Functional Characteristics of a Phospholipase A2Inhibitor from Notechis ater Serum*

A phospholipase A2 inhibitor has been purified from the serum of Notechis ater using DEAE-Sephacel chromatography. The inhibitor was found to be composed of two protein subunits (α and β) that form the intact complex of approximately 110 kDa. The α-chain is a 30-kDa glycoprotein and the β-chain a nonglycosylated, 25-kDa protein. N-terminal sequence analysis reveals a high level of homology to other snake phospholipase A2 inhibitors. The inhibitor was shown to be extremely pH and temperature stable. The inhibitor was tested against a wide variety of phospholipase A2 enzymes and inhibited the enzymatic activity of all phospholipase A2 enzymes tested, binding with micromole to nanomole affinity. Furthermore, the inhibitor was compared with the Eli-Lilly compound LY311727 and found to have a higher affinity for human secretory nonpancreatic phospholipase A2 than this chemical inhibitor. The role of the carbohydrate moiety was investigated and found not to affect thein vitro function of the inhibitor.

A phospholipase A 2 inhibitor has been purified from the serum of Notechis ater using DEAE-Sephacel chromatography. The inhibitor was found to be composed of two protein subunits (␣ and ␤) that form the intact complex of approximately 110 kDa. The ␣-chain is a 30-kDa glycoprotein and the ␤-chain a nonglycosylated, 25-kDa protein. N-terminal sequence analysis reveals a high level of homology to other snake phospholipase A 2 inhibitors. The inhibitor was shown to be extremely pH and temperature stable. The inhibitor was tested against a wide variety of phospholipase A 2 enzymes and inhibited the enzymatic activity of all phospholipase A 2 enzymes tested, binding with micromole to nanomole affinity. Furthermore, the inhibitor was compared with the Eli-Lilly compound LY311727 and found to have a higher affinity for human secretory nonpancreatic phospholipase A 2 than this chemical inhibitor. The role of the carbohydrate moiety was investigated and found not to affect the in vitro function of the inhibitor.
For many years research on PLA 2 1 enzymes has centred on snake venom PLA 2 s, largely because of the fact that PLA 2 enzymes are extremely abundant in snake venoms and are easily purified. PLA 2 enzymes have also been identified and purified from bovine, porcine, and human pancreas (1)(2)(3) and in human synovial fluid aspirates from rheumatoid and osteoarthritis patients (4,5). PLA 2 enzymes have been grouped according to structural and sequence characteristics. Pancreatic PLA 2 enzymes belong to group I, whereas the human secretory nonpancreatic PLA 2 is a group II enzyme. PLA 2 enzymes from snake venom belong to group I or II (6 -8).
Recently PLA 2 enzymes have been implicated as playing a role in a range of diseases including rheumatoid and osteoarthritis, asthma, acute pancreatitis, psoriasis, multiple organ failure, septic shock, and adult respiratory distress syndrome (9 -11). Several review articles (9,10,(12)(13)(14) have examined the physiological and potential disease role of hsPLA 2 -II. Of major importance are the reaction by-products of PLA 2 activity, a free fatty acid and a lysophospholipid. The fatty acid is often arachidonic acid, one of the main constituents of the cell membrane in several tissues (15)(16)(17). Arachidonic acid is the pre-cursor for inflammatory mediators such as thromboxanes, leukotrienes, and prostaglandins (18,19). Lysophospholipids are the precursor of platelet activating factor (20).
The suggestion that PLA 2 enzymes play a role in diseases process has intensified research on PLA 2 inhibitors. Many investigators have purified and characterized PLIs from numerous sources including plant, fungi, and bacteria (21)(22)(23)(24)(25)(26)(27)(28). A large research effort has gone into the design of synthetic compounds, such as the Eli-Lilly compound LY311727, which specifically inhibits hsPLA 2 -II (29). Additionally, a number of PLIs have been purified from the serum of elapid and crotalid snakes. These PLIs have all been shown to be homologous or heterologous complexes that interact with PLA 2 enzymes to inhibit enzymatic activity (30 -43).
We now report the purification and functional characterization of a PLI from Notechis ater (Australian tiger snake) serum. The inhibitor termed, N. ater inhibitor (NAI) has been shown to inhibit the enzymatic activity of all PLA 2 enzymes that it was tested against. Additionally, the affinity of the interaction was in the nanomole to micromole range dependent on the PLA 2 examined.

EXPERIMENTAL PROCEDURES
Materials-Notexin from Notechis scutatus (Common tiger snake) and taipoxin from Oxyuranus scutellatus (Australian taipan) were purchased from Venom Supplies (Tanunda, South Australia). Bee venom PLA 2 (Apis millifera), bovine pancreatic PLA 2 (Bos taurus), and Crotalus atrox PLA 2 (Western diamondback rattlesnake) were purchased from Sigma. Recombinant human secretory human type II PLA 2 was generously donated by the Garvan Institute (Darlinghurst, New South Wales, Australia). The PLA 2 inhibitor LY311727 was generously donated by Eli-Lilly and Co. (Indianapolis, Indiana). Pooled N. ater serum was obtained from World Tiger Snake Farm (Launceston, Tasmania). It should be noted that the serum was pooled from two very similar subspecies of N. ater, namely N. ater serventyi (Chappell Island tiger snake) and N. ater humphreysi (Tasmanian tiger snake).
Purification and N-terminal Sequencing-NAI (which elutes in the 0.5 M peak) was purified from whole N. ater serum using a DEAE-Sephacel (Amersham Pharmacia Biotech) column (1.5 ϫ 15 cm). Serum (10 mg of total protein) was loaded onto the column that had been equilibrated in 0.01 M ammonium acetate, pH 7.0, at a flow rate of 0.5 ml/min. Proteins were eluted with an increasing concentration of 0.1, 0.25, 0.5, and 1.0 M ammonium acetate, pH 7.0, at a flow rate of 0.5 ml/min. The eluant was monitored at 280 nm (ISCO, UA-6 UV-visible Detector), and fractions were collected (ISCO, Retriever II). The 0.5 M peak (referred to as the semi-purified preparation or SPP) was used for functional characterization of NAI. The ␣and ␤-chains of NAI were purified on a Amersham Pharmacia Biotech smart system fitted with a Amersham Pharmacia Biotech Sephasil C8 5 m (2.1 ϫ 100 mm) column. The column was equilibrated in 0.1% (v/v) trifluoroacetic acid in water (buffer A) at a flow rate of 100 l/min. The SPP collected form the DEAE-Sephacel column was loaded onto the C8 column and eluted with a gradient of 0.08% (v/v) trifluoroacetic acid, 80% (v/v) acetonitrile in water (buffer B). The percentage of buffer B was increased as follows: 0 -2 min 0%, 2-10 min 0 -55%, 10 -50 min 55-70%, and 50 -60 min 70 -100%. The eluant was monitored at 214 and 280 nm. Software controlled peak fractionation was used to collect peaks with the integrated fraction collector. Following reduction and alkylation the pro-* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Size Exclusion Chromatography-The molecular mass of NAI was assessed by size exclusion chromatography according to the method of Andrews (44) volume of the peak containing NAI.
Temperature and pH Studies-The pH stability was investigated by altering the pH of the solution in which the SPP (10 g/assay) was dissolved and testing this in the fluorometric PLA 2 assay. The assay was performed using N. scutatus venom as the PLA 2 source (50 ng/ assay with 10pPC as substrate) at room temperature. The pH values tested were 2, 4, 6, 7, 8, 9, 10, and 12. The SPP was made up to the appropriate pH and incubated overnight at 4°C prior to testing. Temperature stability was assessed in the same manner as the pH stability. Samples were heated or cooled at the appropriate temperature for 30 min and immediately tested in the fluorometric PLA 2 assay at room temperature. The temperatures examined were 4, 25, 37, 50, 60, 70, 80, 90, and 100°C. For both experiments the assay was performed at room temperature. Also, samples were not preincubated with venom because the stability of the PLA 2 under varying pH and temperature values could not be assured. All samples were performed in triplicate with appropriate positive and negative controls. Results are expressed as percentage inhibition relative to control values.
Deglycosylation of NAI␣-The SPP was lyophilized to dryness, resuspended in 100 l of 20 mM sodium phosphate solution, pH 7.2, 50 mM EDTA, 0.1% (w/v) SDS, and digested with 0.8 unit of N-glycosidase F or 5.0 milliunits of O-glycosidase (Roche Molecular Biochemicals) for 18 h at 37°C. Following incubation, the SPP was either subjected to SDSpolyacrylamide gel electrophoresis (47) under reducing conditions and silver stained (48) or blotted onto nitrocellulose and carbohydrate detected by the digoxigenin glycan detection system (Roche Molecular Biochemicals). Salts present in the deglycosylation buffer were removed by running samples over a AG11A8/G10 column. The AG11A8 matrix (Bio-Rad) represented 1.5 ϫ 7.5 cm of the column; poured directly on top of this was Sephadex G10 (Amersham Pharmacia Biotech) matrix, representing 1.5 ϫ 6 cm of the column. The column was equilibrated in water at a flow rate of 0.5 ml/min, and samples were eluted under the same conditions. The eluant was monitored at 280 nm (ISCO, UA-6 UV-visible Detector), and fractions were collected. The eluted sample was tested in the fluorometric PLA 2 assay for inhibitory activity against N. scutatus venom, Agkistrodon bilineatus venom (Central American Moccasin) and bee venom PLA 2 . For N. scutatus venom and A. bilineatus venom, 10pPC was used as substrate; 10pPG was used as substrate for bee venom PLA 2 . All PLA 2 sources were dissolved in saline/0.1% (w/v) bovine serum albumin. N. scutatus venom was used at 33.3 ng/ assay, A. bilineatus venom 500 ng/assay and bee venom PLA 2 25 ng/ assay, all in a final volume of 10 l. Native NAI was used as the positive control. Results are expressed as the percentage of inhibition relative to control values. The ␣and ␤-chains were purified by RP-HPLC (see above) after deglycosylation and analyzed by mass spectrometry using a Micromass Platform II single quadrupole instrument. The mass of the reverse phase purified proteins were also measured before deglycosylation.
IC 50 Determination-All enzymes except hsPLA 2 -II and bovine pancreatic PLA 2 were tested in the fluorometric assay using 10pPC as substrate. Bovine pancreatic PLA 2 and hsPLA 2 -II were tested in the fluorometric assay using 10pPG as substrate and in the mixed micelle assay. The amount of enzyme added for the fluorometric assay were as follows; notexin, 16.7 ng; taipoxin, 800 ng; C. atrox PLA 2 , 40 ng; bee venom PLA 2 , 10 ng; bovine pancreatic PLA 2 , 6250 ng. For the mixed micelle assay 100 ng of bovine pancreatic PLA 2 or 5 ng of hsPLA 2 -II were used. All enzymes were dissolved in saline/0.1% (w/v) bovine serum albumin, except bovine pancreatic PLA 2 and hsPLA 2 -II, which were dissolved in 20 mM Tris, pH 8.0, and 1 M NaCl, 10 mM Tris, pH 7.5,

RESULTS
Purification and N-terminal Sequencing-NAI was purified from whole N. ater serum utilizing a DEAE-Sephacel column (Fig. 1). All peaks were collected, concentrated, and tested for inhibitory activity against N. scutatus venom in the fluorometric assay. Peaks 3 and 4 inhibited; however, when samples were diluted to a equal protein concentration, inhibition of PLA 2 enzymatic activity was only detected in peak 4. The purity of NAI is approximately 90% as judged by SDS-polyacrylamide gel electrophoresis analysis (Fig. 2). NAI is composed of two proteins, an ␣-chain (ϳ30 kDa) and a ␤-chain (ϳ25 kDa). These proteins are noncovalently associated because a similar separation pattern is observed when the gel is run under nonreducing conditions (not shown). The native mass of NAI was estimated at 110 kDa by size exclusion chromatography (not shown). The ␣and ␤-chains can be separated by RP-HPLC (Fig. 3), and it is apparent that isoforms of the ␣-chain are present. By measuring the area under the curve for the ␣and ␤-chains at 280 nm and correcting for their relative absorbance using extinction co-efficients calculated from the amino acid sequence, the NAI complex is suggested to be composed of two ␣-chains and one ␤-chain. The N terminus of the purified chains were sequenced (Table I). Two N termini were present in the peak containing NAI␣, demonstrating that isoforms were partially separated under the RP-HPLC conditions utilized (Fig. 3).
Temperature and pH Studies-The temperature and pH stability of NAI was investigated by varying the temperature or pH of the solution in which NAI was dissolved and measuring the inhibitory activity of NAI. The temperature stability of NAI is shown in Fig. 4. As can be seen, NAI is active following exposure to a wide range of temperatures, with 76 Ϯ 13% inhibitory activity still present after heating at 100°C for 30 min. NAI was found to retain greater than 89% of its inhibitory activity following exposure to temperatures between 4 and 90°C. A similar situation was seen for the pH profile of NAI (Fig. 5). NAI retained greater than 94% of its inhibitory activity following overnight incubation at pH 4 -12; at pH 2 the level of inhibition was 35 Ϯ 7%.
Deglycosylated NAI-The glycosylation status of NAI␣ and NAI␤ was investigated with the digoxigenin glycan detection system. After determining that only the ␣-chain was glycosylated, the type of linkage was analyzed, using enzymes specific for N-linked or O-linked sugars (52,53). All carbohydrate moieties were removed by N-glycosidase F but were not affected by O-glycosidase (not shown). Mass spectrophotometric analysis of the native and deglycosylated ␣-chain (not shown) indicated mass differences of 2207 and 2864 Da, which are consistent with carbohydrate masses of biantennary sugars (54). After deglycosylation, NAI (in the form of the SPP) was tested for inhibitory activity against whole N. scutatus venom, A. bilineatus venom, and bee venom PLA 2 using the fluorometric assay (Fig. 6). It is apparent that deglycosylated NAI is as active on

DISCUSSION
Characteristics of NAI-N-terminal sequencing revealed two isoforms of NAI␣ that could be partially separated under the conditions utilized. These isoforms have high homology to other PLIs, with 67-70% homology to the 30-kDa chain of the PLI from Naja naja kaouthia (Thailand cobra), depending on which sequence was used for comparison. Only one sequence was present for NAI␤, which also shows high homology (84%) to the 25-kDa chain of the PLI purified from N. naja kaouthia (31). A number of PLIs have been isolated from either elapid or crotalid serum. Many of these PLIs have been sequenced and show homology to two protein families. We propose that PLIs with homology to urokinase plasminogen activator receptor, Ly-6, and CD59 will be classified as Class A PLIs. Those with homology to the carbohydrate recognition domain of C-type lectins will be classified as Class B PLIs, whereas other PLIs will be classified as Class C PLIs.
Experiments were designed to investigate the pH stability and thermostability of NAI (Figs. 4 and 5). NAI was found to be extremely pH-and temperature-resistant with greater than 89% inhibition recorded for temperatures between 4 and 90°C and greater than 94% inhibitory activity retained between pH 4 and 12. These experiments also serve to demonstrate that NAI is an effective inhibitor whether it is added before the PLA 2 (IC 50 studies) or added after the PLA 2 (pH or temperature studies). When NAI was added after the PLA 2 , inhibition was immediate and as effective as when NAI was preincubated with the PLA 2 enzyme. Preincubation was used for the IC 50 studies because the assay was empirically determined to be more robust when performed in this fashion. Carbohydrate moieties are common on snake PLIs and have been detected or presumptive evidence for their existence has been found on at least one chain for all snake PLIs characterized (30, 33, 34, 36 -43). The common occurrence of carbohydrate moieties raises the question as to its role in vivo or in vitro. Because NAI was found to cross-react with a range of PLA 2 enzymes, deglycosylated NAI was tested for inhibitory activity against group I, II, and III venom PLA 2 enzymes (Fig.  6). The results demonstrated that the inhibitory activity of NAI is not affected in vitro by the glycosylation status of the ␣-chain and that deglycosylated NAI is still as active as native NAI on all PLA 2 enzymes tested. A similar finding was reported for the PLI isolated from Laticauda semifasciata serum (42). From this we can postulate that carbohydrate moieties present on the ␣-chain play little or no role in the in vitro association of NAI with the inhibited PLA 2 enzyme.
IC 50 Determination-To gain quantitative data on the affinity of NAI for PLA 2 enzymes, IC 50 values were determined for notexin, taipoxin, C. atrox PLA 2 , bee venom PLA 2 , bovine pancreatic PLA 2 , and hsPLA 2 -II (Figs. 7-10). It must be considered that some experiments were performed under varying assay conditions. The calculated IC 50 values for bovine pancreatic PLA 2 were vastly disparate depending on the assay utilized (32.3 nM fluorometric assay and 12,050 nM mixed micelle). This in itself reveals something about the interaction between NAI and the inhibited PLA 2 . The difference in the results cannot be explained because of the enzyme concentration used in the assay, because more enzyme was used in the fluorometric assay (6, 250 ng/assay) than the mixed micelle assay (100 ng/assay). If enzyme concentration was the cause, one would expect the opposite finding with the IC 50 values.
However, it is clear that NAI is able to inhibit all PLA 2 enzymes against which it was tested, using a variety of substrates and assay conditions. Direct comparisons can be made for the IC 50 calculated for hsPLA 2 -II and bovine pancreatic PLA 2 (Fig. 9) using the mixed micelle assay, with NAI having a much higher affinity for hsPLA 2 -II. Clearly the only variable is the enzyme itself with hsPLA 2 -II, a basic protein, belonging to group II (55), whereas bovine pancreatic PLA 2 is a group I neutral PLA 2 enzyme (56). As such, it is possible that a charge interaction between NAI and the PLA 2 is important for binding or that the C-terminal extension present in group II PLA 2 enzymes may play a role in increasing the affinity of the interaction.
Examination of IC 50 data obtained with the fluorometric assay reveals a similar pattern. In order of increasing IC 50 values results were: notexin (basic, monomer, group I), bee venom PLA 2 (basic, monomer, group III), taipoxin (heterotrimer, group I), and C. atrox PLA 2 (acidic, homodimer, group II; Refs. [57][58][59][60][61][62][63]. With the exclusion of taipoxin, these results appear to confirm the observation made for hsPLA 2 -II and bovine pancreatic PLA 2 . That is, basic PLA 2 enzymes are bound with high affinity, whereas neutral or acidic PLA 2 s are bound with lower affinity. However, the interaction is not that simple because the three-dimensional structure of bee venom PLA 2 is markedly different from that of notexin or hsPLA 2 -II (64 -66). Additionally, C. atrox PLA 2 is an acidic homodimer, and the affinity of NAI for this enzyme is still quite high. The decrease in affinity from notexin to C. atrox PLA 2 is certainly not comparable with that seen for hsPLA 2 -II (basic) versus bovine pancreatic PLA 2 (neutral). It must be remembered the location of the charges and three-dimensional structure of the inhibited PLA 2 enzyme may be important for the affinity of NAI. Finally, taipoxin, a heterotrimer approximately 46 kDa is size, is strongly bound by NAI, and is composed of basic, acidic, and a neutrally charged chains (59 -61). Clearly, NAI is working via a mechanism that all PLA 2 enzymes have in common.
For comparison with a chemical inhibitor, the IC 50 values for hsPLA 2 -II using NAI or LY311727, were determined (Fig. 10). LY311727 was designed using a structure based approach by crystallizing the inhibitor with hsPLA 2 -II and then examining the structure and identifying interactions between the inhibitor and PLA 2 . After examination, the inhibitor was redesigned to maximize interactions and hence increase the efficacy of binding (29). The IC 50 values were 808 and 287.3 nM for LY311727 and NAI, respectively. As such, NAI represents a potent natural inhibitor of hsPLA 2 -II, exceeding the affinity of LY311727, despite the fact that LY311727 was specifically developed to inhibit hsPLA 2 -II. Additionally, NAI is cross-reactive against all PLA 2 enzymes it has been tested against, whereas LY311727 is specific for hsPLA 2 -II (29).
Generally, Class A and C PLIs tend to be broadly inhibitory, whereas Class B PLIs display a limited inhibition range. Naja PLI (Class A) purified from N. naja kaouthia (Thailand cobra) serum (30) is similarly constructed to NAI (Class A) but lacks the broad inhibitory repertoire, exhibiting poor inhibition against crotalid PLA 2 enzymes (67). NAI is the only PLI to inhibit the enzymatic activity of the hsPLA 2 -II enzyme. Crotalus neutralizing factor, purified from Crotalus durissus terrificus (South American rattlesnake) serum (32)(33)(34), was tested against porcine pancreatic PLA 2 (group I), and Trimeresurus flavoviridis PLI, purified from T. flavoviridis (Habu) serum (37) was tested on bovine pancreatic PLA 2 ; both PLIs were incapable of inhibiting enzymatic activity (33,37).
The mode of action of NAI is as yet unknown, apart from the suggestion that basic charges present on the enzyme and the three-dimensional structure of the inhibited PLA 2 play a role in the interaction with NAI. Nor is it clear why NAI displays such a diverse inhibitory profile while other PLIs tend to be limited in the range of PLA 2 enzymes they inhibit. It is certainly interesting to speculate that NAI or a derivative thereof may prove useful in the treatment of snake bite victims or more importantly in the treatment of the many human diseases in which PLA 2 enzymes have been implicated.