Research Paper Mediators of Inflammation, 8, 53–62 (1999)

Specific binding sites for the anti-inflammatory protein annexin I have been detected on the surface of human monocytes and polymorphonuclear leukocytes (PMN). These binding sites are proteinaceous in nature and are sensitive to cleavage by the proteolytic enzymes trypsin, collagenase, elastase and cathepsin G. When monocytes and PMN were isolated independently from peripheral blood, only the monocytes exhibited constitutive annexin I binding. However PMN acquired the capacity to bind annexin I following co-culture with monocytes. PMN incubation with sodium azide, but not protease inhibitors, partially blocked this process. A similar increase in annexin I binding capacity was also detected in PMN following adhesion to endothelial monolayers. We propose that a juxtacrine activation rather than a cleavage-mediated transfer is involved in this process. Removal of annexin I binding sites from monocytes with elastase rendered monocytes functionally insensitive to full length annexin I or to the annexin I-derived pharmacophore, peptide Ac2-26, assessed as suppression of the respiratory burst. These data indicate that the annexin I binding site on phagocytic cells may have an important function in the feedback control of the inflammatory response and their loss through cleavage could potentiate such responses.


Introduction
Annexin I (also called lipocortin 1) is a member of a family of calcium and phospholipid-binding proteins termed 'annexins', particularly abundant in human and animal leukocytes. A combination of Western blotting, ELISA and flow cytometry analyses have found high levels of this protein in monocytes and polymorphonuclear leukocytes (PMN) whereas lower amounts were detected in lymphocytes. [1][2][3] Annexin I levels in circulating leukocytes are partially under the control of endogenous and exogenous glucocorticoid hormones. A single administration of anti-inflammatory steroids to human volunteers and experimental animals increases annexin I levels associated with circulating leukocytes. 1,4,5 Conversely removal of endogenous corticosterone by means of adrenalectomy or subchronic treatment with a steroid antagonist, mifepristone (RU486), reduces cell-associated annexin I levels by 50%. 5,6 Recent observations have substantiated the role played by annexin I as a mediator of anti-inflammatory glucocorticoid hormones. Glucocorticoiddependent annexin I externalization has been demonstrated in monocytes, macrophages and PMN. 4,7,8 Consistent with this, treatment of experimental ani-mals with anti-annexin I antibodies prevented the anti-inflammatory effect of dexamathasone in several models of acute inflammation. 9,10 Besides glucocorticoid treatment, we and others have recently shown that annexin I secretion from human PMN occurs during the process of extravasation outside the blood vessel. 11,12 Much evidence now exists to support the hypothesis that annexin I interacts with the external face of the leukocyte plasma membrane to exert its anti-inflammatory properties, such as inhibition of superoxide anion generation from activated monocytes. 13 Recent studies have brought to light the existence of specific annexin I binding molecule(s) on the cell surface of peripheral blood leukocytes and several other mono-myelocytic cell lines. 14 Incubation of a mixed population of peripheral blood leukocytes with different concentrations of annexin I resulted in a calcium-dependent and saturable binding to monocytes and PMN, but not to ad-mixed lymphocytes. 15 Monocyte binding capacity has been studied further, observing its sensitivity to trypsin treatment. In addition, monocytes lose annexin I binding capacity following incubation at 37°C in a time-dependent manner. 15 Further to our initial observations, specific annexin I binding to a subset of human T-lymphocytes and on rat anterior pituitary cells has also been described. 16,17 Here we extend our previous observations. It was postulated that annexin I binding to leukocytes could be related to the anti-inflammatory profile of the protein, and that this could be modulated by the status of activation of the cells. To address this hypothesis we have investigated (i) the extent of annexin I binding to monocytes and PMN in relation to culture conditions, (ii) the susceptibility of annexin I binding to proteolytic enzymes, (iii) the connection between annexin I binding to monocytes and inhibition the respiratory burst.

Leukocyte preparation and handling
Human peripheral blood was obtained from healthy volunteers by venepuncture. Healthy donors were selected on the basis of having no current illnesses nor receiving any form of medication. On average between 60 ml and 100 ml of blood was collected into 3.2% sodium citrate (10% v:v). Blood was diluted 1:1 with sterile phosphate-buffered saline (PBS) and separated on a discontinuous histopaque gradient (1077/1119) and centrifuged at 400 3 g for 30 min at room temperature. Peripheral blood mononuclear cells (PBMC) and PMN layers were collected and cells washed three times in PBS. Erythrocytes were removed by hypotonic lysis in cold water. Finally, cells were resuspended in 1 ml RPMI 1640 medium supplemented with 0.2% BSA and 1.8 mM CaCl 2 (RBC). Cell aliquots were counted following 1:10 dilution in Turk's solution using a haematocytometer. PMN population purity was >97% with a distinct population being identified on the basis of the side scatter profile; in addition this population was CD14 negative. PMN contamination in the PBMC pellet was less than 3%. A discrete monocyte population was discerned by positivity for CD14.

Annexin I binding assay and flow cytometry analysis
Cells were resuspended at 50 3 10 6 cells/ml in RPMI 1640 containing 0.2% BSA (bovine serum albumin) and 1.8 mM CaCl 2 (RBC) and 20 m l aliquots were added to 96-well plates (Falcon) and incubated for 1 h at 4°C with 20 m 1 human recombinant full length annexin I at final concentrations ranging from 10 ng/ml to 60 m g/ml. The annexin I was prepared in PBC (PBS containing 0.2% BSA and 1.8 mM CaCl 2 ). Cells were then washed twice with PBC prior to addition of 20 m l human IgG (15 mg/ml in PBC) to block non-specific sites and 20 m l of a specific murine anti-human annexin I monoclonal antibody (mAb) 1B (60 m g/ml in PBC) and 1 h incubation at 4°C. Finally, cells were washed in PBC and incubated with 40 m l (1:30 dilution in PBC) of a F(ab9 ) 2 fragment of a goat anti-mouse IgG conjugated to FITC. After a further 30 min incubation at 4°C the cells were washed and fixed in 1% paraformaldehyde. Samples were analysed within 48 h by flow cytometry (FACScan, Becton Dickinson, Oxford, UK). Cell populations were identified by their forward and 90°s catter characteristics, and fluorescence associated with each population (monocytes within the PBMC population, and PMN) was quantified in the FL1 channel. Cell-associated mean fluorescence intensity was converted into second antibody molecules bound per cell according to a calibration curve using standard fluorescence beads. 18 As alluded to in previous publications, 15,19 this indirect methodology to assess annexin I binding to leukocytes was developed because radiolabelling protocols were found to be unsuitable for the protein, which degrades upon radiolabelling and does not retain either binding capacity or biological activity. An estimation of binding affinity was also made using a flow cytometric approximation of Scatchard analysis, in which free annexin I is calculated from total annexin I added to each tube less the amount bound to the cells.
In selected experiments monocyte populations were enriched (>80% pure) by Nycoprep TM gradient centrifugation (Nycomed Pharma AS, Oslo, Norway), and no difference was observed in the extent of annexin I binding to enriched monocytes when compared with monocytes ad-mixed with lymphocytes in the PBMC preparation (data not shown). For this reason thereafter monocytes were prepared as PBMC.
In some experiments additional cell surface markers on the monocyte plasma membrane were monitored with an anti-CD11b, anti-CD14 or an anti-CD54 mAb. All mAb were used at a 5 m g/ml final concentrations and incubated with the cells prior to staining with the F(ab9 ) 2 fragment of a goat antimouse IgG conjugated to FITC (see above).
Selected samples were also analysed by confocal microscopy. FITC-stained cells were fixed with a equal volume of 2% paraformaldehyde in PBS and pelleted. Propidium iodide (PI) was added as a nuclear counterstain and cells were mounted onto glass slides and sealed with cover slips. A BioRad MRC600 confocal microscope system with 1003 objective lens, 100 mW argon ion laser with an excitation line of 488 nm and COMOS TM analysis software was used to scan the samples and generate TIF format images.

Cell treatment and incubation conditions
Effect of PMN: monocyte co-incubation. PMN and monocyte populations isolated from peripheral blood were washed, counted and adjusted to 5 3 10 7 cells/ H. S. Euzger et al. ml in RBC. Equal numbers of PMN and monocytes were co-incubated in polypropylene tubes. At the same time, PMN and monocytes were incubated separately. After a 15 min incubation period at room temperature, 20 m l of each treatment group (PMNs alone, monocytes alone and a mixed population of the two) were plated into each well (in duplicate). In some cases, either PMN or PBMC were treated with or without 0.2% sodium azide for 30 min at 37°C prior to 15 min co-incubation at room temperature.
Effect of leukocyte: endothelial cell co-incubation. Potential variation(s) in annexin I binding capacity of human PMN or monocytes was tested in an experimental system in which cell-cell interactions certainly occur. As recently described, 11 EA.hy926 cells 20 were grown to confluence in 6-well plates in DMEM-F12 supplemented with 10% FCS and antibiotics. After washing, 1.5 3 10 6 PMNs were added in 1 ml adhesion medium (Hanks's balanced salt solution containing 0.2% BSA, 1.3 mM CaCl 2 and MgCl 2 ; all reagents from Sigma Chemical Co., Poole, UK) and incubated with 0.1 m M platelet-activating factor (PAF) was used as a stimulus to increase the extent of cell adhesion. 21 After 30 min, non-adherent leukocytes were aspirated whereas adherent cells (PMNs + EA.hy926 or monocytes + EA.hy926 cells) were removed using 1 ml of non-enzymatic cell dissociation medium (Sigma).
Effect of protease inhibitors. In another set of experiments the annexin I binding capacity of a mixed population containing equal numbers of PMNs and monocytes were assessed following incubation with either cell type the specific elastase inhibitor ONO-5046 (10 m M) or a cocktail of protease and phosphatase inhibitors (phenylmethylsulphonyl fluoride (PMSF) 1 mM, leupeptin 0.2 mM, aprotinin 0.2 mM and trypsin inhibitor 1 mg/ml) for 30 min in a water bath at 37°C.
Effect of proteases. The effect of purified inflammatory proteolytic enzymes on the extent of annexin I binding was also evaluated. Type I collagenase (0.05, 0.5 and 5 U/ml), neutrophil-derived cathepsin G (1, 10 and 100 m g/ml), neutrophil-derived elastase (1, 10 and 100 m g/ml) or soybean trypsin (250 m g/ml) were added to the cells in polypropylene tubes at 37°C up to 30 min. In some experiments elastase was coincubated with 10 m M ONO-5046, a selective catalytic inhibitor. At the end of the incubation period, proteolytic activity was removed by washing the cells twice in RPMI 1640 containing 4% BSA and 1.8 mM CaCl 2 . Monocytes were subsequently washed once in protein-free RPMI 1640 and resuspended in RBC.
Effect of monocyte supernatants. Monocytes were incubated with or without 10 m g/ml human elastase for 30 min at 37°C. Reaction was stopped by a 5 min addition of 30 m M ONO-5046 followed by centrifugation at 2000 r.p.m. for 5 min. Supernatants were collected and co-incubated with control PMNs for 20 min at 37°C. Cells were spun once more prior to performing the annexin I binding assay.
Effect of monocyte or PMN activation. The effect of monocyte or PMN activation on the annexin I binding capacity displayed by each cell type was also investigated. For this purpose monocytes were incubated with 100 ng/ml monocyte chemotactic protein-1 (MCP-1) and PMNs were activated with 0.1 m M platelet-activating factor (PAF), 1 m M FMLP or 0.1 m M phorbol 12-myristate 13-acetate (PMA).
At the end of all these different treatments, cells were thoroughly washed in PBC, resuspended in RBC (20 m l per 10 6 cells) and the annexin I binding assay performed as described above.

Measurement of respiratory burst activity
Production of reactive oxygen species from activated monocytes was measured using the fluorochrome dihydrorhodamine 6G according to the manufacturer's indications. Monocytes isolated from peripheral blood were adjusted to 2.5 3 10 6 cells/ml in PBSGC (PBS containing 0.9% glucose and 1 mM CaCl 2 ). Aliquots (200 m l) were incubated with 5 mM dihydrorhodamine 6G for 10 min in a water bath at 37°C. After the incubation period monocytes were challenged with 0.1 m M PMA or 1 mg/ml opsonized zymosan for 10 min. The time and the concentrations were chosen on the basis of preliminary experiments. Opsonized zymosan was prepared by incubating 10 mg of boiled zymosan A with 1 ml of serum (prepared from the same donor) for 30 min in a water bath at 37°C, followed by two washes in PBSGC in a microfuge (3 min at 13 000 r.p.m). After the final incubation period the reaction was stopped by adding a final concentration of 0.2% sodium azide, this being followed by immediate incubation of cells on ice, and analysis by flow cytometry.
The biological efficacy of full length annexin I purified from human placenta and the annexin I-derived peptide Ac2-26 22,23 on this parameter of monocyte activation was then related to the binding capacity displayed by the cell. Human native annexin I (0.3 -5 m g/ml) or peptide Ac2-26 (1-10 m g/ml) were added to the cells for 30 min prior to monocyte stimulation for a further 10 min with either PMA (0.1 m M) or opsonized zymosan (1 mg/ml). Also, the ability of annexin I and peptide Ac2-26 to suppress the respiratory burst was monitored in monocytes pretreated with 100 m g/ml elastase (30 min, 37°C), with the proteolytic activity being quenched with 10 m M ONO-5046 prior to washing and addition of peptide Ac2-26 (1 m g/ml) or annexin I (3 m g/ml).

Materials
Human annexin I and mAb IB, 24 used for the binding assays, was provided by Dr J.L. Browning (Biogen

Mediators of Inflammation · Vol 8 · 1999
Corporation, Cambridge, MA). Human biologically active full length native annexin I (used for the superoxide experiments) was purified from placenta and generously supplied by Dr E. Solito (INSERM, Paris, France). Human Ac2-26 (N-acetyl-AMVSEFLK-QAWFIENEEQEYVQTVK; Mr of 3050 Da, prepared at the Advance Biotechnology Centre, The Charing Cross and Westminster Medical School, London, UK) corresponds to a sequence comprising the majority of the human annexin I N-terminus region. The peptide was prepared by use of solid phase step-wise synthesis: amino acid composition and molecular weight were verified by mass spectrometry, and purity (=95%) was assessed by HPLC and capillary electrophoresis (data from manufacturer). Peptide solutions were always prepared in sterile PBS.

Statistics
Data are reported as mean ± SEM of n experiments performed in duplicate or triplicate. Approximate affinity constants (Kd) and maximal binding (Bmax) were determined using an flow cytometric approximation of Scatchard analysis. Briefly, in a seven point Scatchard plot, number of mols of free ligand were calculated as total mols added minus mols bound. Statistical differences between groups were analysed by analysis of variance followed by Bonferroni's test for post-hoc comparisons, or by two-way Student's t-test when only two groups were analysed using the Instat™ software. In all cases, a P value less than 0.05 was taken as the limit of significance.

Monocytes and PMNs bind annexin I
When purified separately, human monocytes displayed a reproducible and significant annexin I binding capacity whereas enriched preparations (>95% pure) of PMN exhibited negligible binding (Fig. 1a). However, significant changes in binding capacities were found when the binding assay was performed with both cell types in co-culture (15 min at room temperature). Figure 1a illustrates a representative experiment where monocytes lost approximately 40% of annexin I binding capacity if cocultured with PMN, and conversely the latter cell type acquired the capacity to bind annexin I. Cumulative data of n = 8 distinct experiments showed 42 ± 5% reduction of the extent of annexin I binding in monocytes which had been co-cultured with the PMN. In the same of set of experiments, PMN showed a substantial increase in annexin I binding such that 20 700 ± 2700 binding sites per cell were detected after co-incubation compared with 2300 ± 600 Expression of another surface protein following monocyte and PMN co-incubation was also monitored. High levels of the CD14 antigen were found on isolated monocytes (293 200 ± 46 000 sites per cell) whereas much lower levels were detected on freshly isolated PMNs (4800 ± 1200 sites per cell). These values remained essentially unaltered following coincubation, with 297 700 ± 32 200 and 5039 ± 2200 CD14 antigen sites per cell on monocytes and PMNs, respectively (data are mean ± SEM of n = 3 experiments performed in triplicate).
The appearance of annexin I binding was investigated by confocal microscopy. Figure 2 shows that exogenous annexin I bound to monocytes with a punctuate distribution on the cell surface. A similar pattern was also visualized on the PMN following incubation with monocytes (data not shown).
Next, we attempted to characterize the process by which monocytes and PMN respectively lost and gained annexin I binding capacity. Pre-incubation of PMN with sodium azide prior to co-incubation with the mononuclear cells prevented the acquisition of annexin I binding capacity by the PMN without altering the reduced binding capacity of the monocytes (Fig. 3a). Pre-incubation of monocytes with sodium azide had no effect. Addition of a specific elastase inhibitor (ONO-5046) or of a mixture of nonspecific protease and phosphatase inhibitors failed to prevent either the reduction in monocyte binding or the PMN increase, in annexin I binding capacity (Fig.  3b and c).

Inflammatory proteases reduce annexin I binding to monocytes
Incubation of monocytes with trypsin abolished subsequent cell binding to exogenous annexin I, thus confirming previous observations. 15 Figure 4 shows that enzymes normally found in inflammatory exudates are also capable of reducing annexin I binding. Incubation of monocytes with elastase, cathepsin G or collagenase produced a significant reduction in the number of annexin I binding sites (Fig. 4a). The specificity of elastase effect was validated by using a selective inhibitor and addition of ONO-5046 prevented the loss of annexin I binding by elastase treatment (Fig. 4b).
The efficacy of elastase prompted the cell transfer experiment. Addition to PMNs of a supernatant derived from elastase-treated monocytes (in which the enzymatic activity had be quenched with ONO-5046) did not promote annexin I binding capacity: 3900 ± 950 and 3200 ± 640 annexin I binding sites per cell were measured after PMN incubation with an elastase-treated monocyte supernatant and control monocyte supernatant, respectively (mean ± SEM of n = 2 experiments performed in triplicate).
The possibility that elastase, which binds to leukocyte plasma membrane, 26 could interfere with the binding by cleaving the ligand, 27 was also taken into account. Not only was ONO-5046 added to the incubation mixture prior to performing the annexin I binding assay (reported to block elastase activity in Fig. 4b), but also an annexin I preparation fully cleaved by elastase was tested. This preparation (33 Kd isoform) produced a curve of binding to monocytes comparable with that obtained using intact full length annexin I. For instance, 24 400 ± 170 and 24 500 ± 300 annexin I molecules bound per monocyte were found following incubation with 20 m g/ml intact full length annexin I or elastasecleaved annexin I, respectively. Figure 5a shows that the modest annexin I binding capacity displayed by freshly prepared PMNs was greatly increased following adhesion to endothelial monolayers. This increase was not modified to any extent by cell-derived proteases which are likely to be present in this microenvironment, since a similar degree of annexin I binding was also attained when a mixture of non-specific protease inhibitors (see above) was added to the adhesion medium: this cocktail of protease inhibitors was internally validated since it prevented the reduced binding of annexin I to post-adherent endothelial cells (data not shown). When monocytes were used in the same experimental procedure, a significant (~50%) reduction in annexin I binding capacity was displayed by postadherent cells (Fig. 5b).

Effect of leukocyte activation
PMN activation either with 0.1 m M PAF, 1 m M FMLP or 1 m M PMA did not modify the annexin I binding capacity displayed by this cell type (Fig. 6). As expected, co-incubation of PMNs with monocytes increased annexin I binding to the neutrophils, and this was used as a positive control in this set of experiments. Similarly, monocyte activation with 100 ng/ml MCP-1 did not cause any reduction on the extent of annexin I binding to the cells (data not shown).

Biological significance of annexin I binding to human monocytes
Monocyte incubation with opsonized zymosan (1 mg/ ml) or PMA (0.1 m M) produced a respiratory burst quantified by flow cytometry (Fig. 7a). This release was selectively inhibition by annexin I. Figure 7b shows a steep concentration-dependent curve for annexin I inhibition with a maximal effect ranging between 20% and 30% at 3 m g/ml. Annexin I inhibitory action was specific because the protein was unable to alter monocyte respiratory burst when cells were activated with PMA (Fig. 7b). The annexin I pharmacophore, peptide Ac2-26, was also able to inhibit superoxide production displaying selectivity towards opsonized zymosan (Fig. 7c). Monocyte pre-treatment with elastase, a procedure known to reduce significantly annexin I binding (see Fig. 4), prevented the inhibitory action of annexin I and peptide Ac2-26 on zymosan-induced superoxide generation (Fig. 7d). The efficacy of elastase in reducing annexin I binding was also monitored. Binding was reduced by >95% following treatment of monocytes with 100 m g/ml elastase (n = 4 experiments, P < 0.05). The elastase effect was specific because the enzyme did not alter the extent of CD11b or CD54 expression on the monocyte plasma membrane (117 400 ± 8900 and 116 200 ± 6200 CD11b molecules per cell, and 6300 ± 300 and 6600 ± 300 CD54 molecules per cell in the absence or presence of elastase, respectively).

Discussion
The main conclusion from this study is that annexin I binding capacity expressed by resting monocytes is proteinaceous in nature and can be dramatically reduced following co-incubation with PMN. We also present strong evidence that occupation of annexin I binding sites on monocytes is a pre-requisite for the inhibitory biological actions exerted by this antiinflammatory mediator. As previously reported, 15,19 human monocytes display a saturable annexin I binding capacity following in vitro incubation with the human recombinant protein. Similarly, we confirmed that trypsin treatment of monocytes abolished their capacity to bind annexin I, and have shown for the first time that this binding is susceptible to the action of other more specific proteolytic enzymes. Both collagenase, cathepsin G and elastase treatment of monocytes resulted in almost complete loss of annexin I binding capacity. This is unlikely to be due to a destruction of the ligand, 27 since (i) cells were extensively washed in high BSA media; (ii) specific inhibitors were added to block the catalytic site of the protease, as in the case of elastase, and (iii) elastase-cleaved annexin I, generating a 33 Kd isoform, was able to bind to monocytes to a similar degree as intact annexin I.
However, this binding is vulnerable in the presence of PMN. In previous studies PMN were always coincubated with monocytes in the binding assay, and showed a binding capacity with a Kd similar to the one calculated for monocytes. 19 Using purified cultures, two distinct processes can be distinguished: monocytes (either pure or together with lymphocytes), but not PMN, express constitutive annexin I binding capacity, which however can be differentially modified in co-culture conditions. PMN acquired annexin I binding capacity only after co-incubation with the monocytes, which in turn showed a comparable degree of loss. A similar Kd could be calculated for both cell types. On this basis, we suggest that annexin I binds to an identical 'site' on both leukocyte types.
There are two potential ways by which the annexin I binding sites on the two cell types could be altered: either the binding site is released (by a proteolytic event) in the culture medium and adheres to the PMN, or it is differentially expressed as a consequence of activation phenomena occurring between the adjacent monocyte and PMN. Examples of the latter, i.e. cell activation through a juxtacrine mechanism, have been recently reported. 28 Based on the experimental data obtained, we believe that the first hypothesis is unlikely for at least three reasons. Firstly, despite the susceptibility showed by the monocyte annexin I binding capacity to proteolytic enzymes, the loss observed after coincubation with the PMNs was not blocked by a mixture of non-specific protease inhibitors or by the specific elastase inhibitor ONO-5046. Secondly, addition to PMNs of a supernatant from elastase-treated monocytes was without effect. Thirdly, monocyte activation with the specific chemoattractant and activator MCP-1 did not reduce the annexin I binding capacity displayed by these cells. Similarly, PMNs did not increase their capacity to bind annexin I following incubation with PAF, FMLP or PMA. All these data rule out the possibility that proteolytic activities, tran-siently exposed on the PMN plasma membrane, 26 could be responsible for this phenomenon.
However, an active role for the PMN was suggested by the experiments with sodium azide. Blockade of PMN, but not monocyte, respiratory pathways prevented the phenomenon of annexin I binding acquisition by the neutrophil without affecting the loss from the monocyte. The hypothesis of a juxtacrine process is clearly reinforced by the experiments of PMN interaction with endothelial cells. Adherent PMNs acquired a considerable binding capacity compared to control cells. This was clearly due to the adhesion process, and not to the addition of FMLP, since this cell activator per se is totally inactive of PMN in suspension. In this context, it is interesting to note that following PMN adhesion to endothelial monolayers, endogenous annexin I is externalized on the surface of the adherent leukocytes. 11 It therefore seems that the entire annexin I pathway can be fully activated at the interface between adherent PMN/ endothelial cells.
In conclusion we propose that PMN co-incubation with monocytes results in acquisition of annexin I binding as a result of an active phenomenon due to a juxtacrine mechanism and involving new protein expression on the plasma membrane. It remains unclear how the annexin I binding protein could disappear from the monocyte plasma membrane. It seems unlikely that this is the result of a proteolytic process, but further studies are necessary to determine if this is the result of internalization following juxtacrine activation in the presence of PMN.
Annexin I possesses the biological profile of an inhibitory protein with the ability to suppress selected parameters of cell activation, including superoxide anion generation and arachidonate, eicosanoids and nitric oxide release. 13,29,30 The annexin I-derived N-terminus peptide, peptide Ac2-26, has been reported to inhibit arachidonate and elastase release from activated PMNs 21 and macrophage phagocytosis. 31 In the present study we report for the first time that annexin I, and its pharmacophore peptide Ac2-26, must interact with proteins associated with the monocyte plasma membrane in order to suppress cell activation, measured as respiratory burst. The inhibition was stimulus-dependent, such that when cells were activated with PMA, which bypasses receptormediated phenomena by activating intracellular protein kinases, no effect was exhibited by annexin I or peptide Ac2-26. A maximal inhibition of 20-30% was reached in this set of experiments which is similar to that obtain with peptide Ac2-26 on murine macrophage phagocytosis using the same protocol. 31 Importantly, annexin I showed an approximate IC50 of 1.2 m g/ml (32 nM) which is close to the Kd calculated from the binding experiments. Pretreatment of monocytes with elastase did not affect the extent of cellular respiratory burst following opso-nized zymosan or the degree of expression of specific surface markers (CD11b and CD54). In contrast, elastase abrogated annexin I binding to monocytes and similarly abolished the inhibitory action displayed by annexin I or its N-terminus peptide. This strongly suggests that annexin I binding to the protein(s) on the monocyte plasma membrane 15 is instrumental in its inhibitory properties. In addition, peptide Ac2-26 action in an in vivo model of experimental inflammation was already proposed to require an interaction with the annexin I binding site, 22 and this hypothesis in reinforced by the data here presented on activated monocytes in vitro.
In conclusion, we report an exquisite sensitivity of the annexin I binding protein(s) present on human monocytes to the phenomena of activation and provide strong evidence that the binding is a prerequisite to the anti-inflammatory action produced by annexin I on this cell type.