Functional site of endogenous phospholipase A 2 inhibitor from python serum Phospholipase A 2 binding and anti-in¯ammatory activity Maung-Maung Thwin 1 , Ramapatna L. Satish 2 , Steven T. F. Chan 2 and Ponnampalam Gopalakrishnakone 1 Venom and Toxin Research Programme, Departments of 1 Anatomy and 2 Surgery, Faculty of Medicine, National University of Singapore, Singapore The functional s ite o f Ôphospholipase A 2 inhibitor from pythonÕ (PIP) was predicted based on the hypothesis of proline brackets. Using dierent sources of secretory phospholipase A 2 (sPLA 2 s) as enzyme, and [ 3 H]arachido- nate-labelled Escherichia coli as substrate, short synthetic peptides representing the proposed site were e xamined f or their secretory phospholipase A 2 (sPLA 2 ) inhibitory activity. A decapeptide P-PB.III proved to be the most potent of t he tested peptides in inhibiting sPLA 2 enzymatic a ctivity in vitro, and exhibited striking anti-in¯ammatory eects in vivo in a mouse paw oedema model. P-PB.III inhibited the enzymatic activity of class I, II and III PLA 2 s, including that of human synovial ¯uid from arthritis patients. When tested by ELISA, biotinylated P-PB.III interacted positively w ith various PLA 2 s, suggesting that the speci®c region of PIP corresponding to P-PB.III, is likely to b e involved in t he PLA 2 ±PLI interaction. The eect of P-PB.III on the peri- toneal in¯ammatory response after surgical trauma in rats was also examined. P-PB.III eectively reduced the extent of postsurgical peritoneal adhesions as compared to controls. sPLA 2 levels at seventh postoperative day i n t he peritoneal tissue of P-PB.III-treated rats were also signi®cantly reduced (P<0.05) in comparison to those of the untreated controls. The p resent results s hed a dditional insight on the essential structural elements for PLA 2 binding, and may be useful as a basis f or the design of novel therapeutic agents. Keywords: anti-in¯ammatory peptide; phospholipase inhibitor from python PIP; protein±protein interaction; phospholipase A 2 inhibitors; p ostsurgical adhesions. Secretory phospholipases A 2 (sPLA 2 s) are enzymes (EC.3.1.1.4) that catalyse the hydrolysis of the sn-2 acyl bond of glycerophospholipids to produce free fatty acids and lysophospholipids [1], and are implicated in a range of diseases associated with in¯ammatory conditions such as arthritis, peritonitis, etc. [2±5]. Furthermore, PLA 2 inhibi- tors (PLIs) have recently become the s ubject of much interest due to the potential bene®ts t hey could offer in th e treatment of in¯ammation and cell injury. A number o f PLIs h ave been puri®ed and characterized from a v ariety of sources, including plant, fungi, and bacteria [6±8]. PLIs that interact with PLA 2 s a nd inhibit their enzymatic activity have been identi®ed in the sera of venomous snakes belonging t o Elapidae and Crotalidae families [9±20]. The discovery of speci®c sPLA 2 inhibitors has a lso been r eported i n the blood serum o f nonvenomous snakes [21,22]. These studies have demonstrated the presence of three d ifferent types of P LIs ( a, b and c)in the s era o f s nakes, which are believed to h ave a natural defensive role against endogenous snake venom sPLA 2 s. Our recent cloning and expression study has revealed that the PLI termed Ôphospholipase inhibitor from p ython (PIP)Õ possesses potent nonspecies speci®c antitoxic and anti- in¯ammatory activities, which have been linked to its ability to inhibit sPLA 2 [22]. This in hibitor signi®es struc tural homology with other c-type snake PLIs [12,14,18] and various mammalian proteins b elonging t o the Ôthree ®ngers Õ neurotoxin superfamily, including the urokinase-type plasminogen-activator receptor, membrane proteins of the Ly-6 family, and a bone-speci®c protein RoBo-1 [12,23]. On the basis of sequence homology study, some groups have been able to identify short peptides t hat act as a surrogate for the larger molecule [24], and their usefulness as potential anti-in¯ammatory agents have been reported [25]. Short peptides called anti¯ammins that are s ynthesized based on t he region of highest homology b etween utero- globin and lipocortin I, have previously been shown to inhibit PLA 2 [24,25], although there are some reports suggesting that t hese anti¯ammins a re devoid of PLA 2 inhibitory activity [26,27]. Recently, the importance of proline b rackets ¯anking protein ±protein interaction sites has been emphasized in i dentifying potential functional s ites in proteins [28]. Following this hypothesis, we were able to identify the active s ite on P IP that bi nds to s PLA 2 s potently in a nonspecies-speci®c manner. In the present study, a short oligopeptide, corresponding to the segment of the hypo- thetical interaction site has been synthesized and e xamined for i ts ant i-in¯ammatory activity and PLA 2 binding, w ith a Correspondence to P. Gopalakrishnakone, Department of Anatomy, Faculty of Medicine, 4 Medical Drive, National University of Singapore, Singapore 117597. Fax: + 65 7787643, E-mail: antgopal@nus.edu.sg Abbreviations:PLA 2 , phospholipase A 2 ; AIP, anti-in¯ammatory peptide; IC 50 , concentration of the inhibitor that inhibits PLA 2 activity by 50%; PIP, phospholipase inhibitor from python; PLI, phospholipase A 2 inhibitor; sPLA 2 , secretory phospholipase A 2 . Note: a web s ite is a vailable at h ttp://w ww.med.nus.edu.sg/ant/ anatomy.htm (Received 2 2 August 2 001, revised 3 0 October 2001, accepted 2 9 November 200 1) Eur. J. Biochem. 269, 719±727 (2002) Ó FEBS 2002 view to locate the particular region on PLIs that is responsible for binding to PLA 2 . EXPERIMENTAL PROCEDURES Materials All the venoms and PLA 2 toxins used in the experiments were available from the VTRP (Venom and Toxin Research Programme) collections, e xcept for the bee (Apis mellifera ) venom P LA 2, which w as purchased from Sigma Chemical Co. (St Louis, MO, USA). Anti-in¯ammatory peptide 2(anti¯amin2),o-phenylenediamine dihydrochloride ([2-(biotinamido)ethylamido]-3,3¢-dithiodipropionic acid N-hyd roxysuccinimide ester, and avidin-peroxidase conju- gate were purchased from Sigma. UniverSol ES liquid scintillation cocktail was from ICN Biomedicals, Inc., USA; Hylan GF 2 0 (Synvisc) gel was purchased from Bayer Pte. Ltd ( Singapore). All other reagents w ere of a nalytical grade or better. Animals Swiss albino mice (20±25 g) used for paw oedema assay and the Sprague±Dawley r ats ( 250±320 g ) used i n t he incisional hernia model were purchased from the Laboratory Animals Centre, Sembawang, Singapore, and housed in the Animal Holding Unit of the Department of Anatomy, National University of Singapore for 2 weeks to acclimatize the animals prior to use. Water and food (Glen Forrest Stockfeeders, WA, Australia) were provided ad libitum and a 12-h light/12-h dark cycle was maintained. The animals were h andled according to t he Guidelines of the National Medical Ethics Committee (Singapore), which conform to the World Health Organization's International Guiding Principles for Animal Research [29]. Peptide synthesis The p eptides w ith t he seq uences LSLQNGLY and PGLPLSLQNG, designated P-PB.II and PB.III, respec- tively, were custom-synthesized at the Biotechnology Pro- cessing Centre, National University of Singapore, by conventional solid phase techniques u sing automated ABI 4338 Peptide Synthesizer. The test peptide, designated P-PB.I with the sequence LPGLPLSLQNGLY, and the control peptide designated SP-PB.III, containing the same amino-acid composition as that of P-PB.III, but with the scrambled sequence, QLNPLP GLGS, were synthesized at the Fukuoka Women's University, Japan. All the s ynthetic peptides were puri®ed by RP-HPLC to more than 95% purity, with yields between 85 and 90%. The sequences were validated by MALDI-MS (Voyager-DESTR BioSpectrom- etry Workstation). SPLA 2 assay with [ 3 H]arachidonate-labelled E. coli Enzyme activity was assayed according to the described method [30] with minor modi®cations. B rie¯y, the reaction mixture contained 200 lL of assay buffer ( 100 m M Tris/ HCl pH 7.5, 25 m M CaCl 2 ), 20 lLof[ 3 H]arachidonate- labelled E. coli suspension (0.005 mCiámL )1 ;5.8lCiá lmo l )1 ,NEN)and30lL ( 10 ng) o f daboiatoxin, crotoxin subunit B , b-bungarotoxin, bee venom PLA 2 (Sigma, 1360 Uámg )1 ), or human synovial ¯uid, in a t otal volume of 250 lL. After incubation of the m ixture (37 °C, 1 h) and termination of the reaction with 750 lL of chilled NaCl/P i containing 1% BSA, the microfuge tubes c ontaining the samples w ere centrifuged (10 000 g,4°C) for 15 m in, and 500-lL aliquots of t he supernatant taken to measure t he amount of 3 H-labelled arachidonate released from the E. coli membrane using liquid scintillation counting (Mul- tipurpose Scintillation Counter LS 6500; Beckman). Appropriate controls without PLA 2 were also included in the assays. To determine the inhibitory activity, daboiatoxin or different source of PLA 2 s was preincubated for 1 h at 37 °C with each peptide at varying concentrations (1±250 l M ), before addition of the E. coli substrate suspen- sion. As controls for the inhibition assays, PLA 2 was preincubated with t he assay buffer. All samples, including the c ontrols, were perfo rmed in triplicate and plotted a s the percentage inhibition relative to negative controls. IC 50 determination IC 50 was determined by preincubating varying concentra- tions (1±250 l M ) of peptides in a constant volume, against a constant amount of enzyme as described earlier. A sigmoid dose±response curve was generated to allow calculation of the IC 50 values. All samples were performed in triplicate. Results were analyzed by nonlinear regression w ith G rap h- Pad PRISM (version 2.01) and expressed as t he percentage of inhibition relative to control values. Biotinylation of peptide Five-hundred micrograms of peptide P-PB.III (0.36 lmol) was d issolved in 1 mL o f 0.1 M NaHCO 3 pH 7.5, and t he biotinylation r eaction w as initiated by a ddition of 60 lL (1.08 lmol) of the biotin d isul®de N-hydroxysuccinimide ester solution to the peptide solution. The molar ratio of the peptidetobiotinusedinthereactionwas1:3.After incubation of the reaction mixture at 25 °C for 1 h, the reaction was stopped, and unreacted biotinylating agent was removed by dialyzing against 2-L volumes of NaCl/P i (three changes) at 4 °C using Spectra/Por6 membrane (molecular mass cutoff 1000; SPECTRUM Medical Industries, Inc.). To check the purity, the b iotinylated P-PB.III was injected onto a Vydac C 18 RP-HPLC column and eluted with a linear gradient of s olvent A ( 0.1% tri¯uoroacetic acid) a nd solvent B (100% acetonitrile/0.1% tri¯uoroacetic acid) at a ¯ow rate of 1 mLámin )1 . T he column eluate was monitored at 215 nm and 1 -min fractions were collected. In addition, HPLC-puri®ed b iotinylated P-PB.III was subjected to MS analysis. ELISA Wells of microtitre plates (Dynex Technologies, Inc., USA) were coated overnight at 4 °C with 100 lLofdifferent sources of e ither the venom (5 lgámL )1 )orPLA 2 (1 lgámL )1 ) i n 100 m M carbonate/bicarbonate buffer, pH 9.6. The c ontrols wells were coated with buffer only. All washing steps were carried out at leas t three times w ith NaCl/P i /Tween throughout. T he coated plates were washed, a nd unbound sites were s aturated by incubating 720 M M. Thwin et al. (Eur. J. Biochem. 269) Ó FEBS 2002 for 1 h at 37 °Cwith150lL of 3% fat-free milk powder (Bio-Rad) i n NaCl/P i /Tween. A ft er wa shing, the w ells w ere incubated with 100 lL of b iotinylated P-PB.III (1 lgámL )1 ) in NaCl/P i /Tween for 1 h at 37 °C, washed again and incubated f urther for 1 h a t 37 °C with 100 lL o f Avidin- peroxidase conjugate ( Af®nity puri®ed, S igma) at a dilution of 1 : 2000 in NaCl/P i /Tween. After washing, 100 lLof substrate solution (0.5 g áL )1 o-phenylenediamine di-HCl/ 0.02% H 2 O 2 ;Sigma)wasaddedtoeachwellandthe enzymatic r eaction stopped by adding 50 lLof2 M H 2 SO 4 prior t o reading the absorbance a t 490 nm (Emax P recision Microplate Reader, Molecular Devices). Effect of active peptide on PLA 2 -induced mouse paw oedema The o edema produced by the crude venom or puri®ed PLA 2 sfromDaboia r usselli siamensis venom or bee venom, was a ssayed according to the m ethod described [31]. Male Swiss a lbino mice ( 20±25 g) in g roups of four were injected subcutaneously into the footpad of the left hind paw with the indicated a mounts of venom or PLA 2 s(5lg venom; 1 lg d aboiatoxin or bee venom PLA 2 ) in a total volume o f 25 lL of sterile NaCl/P i . At 45 min thereafter, the mice were euthanized using C O 2 insuffulation, and both h ind limbs disarticulated at the a nkle joint were individually w eighed. The increase i n weight due to oedema was c alculated by subtracting t he weight of each nontreated right hind limb. To study the effect on PLA 2 -induced paw o edema, venom (5 lg) or PLA 2 s(1lg) were preincubated with varying concentrations of the i nhibitors (PIP, 0 .5, 1 nmol; P-PB.III, 50,100nmol;AIP-2,92nmol),inatotalvolumeof25lL prior to injection. Inhibitory effects were assessed by comparing the paw o edema o f inhibitor-treated groups to that of nontreated groups. Inhibitors alone or NaCl/P i alone were injected as controls. Effect of active peptide on postsurgical peritoneal adhesions An in vivo incisional hernia model [ 32] was used to assess the potential therapeutic application of the active peptide P-PB.III in reducing the formation of postsurgical perito- neal adhesions in male Sprague±Dawley rats (250±320 g ). Under light ether anesthesia and by means of a midline laparotomy incision, a ventral abdominal defect (15 ´ 25 mm) was created in each of the 30 r ats, which were divided into four g roups. T he caecum was located, externalized and the serosal surface abraded, using dry gauze until subserosal punctate hemorrhage was seen. A polypropylene mesh (Surgipromesh, Autosuture Co.) was then sutured to the abdominal defect. Prior to closure of the abdominal skin, a hyaluronate-based gel (Hylan GF 20), e ither a lone or with an anti-in¯ammatory peptide, was administered intraperitoneally over the abraded cae- cum. Group I (n  12) contained only the mesh to serve as a c ontrol; group II (n  6) contained exclusively the gel, while groups III (n  6) and IV (n  6) contained the gel spiked with 0.16 lmol each of the anti-in¯amma- tory peptides, P -PB.III and AIP-2, r espectively. On post- operative day 7, a re-laparotomy was performed and peritoneal adhesions were graded using a method previ- ously described [33]. Peritoneal tissue sPLA 2 activity The peritoneal tissue s pecimens collected from each rat at day 0 and o n p ostoperative day 7 were stored immediately at )80 °C until the time of analysis. Approximately 150± 250 mg (wet weight) o f the peritoneal tissues were weighed and homogenized in 2 mL of N aCl/P i using Heidolph DIAX900 homogeniser (Germany). Supernatant collected after centrifugation (20 000 g)at4°C for 20 min w as used for measurement of total protein [34] and PLA 2 activity [30]. For each sample, the mean and standard deviations were obtained for replicates (n  3). Statistical analysis The results from the paw oedema experiment in mice were analyzed by a one-tailed Student's t-test for groups of unpaired observations. S igni®cance was t ak en at P < 0.05. The statistical signi®cance of the effects of the peptides was also con®rmed by one-way ANOVA . Wilcoxon rank sum test was u sed for a nalyzing differ- ences in peritoneal tissue P LA 2 activity at two different time points, day 0 (at the time of surgery) and day 7 (after surgical trauma). The signi®cance of the difference in the postoperative peritoneal tissue PLA 2 activity at day 7 between the P-PB.III-treated and untreated groups were analyzed by nonparametric Mann±Whitney U-test. A P value less than 0.05 was considered statistically signi®cant. RESULTS PIP has signi®cant amino-acid sequence homology with other snake PLIs The nonredund ant BLASTP alignment o f t he amino-acid sequence o f a mature PIP m onomer with the database sequences whose match satis ®es the p reset E value of 0.001 is shown in F ig. 1 . The mature PIP protein contains 16 cysteine residues all of which align p erfec tly in the d atabase matched sequences. It has the highest sequence i dentity (57± 61%) to the mature PLIs from the sera of Crotalidae snakes, Agkistrodon blomhoi s initicus [14], Crotalus durissus ter- ri®cus [11,13], and Trimere surus ¯avoviridis (Protobothrops ¯avoviridis) [9,15], w ith s equence identities o f 6 1, 60 and 57%, r espectively. PIP a lso has a signi®cant ( 57%) homo- logy to the sequences of mature PLIs of a nonvenomous snake Elaph e quadrivirgata [21], and also to those of t he PLIs from the sera o f Australian E lapidaes, Notechis ater, Notechis scutatus,andOxyuranus scutellatus [19], with sequence identities in the vicinity of 56%. The potential interaction site on PIP is predicted by searching for proline residues that mark the ¯anks of protein±protein interaction sites The amino-acid sequence of PIP (Fig. 1) shows four proline residues at posit ions 85, 87, 90 and 100. As the residues at position 8 5 and 100, and 90 and 100, respectively, served as the ¯anking prolines enclosing a small segment of the PIP in each case, we p redicted that the s egments, LPGLPLSLQN GLY (P-PB.I) and/or LSLQNGLY (P-PB.II), m ight indi- cate possible interaction site for PIP with sPLA 2 .Boththese peptides displayed in vitr o PLA 2 inhibitory activity, but Ó FEBS 2002 PLA 2 interaction site of the python serum inhibitor (Eur. J. Biochem. 269) 721 P-PB.I was the only peptide that was found to possess remarka ble in vivo anti-in¯ammatory activity, while P-PB.II was less active. With P-PB.I being more active than P-PB.II, it was a ssumed t hat the bioactivity might be related m ainly to that particular segment of PIP. The third peptide P-PB.III with the sequence PGLPLSLQNG, which rep re- sents the shorter segment of the p roposed site, exh ibited the strongest anti-PLA 2 and an ti-in¯ammatory activities, while the scrambled peptide SP-PB.III was found to be totally devoid of PLA 2 -inhibitory activity (Table 1). Synthetic peptides derived from the hypothetical interaction site inhibit PLA 2 enzyme activity and bind to sPLA 2 s The dose±response relationships for t he synthetic peptides and the full-length recombinant P IP were determined and are s hown in Fig. 2. The 13-residue peptide P-PB.I, which corresponds to PIP residues 86±98, is a strong inhibitor against the PLA 2 activity of daboiatoxin (IC 50 37.82  2.40 l M ), while the o ctapeptide P-PB.II, is less potent (IC 50 45.09  1.14 l M ). Among the three synthetic peptides examined for PLA 2 inhibitory activity, the deca- peptide P-PB.III, corresponding to PIP residues 87±96, is the strongest inhibitor that possesses PLA 2 inhibitory potency (IC 50 22.65  2.91 l M ) equivalent to that of the r ecombinant inhibitor P IP (IC 50 19.51  2.06 l M ). P-PB.III dose-dependently inh ibits the e nzyme a ctivity o f a variety of sPLA 2 sources from snakes, b ee and human, over a wide concentration range (1±250 l M ), while the scrambled peptide SP-PB.III, fails to inhibit sPLA 2 at any concentra- tion tested. Biotinylation of the active peptide, P-PB.III does not seem to result in considerable loss of inhibitory potency as judged by similar IC 50 values obtained for the native P-PB.III (IC 50  22.6  2.9 l M ) and its biotinylated product (IC 50  25.8  3.1 l M ) in the binding assays. The experimental evidence of the fact that P-PB.III interacts with s PLA 2 is demonstrated by EL ISA and shown in Fig. 3. The purity of the biotinylated P-PB.III as evaluated by RP-HPLC was 95%, and the determined Table 1. Amino-acid sequences and pro perties of pe ptides d erived from the predicted site. Test peptides P-PB-I, II, III a nd the control scrambled peptide S-PB.III were synthesized by solid phase techniques. Exper imental details are described in the Experimental p rocedures. PLA 2 inhibition indicates m aximal e nz yme i nhib ition t owards daboiatoxin seen at a ®xed peptide concentration (100 l M ). IC 50 values were c alculated f rom t he corresponding dose ±respon se curves shown in Fig. 2, by n onlinear re gre ssion an alysis with GraphPad PRISM (version 2.01). An ti-in¯am matory activity was a ssessed by d ab oiatoxin-induced mouse paw o edema experiments. Values re ported are the mean of triplicate e xperimen ts. Code no. Sequence Length M r PLA 2 inhibition (%) IC 50 (l M ) Anti- in¯ammatory activity P-PB.I LPGLPLSLQNGLY 13 1385 70.71 37.82 (+) moderate P-PB.II LSLQNGLY 8 1018 51.69 45.09 (±) negative P-PB.III PGLPLSLQNG 10 995 91.60 22.60 (+ +) strong SP-PB.III GLNPLPGLQS 10 995 0 ± Not tested Fig. 1. Alignment of t he mature PIP monomer with the database sequences. Th e E value was preset at 0.001 f or matching the amino-acid sequences. The shaded boxes indicate residues identical to those of PIP. (1) Python reticulatus PIP; (2) Agkistrodon b lomhoi siniticus PL Ic; (3) Cro talus d. te rri®cus CNF; (4) Proto- bothrops ¯avoviridis PLIc; (5) Elaphe quadri- virgata PL Ic;(6)Notechis a te r a subunit isoform NAI-3 A; (7) Note chis scutatus a chain iii; (8) Oxyuranus scutellatus a subunit isoform OSI-1 A. 722 M M. Thwin et al. (Eur. J. Biochem. 269) Ó FEBS 2002 mass was 1118 Da. Based on the mass (998 Da) of the intact peptide and that of the biotinylated P -PB.III, 0.12 mol biotin was apparently bound to 1 m ol of P-PB.III. Most veno ms and PLA 2 s examined, reacted positively with the biotinylated P-PB.III, although the results vary depending upon the t ype of PLA 2 used. P -PB.III i nteracts very strongly with group I P LA 2 toxin, b-bungarotoxin, but binds moderately to group II PLA 2 toxins like daboiatoxin, m ojave toxin subunit B , ammodytoxin A and crotoxin. It gives s trong positive ELISA reaction with the enzymatically active basic s ubunit of crotoxin while its binding to the non-PLA 2 acidic subunit of crotoxin is negligible. I nterestingly, t he biotinylated peptide P-PB.III also reacted strongly with the human synovial ¯uid collected from arthritic patients. The synthetic peptide corresponding to the active site has marked anti-in¯ammatory activity The anti-in¯ammatory effects o f P-PB.III, in comparison to those of the full-length recombinant PIP and the anti- in¯ammatory peptide (anti¯ammin 2) is reported in Table 2. Co-injection of P-PB.III, either wi th the venom, t oxic PLA 2 (daboiatoxin), or the bee venom PLA 2 into the mouse footpad signi®cantly (P < 0.01) inhibits the formation of in¯ammatory oedema over two different dose ranges (50, 100 nmol), with a higher s uppression of the in¯ammatory response seen at a higher dose. In contrast, AIP-2, is less potent t hen P-PB.III. Comparison of the dose±responses of the recombinant PIP (0.5, 1 nmol) and P-PB.III (50, 100 nmol) by one-way ANOVA shows that there is no signi®cant difference (P < 0.0 5) between the two forms of inhibitor, thus providing evidence t hat t he peptide P-PB.III retains much of the anti-in¯ammatory property of the intact parent PIP molecule. Although P-PB.III (100 lg) is as potent a s P IP (100 lg) on basis of mass, it is much less potent ( 100 f old) on a molar basis. Intraperitoneal administration of P-PB.III reduces peritoneal tissue PLA 2 activity and modulates peritoneal in¯ammatory response after surgical trauma With the aim of investigating the potential therapeutic application of P -PB.III, the e ffect of the peptide in reducing peritoneal in¯ammatory response was studied in an in vivo Fig. 3. Binding of P-PB.III to various sources of sPLA 2 in ELISA. Biotinylated P-PB.III was directed against dierent sources o f sPLA 2 or crude venom coated o n microtitre plate wells. Bound b iotinylated peptide in each w ell was detected with avidin±perox idase conjugate and color development with a substrate solution as d e scribed in the Experimental procedures. All samples were measured in triplicates and the mean signals (A 490 ) are shown over each w ell a rea. Rows (A±C), from left to right: b-bungarotoxin, mulgatoxin, taipoxin, crotoxin, crotoxin B, crotoxin A, ammodytoxin A, daboiatoxin, mojave toxin B, bovine pancreatic PLA 2 , human synovial ¯uid, blank. Rows (E±G), from left to right: venoms of N. siam ensis, P. au stralis, O. hannah, N. kaouthia, B. multicinctus, E. c arinat us, C. rhodostoma, D. siamensis (Myanmar), D. russelli (In dia), D. pulchella (Sri Lanka), D. siamensis (Thailand), blank. Fig. 2. Phospholipase A 2 inhibition curves for PIP and various synthetic peptides. (A) Inhibition pro®les against daboiatoxin PLA 2 activity: PIP (.); PI P -derived test p eptides, P-PB.III (s), P-PB.I ( h), P-PB.II (n); control s cram bled pe ptid e, SP-PB.III (d); Biotinylated P-PB.III (j). (B) Inhibition pro®les of the active peptide P-PB.III against enzymatic activity of various sources of sPLA 2 s±b-bu ngarotoxin (n), crotoxin B (.), bee venom PLA 2 (j), hum an synovial ¯u id (d). Results a re the mean  S D. IC 50 values were graphically determined from the inhi- bition curves, constructed on the basis of the in vitro results of 3 H-labelled E. coli m embrane assays. Ó FEBS 2002 PLA 2 interaction site of the python serum inhibitor (Eur. J. Biochem. 269) 723 incisional h ernia m odel in rats. Most animals in the control group (group I) developed d ense adhesions, while in the experimental groups (groups II±IV), relatively fewer post- surgical peritoneal adhesions were seen. Intraperitoneal administration of P-PB.III along with the gel to the site of peritoneal injury signi®cantly reduced the peritoneal in¯am- matory response with fewer postsurgical adhesions (P < 0.05), whereas either the gel alone or the g el w ith AIP-2, was f ound to be relatively less potent i n reducing the postsurgical peritoneal adhesions. Table 3 d epicts adhesion grades in individual rats as analyzed by an independent observer w ho was blinded about the treatment and nontreatment groups. At day 7 following s urgical trauma, the PLA 2 activity of the peritoneal tissue extracts of control rats markedly increased ( P  0.028) o ver t he basal levels f ound at day 0 (Fig. 4 A). In contrast, no signi®cant difference in the peritoneal PLA 2 activity (P > 0.05) was found between those two levels (day 0 vs. day 7) in the P-PB.III-treated rats (Fig. 4 B). Moreover, when the p eritoneal tissue P LA 2 levels of P-PB.III-treated (Fig. 4B) and untreated (Fig. 4 A) rats at day 7 following surgical trauma were compared, there was a highly signi®cant difference ( P  0.025) observed between the c ontrols and the inhibitor-treated animals (Fig. 4 A v s. 4B). These results suggest that the active peptide P-PB.III can afford an effective in vivo inhibition of total catalytic PLA 2 activity which is apparently increased as a result of trauma after surgery. DISCUSSION Identi®cation o f a protein±protein interaction site is an important step that has signi®cant potential to clarify structure±function relationships of protein and drug designs. Through a survey of a database of p rotein±protein Table 2. Anti-in¯ammatory eect of inhibitors on PLA 2 -induced mouse paw oedema. Experimental details are described in the Experimental procedures. Inhibitory eects were expressed as percentage inhibition of paw oedema, and were assessed by comparing the paw oedema (increase of wt. in m g) of mice receiving (PLA 2 + inhibitor) to those receiving PLA 2 alone. The results (mean  SD; n  4) were analyzed by a o ne-tailed Student's t-test for groups of unpaired observations (signi®cance taken at minimum of P < 0.05). PIP, phosp holipa se inhibitor from python; P-PB.III, active peptide; AIP-2, anti-in ¯ammatory peptide-2 from Sigma. Treatment nmol (lg) Oedema (mg) % Inhibition D.r. siamensis venom ± (5) 117  20 ± + PIP 1 (100) 29  1 (P < 0.01) 74.7  0.8 + P-PB.III 100 (100) 33  6 (P < 0.01) 71.1  5.5 + AIP-2 92 (100) 76  11 (P < 0.05) 35.0  9.7 Daboiatoxin PLA 2 (1) 166  9 ± + PIP 0.5 (50) 18  5 (P < 0.01) 89.2  3.0 + PIP 1.0 (100) 13  3 (P < 0.01) 92.2  1.8 + P-PB.III 50 (50) 63  7 (P < 0.01) 62.1  4.0 + P-PB.III 100 (100) 35  6 (P < 0.01) 79.0  3.6 + AIP-2 92 (100) 108  11 (P < 0.01) 35.3  6.5 Bee venom PLA 2 (1) 89  6 ± + PIP 0.5 (50) 52  3 (P < 0.01) 39.9  3.8 + PIP 1.0 (100) 19  2 (P < 0.01) 78.1  2.1 + P-PB.III 50 (50) 42  4 (P < 0.01) 51.7  4.2 + P-PB.III 100 (100) 31  4 (P < 0.01) 63.8  4.8 + AIP-2 92 (100) 60  7 (P < 0.01) 33.6  3.9 PIP alone 1.0 (100) 9  8 ± P-PB.III alone 100 (100) 13  2 ± AIP-2 alone 92 (100) 9  4 ± Table 3. Eect of anti-in¯ammatory peptides on peritoneal in¯amma- tory response i n individual ra ts after surgical t rauma. Experimental details are described in the Experimental p rocedures. Values reported are the means  SD, where n  6±12 rats. One-tailed Student's t-test for groups of u npaired observations was done with signi®cance tested at P < 0 .05: a vs. b, not signi®cant ( P >0.05);avs.c,signi®cant (P < 0.05); a vs. d, not signi®cant (P > 0.05). The eects of P-PB.III and AIP-2 we re con®rmed by one-way ANOVA. Group no. Rat no. Adhesion score Grade Mean  SD I (control) 255±266 4 4.0  0 a (n  12) II (with gel only) 267 4 3.16  2.82 b (n  6) 268 4 271 4 270 3 269 2 272 2 III (with gel + P-PB.III) 275 1 2.00  0.82 c (n  6) 276 1 278 2 273 2 274 3 277 3 IV (with gel + AIP-2) 279 4 3.30  1.07 d (n  6) 284 4 281 4 280 3 283 3 282 2 724 M M. Thwin et al. (Eur. J. Biochem. 269) Ó FEBS 2002 interaction sites, a unique prediction method to identify those s ites has previously been proposed, based on the observation t hat proline is the m ost common r esidue found in the ¯anking segments of interaction sites [28]. In the present study, we have r ecognized a p roline-rich cluster corresponding to residues 85±100 of PIP and other database sequences in the a lignment. Because this proline-rich segment i s highly c onserved amongst members of t he snake serum P LI family, it is a distinguishing feature, and is therefore believed t o contribute t o t he biological activity speci®cally associated with the snake PLI family. Hence, using the proline bracket method for predicting interaction sites, we have been able to identify the functional site of PIP belonging to the three ®ngers neurotoxin superfamily. The present ®ndings p rovide evidence that the mode of interaction between the PLI and the PLA 2 occurs via a common sequence motif represented by t he peptide P-PB.III. This decapeptide displays a diverse inhibitory pro®le against the enzymatic activity of all types of PLA 2 s e xamined, including that of human secr etory PLA 2 present in the synovial ¯uid o f subjects suffering from arthritis. Using a monoclonal antibody speci®c against human synovial s PLA 2 (Calbiochem, USA), w e f ound that sPLA 2 activity d etected in the synovial ¯uid was inhibited (data not shown), thus con®rming t hat t he enzyme contained in the synovial ¯uid was in fact, a human group II sPLA 2 . To e nsure that t he inhibition displayed by the active peptide against sPLA 2 s was speci®c and not artefac- tual, a dose±response experiment was performed with the peptide P-PB.III, as well as with the control peptide S-P- PB.III, that had scrambled sequence. The peptide P-PB.III inhibited m ost types of sPLA 2 s e xamined, including human synovial s PLA 2 , while the scrambled peptide was nonin- hibitory, con®rming that the i nhibition was not nonspeci®c. The active peptide also bind s t o different sources of PLA 2 s tested in ELISA. Whatever t he species of sPLA 2 origin, t he wide sp ectrum o f b inding to sPLA 2 s a nd inhibitio n o f the enzyme activity displayed by the active peptide, coupled with the striking anti-in¯ammatory effects it possessed, outlines the potential therapeutic usefulness of t his inhibitor as an anti-in¯ammatory agent. The domain of sPLA 2 or PLI involved in inhibitor binding has yet to be fully elucidated although some structural information suggests that the three-®nger motifs of PLIs are important for interaction between c-type i nhibitors and PLA 2 s [35,36]. The b road spectrum of inhibition seen with the P IP-derived peptide in this study suggests that like P IP and o ther c-type inhibitors, it could p robably recognize the Ca 2 + -binding loop, which i s a common structural e lement conserved among all groups of secretory PLA 2 s, including human synovial sPLA 2 [37]. P revious data on epitope mapping and studies with synthetic peptides a lso suggest that the conserved core region of PLA 2 including most of the Ca 2 + -binding loop may be a potential target for developing selective inhibitors of sPLA 2 s[38].Basedon ELISA results, it appears t hat P -PB.III binds directly to sPLA 2 , perhaps thr ough t he residues on the Ca 2 + - binding loop. However, it is highly unlikely that its binding to sPLA 2 could involve nonspeci®c electrostatic interaction, as no charged amino-acid residues, other than the polar and nonpolar residues, are present in the sequence of P -PB.III. The present results show t he oedema-reducing a ctivity of the active p eptide, which appears to a ct via inhibition of PLA 2 activity, and con®rms the decapeptide P-PB.III as a poten t anti-in¯ ammatory peptide t hat h as poten tial therapeutic applications, e specially for P LA 2 -related in¯ammatory conditions. The in vivo postsurgical peritoneal adhesion model in rats indicates that the intraperitoneal administration of the peptide directly to the site of peritoneal injury can reduce t he formation of postsurgical adhesions by a mechanism that could involve inhibition of the activation of endogenous sPLA 2 [2] and through reduction in the peritoneal in¯ammatory response that occurs after surgery. T hese results strongly support that t he predicted region i ndeed plays an important role in the interaction between sPLA 2 and the endogenous PLIs of snakes. At p resent, a crystallographic s tudy is in progress to understand the structural details of PLA 2 ±PLI interaction. ACKNOWLEDGEMENTS This work was supported by t he Research Grant (R-181-000-025-112) from the National University of Singapore. We are very g rate ful to Professor Shamal Das De, Department of O rthopaedic Surgery, National University of Singapore, Republic of Singapore, for providing synovial ¯uid specimens, a nd also to Professor K azu ki Sato, Fukuoka Women's U niversity, Kasumigaoka, H igashi-ku, Fu kuoka, 813±8529, Japan, for the peptides (P-PB.I and S-P-PB.III) used i n our study. REFERENCES 1. Balsinde, J., Balboa, M.A., Insel, P.A. & Dennis, E.A. (1999) Regulation and inhibition of phospholipase A 2 . Annu. Rev. Pharmacol. Toxicol. 39, 175±189. 2. Yedgar, S., Licht enberg, D. & Schnitzer, E. (2000) Inh ibition of phospholipase A 2 as a therapeutic target. Biochem. Biophys. Acta. 1488, 1 82±187. Fig. 4. Box plots of the PLA 2 activity in extracts of rat peritoneal tissue obtained at the time of surgery (basal) and at day 7 following surgical trauma. (A) Control group (N  6) without P -PB.III treatment; ( B) Experimental group (N  6) with P-PB.III treatment. Data are presented a s median (solid li ne across the b ox), 25th and 75th per- centiles. Signi®cant d ierences a re indicated b y the P values. Outliers are represented by circles; open and hatched boxes represent basal and day 7 samples, respectively. Ó FEBS 2002 PLA 2 interaction site of the python serum inhibitor (Eur. J. Biochem. 269) 725 3. Vadas, P. & Pruzanski, W . (1986) Biology of d isease: role of secretory pho spho lipase A 2 in the p athology of disease. Lab. Invest . 55, 3 91±404. 4. Murakami, M ., Nakatani, Y ., Atsumi, G., Inoue, K. & Kudo, I. (1997) Regulatory func tions of phospholipase A 2 . Crit. Rev. Immunol. 17, 2 25±283. 5. Vadas, P., Browning, J., Edelson, J. & Pruzanski, W. (1993) Extra- cellular p hosp holipase A 2 expression and in¯ammation: t he rela- tionship with associated disease states. J. Lipid Med. 8, 1±30 . 6. Cuella, M.J., Giner, R.M., Recio, M.C., Jus t, M.J., Manez, S. & Rios, J. (1996) L) Two fungal lanostane de rivatives a s phosphol- ipase A 2 inhibitors. J. Nat. Prod. 59 , 977±979. 7. Matsumoto, K., Tanaka, K., Matsutani, S., Sakazaki, R., Hinoo, H., Uotani, N., Tanimoto, T., Kawamura, Y., Nakamoto, S. & Yoshida, T. (199 5) Is olation a nd biol ogical ac tivity of thielocins: novel phospholipase A 2 inhibitors produced by Thielavia terricola RF- 1 43. J. Antibiotics 48, 1 06±112. 8. Lindahl, M. & Tagesson, C. (1997) Flavonoids as phospho- lipase A 2 inhibitors: impo rtance o f t heir structures for selective inhibition of grou p II pho spho lipase A 2 . In¯ammation 21 ,347± 356. 9. Kogaki, H. , Inoue, S., Ikeda, K., S amejima, Y., Omori-Satoh, T. & Hamaguchi, K. (1989) Isolation and fundamental properties of a phospholipase A 2 inhibitor from the blood plasma of Trimeresurus ¯avoviridis. J. Biochem. (Tokyo) 106 , 966 ±971. 10. Inoue, S., Kogaki, H., Ikeda, K., Samejima, Y. & Omori-Satoh, T. (1991) Aminoacid s equences of the two subunits of a phospholi- pase A 2 inhibitor from the bloodplasma of Trimeresurus ¯a vovir- idis. J. Biol. Chem. 266, 1001±1007. 11. Fortes-Dias, C.L., Lin, Y ., Ewell, J., Diniz, C.R. & Liu, T Y. (1994) A phospholipase A 2 inhibitor from the plasma of the South American rattlesnake (Crotalus d urissus terri®cus). J. Biol. Chem. 269, 15646±15651. 12. Ohkura, N., Inoue, S ., Ikeda, K. & Hayashi, K . ( 1994) The two subunits of the phospholipase A 2 inhibitor from the plasma of Thailand cobra having structural similarity to urokinase-type plasminogen activator receptor and Ly-6 related proteins. Biophys. Biochem. Re s. Commun. 204, 1212±1218. 13. Perales, J., Villela, C ., Domont, G .B., Choumet, V., Saliou, B., Moussatche, H., Bon, C. & Faure, G . (1995) Molecular structure and mechanism of action of the c rotoxin inhibitor fro m Crotalus durissus terri ®cus serum. Eur. J. Biochem. 227 , 19±26. 14. Ohkura, N., Okuhara, H ., I noue, S., Ikeda, K. & H ayashi, K . (1997) Puri®cation and characterization of three distinct types of phospholipase A 2 inhibitors from the blood plasma of the Chinese mamushi, Agkistrodon b lomhoi sinictus. Biochem. J. 325,527± 531. 15. Nabuhisa, I., Inamasu, S., Nakai, M., Tatsui, A., Mimori, T., Ogawa, T., Shimohigashi, Y., Fukumaki, Y., Hattori, S., Kihara, H. & Ohno, M. (1997) Characterization and evolution of the gene encoding a Trim eresu rus ¯ avoviridis serum protein that inhibits basic p hospholipase A 2 isozymes in the sn ake's venom. Eur. J. Biochem. 249, 838±845. 16. Lizano, S., Lomonte, B., F ox, J.W. & Gutierrez, J.M. (1997) Biochemical characterization and pharmacological properties of a phospholipase A 2 myotoxin inhibitor f rom the p lasma of the snake Bothrops asper. Biochem. J. 326, 853 ±859. 17. Okumura, K., Ohkura, N. , Inoue, S., Ikeda, K. & Hayashi, K. (1998) A novel phospholipase A 2 inhibitor with leucine-rich repeats from t he blood p lasma of Agkis trodon blomhoi s inictus. Sequence homologies with hu man l eucine-rich A 2 ±glycoprotein. J. Biol. Chem. 273 , 19469±19475. 18. Ohkura, N., Kitahara, Y., Ino ue, S., Ikeda, K. & Hayashi, K. (1999) Is olation and amino acid s equencing o f the phospholipase A 2 inhibitor from the blood plasma of t he sea krait, Laticauda semifasciata . J. Bioc hem. (Tokyo) 125, 375±382. 19. Hains, P.G., Sung, K L., Tseng, A. & B roady, K.W. (20 00) Functional characteristics of a phospholipase A 2 inhibitor from Notechis ater serum. J. Biol. Chem. 275 , 983±991. 20. Hains, P.G . & Broady, K.W. (2000) Puri®cation and inhibitory pro®le of phospholipase A 2 inhibitors from Australian elapid sera. Biochem. J . 346, 139±146. 21. Okumura, K., Masui, K., Inoue, S., Ikeda, K. & Hayashi, K. (1999) Puri®cation,characterization and cDNA cloning of a phospholipase A 2 inhibitor from the seru m of the non-venomous snake Elaphe quadrivirgata. Biochem. J. 341, 1 65±171. 22. Thwin, M M., Gopalakrishnakone, P., Kini, R.M., Armugam, A. & Jeyaseelan, K. (2000) Recombinant antitoxic and anti- in¯ammatory factor fro m the no nvenomou s snake Py thon retic- ulatus. phospholipase A 2 inhibition and venom ne utralizing potential. Biochemistry 39 , 9604±9611. 23. Noel, L.S., Champion, B.R., Holley, C.L., Simmons, C.J., Morris, D.C., P ayne, J .A., Lean, J.M., Chamber, T.J., Zaman, G., Lanyon, L.E., Suva, L.J. & Miller, L. R. (1998) RoBo-1, a n ovel member of the u rokinase plasm ino gen a ctivator rece ptor/CD59/ Ly-6/snake toxin family selectively expressed in rat bone a nd growth plate c ar tilage. J. Biol. C he m. 273, 3 878±3883. 24. Miele, L., Cordella-Miele, E., Facchiano, A. & Mukherjee, A.B. (1988) Nove l anti-in¯ammatory p eptides from t he region of highest similarity between uteroglobin and lipocortin I. Nature 335, 726±730. 25.Rodgers,E.E.,Girgis,W.,Campeau,J.D.&diZerega,G.S. (1997) Reduction of adhesion formation by intraperitoneal administration of anti-in¯ammatory peptide 2. J. Invest. Surg. 10, 31±36. 26. Marastoni, M., Scaranari, V., Romualdi, P., Donatini, A ., Fer ri, S. & Tomatis, R. (1993) Studies on the anti-phospholipase A 2 and anti-in¯ammatory activities of a uteroglobin fragment and related peptides. Arzneim. Fors ch/Drug Res. 43, 997±1000. 27. Hope, W.C., Patel, B.J. & Bolin, D.R. ( 1991) Anti¯amin-2 (HDMNKVLDL). does not inhibit p hospholipase A 2 activities. Agents Actions 34, 7 7±80. 28. Kini,R.M.,Caldwell,R.A.,Wu,Q.Y.,Baumgarten,C.M.,Feher, J.J. & Evans, H.J. ( 1998) Flanking proline r esidue s identify t he 1-type CA 2 + channel bindingsite of calciseptine and FS2. Bio- chemistry 37, 9058±9063. 29. Council fo r International Organizations of M edical Sciences (CIOMS) ( 1985) International Guiding Principles for Bio- medical Research involving Animals. The WHO Chronicle 39, 51±56. 30. Elsbach, P. & Weiss, J. (1991) Utilization of labeled Escherichia coli as phospholipase substrate . I n Methods i n Enzymology. Phospholipase s (Dennis, E.A., ed.), Vol. 197, pp. 24±31. Academic Press Inc., San Diego, C A. 31. Yamakawa, M., Nozaki, M. & Hokama, Z. (1976) Fractionation of Sakishmma-habu (Trimeresurus elegans) venom and lethal hemorrhage and oedema f orming activity of the fractions. In Animal, Plant a nd M icrobial Toxins (Ohsaka, A., Hayashi, K. & Sawai, Y., eds), Vol. 1, pp. 97±103. Ple num Press, N ew York. 32. diZerega, G.S. & Rodgers, K.E. (1992) Intraperitoneal adhesions. In The p eritoneum, pp. 274±306. Springer, New York. 33. Alponat, A., Lakshminarasappa, S.R., T eh, M., Rajnako va, A., Moochhala, S., Goh, P.M. & Chan, S.T. (1997) Eects of physical barriers in pre ventio n of adhesions: an incisional hernia m ode l in rats. J. Su r g. Res. 68 , 126±132. 34. Bradford, M.M. ( 1976) A rapid and se nsitive m ethod for quant- itation of microgram quantities of protein utilizing the principle of protein dye binding. Anal . Biochem. 72, 248±254. 35. Nabuhisa, I., Chiwata, T., Fukumaki, Y., H attori, S., Shimohigashi, Y. & Ohno, M. (1998) Structural e lements o f Trimeresurus ¯avo- viridis serum inhibitors for recognition of its venom phospholipase A 2 isozymes. FEBS Lett. 429, 385±389. 726 M M. Thwin et al. (Eur. J. Biochem. 269) Ó FEBS 2002 36. Faure, G. ( 2000) Natural inhibitors o f t oxic phospholipase A 2 . Biochimie 82, 833±840. 37. Inoue, S., Shimada, A., Oh ku ra, N., Ikeda, K ., Samejim a, Y., Omori-Satoh, T. & Hayashi, K. ( 1997) Speci®city of two types of phospholipase A 2 inhibitors from the plasma of venomous snakes. Biochem. Mol. Biol. Int. 41, 529 ±537. 38. Cordella-Miele, E., Miele, L. & Mukherjee, A.B. (1993) Identi®- cation o f a s peci®c region o f low molecular weight phospholipase A 2 (residues 21±40) as a potential target for structure-based design of inhibitors of these enzymes. Proc. Natl Acad. Sci. USA 90, 10290±10294. Ó FEBS 2002 PLA 2 interaction site of the python serum inhibitor (Eur. J. Biochem. 269) 727 . Functional site of endogenous phospholipase A 2 inhibitor from python serum Phospholipase A 2 binding and anti-in¯ammatory activity Maung-Maung. University of Singapore, Singapore The functional s ite o f phospholipase A 2 inhibitor from python (PIP) was predicted based on the hypothesis of proline