectable, and these cells were used. Hypoxia augmented activity of the 531-bp segment of the COX-2 promoter in U937 cells (Fig.  3). Another mode of regulation of COX-2 levels can occur posttranscriptionally with stabilization of mRNA in response to LPS or interleukin-1␤ (23), although this has not been examined in hypoxia. Therefore, the effect of hypoxia on COX-2 mRNA stability in monocytes was examined. Monocytes were transiently stimulated with LPS for 15 min in normoxia or hypoxia and then washed and incubated in fresh normoxic or hypoxic medium for 3 h to allow synthesis of COX-2 mRNA. Actinomycin D was added to inhibit further transcription, and the level of COX-2 mRNA was measured for a further 3 h. COX-2 mRNA levels decreased in normoxia by more than 90% within 3 h after the addition of actinomycin D (Fig. 4). By comparison, COX-2 mRNA levels decreased in hypoxia by less than 20% within 3 h after the addition (Fig. 4).
Effect of Hypoxia on Monocyte COX-2 Activity-Fresh monocytes were treated with LPS for 18 h at 37°C in normoxia or hypoxia, and the accumulation of PGE 2 and TXA 2 in the cell supernatants was measured. There was a marked reduction in the accumulation of PGE 2 and TXA 2 synthesis in hypoxia (Fig.  5). Similar time courses and reduced eicosanoid synthesis in hypoxia were also observed when LPS was transient, with LPS removal after 15 min (data not shown). The reduced synthesis of these eicosanoids in hypoxia did not correlate with the increased expression of COX-2 protein in hypoxia, described above. Possible explanations for the disparate hypoxia-induced changes in COX-2 expression and eicosanoid synthesis were sought.
Effect of Heme on COX-2 Activity in Hypoxia-COX-2 is a heme-containing enzyme, and cellular heme levels can be reduced by heme oxygenase, including the inducible isoform, heme oxygenase-1, which may be up-regulated during hypoxia (24 -26). Therefore, the cellular levels of heme may become limiting for adequate COX-2 constitution in hypoxia and may explain the dissociation of up-regulated COX-2 protein and activity. The addition of heme or a heme oxygenase inhibitor, zinc-protophorhyrin IX, under previously reported conditions (25) did not affect the amount of COX-2 protein (Fig. 6a) or the production of PGE 2 and TXA 2 (Fig. 6b) protein in normoxia or hypoxia under these experimental conditions, suggesting that heme oxygenase activity or heme levels are not responsible for reduced COX-2 activity in hypoxia.
Dependence of COX-2 Activity on O 2 as a Cosubstrate in Hypoxia-COX-2 utilizes oxygen as a cosubstrate during the conversion of AA to prostaglandin H 2 , the common precursor of PGE 2 and TXA 2 . In this study, O 2 in the incubation chamber was set at 1%; cf. ϳ20% for air at sea level. This level of hypoxia reduced dissolved oxygen in the incubation medium to 33 mm Hg. To determine whether these levels of O 2 were rate-limiting for eicosanoid synthesis, monocytes were first incubated in hypoxia with LPS to induce COX-2. After 18 h, cells were washed twice and incubated in fresh hypoxic or normoxic medium with 10 M exogenous AA for 15 min. Oxygenation of the medium had no effect on the production of PGE 2 and TXA 2 (Fig.  7). These results indicated that dissolved O 2 at the levels of hypoxia used in this study were not rate-limiting for COX activity.
Effect of Exogenous AA on COX-2 Activity in Hypoxia-Monocytes were incubated with LPS for 18 h in the absence or presence of hypoxia to induce COX-2. The following day cells were washed twice and incubated with fresh normoxic or hypoxic medium and 10 M exogenous AA for 15 min. In hypoxia, there was an increase in PGE 2 and TXA 2 synthesis (Fig. 8). This contrasted with results above (Fig. 5)   had been hypoxic for 9 h were returned to oxygenated conditions for the next 21 h. In normoxia, there was a time-dependent increase in the release of labeled AA from monocytes when stimulated with LPS ( Fig. 9). By comparison, there was a marked reduction in the release of AA from monocytes stimulated with LPS in hypoxia (Fig. 9). Reoxygenation after 9 h of hypoxia resulted in a gradual restoration of AA release from cells to rates that were similar to those observed in normoxic cells (Fig. 9). Because cPLA 2 is prominently involved in the release of AA from membrane phospholipids, the effects of hypoxia on cPLA 2 phosphorylation were examined.
Effect of Hypoxia on the Phosphorylation of cPLA 2 -After stimulation with 1 M A23187, the phosphorylation of cPLA 2 appeared to be maximal at 30 min, and dephosphorylation occurred at times after 30 min (Fig. 10). In contrast, phosphorylation of cPLA 2 in hypoxia appeared to be reduced at 10 min and showed an accelerated dephosphorylation of the enzyme at later times (Fig. 10). MAP kinases may regulate the phosphorylation and activation of cPLA 2 (27)(28)(29)(30)(31)(32). Therefore, the effects of hypoxia on the phosphorylation of p44/42 MAPK (ERK1/2) and p38 MAPK were examined.
Effect of Hypoxia on the MAPK Pathways-In normoxia, the phosphorylation of p44/42 was maximal at 30 min followed by dephosphorylation up to 240 min (Fig. 11). In hypoxia, there was a reduction in the phosphorylation at 30 min and accelerated dephosphorylation at later times (Fig. 11).
In normoxia, the phosphorylation of p38 MAPK was maximal at 30 -60 min followed by dephosphorylation at later times. Hypoxia had no effect on the time course of phosphorylation of p38 MAPK or the time course of decay in the amount of phosphorylated enzyme (Fig. 11). In addition, some prelabeled cells that had been hypoxic for 9 h were returned to oxygenated conditions with or without PD98059, an inhibitor of p44/42 MAPK phosphorylation. PD98059 inhibited the restoration of AA release from reoxygenated cells to levels similar to those observed in cells maintained in hypoxia (Fig. 12).
Effect of Hypoxia on Monocyte TNF-␣ Synthesis-Fresh monocytes were treated with LPS in normoxia or hypoxia, and the accumulation of TNF-␣ in the cell supernatants was measured. Hypoxia markedly increased TNF-␣ synthesis (Fig. 13). Similar time courses and increased TNF-␣ synthesis in hypoxia were also observed when LPS stimulation was transient (data not shown). To examine the possibility that the reduction in PGE 2 synthesis in hypoxia may be related in part to the augmentation of TNF-␣ production, the effects of hypoxia on TNF-␣ synthesis were examined in the presence of COX inhibitors.
Monocytes were preincubated for 15 min at 37°C with a selective COX-2 inhibitor, NS398 (1 M) or a general COX inhibitor, indomethacin (10 M) prior to LPS stimulation in normoxia and hypoxia. In normoxia and hypoxia, both NS398 and indomethacin resulted in a marked reduction in PGE 2 production and a significant increase in TNF-␣ synthesis (Fig. 14). Furthermore, in the presence of both NS398 and indomethacin, TNF-␣ synthesis was similar under normoxic and hypoxic conditions (Fig. 14). In addition, exogenous PGE 2 dose-dependently inhibited TNF-␣ synthesis in both normoxia and hypoxia (Fig. 15).
Effect of Monocyte-derived TNF-␣ on COX-2 Induction and Activity-Because hypoxia increased TNF-␣ synthesis, it was important to determine whether endogenous TNF-␣ could have an autocrine effect on COX-2 expression in monocytes. To examine this, fresh monocytes were preincubated with a neutralizing antibody against TNF-␣ or with an isotype-matched control antibody (1B5) prior to LPS stimulation in normoxia. Neutralizing TNF-␣ activity significantly inhibited COX-2 expression (Fig. 16a) and activity (Fig. 16b). DISCUSSION To date, in vitro studies of inflammatory mediator production by human monocytes/macrophages have been well characterized in normoxic conditions (20% O 2 ). However, this is unlikely to reflect conditions of oxygenation which monocytes encounter in monocyte-containing lesions such as inflamed joints, atheromatous lesions, and solid tumors. Joints with effusions can be chronically hypoxic (14 -16, 33), and dissolved O 2 levels in the range of ϳ8 -80 mm Hg have been observed (14 -16). The presence of an effusion can readily increase intraarticular pressure to levels above capillary closure pressure, particularly during everyday activities such as standing, walking, and even modest flexion (33). Similarly, several studies have demonstrated a decreased oxygen concentration in the media of atherosclerotic arteries, ranging between 2 and 50 mm Hg (17, 34 -36), leading to the hypothesis that hypoxia is a component of the pathology of atherosclerotic plaques (37)(38)(39). In addition, regions of reduced oxygen have been reported in cancers, including breast (7,8), prostate (9), melanoma (10, 11), and cervical cancers (12,13), although the oxygen levels are very heterogeneous within individual tumors. Thus examination of the effects of hypoxia on monocyte inflammatory mediator production has relevance to many pathological situations in which monocytes are present.
It has been demonstrated in human umbilical vein endothelial cells that hypoxia increased COX-2 expression and that an increase in transcription was involved (21). However, the effect on prostaglandin production of this COX-2 response to hypoxia was not measured (21). We observed that hypoxia caused a marked up-regulation of COX-2 mRNA and protein in fresh human monocytes and that this may be explained in part at least by an increase in COX-2 transcription and in COX-2 mRNA stability. The 3Ј-untranslated region of the COX-2 gene contains 22 copies of the AUUUA motif, which is related to mRNA stability (41)(42)(43). The response to hypoxia of COX-2 mRNA observed in this study may be a more general phenomenon because the gene for vascular endothelial growth factor also contains instability motifs in its 3Ј-untranslated region, and the mRNA is stabilized under hypoxic conditions (44,45).
Although we observed an up-regulation of monocyte COX-2 in hypoxia, this was accompanied by a decrease in PGE 2 and TXA 2 production. This appears to be a paradoxical response. Rat lung exposed to hypoxia had increased COX-2 levels and increased prostaglandin production (46). Similarly, ischemia induced an increase in COX-2 mRNA and an increase in PGE 2 synthesis in mouse cerebrum (47). Nevertheless, an increase in COX-2 expression and a decrease in PGE 2 synthesis were reported for the effects of hypoxia on a rabbit corneal epithelial cell line (25).
In the latter case, Bonazzi et al. (25) attribute this dissociation to increased activity of heme oxygenase in hypoxia with resultant decreased heme available for the activity of COX-2, a heme-containing protein. The inducible isoform of heme oxygenase, heme oxygenase-1, is increased in hypoxia and may be responsible for heme depletion in hypoxic cells (25,26). However, when we repeated the procedures of Bonazzi et al. (25) of adding heme or an inhibitor of heme oxygenase, there was no restoration of prostaglandin synthesis like that reported in the rabbit corneal epithelial cell line. It is possible that this is an intrinsic difference between systems in monocytes and rabbit corneal epithelium, or it is possible that the rabbit corneal epithelial cells are depleted in heme because of long term culture, whereas fresh human monocytes have adequate heme stores.
Another possible explanation for decreased eicosanoid synthesis in hypoxia at 1% O 2 is that availability of O 2 , which is a cosubstrate for COX activity, is rate-limiting. However, we observed that this was not the case because in the presence of exogenous AA, eicosanoid synthesis was similar in hypoxic and reoxygenated monocytes.
We observed that AA release was considerably depressed in hypoxia and that reoxygenation restored the release of AA. These results correlated with a reduction in both the phosphorylation of cPLA 2 and of the p44/42 MAPK in hypoxia. The latter enzyme is reportedly involved in regulating the phosphorylation of cPLA 2 in macrophages (27,28), neutrophils (29 -31), and basophils (32). The involvement of the p44/42 MAPK in the reduced AA release that we observed in hypoxic monocytes is corroborated by the observation that inhibition of p44/42 MAPK activation by use of the MEK-1 inhibitor, PD98059, inhibited the restoration of AA release after reoxygenation.
Although there was reduced eicosanoid synthesis in hypoxic monocytes, there was simultaneously a marked increase in TNF-␣ production in hypoxia compared with normoxia. In normoxia, the increased TNF-␣ production in the presence of NS398 indicated that a COX-2 product is responsible for normal autocrine suppression of TNF-␣ synthesis. For two reasons, the responsible COX-2 product is likely to be PGE 2 . First, COX-2 induction in monocytes is associated with greatly increased PGE 2 synthesis relative to that of TXA 2 (48). Second, exogenous PGE 2 suppressed TNF-␣ synthesis in a dose-dependent manner. Therefore, the hypoxia-induced increase in TNF-␣ synthesis may be caused by the concomitant hypoxiainduced reduction in PGE 2 synthesis. This is not the only possible mechanism. It has been reported that hypoxia-induced increases in TNF-␣ synthesis in J774.1 murine macrophage cell line are attributable to hypoxia-induced mitochondrial production of reactive oxygen species (49).
Although the reduction in PGE 2 synthesis observed in hy-poxia was caused by reduced p44/42 MAPK activation and consequent reduced AA release from membrane phospholipids, there was increased COX-2 expression. In normoxia, the suppression of LPS-stimulated COX-2 expression by the addition of neutralizing anti-TNF-␣ antibody indicated that endogenous TNF-␣ is normally involved as a mediator of COX-2 up-regulation. Therefore, the overexpression of COX-2 in hypoxia may result from increased TNF-␣ production in hypoxia.
Overall, the observations in stimulated monocytes in normoxia suggest an autocrine loop in which production of TNF-␣ up-regulates COX-2 and synthesis of PGE 2 , which in turn suppresses TNF-␣ production. We have demonstrated previously in stimulated human monocytes that although TXA 2 synthesis is an early COX-1-dependent response, synthesis of PGE 2 is delayed and is dependent on induction of COX-2 (2). Thus, it is possible that COX-2 up-regulation and consequent PGE 2 synthesis provides a system for a "self-limited" monocyte response with regard to TNF-␣ production (Fig. 17). This proposition is supported by the increased TNF-␣ synthesis observed with COX-2 inhibition. In hypoxia, the system is dysregulated with regard to TNF-␣ synthesis possibly because PGE 2 synthesis is reduced as a result of reduced AA release, and a consequence of the unrestricted TNF-␣ synthesis is overexpression of COX-2 (Fig. 17).
Despite a decrease in monocyte PGE 2 synthesis by hypoxia, the consequences for PGE 2 concentrations in inflamed lesions such as a rheumatoid joint remain speculative. The rate of monocyte PGE 2 synthesis in a hypoxic rheumatoid joint may be reduced compared with a normoxic rheumatoid joint. However, the synovial concentration of PGE 2 derived from hypoxic monocytes would still be well above those in a healthy joint simply because of their presence in an inflamed synovium. In addition, there is likely to be a contribution to total joint PGE 2 levels from other cell types. Thus, even in a hypoxic joint, it is probable that the PGE 2 concentration is sufficient to contribute to the signs and symptoms of swelling and pain which are alleviated by nonsteroidal anti-inflammatory agents, including COX-2 inhibitors. Furthermore, it is possible that any increase in monocyte TNF-␣ synthesis which may result from reduced monocyte PGE 2 synthesis could be important for other joint pathology.
Whatever may be the pathological consequences, it is clear that hypoxia is an important but often neglected determinant of inflammatory mediator production and one that potentially may influence a broad range of events that occur in monocytecontaining lesions. Therefore, the effect of hypoxia on the activities of other cells at sites of inflammation or ischemia also warrants investigation.