Research Paper Mediators of Inflammation, 9, 125–132 (2000)

Annexin I is a glucocorticoid-induced mediator with anti-inflammatory activity in animal models of arthritis. We studied the effects of a bioactive annexin I peptide, ac 2-26, dexamethasone (DEX), and interleukin-1beta (IL-1beta) on phospholipase A2 (PLA2) and cyclooxygenase (COX) activities and prostaglandin E2 (PGE2) release in cultured human fibroblast-like synoviocytes (FLS). Annexin I binding sites on human osteoarthritic (OA) FLS were detected by ligand binding flow cytometry. PLA2 activity was measured using 3H-arachidonic acid release, PGE2 release and COX activity by ELISA, and COX2 content by flow cytometry. Annexin I binding sites were present on human OA FLS. Annexin I peptide ac 2-26 exerted a significant concentration-dependent inhibition of FLS constitutive PLA2 activity, which was reversed by IL-1beta. In contrast, DEX inhibited IL-1beta-induced PLA2 activity but not constitutive activity. DEX but not annexin I peptide inhibited IL-1beta-induced PGE2 release. COX activity and COX2 expression were significantly increased by IL-1beta. Annexin I peptide demonstrated no inhibition of constitutive or IL-1beta-induced COX activity. DEX exerted a concentration-dependent inhibition of IL-1beta-induced but not constitutive COX activity. Uncoupling of inhibition of PLA2 and COX by annexin I and DEX support the hypothesis that COX is rate-limiting for PGE2 synthesis in FLS. The effect of annexin I but not DEX on constitutive PLA2 activity suggests a glucocorticoid-independent role for annexin I in autoregulation of arachidonic acid production. The lack of effect of annexin I on cytokine-induced PGE2 production suggests PGE2-independent mechanisms for the anti-inflammatory effects of annexin I in vivo.


Introduction
Prostaglandins, such as prostaglandin E 2 (PGE 2 ), mediate the pain and edema associated with arthritis, leading to the widespread use of non-steroidal antiinflammatory drugs in the treatment of arthritis. Within the arthritic synovial lesion, fibroblast-like synoviocytes (FLS) have been implicated as a primary source of PGE 2 . 1 In eicosanoid generation, phospholipase A 2 (PLA 2 ) and cyclooxygenase (COX, or prostaglandin H synthase) have been described as important regulatory enzymes. The hydrolytic release of arachidonic acid from membrane phospholipids is initiated by PLA 2 , and arachidonic acid is then catalysed by COX enzymes for the subsequent production of PGE 2 . PLA 2 exists as a secretory group II isoform (sPLA 2 ) and an arachidonic acid-selective cytosolic form (cPLA 2 ). 2 Two isoforms of COX have also been identified: a constitutive (COX1) and a mitogen/ growth factor-inducible (COX2) form. 3 Studies in a COX2 gene knock-out animal model demonstrate that joint inflammation still persists in the absence of this enzyme. 4 At present, the enzymatic events responsible for the enhanced production of PGE 2 in rheumatoid arthritis (RA) are not fully understood. Knowledge pertaining to which enzyme is rate-limiting in eicosanoid generation in synovium is not fully elucidated. It is known that expression of the cPLA 2 and COX2 are increased by interleukin-1b (IL-1b) 5,6 and that this induction is inhibited by the anti-inflammatory action of dexamethasone (DEX). 3,6,7 Annexin I (previously known as lipocortin-1) is a 37 kDa protein which belongs to a family of at least 13 structurally related proteins that bind to anionic phospholipids in a Ca 2+ -dependent manner. 8 Annexin I was originally defined as a glucocorticoid-inducible inhibitor of PLA 2 activity. 9 The concept that annexin I may modulate its anti-inflammatory actions via cell surface binding sites has been suggested from the identification of extracellular annexin I binding sites on peripheral blood leukocytes. 10 The involvement of annexin I in the regulation of joint inflammation has now been demonstrated in several rat models of arthritis, in which a range of biological actions extending beyond effects on PLA 2 activity has been suggested. [11][12][13] The possibility that annexin I has antiinflammatory activity in human arthritis has not been previously explored. We sought to further define the effects of annexin I on PLA 2 , COX and prostaglandin production in human FLS, and use the known PLA 2 inhibitory properties of annexin I to investigate the rate-limiting step in synoviocyte prostaglandin production.

Purification of annexin I protein
Recombinant human annexin I, a generous gift from Dr Yuko Giga-Hama (Research Centre, Asahi Co. Ltd, Japan), 15 was purified by affinity gel chromatography, as described by Sakata et al., 16 using purified monoclonal mouse anti-human annexin I antibody. To confirm purification, samples of purified annexin I dissolved in sample buffer were separated on a 12% gel using SDS-polyacrylamide gel electrophoresis techniques. A single 37 kDa protein band was detected by western blotting using a specific antiannexin I mAb.
Isolation and culture of FLS FLS were obtained from synovium of osteoarthritic (OA) patients (n = 8) undergoing joint replacement surgery. All patients satisfied the American College of Rheumatology criteria for the classification of OA. 17,18 FLS were isolated and cultured as described by Koch. 19,20 In brief, the synovial-lining tissue was dissected and minced into 2-3 mm pieces and rotated in 10 ml/2 g of enzyme solution containing 2.4 mg/ml dispase (grade II, 5 U/mg), 1 mg/ml collagenase (type II) and 1 mg/ml DNase (type I) in Ca 2+ and Mg 2+ free Hank's balanced saline solution. FLS were cultured in 10 cm culture plates in RPMI/10% foetal calf serum at 37°C in a 5% CO 2 humidified incubator. Cells at 3rd passage were more than 99% FLS, as demonstrated by dendritic, spindle morphology, and negative staining for the pan-leukocyte antigen CD45 by flow cytometry. Cells were used between passages 4 and 9. For all experiments, FLS were seeded at 1 ´10 5 cells per well in 24-well culture plates in RPMI/10% FCS and allowed to adhere overnight, prior to medium being replaced with RPMI/0.1% BSA for experimental purposes. This study was approved by the institutional ethics board.

Detection of annexin I binding sites
FLS were examined using a method that detects specific saturable annexin I binding sites, as described by Perretti e t a l. 21 Briefly, surface bound annexin I was removed by washing with 1 mM EDTA/PBS. FLS were then washed with 0.1% BSA/phosphate buffered saline (PBS)/1.3 mM CaCl 2 and resuspended in 20 ml 0.2% BSA/RPMI at 4°C and incubated with 0-10 mM of human recombinant annexin I. Annexin I binding sites were detected by sequential incubation with 60 mg/ml annexin I monoclonal antibody 1B or matched isotype control mAb, IgG 2a and FITC-conjugated anti-mouse IgG. Fluorescence was analysed on a MO-FLO flow cytometer (Cytomation, Ft Collins, CO). Cells were gated according to forward and right angle scatter, using 10 000 cells for each determination. Each determination was performed in duplicate, with Assessment of PLA 2 activity PLA 2 activity in FLS was determined according to Croxtall e t a l. 22 Briefly, FLS were incubated for 18 h with 1 mCi/ml [ 3 H]arachidonic acid in 0.1% BSA/ RPMI. Cells were treated with 0-100 mg/ml annexin I peptide, 0, 0.1 ng/ml human recombinant IL-1b, 10 -9 -10 -7 M DEX for 8 h. Triplicate cultures were used for each determination. Radioactivity in the supernatant was determined using a Wallac 1409 liquid scintillation counter (Pharmacia, Finland), with results expressed as tritiated arachidonic acid released, or as a percentage of the mean result of unstimulated cells.
Determination of PGE 2 levels PGE 2 release was determined by assaying treated supernatants, using ELISA. Each determination was performed in duplicate, with results expressed as a percentage of the mean result of the unstimulated cells. The detection limit for this assay was <0.1 ng/ml.

Assessment of COX activity
COX activity was measured according to Wilborn e t a l. 7 In brief, FLS were incubated for 30 min with 10 mM exogenous arachidonic acid. Supernatants were aspirated and assayed for PGE 2 content using ELISA. Duplicate cultures were used for each determination. COX activity was expressed as a percentage of the mean result of the unstimulated cells.

Determination of intracellular COX2
Intracellular expression of COX2 in FLS was determined using permeabilization flow cytometry, as described by Morand e t a l. 23 Briefly, cells were fixed by suspension in 2% paraformaldehyde/PBS, followed by permeabilization using 0.2% saponin/PBS. FLS were then sequentially incubated with 20 mg/ml COX2 mAb (or isotype-matched mAb control, IgG 1 ) and FITC-conjugated anti-mouse IgG. Permeabilization of the cells was reversed with PBS. Labelled FLS were analysed using flow cytometry. 10 000 cells were used for each determination. Intracellular COX2 protein was expressed as mean fluorescence intensity after subtraction of mean fluorescence intensity obtained with negative control mAb.

Statistical analysis
Results are expressed as the mean ± standard error of the mean (SEM). Statistical analysis was performed using the Student's t-test, with values of p <0.05 regarded as statistically significant.

Results
We initially sought to determine the presence of annexin I binding sites on human FLS. Annexin I binding sites as determined by mean fluorescence intensity were detected using a concentration range of annexin I (0-10 mM). Results demonstrated concentration-dependent annexin I binding, with more than 99% of cells demonstrating annexin I binding at 10 mM (Fig. 1, Table 1).
To define the effect of annexin I on synoviocyte PLA 2 activity, FLS were treated with annexin I N-terminal peptide ac 2-26 (0-100 mg/ml). Annexin I peptide induced a significant concentration-dependent inhibition of constitutive PLA 2 activity ( Fig. 2(a)).  Table 1. Inhibition of PLA 2 activity by annexin I was reversed when cells were co-stimulated with IL-1b ( Fig.  2(b)). We subsequently assessed the effects of DEX on synoviocyte PLA 2 activity. DEX treatment did not significantly inhibit constitutive PLA 2 activity over the concentration range 10 -9 -10 -7 M. In contrast, DEX exerted a significant concentration-dependent reduction of PLA 2 activity in IL-1b-stimulated FLS (Fig. 3).
Since annexin I peptide inhibited constitutive PLA 2 activity, we next determined whether treatment with annexin I peptide affected PGE 2 release. Constitutive PGE 2 synthesis was detected in FLS. Annexin I peptide did not significantly reduce constitutive PGE 2 synthesis ( Fig. 4(a)). A significant increase in PGE 2 release was observed following IL-1b stimulation (p =0.001), and annexin I peptide did not inhibit IL-1b-stimulated PGE 2 release. In contrast, DEX had no effect on constitutive PGE 2 release but significantly inhibited IL-1b-induced PGE 2 release in a concentrationdependent manner (Fig. 4(b)).
To further address the differential inhibitory actions of annexin I on PLA 2 activity and PGE 2 synthesis, studies were designed to evaluate the effects of annexin I peptide and DEX on constitutive and IL-1b-induced synoviocyte COX activity. Since the synthesis of PGE 2 from endogenous arachidonic acid results from the activities of both PLA 2 and COX, to estimate maximal COX activity, PLA 2 activity was bypassed by addition of exogenous arachidonic acid. 7 Low levels of constitutive COX activity were detected in FLS. IL-1b significantly increased COX activity (p=0.0004) ( Fig.  5(a)). Annexin I peptide exerted no significant effect on constitutive or IL-1b-stimulated COX activity ( Fig.  5(a)). DEX exerted no significant effect on constitutive COX activity in DEX treated FLS. In contrast, DEX resulted in a concentration-dependent reduction of IL-1b-induced COX activity (Fig. 5(b)).

Mediators of Inflammation · Vol 9 · 2000
The question of whether modulation of COX activity reflected its intracellular expression was investigated using permeabilization flow cytometry. Constitutive expression of intracellular COX2 protein was detected in FLS, and was increased by IL-1b. Annexin I peptide did not demonstrate any significant effect on constitutive or IL-1b-stimulated COX2 protein in FLS (Fig. 6(a)). DEX did not significantly inhibit constitutive intracellular expression of COX2. However, DEX exerted a concentration-dependent inhibition of intracellular COX2 expression in IL-1b-stimulated FLS (Fig. 6(b)). These results demonstrate consistency between modulation of intracellular expression of COX2 and its activity.

Discussion
The hypothesis that annexin I is an anti-inflammatory mediator in arthritis originated from the detection of annexin I in human rheumatoid synovium. 24 Animal models of arthritis have subsequently demonstrated that annexin has important inhibitory effects in v ivo . For example, administration of annexin I peptide substantially inhibited carrageenan-induced arthritis, whilst anti-annexin I antibody exacerbated arthritis severity and reversed the effect of exogenous DEX in this model. 12 Similarly, exacerbation of disease and increased synovial production of tumour necrosis factor-a and PGE 2 were observed in adjuvant arthritic rats administered anti-annexin I antibody. 13 Moreover, annexin I neutralization reversed the effects of DEX on disease severity in adjuvant arthritis 13 and on rat synovial macrophage nitric oxide production. 11 In human disease, histological studies demonstrate annexin I in the synovial-lining layer of RA synovium. 24 Annexin I expression in human peripheral blood leukocytes has been described, with maximal annexin I content in neutrophils and monocytes. 23 Moreover, altered annexin I biology is seen in subjects with RA, who exhibit reduced numbers of annexin I binding sites on leukocytes compared to controls. 25 Impaired glucocorticoid induction of annexin I production has also been noted in leukocytes of RA patients, 26 and RA patients with high titre antiannexin I antibodies exhibit impaired glucocorticoid responsiveness. 27 Treatment of undifferentiated cells with glucocorticoids increases annexin I transport to the cell surface. 28 The mechanisms whereby annexin I crosses the membrane are unclear since the protein lacks a signal sequence and is therefore unlikely to access secretory vesicles for release via the conventional method of exocytosis. The translocation of annexin I protein from intracellular stores to cell surface binding sites is believed to be imperative for its function. 29,30 Certainly, extracellular anti-annexin I antibodies significantly reverse the anti-inflammatory effects of glucocorticoids and simultaneously deplete intracellular cellular annexin I. 11,13,31,32 Evidence that exogenous annexin I modulates its biological effects through specific cell surface annexin I binding sites originated from the identification of these sites on peripheral blood leukocytes. 10 Given that annexin I is expressed in RA synovium, it is conceivable that in FLS, annexin I mediates its biological effects via binding to cell surface sites. No previous study has examined the presence of binding sites or the biological function of annexin I in human synovial cells.
The current data establish the presence of annexin I binding sites on human FLS. The factors influencing annexin I binding site expression are unclear, and the absence of molecular identification of the binding site limits study of its regulation. Nonetheless, having established the presence of potentially biologically active annexin I binding sites in FLS, we sought to assess the effect of annexin I on these cells. An annexin I N-terminal peptide of 25 amino acids, annexin I ac 2-26, has been shown to mimic antiinflammatory actions of annexin I, 33,34 including inhibition of arthritis models. 11 Perretti e t a l. have reported that the biological activity of annexin I ac 2-26 is comparable with that of recombinant human annexin I, albeit at a lower molar potency, and that it requires binding to cell surface binding sites to exert its biological effects. 34 Previous investigations have evaluated the effects of glucocorticoids and cytokines on the regulation of PLA 2 and COX at the levels of transcription 6,35 or translation. 36 Studies have also compared the level of activity of these enzymes in various cell types, following IL-1 37 and DEX treatments, 7 but studies of the rate-limiting step at the enzyme activity level in human FLS have not been undertaken. Our results indicate a significant concentration-dependent inhibition of constitutive but not IL-Ib-induced PLA 2 activity by annexin I ac 2-26. In A549 cells, inhibition of arachidonic acid release by annexin I peptide is believed to be mediated through inhibition of the activation of cPLA 2 whereby cPLA 2 is not phosphorylated in the presence of annexin I. 38 Details of how this is accomplished are still uncertain, but binding site-dependent effects of ac 2-26 are unlikely to relate to the previously proposed 'substrate binding' hypothesis for the effects of annexin I on PLA 2 . The selective inhibitory action of DEX on IL-1b-stimulated but not constitutive cPLA 2 transcription and expression has been described in other investigations 6,35 but is confirmed at the enzyme activity level for the first time in the current study. The uncoupling of the effects of annexin I and glucocorticoids on PLA 2 supports the contention that the constitutive antiinflammatory effects of annexin I are only in part related to mediating the effects of glucocorticoids.
As previously demonstrated, constitutive PGE 2 synthesis was detected in cultured FLS. Cytokine upregulation of PGE 2 synthesis was observed, again consistent with previous findings. 6,36,37 Annexin I peptide did not demonstrate any significant inhibitory effect on constitutive or IL-1-stimulated PGE 2 production. In rat astrocytes, annexin I peptide reduced but did not abolish endotoxin-induced PGE 2 release. 38 Consistent with its effects on PLA 2 activity, DEX significantly reduced IL-1b-stimulated but not constitutive PGE 2 release. 7,37,38 Annexin I mAb, however, has been observed to reverse the effects of DEX inhibition on PGE 2 release, and this study also suggested that glucocorticoid suppression of COX occurs via an annexin I-independent mechanism. 39 Glucocorticoid-induced suppression of COX2 appears to be independent of annexin I and is almost certainly explained by the direct interaction of the steroid-receptor complex with nuclear factorkappa B. 40 DEX inhibition of cytokine-or mitogeninduced COX regulation has been previously reported. 6,7,41,42 In this study, regulation of intracellular COX2 expression was found to be consistent with regulation of its activity, suggesting that the level of COX activity is dependent upon intracellular levels of COX2. 43 The differential effects of annexin I peptide and DEX on PLA 2 activity, COX, and PGE 2 production support the hypothesis that COX is the rate-limiting enzyme in IL-1b-induced PGE 2 synthesis in FLS. In contrast, the lack of effect of glucocorticoids on constitutive PLA 2 and COX2 activity suggests that annexin I may have a glucocorticoidindependent role in constitutive regulation of PLA 2 activity. Biological activity of annexin I peptide has been demonstrated in v ivo in rat carrageenan arthri-tis, 12 a model mediated by eicosanoids in addition to nitric oxide and reactive oxygen species. That annexin I peptide has previously been demonstrated to have inhibitory effects on nitric oxide and reactive oxygen species production in other systems highlights the importance of viewing annexin I not solely as an inhibitor of PLA 2 activity.
In summary, three novel findings are reported in this study. Firstly, the presence of annexin I binding sites on synoviocytes is demonstrated. Secondly, uncoupling of the effects of annexin I and glucocorticoids on constitutive and cytokine-induced arachidonic acid generation suggests that annexin I has glucocorticoid-independent regulatory activities in inflammation. Our results are consistent with the conclusion that COX2 is the rate-limiting step in synoviocyte PGE 2 synthesis, suggesting that PLA 2directed strategies may not successfully inhibit eicosanoid production in arthritis. Thirdly, the lack of effect of annexin I on PGE 2 synthesis, despite clear effects on inflammation in v ivo , suggests the antiinflammatory effects of annexin I are eicosanoidindependent. For example, annexin I may influence arachidonic acid-mediated intracellular signal transduction. Arachidonic acid influences activation of the Jun-N terminal kinase/stress activated protein kinase subgroups of the membrane activated protein kinase family of signal transduction enzymes, via an eicosanoid-independent pathway. 44 This important pro-inflammatory signal pathway is also influenced by reactive oxygen species, 45 also known to be inhibited by annexin I. 46 The possibility that annexin I modulates inflammation via eicosanoid-independent mechanisms remains to be further investigated. Studies directed at understanding the role of annexin I in such pathways may have therapeutic benefit in the treatment of arthritis.