Annals ofthe Rheumati Diseases 199; 49: 747-752 Measurement of the hemotati omplement fragment C5a in rheumatoid synovial fluids by radioimmunoassay: role of C5a in the aute inflammatory phase 747 P J Jose, I K Moss, R N Maini, T J Williams Setion of Vasular Biology, MRC Clinial Researh Centre, Harrow, Middlesex P J Jose T J Williams Department of Clinial Researh, Charing Cross and Westminster Medial Shool, London W6 I K Moss Department of Immunology of Rheulmati Diseases, C g Cross and Westmister Medial Shool, London W6 R N Maini Correspondene to: Dr P J jose, Departnent of Applied Pharmalogy, National Heart and Lung Institute, Dovehouse Street, London SW3 6LY. Aepted for publiation 21 September 1989 Abstrat videne suggests that omplement is ativated in rheumatoid joints. A sensitive radioimmunoassay for the ativation fragment of C5, C5a, whih is a potent hemoattratant for neutrophils, was therefore developed. A mean CSa onentration in 22 rheumatoid joint fluids of about 2-5xlO-9 mol/l was found. This onentration of C5a is suffiient to indue two of the harateristi features of the aute inflammatory phases of rheumatoid arthritis: neutrophil aumulation and mirovasular plasma protein leakage. In animal models it has been shown that C5a is a potent induer of inflammatory oedema by a neutrophil dependent mehanism. A striking feature of the aute inflammatory phases of rheumatoid arthritis is the appearane of high numbers of neutrophils in the synovial fluid. It is suggested that CSa might have a role in mediating neutrophil aumulation and, as a onsequene, may be important in aute joint swelling and pain. In addition to the hroni synovitis of rheumatoid arthritis (RA) are the aute flare-up reations haraterised by extravasation of neutrophils and plasma proteins. videne suggests that the omplement system is ativated in rheumatoid arthritis.' One produt of omplement ativation is CSa, a potent mediator of neutrophil infiltration and oedema.2'1 The binding of CSa to reeptors on leuoytes often renders it undetetable in the fluid phase even when omplement ativation has been learly shown. We developed " a sensitive radioimmunoassay for the measurement of CSa in inflammatory exudate fluids, and in this paper we report for the first time as far as we know, the C5a onentrations in rheumatoid synovial fluid. Ativation of the omplement system by either the lassial or alternative pathway leads to the generation of C5a.'2 This 74 amino aid glyoprotein is rapidly onverted to the physiologially more stable C5a des Arg by the ation of arboxypeptidase N. C5a is a potent anaphylatoxin, releasing histamine from mast ells and basophils; CSa des Arg is virtually inative in this respet (in man, but not in all speies). In ontrast, both forms of CSa stimulate leuoyte loomotion in vitro and emigration in vivo. The rabbit skin provides a onvenient system to test the biologial ativity of C5a in vivo. In this system both rabbit and human C5a indue inreased mirovasular permeability that is independent of the arboxyl terminal arginine and histamine release.5 6 Plasma protein leakage indued by CSa injeted intradermally is rapid in onset: radioisotopi studies showed leakage five to six minutes after injetion.7 This response, however, is entirely dependent on irulating neutrophils: neutrophil depletion abolished leakage indued by CSa but had no effet on leakage indued by histamine and bradykinin.6 7 Thus a rapid interation between neutrophils and mirovasular endothelial ells triggered by CSa gives rise to inreased endothelial permeability to plasma proteins. 3 These observations an explain the dependene of oedema on the presene of irulating neutrophils in animal models of immune vasulitis.'4 In one of these models (an Arthus-type reation in rabbits) we have shown that oedema is assoiated with the extravasular generation of OSa'5 and that the response is inhibited by anti- C5a antibodies.'6 Cleavage of the third and fourth omponents of omplement leads to the formation of C3a and C4a. Although struturally similar to CSa, these proteins interat with ells at a site whih is distint from the CSa reeptor and promote mast ell or basophil histamine release, but not leuoyte ativation. Human C3a is about a thousand times less ative than human C5a in inreasing mirovasular permeability when tested in rabbit skin.6 Synovial fluid ontains omplement,'7 18 though at lower onentrations than is found in plasma. There is evidene for both lassial and alternative pathway ativation in rheumatoid arthritis, though ativation of the lassial pathway by IgG aggregates and immune omplexes is likely to be more important. "' Conentrations of several C3 fragments are inreased in rheumatoid synovial fluids.22-24 Based on their in vitro ativity, none of these C3 fragments is likely to mediate neutrophil aumulation and neutrophil dependent oedema formation. Ward and Zvaifler reported hemotati ativity for neutrophils in 38 of 54 RA synovial fluids tested.25 Anti-C5 antibodies inhibited the ativity (usually by 8-9%) in 14 out of 15 hemotati rheumatoid synovial fluids. Inhibition by anti-c3 antibodies was deteted in only seven of the 15 fluids and was muh weaker than the inhibition by anti-c5 antibodies. These authors attributed hemotati ativity to the presene of C567 and C5a,25 although subsequent work showed that the ativity asribed to C5a had a very different eletrophoreti mobility to the purified protein.26 In
748 78ose, Moss, Maini, Williams view of its potent biologial ativity there have been surprisingly few attempts to measure CSa in synovial fluids. Wagner and Hugli studied a total of 21 patients with gout, RA, and osteoarthritis: C3a and C4a were deteted but C5a was not." Moxley and Ruddy, who found muh higher C3a onentrations in 41 patients with RA than in those with degenerative joint disease (osteoarthritis), reported the detetion of C5a in only one patient with RA.24 We developed a more sensitive radioimmunoassay for CSa than that available ommerially, and in this paper we report the measurement of biologially ative onentrations of CSa in rheumatoid joint fluids; there was no measurable CSa in the plasma of these patients. A small number of synovial fluid samples from patients with systemi lupus erythematosus, osteoarthritis, Reiter's disease, and mehanial joint trauma were also investigated. We also measured onentrations of the more abundant, but muh less inflammatory C3a. Patients and methods The patients in this study fulfilled Amerian Rheumatism Assoiation riteria for definite or lassial rheumatoid arthritis28 (2 women aged 26-69, two men aged 4 and 57) and systemi lupus erythematosus29 (five women aged 2-36). Osteoarthritis was diagnosed radiologially (three patients). The four patients with Reiter's syndrome were all men whose arthritis was assoiated with reent non-speifi urethritis. 'Traumati' synovial fluids were obtained from patients with mehanial injuries (two women aged 21 and 47, two men aged 21 and 58). All fluids were derived by arthroentesis of knees arried out for relief of symptoms. To prevent omplement ativation after olletion blood or joint fluid was olleted in tubes ontaining suffiient dry sodium DTA to give a minimum final onentration of 2 mmolll. Samples were entrifuged (15 g for 1 minutes) as soon as possible after olletion. Care was taken not to disturb the ell pellet when removing the supernatant, whih was then stored frozen in aliquots while awaiting assay. RADIOIMMUNOASSAY The stable des Arg metabolites of C5a and C3a were purified from human plasma ativated by zymosan using methods desribed by Fernandez and Hugh3 followed by hromatofousing (in the ase of C5a des Arg only)'5 and ation exhange high performane liquid hromatography (HPLC). Immunisation with these proteins was used to obtain goat antihuman C5a and rabbit antihuman C3a. The goat antiserum reated equally with CSa and C5a des Arg but did not ross reat with C3a or C3a des Arg (< 3%). The rabbit antiserum reognised C3a and C3a des Arg but not CSa or CSa des Arg (<e 1%). As has been reported by others,3' 32 C5 ross reated with anti-c5a (23% on a molar basis) and C3 ross reated with the anti-c3a (42%). For this reason, it was essential to remove the C5 and C3 preursors before measuring the onentrations of their respetive split produts of ativation. A ommerially available preipitating agent supplied with radioimmunoassay kits32 did not ompletely remove these preursors from samples ontaining varying protein onentrations suh as those found in inflammatory exudate fluids. We therefore developed a system using a ombination of polyethylene glyol 6 (PG) and protamine sulphate whih effetively removes C5 and C3 under these onditions (fig 1). Samples of plasma or joint fluids (-2 ml undiluted for CSa assays, and diluted twofold for C3a assay) (fig 1) were mixed with an equal volume of a mixture of 22% PG and 1% protamine sulphate in 1 mm sodium phosphate buffered saline (PBS), ph 7 4, ontaining -1% sodium azide and 1 mm DTA. After inubation at 4 C for one hour and entrifugation at 53 g for 1 minutes the supernatants were removed for assay. Under these onditions the reovery of CSa des Arg and C3a des Arg in the supernatants was at least 8% while no preursor C5 and C3 was detetable (< 1%) (fig 1). Furthermore, reovery of IgG was far too low (1-2%) to interfere with the protein A separation step desribed below. The ompetitive binding radioimmunoassay proedure was modified from that desribed previously for rabbit CSa. '5 Radiolabelled traer ligands were prepared by inubating 1 [tg C5a C3a des Arg) with 37 MBq Na'251 in the presene of a solid phase gluose oxidase-latoperoxidase system (nzymobeads; Rio-Rad Laboratories). The speifi ativities obtained were 59-78 kbq/4g CSa des Arg and 48-63 kbq/tg C3a des Arg. Using the proedure desribed below, binding of radiolabel in the presene of exess antiserum was 93% for CSa des Arg and 97% for C3a des Arg. Standards and, where neessary, sample supernatants were diluted with a mixture of 11% PG and 5% protamine sulphate in PBS, 1% azide, 1 mm DTA. 1251 labelled ligands, antisera, and protein A baterial adsorbent (PABA; Miles Sientifi) were diluted with PBS ontaining -5% protamine sulphate, -2% gelatin, -1% azide, and 1 mm DTA. Sample supernatants (1 1d) were mixed with 1 1t radiolabelled ligand ( 5 ng '251-CSa des Arg or 2- ng 1251_ des Arg (or 2 [ig C3a des Arg) and 1 [L antiserum (goat anti- C5a diluted 1/34 or rabbit anti-c3a diluted 1/22 ). After inubation for 4 hours at 2-24C an exess of PABA (25,tl of a 5% suspension) was added. After a further one hour 1 ml of PBS (- 1% azide) was added, the tubes were entrifuged (53 g for 1 minutes) and the supernatants were removed by sution. The antibody bound radioativity in the pellets was ounted in a multiwell gammaounter (LKB Walia 126, Multigamma II) for a time suffiient to register at least 1 ounts in samples to whih no unlabelled higand was added. Conentrations were determined from standard urves onstruted by use of a omputer assisted spline fit programme. The limits (mean (SM)) of adequate measurements (defined as the onentration required to give 2% inhibition of binding of the radiolabelled ligand) were 6- (-2) ng C5a/ml plasma or joint fluid and 22-1 (1 4) ng
Measurement ofcsa in rheumatoid syovialfluids by radioimmunoassay Q C L. C 1-8- "IS 6-2> 4- u, LI 4, 2- O- C5a C5a Is I r,, 1 1 5 1 1 1 Plasma onentration (%J Figure I ffets ofprotein onentration on the preipitation ofc5 and C3 and the reovery ofcsa and C3a. Reovery of 125I traer in the supernatant ofplasma samples treated with PGlprotamine sulphate (a) and, for omparison, the preipitation agent supplied with the Upjohn radioimmunoassay kits (). The left handpanel shows resultsfor CSa des Arg and CS; the right hand panel shonvs resultsfor C3a des Arg and C3. Samplesfor CSa radioimmunoassay were used undiluted and samples for C3a assay were diluted twofold before addition ofthe PGlprotamine sulphate preipitating agent to ensure that reovery ofthe split fragments was at least 8%. ah point is the mean ofthree determinations. C3a C3a G3a/ml (n= 15 assays). These assays were about twie as sensitive as the ommerial kits. The interassay oeffiients of variation in the C5a assays were 21% for a rheumatoid joint fluid and 9/o for a partially ativated plasma sample. In the C3a assays these values were 7% for the rheumatoid joint fluid and 14% for the ativated plasma sample. Intra-assay oeffiients of variation for rheumatoid joint fluid were 4% in both radioimmunoassays. IN VITRO STUDIS Peripheral venous blood was withdrawn into heparin (1 U/ml) for preparation of plasma. Plasma was ativated at 37 C for 3 minutes using a dose range of ativator suspensions in saline. Sodium DTA (ph 7-4) and magnesium GTA (ph 7-4) were added to a final onentration of 1 mmol/l, as apppropriate. Inubations were terminated by the addition of sodium DTA, ooling in ie-old water, and entrifugation at 2 C. The ativators used were heat aggregated gammaglobulin (HAGG: human IgG (Sigma) heated to 63 C for one hour) and yeast ell walls (zymosan A, Sigma). HPLC ANALYSIS For HPLC analysis of immunoreativity in rheumatoid joint fluid and ativated plasma samples, PG/protamine sulphate supernatants were mixed with an equal volume of 25 mm sodium phosphate buffer, ph 5-8, and applied to a TSK 535 CM olumn (7-5x15 mm). Proteins were eluted using a 1-9 mm sodium hloride gradient in the ph 5 8 buffer at a flow of 1 ml/minute. Frations (1 ml) were olleted into -2 ml of a solution designed to adjust the protamine sulphate onentration to 749-5% and the ph to about 7-4 for subsequent radioimmunoassay. Results GNRATION OF C5a AND C3a DURING ACTIVATION OF CLASSICAL AND ALTRNATIV PATHWAYS IN VITRO To investigate omplement ativation in vitro, plasma was inubated with varying doses of HAGG and zymosan. Generation of immunoreative CSa and C3a was related to the dose of ativator used (fig 2). Magnesium GTA, whih is known to helate alium ions and inhibit lassial pathway ativation, onsiderably redued ativation by HAGG but had no inhibitory effet on ativation by zymosan. Sodium DTA, whih is known to helate magnesium as well as alium ions and therefore by zymosan. These results show that the - I rr. -- TT A tn -_. C _r. prmipal eiiet O HAUA was ativation O tme lassial pathway while zymosan was effetive in ativating the alternative pathway of omplement, and that ativation of either pathway leads to the generation of CSa and C3a immunoreativity. 1.5 - p Zym + MgGTA -R 1.-. i; I/ 'HAGG.5- LA - -_ HAGG + MgGTA i 1-I lv + DTA 4-1 3- t. 2- N - HAGG + MgGTA 1- O- -T- r-m-t + DTA.6- * Zym Zym + MgGTA.4-.~- Ln ue N m- - - _ - - -O.2- - HAGG + MgGTA.2 - "_ -*HAGG - --',.2.2 2. Ativator onentration (mg/ml plasma) Figure 2 Complement ativation in vitro. Both heat aggregated gammaglobulin (HAGG) and zymosan produed a dose related generation ofimmunoreative CSa (upper panel) and C3a (middle panel). Magnesium GTA (open symbols) inhibited ativation by HAGG but not by zymosan. The lowerpanel shows that HAGG produes less CSa relative to C3a than does zymosan. This may reflet differenes in the physial nature ofthe ativator or differenes in the abtlity oflassial and alterativepathway enzymes to leave C3 and C5. ah point is the mean of dupliate inubations. C3 to inhibit both pathways of omplement, 5 1 ompletely inhibited ativation by HAGG and
75 7ose, Moss, Maini, Williams 4- t 2- U-, - The results shown in fig 2 also show that the ratio of C5a:C3a generated depends on the ativator used as zymosan produed a higher ratio than HAGG. Thus measurements of C3a annot be used to predit aurately the onentrations of CSa when the nature of the stimulus is unknown. C5a AND C3a CONCNTRATIONS IN JOINT FLUIDS Joint fluids from patients with RA, systemi lupus erythematosus (SL), osteoarthritis, Reiter's disease, and mehanial trauma were analysed for immunoreative C5a and C3a (fig 3). For omparison, plasma samples from 11 of the 22 patients with RA were also analysed. CSa onentrations were well below the limit of adequate measurement (6 ng/ml) in the rheumatoid plasma samples. Joint fluids from patients with RA ontained 4-86 ng C5a/ml (mean=21). Fluids taken from mehanially traumatised joints ontained low C5a onentrations. Although the sample numbers were low, the onentrations of C5a in synovial fluid samples from patients with SL, osteoarthritis, and Reiter's disease were inreased but not to the degree seen in RA. Rheumatoid plasma samples ontained C3a (-26 ( 7) jig/ml). G3a onentrations in joint fluid were greatly inreased in RA (mean 4-6 jg/ml); the other onditions showed a range of C3a onentrations above that in plasma. :. *86 4. 69 @ : * * -4. The onentrations of CSa and C3a in the 22 individual rheumatoid joint fluids are orrelated in fig 4. Although there is a positive orrelation, measurement of C3a was a very poor preditor of the onentration of C5a in individual joint fluids. The two patients with RA with the highest onentrations of synovial C5a had normal onentrations of irulating C3, C4, and IgG. One of these patients, whose joint fluid ontained 86 ng CSa/ml and 7A46,ug G3a/ml, was strongly seropositive for rheumatoid fator and was on a repeat ourse of gold treatment at the time of sample olletion. The other, whose joint fluid ontained 69 ng CSa/ml and 7'76 tig G3a/ml, was seronegative and was being treated with indomethain and prednisolone at the time of sample olletion. Both samples were taken during aute flare-up reations. HPLC OF IMMUNORACTIVITY IN RHUMATOID JOINT FLUID COMPARD WITH ACTIVATD PLASMA A sample of rheumatoid joint fluid (1P8 ml ontaining 3 ng C5a and 14 5 pg C3a) was mixed with the PG/protamine sulphate preipitation agent. The supernatant was then mixed with olumn buffer and applied to a arboxymethyl ation exhange HPLC olumn. The olumn was eluted with a salt gradient and frations olleted for radioimmunoassay. To ompare the elution profiles this was followed by injetion of the C5a des Arg and C3a des Arg standards. The profile of immunoreativity in the rheumatoid joint fluid is shown in fig 5. Immunoreative CSa eluted as a disrete peak at fration 18, orresponding to the position of standard C5a des Arg. Some apparent immunoreativity 12 Rheumatoid joint fluids m 12-8-. S -o le U u 1 - s I 4- -. *. RA RA SL OA Reiter's Trauma plasma disease FWi 3 Conenaio ofimmnwreative CSa (upperpanel) and C3a (lower panel) in jointfluids andplasma.jointfluidsfirom 22 patents with rheuatoid arthriis (RA) ontained a wide range ofcsa onentrations with a mean value of21 ng/ml. In ontrast, plasma samples (shown as smallr ymbos) taknfrom II ofthese patients had virtually dele onentrations ofcsa (I 3 (-3) nglml). The mean values ofc3a in rheumatoid jointfluids was 4-6 g/mi, whih is muh higher than onentrations found in rheumatoid plasma samples (-26 (7) pglml).. * 1-12 13 14 C3a onentratlon (ng/mil Figue 4 Correlaon ofcsa and C3a onentrations in indiviual rhewnatoidjointfluids. Linear regression analysis oflog tasfed data showed a positive orrelation (r=-7) between CSa and C3a onentations. The large deviation ofdata pot fnmn the predited line shows that m oreative C3a ement is a poorpreditor ofthe onentration ofthe biogially more relevant CSa in individual sampes.
Measurement ofcsa in rheumatoid synovialfluids by radioimmunoassay u Ln 2 1- - 1 Rheumatold joint fluid Ativated plasma CSa desarg -- I- I C3a desarg 1 2 3 Retention time (minutes) Figure S Radioimmunoassays for CSa and C3a after ation exhange high performane liquid hromatography. Immunoreative CSa () in rheumatoidjointfluid eluted as a single peak, orresponding to the retention time ofthe CSa des Arg standard (upper panel). C3a immunoreativity () eluted as two peaks, the major one orresponding to the retention time ofc3a des Arg. For omparison, plasma was ativated in vitro by inubation with heat aggregated gammaglobulin and hromatographed (lower panel). Immunoreative C3a eluted as a single peak orresponding to the C3a des Arg standard. In ontrast with thejointfluid, the plasma sample did not ontain the peak ofunidentified C3a immunoreativity. Immunoreative CSa showed a similar elution profile to that seen with the jointfluid. was seen in later frations; this was oelow the limit of adequate measurement and was shown to be a refletion of the high salt onentrations (up to 9 mmol/l NaCl) in these frations. Frations 17-19 ontained 83% of the immunoreativity applied to the olumn, whih is a high reovery for suh a low C5a onentration. The major peak of immunoreative C3a eluted in fration 25, orresponding to the position of standard C3a des Arg. There was, however, a smaller peak of immunoreativity in fration 23. This profile was onfirmed in a subsequent assay and both peaks gave omplete displaement of radiolabelled ligand at low dilutions and showed parallels with the standard when tested in serial dilutions. The early eluting peak of C3a immunoreativity may represent a breakdown produt of C3a or C3a des Arg or a slightly different leavage produt of C3, perhaps generated by proteases released from tissue ells or the aumulating leuoytes. When plasma samples ativated by HAGG (fig 5) and zymosan (data not shown) were analysed, immunoreative C3a eluted with the same retention time as the standard C3a des Arg. There was no evidene, however, for the smaller early eluting peak of immunoreative C3a previously seen in the rheumatoid joint fluid. As was found with the joint fluid, immunoreative C5a in the ativated plasma samples eluted as a single peak orresponding to the mobility of CSa des Arg (fig 5). I Disussion Complement ativation has been well doumented in the inflamed joints of rheumatoid arthritis.' Of the produts of omplement ativation, CSa and its more stable des Arg metabolite have potent proinflammatory ativities whih might aount for some of the features of the aute flare-up phases of rheumatoid arthritis, haraterised by plasma protein and neutrophil extravasation. Moxley and Ruddy, who used ommerial radioimmunoassay kits and found high onentrations of C3a in rheumatoid joint fluids, onluded that the onentrations of CSa were either below or barely above the limit of detetion of the CSa kit.24 We developed a CSa radioimmunoassay of greater sensitivity and with more reliable elimination of the rossreating C5 preursor, espeially when using samples of variable protein onentrations. The results show that C5a was present in rheumatoid joint fluids, but not in the plasma of these patients (fig 3). The immunoreative C5a in joint fluids is assumed to be C5a des Arg owing to the ation of endogenous arboxypeptidase N. Moreover, the immunoreativity hromatographed as a single peak, orresponding to the position of CSa des Arg on ation exhange HPLC (fig 5). The system used will distinguish between CSa and CSa des Arg (data not shown). The mean value found in rheumatoid joint fluids was 21 ng/ml (about 2-5x1-9 mol/l) and this onentration of CSa/CSa des Arg is suffiient to promote neutrophil hemotaxis in vitro,33 oedema formation, and neutrophil aumulation in rabbit skin 7 ' and wheal and flare reations in human skin.34 The detetion of high onentrations of immunoreative C3a agrees with the results of Moxley and Ruddy.24 C3a is not a mediator of neutrophil aumulation and assoiated oedema formation,6 although it an ause mast ell degranulation and a wheal and flare reation in human skin, provided that the arboxyl terminal arginine is present. Rheumatoid synovial tissues ontain mast ells whih an be stimulated to release histamine.35 It is doubtful, however, that histamine release by anaphylatoxins is important in rheumatoid arthritis beause treatment with antihistamines is ineffetive and C3a is even more suseptible to onversion to its des Arg metabolite than C5a.36 In this ontext it is interesting to note that oedema formation and neutrophil aumulation indued by injetion of C5a and CSa des Arg in rabbit and human skin are largely resistant to antihistamines. 5 6 34 37 38 The orrelation between C5a and C3a onentrations in individual fluids was poor (fig 4). This may be explained by in vitro studies. There is general agreement that various types of leuoytes bind C5a but not C3a."1 751 39 4 When omplement was ativated in vitro in the absene of ells the ratio of immunoreative C5a:C3a was also found to vary with the stimulus (fig 2). Heat aggregated gammaglobulin, used as a model for rheumatoid fator omplexes with IgG, generated a lower C5a:C3a ratio than zymosan. This may reflet differenes in the physial nature of the ativator as well as differenes in the relative effiienies of the C3 and C5 onvertases generated during ativation of the lassial and alternative pathways. Furthermore, Ng et al found that immune omplexes ontaining IgM x rheumatoid fator ativate C3 in the fluid phase yet fix C3 poorly4"; this would be expeted to generate lower than normal
752 7ose, Moss, Maini, Williams C5 onvertase ativity and thus to yield low C5a:C3a ratios. Taken together, the variable C5a:C3a ratios generated during ativation and the subsequent removal of C5a by leuoytes might aount for the variable C5a:C3a ratios observed in the rheumatoid joint fluids. Chemotati ativity related to C5 was found in rheumatoid joint fluid in another study: this ativity was attributed to the presene of the C567 omplex and C5a.26 Subsequent work, however, ast doubt on the identifiation of C5a beause the hemotati ativity asribed to it migrated anodally on Pevikon blok eletrophoresis.26 The generation of C5a an be inferred from the detetion of inreased onentrations of the fluid phase terminal omplement omplex in rheumatoid joint fluids,42 as leavage of C5 with the liberation of C5a is an essential step in assembly of this omplex. Previous attempts to measure C5a in synovial fluids have been unsuessful.24 27 Using a more sensitive radioimmunoassay, we have now shown that rheumatoid joint fluids ontain biologially ative onentrations of GSa. In animal models we have shown that C5a (and its des Arg metabolite) is a potent induer of oedema by a neutrophil dependent mehanism.6 7 A striking feature of the aute inflammatory phases of rheumatoid arthritis is the appearane of high numbers of neutrophils in synovial fluid. We suggest that C5a might have an important role in mediating neutrophil aumulation and, as a onsequene, may be important in aute joint swelling. This work was supported by a grant from the Arthritis and Rheumatism Counil. We thank Dr R A Harrison, Mehanisms in Tumour Immunity Unit, MRC Centre, Cambridge for the generous gifts of C3 and C5. I Goldstein I M. Clinial appliations of omplement measurements in rheumati diseases. Am J Med Si 1975; 269: 172-82. 2 Snyderman R, Phillips J K, Mergenhagen S. Biologial ativity of omplement in vivo. Role of C5 in the aumulation of polymorphonulear leukoytes in inflammation. i xp Med 1971; 134: 1131-43. 3 Damerau B, Vogt W. ffet of hog anaphylatoxin (CSa) on vasular permeability and leukoyte emigration in vivo. Naunyn Shmiedebergs Arh Pharmaol 1976; 255: 237-41. 4 Issekutz A C, Movat K W, Movat H Z. nhaned vasular permeability and hemorrhage-induing ativity of rabbit CSa des arg: probable role of polymorphonulear leukoyte lysosomes. Cl:n xp Immunol 198; 41: 512-2. 5 Williams T J, Jose P J. Mediation of inreased vasular permeability after omplement ativation: histamineindependent ation of rabbit COa. _7 xp Med 1981; 153: 136-53. 6 Jose P J, Forrest M J, Williams T J. Human CSa des Arg inreases vasular permeability. J Immunol 1981; 127: 2376-8. 7 Wedmore C V, Williams T J. Control of vasular permeability by polymorphonulear leukoytes in inflammation. Nature 1981; 289: 646-5. 8 Jose P J. Complement-derived peptide mediators of inflammation. Br Med Bull 1987; 43: 33649. 9 Yaney K B. Biologial properties of human CSa: seleted in vitro and in vivo studies. Clin xp Immunol 1988; 71: 27-1 1 Rampart M, Williams T J. videne that neutrophil aumulation indued by interleukin-1 requires both loal protein biosynthesis and neutrophil CD18 antigen expression in vivo. Br]7 Pharmaol 1988; 94: 1143-8. 11 Chenoweth D, Cooper S W, Hugli T, et al. Complement ativation during ardiopulmonary bypass. videne for generation of C3a and CSa anaphylatoxins. N nglj7 Med 1981; 34: 493-53. 12 Hugli T, Muller-berhard H J. Anaphylatoxins: C3a and CSa. Adv Immunol 1978; 26: 1-53. 13 Williams T J. Inreased vasular permeability following interation between polymorphonulear neutrophils and vasular endothelial ells. In: Bagge U, Born G V R, Gaehtgens P, eds. Morphology and rheology as related to funtion. The Hague, Holland: Martinus Nijhoff, 1982: 14 Humphrey J H. The mehanism of Arthus reations. I. The role of polymorphonulear leuoytes and other fators in reversed passive Arthus reations in rabbits. Br 7 xp Pathol 1955; 36: 268-89. 15 Jose P J, Forrest M J, Williams T J. Detetion of the omplement fragment C5a in inflammatory exudates from the rabbit peritoneal avity using radioimmunoassay. 7 xp Med 1983; 158: 2177-82. 16 Hellewell P G, Jose P J, Williams T J. ffet of anti-csa antibodies on oedema formation and PMN leukovte aumulation in allergi inflammation in the rabbit. Br ] Pharmaol 1988; 95: 531. 17 Hedberg H. Studies on the depressed hemolyti omplement ativity of synovial fluid in adult rheumatoid arthritis. Aia Rheumatologia Sandinavia 1963; 9: 165-93. 18 Pekin T J, Zvaifler N J. Hemolyti omplement in synovial fluid.] Clin Invest 1964; 43: 1372-82. 19 Winhester R J, Agnello V, Kunkel H G. Gamma globulin omplexes in synovial fluids of patients with rheumatoid arthritis. Clin xp Immunol 197; 6: 689-76. 2 Ruddy S, Austen K F. The omplement system in rheumatoid synovitis. I. An analysis of omplement omponent ativities in rheumatoid synovial fluids. Arthritis Rheum 197; 13: 713-23. 21 l-ghobarey A F, Whaley K. Alternative pathway omplement ativation in rheumatoid arthritis. ] Rheumatol 198; 7: 453-6. 22 Nydegger U, Zubler R H, Gabav R, et al. Cirulating omplement breakdown produts in patients with rheumatoid arthritis: orrelation between plasma C3d, irulating immune omplexes, and linial ativity. 7 C/in Invest 1977; 59: 862-8. 23 Bourke B, Moss I K, Maini R N. Measurement of the omplement C3 breakdown produt C3d bv roket immuno-eletrophoresis. 7 Immunol Methods 1982; 48: 97-18. 24 Moxley G, Ruddy S. levated C3 anaphvlatoxin lvels in] synovial fluids from patients with rheumatoid arthritis. Arthnrtis Rheum 1985; 28: 189-95. 25 Ward P A, Zvaifler N J. Complement-derived leukotati fators in inflammatory synovial fluids of humans. 7 Clin Invest 1971; 5: 66-16. 26 Ward P A. Complement-dependent phlogisti f'ators in rheumatoid synovial fluids. Ann NY Aad S:i 1975; 256: 169-76. 27 Wagner J L, Hugli T. Radioimmunoassav for anaphvlatoxins: a sensitive method for determining omplement ativation produts in biologial fluids. Anal B:ohem 1984; 136: 75-88. 28 Ropes M W, Bennett G A, Cobb S, et al. Diagnosti riteria for rheumatoid arthritis: 1958 revision. Ann Rheum )is 1959; 18: 49-53. 29 Tan M, Cohen A S, Fries J F, et al. 'I'he 1982 revised riteria for the lassifiation of svstemi lupus erythematosus. Arthritis Rheum 1982; 25: 1271-7. 3 Fernandez H N, Hugli T. Primary strutural analysis of the polypeptide portion of human CSa anarhylatoxin: Polvpeptide sequene determination and assignment of the oligosaharide attahment site in C5a..7 Biol Chemn 1978; 253: 6955-64. 31 Hugli T, Chenoweth D. Biologially ative peptides of omplement: tehniques and signifiane of C3a and C5a measurement. In: Nakamura R M, Dito W R. Tuker S, eds. Immunoassays: linial laboratory tehniques for the 198s. New York: Alan R Liss, 1981: 443-6. 32 Satoh P S, Yonker T C, Kane D P, et al. Masurmnt of anaphylatoxins: an indx for ativation of omplmnt asades. BioTehniques 1983; 1: 9-5. 33 Beker L. Chemotati fators in inflammation. Trends Pharmaol Si 1983; 4: 223-5. 34 Swerlik R A, Yaney K B, Lawley T J. A diret in vivo omparison of the inflammatory properties of human CSa and CSa des Arg in human skin. ] Immunol 1988: 14: 2376-81. 35 Gruber B, Poznansky M, Boss, et al. Charaterization and funtional studies of rheumatoid synovial mast ells: ativation by seretagogues, anti-ig, and a histaminereleasing lymphokine. Arthritis Rheum 1986; 29: 944-55. 36 Hugli T, Gerard C, Kawahara M, et al. Isolation of three separate anaphylatoxins from omplement-ativated human serum. Mol Cell Biohem 1981; 41: 59-66. 37 Yaney K B, Hammer C H, Harvath L, et al. Studies of human C5a as a mediator of inflammation in normal human skin.]i Clin Invest 1985; 75: 486-95. 38 Williamson L M, Sheppard K, Davies J M, et al. Neutrophils are involved in the inreased vasular permeability produed by ativated omplement in man. Br] Haematol 1986; 64: 375-84. 39 Chenoweth D, Hugli T. Demonstration of speifi C5a reeptor on intat human polymorphonulear leukoytes. Pro Natil Aad Si USA 1978; 75: 3943-7. 4 Solomkin J S, Cotta L A, Satoh P S, et al. Complement ativation and learane in aute illness and injury: videne for CSa as a ell-direted mediator of the adult respiratory distress syndrome in man. Surgery 1985; 97: 668-78. 41 Ng Y C, Peters D K, Walport M J. Monolonal rheumatoid fator-igg immune omplexes: poor fixation of opsoni C4 and C3 despite effiient omplement ativation. Arthritis Rheum 1988; 31: 99-17. 42 Mollnes T, Lea T, Mellbye J, et al. Complement ativation of rheumatoid arthritis evaluated by C3dg and the terminal omplement omplex. Arthnrtis Rheum 1986; 29: 715-21.