Mapping the Region of the α-Type Phospholipase A2 Inhibitor Responsible for Its Inhibitory Activity*

α-Type phospholipase A2 inhibitory protein (PLIα) from the serum of the venomous snake Gloydius brevicaudus, GbPLIα,isone of the protective endogenous proteins that neutralizes its own venom phospholipase A2 (PLA2), and it is a homotrimer of subunits having a C-type lectin-like domain. The nonvenomous snake Elaphe quadrivirgata has a homologous serum protein, EqPLIα-LP, that does not show any inhibitory activity against various snake venom PLA2s (Okumura, K., Inoue, S., Ikeda, K., and Hayashi, K. (2003) IUBMB Life 55, 539–545). By constructing GbPLIα-Eq- PLIα-LP chimeric proteins, we have mapped the residues important in conferring GbPLIα inhibitory activity on region 13–36 in the primary structure of GbPLIα. Noninhibitory EqPLIα-LP showed comparable inhibitory activity only when this region was replaced with that of GbPLIα. Further, mutational analysis of the candidate residues revealed that the individual GbPLIα to EqPLIα-LP residue substitutions N26K, K28E, D29N, and Y144S each produced a mutant GbPLIα protein with reduced inhibitory activity, with the single N26K substitution having the most significant effect. Residues 13–36 were suspected to be located in the helical neck region of the GbPLIα trimer. Therefore, the region of GbPLIα responsible for PLA2 inhibition was distinct from the carbohydrate-binding site of the homologous C-type lectin.

␣-Type phospholipase A 2 inhibitory protein (PLI␣) from the serum of the venomous snake Gloydius brevicaudus, GbPLI␣, is one of the protective endogenous proteins that neutralizes its own venom phospholipase A 2 (PLA 2 ), and it is a homotrimer of subunits having a C-type lectin-like domain. The nonvenomous snake Elaphe quadrivirgata has a homologous serum protein, EqPLI␣-LP, that does not show any inhibitory activity against various snake venom PLA 2 s (Okumura, K., Inoue, S., Ikeda, K., and Hayashi, K. (2003) IUBMB Life 55, 539 -545). By constructing GbPLI␣-Eq-PLI␣-LP chimeric proteins, we have mapped the residues important in conferring GbPLI␣ inhibitory activity on region 13-36 in the primary structure of GbPLI␣. Noninhibitory EqPLI␣-LP showed comparable inhibitory activity only when this region was replaced with that of GbPLI␣. Further, mutational analysis of the candidate residues revealed that the individual GbPLI␣ to EqPLI␣-LP residue substitutions N26K, K28E, D29N, and Y144S each produced a mutant GbPLI␣ protein with reduced inhibitory activity, with the single N26K substitution having the most significant effect. Residues 13-36 were suspected to be located in the helical neck region of the GbPLI␣ trimer. Therefore, the region of GbPLI␣ responsible for PLA 2 inhibition was distinct from the carbohydrate-binding site of the homologous C-type lectin.
Phospholipases A 2 (PLA 2 s, EC 3.1.1.4) 2 catalyze the hydrolysis of the acyl-ester bond at the sn-2 position of glycerophospholipids to yield fatty acids and lysophospholipids. Secretory PLA 2 s are a growing family of low molecular weight, highly disulfide-linked, Ca 2ϩ -requiring secretory enzymes with a His-Asp catalytic dyad and are classified into six main groups (I, II, III, V, X, and XII) according to their primary structures (1). Snake venom is one of the most abundant sources of secretory PLA 2 s, which exhibit a wide variety of pharmacological effects including neurotoxicity and myotoxicity (2). Elapidae venom contains group I PLA 2 s, and Viperidae venom contains group II ones (3). Venomous snakes have three distinct types of PLA 2 inhibitory proteins (PLI␣, PLI␤, and PLI␥) in their blood to protect themselves from the leakage of their own venom PLA 2 s into the circulatory system (4 -6). PLI␣ has only been identified in the blood of Viperidae snakes, such as Protobothrops flavoviridis (renamed from Trimeresurus flavoviridis according to the present taxonomy) (7), Gloydius brevicaudus (renamed from Agkistro-don blomhoffii siniticus) (8), Bothrops asper (9), and Cerrophidion godmani (10). It is a 75-kDa trimeric glycoprotein of 20-kDa subunits having a C-type lectin-like domain (CTLD), which is homologous to that of collectins, such as serum mannose-binding protein (MBP) and lung surfactant-apoproteins (11). G. brevicaudus, B. asper, and C. godmani PLI␣s are composed of three identical subunits, whereas P. flavoviridis PLI␣ is a trimer of two homologous subunits. P. flavoviridis and G. brevicaudus PLI␣s inhibit specifically the group II acidic PLA 2 s from their own venom (12). On the contrary, B. asper PLI␣ (BaMIP) and C. godmani PLI␣ (CgMIP-II) selectively inhibit the group II basic myotoxic PLA 2 s from their own venom (9,10). Recently, we identified a PLI␣ homolog (PLI␣-LP) from the serum of nonvenomous snake Elaphe quadrivirgata (13). This protein had 70% sequence identity with G. brevicaudus PLI␣ but did not show any inhibitory activity against various snake venom PLA 2 s.
CTLDs were first identified as carbohydrate recognition domains of C-type animal lectins that bound carbohydrates in a Ca 2ϩ -dependent manner (14). CTLDs have been identified in a variety of proteins that interact with protein ligands without carbohydrate binding activity (15). The M-type PLA 2 receptor is a type I transmembrane glycoprotein having a tandem repeat of eight CTLDs (16,17), and its soluble form was shown to be a mammalian endogenous PLA 2 inhibitor (18). Furthermore, SP-A was reported to have inhibitory activity against group IIA and group X PLA 2 s (19). Therefore, CTLDs might play a very important role as a common structural basis for the PLA 2 binding and inhibition.
In the present study, we mapped the region of G. brevicaudus PLI␣ important for the inhibition of group II acidic PLA 2 by analyzing the inhibitory and binding activities of various chimeric proteins made with the inactive E. quadrivirgata PLI␣-LP and various point-mutated proteins.
Construction of PLI␣ and PLI␣-LP Expression Plasmid-The construction of the G. brevicaudus PLI␣ (GbPLI␣) expression plasmid, ␣BB105/16b, was described previously (20). For expression of E. quadrivirgata PLI␣-LP (EqPLI␣-LP), two synthetic oligonucleotides were designed and used for the PCR amplification with a cloned PLI␣-LP cDNA used as a template. The sense primer, 5Ј-GGGAATTCCATAT-GGACGACGACGACAAGCATGAAACGGACCCTGAAGG-3Ј, contained a NdeI site (underlined) and 15 bases encoding an in-frame enterokinase cleavage site, which was followed by 20 bases encoding N-terminal residues 1-7 of the mature PLI␣-LP. The antisense primer (BamHI-␣-3/EQ), 5Ј-GGGGGGATCCCACAGGATGCTAGCCTTC-C-3Ј, contained a BamHI site (underlined), followed by 19 bases that were complementary to the nucleotide sequence of 546 -564 of the cDNA. After the PCR amplification, the amplified product was cut with NdeI and BamHI and then ligated between NdeI and BamHI sites of the bacterial expression vector pET-16b (Novagen). The ligated products were used to transform Escherichia coli XL1-Blue competent cells. Plasmid DNAs were recovered, and the nucleotide sequence of the insert was verified by both restriction analysis and DNA sequencing. The resulting plasmid, Eq␣NB/16b, was then used to transform E. coli strain BL21(DE3)pLysS.
Chimeric and Deletion Mutation-Chimeric PLI␣s were constructed by fusing appropriate expression plasmid segments. To construct a pair of expression plasmids for the N-terminal hybrid proteins, we digested ␣BB105/16b and Eq␣NB/16b with EcoRI and replaced the purified small EcoRI fragment of one with the other to yield Eq13Gb and Gb13Eq. For construction of a pair of expression plasmids for the C-terminal hybrid proteins ␣BB105/16b and Eq␣NB/16b were double-digested with AatI and HindIII, respectively, and the resultant small AatI-HindIII fragment of one was replaced with that of the other to yield Eq84Gb and Gb84Eq.
To prepare a pair of plasmid DNAs for CTLD hybrid proteins, we introduced SnaBI sites into ␣BB105/16b and Eq␣NB/16b by using the unique-site elimination (USE) mutagenesis kit (Amersham Biosciences) with mutagenetic primers 5Ј-CGACTTCCTTGTTGGTTACGTACA-ATCTTTCACTGCC-3Ј and 5Ј-CCTACTTGCTTGTTGGTTACGT-ATAATCTTTCACTGCC-3Ј (replaced nucleotides are underlined). The obtained plasmid DNAs were double-digested with ApaI and SnaBI, and the resultant small ApaI-SnaBI fragment of one was replaced with that of the other to yield Eq37Gb and Gb37Eq.
To prepare deletion mutants, GbCTLD, having only the CTLD of GbPLI␣, we designed the following sense primer: 5Ј-GGGAATTCCA-TATGGACGACGACGACAAGGTGACCAACAAGGAAGTCGG-G-3Ј containing a NdeI site (underlined) and 15 bases encoding an in-frame enterokinase cleavage site, which was followed by 21 bases encoding residues 49 -55 of GbPLI␣. The BamHI-␣-3 primer, which was previously designed for cloning of GbPLI␣ cDNA (20), was used as Primers used for the site-directed mutagenesis of GbPLI␣ Sequences of the synthetic oligonucleotide primers with the mutated codons in bold type are shown.

Mutant
Mutagenetic primer the antisense primer. The amplified DNA product was digested with NdeI and BamHI and ligated into the pET-16b vector. Similarly, the deletion mutant, EqCTLD, was prepared using the sense primer 5Ј-G-GGAATTCCATATGGACGACGACGACAAGGTGACCAACAAG-CAAGTAGGG-3Ј, and the antisense primer BamHI-␣-3/EQ. The nucleotide sequences of these chimeric and deletion mutants were confirmed by automated sequencing.
Site-directed Mutagenesis-A USE mutagenesis kit (Amersham Biosciences) was used to introduce point mutations into the GbPLI␣ coding region of ␣BB105/16b expression plasmid (20) according to the manufacturer's instructions. The complementary mutagenetic oligonucleotides used are shown in TABLE ONE. DNA sequences were confirmed by automated sequencing of the entire coding sequences.
Expression and Purification of the Recombinant Proteins-The expression plasmid DNAs were used to transform competent E. coli strain BL21(DE3)pLysS. Because all of the recombinant proteins accumulated as insoluble inclusion bodies upon the addition of isopropyl ␤-d-thiogalactopyranoside, the inclusion bodies were solubilized in 50 mM Tris-HCl buffer (pH 8.0) containing 8 M urea and 0.1 M NaCl and purified by using TALON metal affinity resin (BD Biosciences) as described previously (20). The purified recombinant proteins were denatured in 50 mM Tris-HCl buffer (pH 8.0) containing 6 M guanidine-HCl and 1 mM EDTA and then gradually refolded by dialysis for 4 days against 50 mM Tris-HCl buffer (pH 8.0) containing 1 mM EDTA. After the insoluble unfolded protein had been removed by centrifugation, the supernatant was further purified by gel filtration chromatography on a HiLoad 10/30 Superdex 200-pg column (Amersham Biosciences).
DNA Sequencing and Analysis-Both strands of DNAs were sequenced with a Thermo Sequenase core sequencing kit with 7-deaza-dGTP (Amersham Biosciences) on an SQ-3000 DNA sequencer (Hitachi) or an ABI Prism 310 genetic analyzer (Applied Biosystems). Analysis of DNA sequencing data were performed by use of the DNASIS software package (Hitachi Software Engineering).
Inhibition of PLA 2 Enzymatic Activity-Enzymatic activities of G. brevicaudus acidic PLA 2 were measured fluorometrically using 1-palmitoyl-2-pyrenedecanoylphosphatidylcholine (Cayman Chemical) as a substrate, according to the method of Radvanyi et al. (21) in the presence of various concentrations of the recombinant PLI␣s. The values of the apparent inhibition constant (K i ) were determined by nonlinear least squares analysis of the relative PLA 2 activities by using the equation described previously (20).
Binding Analysis by Surface Plasmon Resonance-Bindings of PLI␣s to PLA 2 were analyzed with a BIAcore X system (Biacore, Uppsala, Sweden). G. brevicaudus acidic PLA 2 was directly linked to the carboxymethylated dextran matrix of a CM5 sensor chip (Biacore) surface. For protection of the active site of PLA 2 from the modification, 11.8 g of PLA 2 was dissolved in 100 l of 9 mM sodium acetate buffer (pH 4.0) containing 6.66 mM CaCl 2 and 10 mM n-dodecyl phosphorylcholine and was then covalently bound to the CM5 chip surface at a flow rate of 5 l/min by using an amine coupling kit (Biacore) according to the manufacturer's instruction. Various concentrations of the recombinant proteins were injected at a flow rate of 10 l/min at 25°C with running buffer (50 mM Hepes buffer (pH 7.5) containing 0.05% Tween 20 and 1 mM EDTA). Sensor chips were regenerated at the end of each run by the injection of 10 mM HCl. Analysis of the association and dissociation curves was performed with the BIAevaluation 3.0 software (Biacore) using the 1:1 Langmuir binding model with drifting base line (global fitting), and the mean values of the apparent dissociation constants (K d ) were calculated from the association rate constant (k a ) and the dissoci- Homology Modeling of GbPLI␣-Protein homology modeling was performed by using the Molecular Operating Environment (version 2004.03; Chemical Computing Group Inc., Montreal, Canada). The amino acid sequence of GbPLI␣ (NCBI accession code BAA86972) was aligned to the template sequence (rat SP-A; Protein Data Bank code 1R13) by using the protein alignment tools in the Molecular Operating Environment. Panels A and B of Fig. 6 were prepared by using Viewer-Lite Version 4.2 (Accelrys Inc.).

RESULTS
Preparation and Characterization of Mutated PLI␣s-We previously established the system for bacterial expression of the recombinant G. brevicaudus PLI␣ (GbPLI␣) as a His 10 -tagged fusion protein and showed the recombinant protein to have strong inhibitory activity against acidic PLA 2 comparable with that of the natural PLI␣ (20). Using this expression system, we produced recombinant E. quadrivirgata PLI␣ homolog (EqPLI␣-LP) having 70% sequence identity to GbPLI␣ (Fig. 1). Furthermore, we produced various recombinant chimeric proteins between GbPLI␣ and EqPLI␣-LP, as shown in Fig. 2. The farultraviolet CD spectra of the purified recombinant proteins were very similar to that spectrum of the native PLI␣ (data not shown), indicating that they had correctly refolded. The trimeric structure of the recombinant GbPLI␣ was already confirmed by a chemical cross-linking experiment (20). Similarly, the recombinant EqPLI␣-LP was found to form the trimeric structure (data not shown). However, the truncated recombinant proteins, GbCTLD and EqCTLD, lacking the respective N-ter-  minal 48 residues of GbPLI␣ and EqPLI␣-LP were found to be monomers because they were eluted with a 25-kDa peak corresponding to the monomeric form by Superdex 200 column chromatography. Therefore, it is conceivable that the N-terminal regions preceding the CTLD of GbPLI␣ and EqPLI␣-LP were required for the formation of the trimeric structure. Fig. 3A shows the PLA 2 inhibitory activities of four recombinant proteins, GbPLI␣, EqPLI␣-LP, GbCTLD, and EqCTLD. As expected, the recombinant EqPLI␣-LP did not inhibit acidic PLA 2 just like the natural EqPLI␣-LP, whereas the recombinant GbPLI␣ inhibited it with an apparent inhibition constant (K i app ) value of 3.05 nM. The truncated recombinant proteins, GbCTLD and EqCTLD, did not show any inhibition against acidic PLA 2 . The lack of inhibitory activity of monomeric GbCTLD against acidic PLA 2 raises two possibilities, that the PLA 2 binding site is not in the CTLD or the trimeric structure of CTLD is essential for the complex formation with PLA 2 , because one trimeric PLI␣ molecule was previously shown to bind stoichiometrically to one PLA 2 molecule (7). Therefore, the loss-of-function experiments should be designed under conditions where the trimeric structure is retained, otherwise, the obtained results would be complicated. All of the chimeric proteins and point-mutated PLI␣s used in the present study proved to form a trimeric structure, like the native PLI␣, as judged from the elution profiles of gel filtration on a Superdex 200 column, which showed peaks corresponding to the apparent molecular mass of 70 -75 kDa.
Chimeric Mutagenesis between GbPLI␣ and EqPLI␣-LP-To specify the regions responsible for the PLA 2 inhibition in the GbPLI␣, we created various sets of chimeric proteins between GbPLI␣ and EqPLI␣-LP (Fig. 2). Fig. 3B shows the inhibitory activity of these chimeric proteins against G. brevicaudus acidic PLA 2 , and their estimated inhibition constants (K i app ) are summarized in TABLE TWO. The chimeric proteins Eq13Gb, Gb37Eq, Gb84Eq, and Eq13Gb37Eq were inhibitory against acidic PLA 2 , whereas the others, Gb13Eq, Eq37Gb, Eq84Gb, and Gb13Eq37Gb, showed little or no significant inhibition. These results indicate that residues 13-36 of GbPLI␣ were the critical region for the PLA 2 inhibition. Most strikingly, Eq13Gb37Eq showed significant inhibition toward the acidic PLA 2 , with a K i app value of 9.38 nM, which was comparable with that of GbPLI␣ (3.05 nM), that is, noninhibitory EqPLI␣-LP gained inhibitory activity only when its residues 13-36 were replaced with those of GbPLI␣. In contrast, the inhibitory activity of GbPLI␣ was lost when these residues were replaced with those of EqPLI␣-LP (Gb13Eq37Gb). Because all of these chimeric proteins retained their trimeric structure and showed backbone CD spectra similar to that of the native PLI␣, the major conformation would not be affected by the replacement of these residues.
Detailed comparisons of apparent inhibition constants between GbPLI␣ and Eq13Gb, and between Gb37Eq and Eq13Gb37Eq suggested that residues 1-12 of EqPLI␣-LP were preferred for the inhibition toward the acidic PLA 2 . Between the amino acid sequences of GbPLI␣ and EqPLI␣-LP shown in Fig. 1, there are only three amino acid replacements in residues 1-12. This indicates that the replacements, Asp to Glu at residue 6, His to Gln at residue 8, and Val to Ile at residue 9, can affect the inhibitory activity. Furthermore, the comparison of K i app values of GbPLI␣ and Gb37Eq and of Eq13Gb and Eq13Gb37Eq suggested that residues 37-147 of GbPLI␣ were preferred for the PLA 2 inhibition rather than those of EqPLI␣-LP. These preferable residues could be further restricted to residues 84 -147 from the comparison of the K i app value between GbPLI␣ and Gb84Eq.
To study the effect of these chimeric mutations on the binding kinetics of the PLA 2 -PLI␣ interaction, we immobilized G. brevicaudus acidic PLA 2 on a biosensor chip in the presence of the micellar substrate ana-log (n-dodecyl phosphorylcholine) and measured the interaction with various chimeric proteins in real time by monitoring the changes in surface plasmon resonance after the injection of various concentrations of the chimeric PLI␣s. Because Ca 2ϩ had no apparent effect on the binding curves (data not shown), the formation of PLI␣-PLA 2 complex was considered to be independent of Ca 2ϩ . Previously, no significant effect of Ca 2ϩ was observed in the gel filtration experiments using a mixture of P. flavoviridis PLI␣ and its venom PLA 2 (7). The Ca 2ϩ -independent binding of PLI␣ to PLA 2 is in contrast with the Ca 2ϩ -dependent binding of C-type lectins to carbohydrates. Typical binding and dissociation curves at various concentrations of GbPLI␣ and those obtained at about 20 nM concentrations of the chimeric proteins in the presence of 1 mM EDTA are shown in Fig. 4. GbPLI␣ formed a stable complex with acidic PLA 2 , showing fast association and slow dissociation. The dissociation constants (K d ) and the association and dissocia- tion rate constants (k a and k d ) were calculated from the curve fitting analysis of the sensorgrams (TABLE TWO). The K d value of GbPLI␣ was calculated to be 4.53 nM, which was comparable with the K i app value (3.05 nM) obtained from the measurement of its inhibitory activity. The chimeric proteins Gb37Eq and Eq13Gb37Eq formed a complex with the immobilized PLA 2 with K d values of 24.5 and 6.5 nM, respectively. However, the chimeric proteins Eq37Gb and Gb13Eq37Gb did not bind to the immobilized acidic PLA 2 This result is consistent with that obtained on the inhibitory activity described above, confirming that residues 13-36 of GbPLI␣ were the critical region for PLA 2 binding and PLA 2 inhibition. The replacement of residues 37-147 of GbPLI␣ with those of EqPLI␣-LP caused a 3-fold increase in the dissociation rate constant and a decrease in the association rate constant. Further replacement of residues 1-12 of Gb37Eq with those of EqPLI␣-LP caused a 2-fold increase in the association rate constant. This increase in the association rate constant might explain the preference of residues 1-12 of EqPLI␣-LP in PLA 2 inhibition described above.
Single Amino Acid Substitution of GbPLI␣-When the amino acid sequence of EqPLI␣-LP was compared with those sequences of PLI␣s from G. brevicaudus, P. flavoviridis, and C. godmani (Fig. 1), there were some unique residues only found in the EqPLI␣-LP sequence, including residues 21, 23, 25-29, and 59 -63, the insertion of two residues at position 118, and the deletion of the C-terminal residue. Therefore, we performed site-directed mutagenesis to substitute the corresponding amino acid residues of GbPLI␣ with these unique residues of EqPLI␣-LP to specify the residues responsible for the PLA 2 inhibition. The inhibitory activities of these mutant proteins are shown in Fig. 5, and the estimated apparent inhibition constants are summarized in TABLE THREE. Because all of these mutants showed significant inhibitory activity, no inhibitory activity of EqPLI␣-LP was not ascribable to the replacement of a single critical residue with one of these unique residues. Therefore, multiple mutations might be required for the complete loss of the inhibition. Among the mutant proteins substituted within residues 13-36, N26K showed the weakest inhibition toward the acidic PLA 2 with a K i app value of 80 nM, which was a value 26-fold higher than that of GbPLI␣. From the chimeric mutagenesis experiments described above, residues 13-36 of GbPLI␣ were suggested to be essential for both the inhibition and binding of the acidic PLA 2 . Because other mutant proteins, Q21E, E23K, S25D, K28E, and D29N, showed decreased inhibitory activities, multiple residues including Asn-26 were probably responsible for PLA 2 binding and inhibition. Further, binding kinetics of E23K, S25D, N26K, K28E, and D29K with respect to the immobilized acidic PLA 2 were investigated by surface plasmon resonance analysis, and the values obtained are also shown in TABLE THREE. The calculated K d value of N26K was 4-fold higher than that of GbPLI␣. The disagreement between the K d and K i app values might reflect the difference between the free and immobilized forms of PLA 2 , or the difference in the experimental conditions between the presence and absence of phospholipid substrate. N26K caused a 3.3-fold increase in the dissociation rate constants, resulting in decreased inhibitory activity. The introduction of a positive charge at position 26 is likely to have affected the interaction with PLA 2 . Despite the drastic change of the residues to the oppositely charged residues, neither E23K nor K28E caused any significant reduction in the inhibitory and binding activities, suggesting that Glu-23 and Lys-28 were unimportant to the interaction with the acidic PLA 2 . But there was a remarkable feature found only in the sensorgrams of K28E, i.e. the K d value for K28E increased with a decrease in its concentration, whereas that for other mutant proteins was independent of their concentrations (data not shown). The K28E protein might be structurally unstable or might dissociate into monomer at its low concentrations. Although D29N had an inhibitory activity weaker than the intact protein against acidic PLA 2 , it showed association and dissociation rate constants of 2.5-and 2.1-fold, respectively, indicating that the binding of PLA 2 to D29N was slightly stronger than that for GbPLI␣. The negative charge at position 29 might be important in the inhibitory activity but not in the binding activity. Fig. 5B shows the inhibitory activities of GbPLI␣ with point mutation in its CTLD. The K i app values of Q79E, Q111E, T118K, T118NKL N136D, and L148⌬ toward acidic PLA 2 were 10. 5, 9.09, 11.8, 3.87, 4.72, and 10.8 nM, respectively, values comparable with that value of GbPLI␣ (3.05 nM). However, the K i app value of Y144S was 43.0 nM, which was 14-fold higher than that of GbPLI␣. Therefore, Tyr-144 might be important in the PLA 2 inhibition. This substitution at residue 144 may account for the weak inhibitory activity of Gb84Eq. Furthermore, the results obtained for the point mutants H129N and S130D, in which the unique residues in PLI␣ sequences were substituted by the conserved residues among the extracellular collectins, showed that these unique residues were not involved in PLA 2 inhibition.

DISCUSSION
Earlier, Nobuhisa et al. (22) investigated the binding of various fragments of P. flavoviridis PLI␣ to the venom PLA 2 isozymes and reported that hydrophobic residues 136 -147 of P. flavoviridis PLI␣ were critical for binding to the PLA 2 isozymes. In the present study, Tyr-144 of GbPLI␣ was found to contribute to the PLA 2 inhibition to some extent. However, the loss of PLA 2 inhibitory activity in EqPLI␣-LP could not be rationalized by the replacement of residues 136 -147, because EqPLI␣-LP was found to exhibit the inhibitory activity solely by replacing residues 13-36 with those of GbPLI␣. Moreover, they reported that the N-terminal fragment (residues 1-37), and the CTLD fragment (residues 38 -147) of P. flavoviridis PLI␣ also bound to the PLA 2 isozymes, although their affinities and stoichiometry were not quantified (22). Because Eq37Gb and GbCTLD lost the binding and inhibitory activities toward PLA 2 , the role of CTLD in the PLA 2 binding appears to be inconsistent with their results for the P. flavoviridis CTLD fragment.
We constructed a three-dimensional structural model of GbPLI␣ by homology modeling on the basis of the crystal structure of rat SP-A (23). Fig. 6 compares the overall fold of the homology-built model of GbPLI␣ with that of rat SP-A. The surface loops and calcium binding sites of SP-A CTLD, which are responsible for the carbohydrate binding, were drastically changed in the GbPLI␣ CTLD, presumably because of the deletion of 13 amino acid residues starting at residue 97 of GbPLI␣ (Fig.  1). This finding may agree with the previous result that GbPLI␣ does not have any Ca 2ϩ -dependent carbohydrate binding activity (8). Residues 13-36 of GbPLI␣, which were found to be critical for PLA 2 binding and inhibition in the present study, are located at the helical neck region in this model. This ␣-helical coiled-coil neck region is known to mediate the trimerization of CTLD in many C-type lectins, including human MBL (24), rat MBP-A (25), human SP-D (26), rat SP-A (23), and tetranectin (27). This is likely to apply to GbPLI␣ as well because the

Inhibition constants (K i app ) of recombinant proteins toward G. brevicaudus acidic PLA 2 and dissociation constants (K d ) and association and dissociation rate constants (k a and k d ) for the interaction of the recombinant proteins with immobilized G. brevicaudus acidic PLA 2
truncated recombinant protein GbCTLD lacking the N-terminal 48 residues of GbPLI␣ was in monomeric form. Therefore, residues 13-36 of GbPLI␣ are important both in binding of acidic PLA 2 and in trimerization of PLI␣ subunits. It should be noted that the N-terminal 37-amino acid fragment of P. flavoviridis PLI␣ was reported to bind directly to its venom PLA 2 isoenzymes (22). As was depicted in Fig. 6C, we suppose that the central pore formed by the trimerization of GbPLI␣ would be the PLA 2 -binding site, because one trimeric PLI␣ stoichiometrically binds one PLA 2 molecule. In this model, the region of residues 13-36 may not function as a direct binding site of PLA 2 but serve as a dominant determinant of the central pore structure formed by the trimerization through the interactions of these ␣-helical neck regions. The central pore formed by residues 13-36 of EqPLI␣-LP might be too small to interact with G. brevicaudus acidic PLA 2 , and their replacement with GbPLI␣ residues would enlarge the pore to permit the interaction. Because the region of residues 13-36 are the most variable region among various PLI␣s, the amino acid substitutions in this region might contribute to the high specificity of PLI␣ toward various PLA 2 s. This possible PLA 2 -binding site in the central pore is distinct from the carbohydrate-binding sites of the homologous SP-A trimer (Fig. 6D). Because Tyr-144 is expected to be located in the central pore, this residue might be one of the residues responsible for the direct interaction with to the PLA 2 molecule. Further studies using site-directed mutagenesis targeting the residues in the central pore will be required to confirm this model.
Recently SP-A was reported to be an endogenous inhibitor of groups IIA and X PLA 2 s (19). The inhibition of P. flavoviridis PLA 2 by SP-A was not affected by the presence of various monosaccharides (28), suggesting that the carbohydrate-binding domain of SP-A is not involved in the PLA 2 inhibition. Therefore, the PLA 2 -binding site of SP-A is likely to reside apart from the carbohydrate-binding sites, as similarly suggested for PLI␣. But there are some differences between SP-A and PLI␣ in the interactions with the PLA 2 molecule; i.e. SP-A interacts directly with group IIA PLA 2 in a Ca 2ϩ -dependent manner (29), whereas PLI␣ in a Ca 2ϩ -independent manner. This difference might be due to the loss of the calcium-binding site in PLI␣. Another endogenous inhibitor of group IB and X PLA 2 s is a soluble form of the PLA 2 receptor, which contains eight tandem CTLDs (18). Because three CTLDs (domains 3-5) of the PLA 2 receptor were shown to be PLA 2 -binding region (30), these three CTLDs might bind one PLA 2 molecule in a way similar to that of trimeric PLI␣. Further information about the role of CTLDs in PLA 2 inhibition will be obtained by examining the crystal structure of the PLI␣-PLA 2 complex.