HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

J. Biol. Chem., Vol. 277, Issue 26, 23651-23657, June 28, 2002

This Article Abstract Full Text (PDF) All Versions of this Article: 277/26/23651    most recent M111120200v1 Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via Google Scholar Google Scholar Articles by Fujitani, N. Articles by Kawano, K. Search for Related Content PubMed PubMed Citation Articles by Fujitani, N. Articles by Kawano, K. Social Bookmarking                      What's this?

Structure of the Antimicrobial Peptide Tachystatin A^*

Naoki Fujitani, Shun-ichiro Kawabata^§¶, Tsukasa Osaki^§, Yasuhiro Kumaki, Makoto Demura, Katsutoshi Nitta^**, and Keiichi Kawano

From the  Division of Biological Sciences, Graduate School of Science, Hokkaido University, Sapporo 060-0810, Japan, the ^§ Department of Molecular Biology, Graduate School of Medical Science, Kyushu University, Fukuoka 812-8582, Japan, the ^¶ Department of Biology, Kyushu University, Fukuoka 812-8581, Japan, the  Faculty of Pharmaceutical Sciences, Toyama Medical and Pharmaceutical University, Toyama 930-0194, Japan, and  Core Research for Evolutional Science and Technology (CREST), Japan Science and Technology Corp., Tokyo 101-0062, Japan

Received for publication, November 20, 2001, and in revised form, April 14, 2002

	ABSTRACT

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

The solution structure of antimicrobial peptide tachystatin A from the Japanese horseshoe crab (Tachypleus tridentatus) was determined by two-dimensional nuclear magnetic resonance measurements and distance-restrained simulated annealing calculations. The correct pairs of disulfide bonds were also confirmed in this study. The obtained structure has a cysteine-stabilized triple-stranded -sheet as a dominant secondary structure and shows an amphiphilic folding observed in many membrane-interactive peptides. Interestingly, tachystatin A shares structural similarities with the calcium channel antagonist -agatoxin IVA isolated from spider toxin and mammalian defensins, and we predicted that -agatoxin IVA also have the antifungal activity. These structural comparisons and functional correspondences suggest that tachystatin A and -agatoxin IVA may exert the antimicrobial activity in a manner similar to defensins, and we have confirmed such activity using fungal culture assays. Furthermore, tachystatin A is a chitin-binding peptide, and -agatoxin IVA also showed chitin-binding activities in this study. Tachystatin A and -agatoxin IVA showed no structural homology with well known chitin-binding motifs, suggesting that their structures belong to a novel family of chitin-binding peptides. Comparison of their structures with those of cellulose-binding proteins indicated that Phe⁹ of tachystatin A might be an essential residue for binding to chitin.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

There is no acquired immunity in invertebrate animals, which lack a system for the production of antibodies. Therefore, antimicrobial proteins, lectins, and phenoloxidases play the most important role in the host-defense activities of invertebrates (1-3). Although vertebrate animals have acquired innate immunities and are defended by their complex defense activities, invertebrate animals cannot use the diversification of antibodies, and the defending molecules must distinguish self from nonself. Self-defense by antimicrobial substances involves a direct attack on infectious agents. This is the fundamental mechanism of the host-defense systems identified in all species, including both invertebrate and vertebrate animals, and is an important object of study with regard to biological and immunological self-defense reactions.

The defense actions of the horseshoe crab have been found to be carried out by innate immune substances derived from hemocytes in hemolymph plasma (2, 4-6). Whereas the hemolymph of crustacean animals in general have three kinds of hemocytes (granular, small granular, and nongranular), in the case of the horseshoe crab, granular hemocytes comprise 99% of all hemocytes. The granular hemocytes contain two kinds of granules, large and small, which store the peptides and proteins related to self-defense activities (7). The clotting factors are stored in the large granules, and most antimicrobial substances are present in the small ones (7). The hemocytes are sensitive to lipopolysaccharides, which are major components of the outer membrane of Gram-negative bacteria, and defense molecules are secreted by exocytosis to clot body fluid and to kill infectious pathogens in response to lipopolysaccharide stimulation (2, 7-8). The following four antimicrobial peptides have been characterized in the small granules: tachyplesins (4), big defensin (5), tachycitin (6), and tachystatins (9).

Tachystatin A consists of two isopeptides made up of 44 amino acid residues; tachystatin A1 contains phenylalanine at the C-terminal end, and tachystatin A2 contains tyrosine at the corresponding site (9). Tachystatin A inhibits the growth of Gram-negative bacteria, Gram-positive bacteria, and fungi (9). Whereas tachystatin A has no similarities with the other three peptides in terms of molecular mass or amino acid sequence, tachystatin A shows weak sequence similarity to -agatoxin IVA (-aga IVA)¹ (10), a P-type calcium channel (11, 12) antagonist found in the venom of the funnel web spider (Agelenopsis aperta). Because the horseshoe crab is an invertebrate animal phylogenetically close to spiders and scorpions, it is possible that both molecules have a common ancestor. Furthermore, tachystatin A has a chitin-binding property that is a common feature of the components of small granular derived antimicrobial peptides. It is likely that this property has two explanations. First, the property may be important for killing fungal pathogens, because chitin is one of the major components of the fungal cell wall. Second, this property may contribute to the healing of the damaged exoskeleton of the horseshoe crab, since chitin is the major component of this exoskeleton.

In this study, we determined the structure of tachystatin A in an aqueous solution using two-dimensional ¹H NMR spectroscopy and distance geometry-simulated annealing calculations. The solution structure involved a triple-stranded -sheet and two -turns. This folding differed from that of the antimicrobial peptides, tachyplesins (13) and tachycitin (14), isolated from the horseshoe crab. With respect to the chitin-binding property, no chitin-binding domains with the structural motif of tachystatin A have yet been found. In this paper, we show that the folding of tachystatin A produces a unique motif not only as an antimicrobial peptide but also as a chitin-binding peptide.

	EXPERIMENTAL PROCEDURES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

NMR Spectroscopy-- Hemocyte debris from the Japanese horseshoe crab (Tachypleus tridentatus) was prepared by the methods of Nakamura et al. (4), and tachystatin A was purified by the methods of Osaki et al. (9). Samples were prepared for NMR spectroscopy by dissolving tachystatin A in 10% D₂O, 90% H₂O or 99% D₂O at a final concentration of 2 mM. The sample pH was adjusted to 3.5 to observe amide protons by decreasing the hydrogen-deuterium exchange rates.

All NMR spectra were observed on a JEOL JNM-alpha 500 spectrometer operating at a proton frequency of 500 MHz with a sample temperature of 298 K. Two-dimensional DQF-COSY (15), TOCSY with MLEV-17 spin lock sequence (16, 17), NOESY (18), and E. COSY (19) measurements were recorded in a phase-sensitive mode. TOCSY spectra were obtained with mixing times of 80 and 100 ms. NOESY spectra were recorded with mixing times of 100, 150, and 300 ms. For water signal suppression, the DANTE pulse method (20) was applied to all NMR experiments. The proton chemical shifts were referenced to external sodium 3-(trimethylsilyl) propionate-2,2,3,3-D₄. The spectral width of all measurements was 6002.40 Hz. All two-dimensional measurements were recorded with 1024 × 512 frequency data points and were zero-filled to yield 1024 × 1024 data matrices except for the high resolution DQF-COSY and E. COSY. High resolution DQF-COSY and E. COSY were recorded with 4096 × 512 data points in the t₂ and t₁ dimensions, respectively, and zero-filled to 8192 × 8192 points to measure the coupling constants. All NMR data were processed by NMRPipe software (21). Time domain data in both dimensions were multiplied by a sine bell window function with a 90° phase shift prior to Fourier transformation. Base-line correction was applied using a fifth order polynomial.

All NMR spectra were assigned with the program XEASY (22). DQF-COSY and TOCSY spectra recorded with the samples dissolved in H₂O were used to identify the spin systems of all residues. In addition, one-dimensional and DQF-COSY spectra were used to identify the amide protons protected from solvent exchange using a sample that had been lyophilized from H₂O solution and resuspended in D₂O.

Structure Calculations-- Three-dimensional structures of tachystatin A were calculated with the program X-PLOR 3.1 (23) on the R5000 processors of a Silicon Graphics Indy work station.

Distance restraints for calculations were estimated from the cross-peak intensities in NOESY spectra with a mixing time of 150 ms; the estimated restraints were then classified as strong, medium, or weak and assigned upper limits of 2.7, 3.5, and 5.0 Å, respectively. Pseudo-atom corrections were used for unresolved NOE cross-peaks and those of methyl protons, nonstereospecifically assigned methylene, and aromatic protons according to the protocol of X-PLOR. In addition, 0.5 Å was added to the upper limit of the distance constraints of only the involved methyl protons according to the report by Clore et al. (24). The restraints of the dihedral angle  were based on ³J_HN coupling constants measured in high resolution DQF-COSY and E-COSY. When ³J_HN was more than 8.0 Hz, the dihedral angle  was constricted to 120 ± 40°. Hydrogen bond restraints were used as distance constraints of 1.5-2.5 Å between amide protons and carbonyl oxygens and 2.5-3.5 Å between amide nitrogens and amide protons. All analyses of r.m.s. of difference values as well as secondary and tertiary structures of tachystatin A were performed with the program MOLMOL (25).

Determination of Disulfide Linkages-- Tachystatin A was dissolved in 0.1 M Tris-HCl, pH 6.8, containing 2 M urea, and digested with trypsin and thermolysin (E/S ratio = 1:20, w/w) at 37 °C for 16 h. The digest was separated by reverse-phase high pressure liquid chromatography on a Chemcosorb 5-ODS-H column (2.1 × 150 mm; Chemco Scientific Co., Ltd., Osaka, Japan). Peptides containing disulfide bonds were identified by amino acid analysis after performic acid oxidation.

Antimicrobial Activity and Chitin-binding Assay for -aga IVA-- Antimicrobial activity of -agatoxin IVA was assayed as described by Saito et al. (5). Pichia pastoris was plated on nutrient agar plates. Fungal culture was collected at the logarithmic phase of growth, washed twice with 10 mM phosphate buffer (pH 7.0), and adjusted to 5000-10000 cells/ml with the same buffer. The peptide solution (50 µl) was added to 450 µl of the fungal suspension, and the mixture was incubated at 30 °C for 1 h. An aliquot of the reaction mixture (100 µl) was then plated onto the agar plate. After 24 h of incubation at 30 °C, the number of colonies on the plates was counted. As a control experiment, the phosphate buffer was added to the fungal suspension, and the mixture was incubated for 1 h, plated on agar, and cultured.

Chitin (0.5 mg) was mixed with -aga IVA in 100 µl of 20 mM Tris-HCl buffer (pH 7.5) containing 0.15 M NaCl and 2 mM CaCl₂ and then incubated at room temperature for 15 min and centrifuged at 15,000 rpm for 2 min. The supernatant was removed, and the precipitate was washed with 1 ml of the same buffer and then eluted with 100 µl of 0.1 M HCl. The peptide concentrations of the eluate were determined using a micro BCATM protein assay kit from Pierce.

	RESULTS

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Identifications of Secondary Structures-- Well resolved spectra were obtained for the sample conditions and methods described under "Experimental Procedures," and the cross-peaks observed in all of the two-dimensional ¹H NMR spectra were assigned completely. Fig. 1 summarizes the sequential assignments derived from NOESY with a mixing time of 150 ms. The sequential NOE connectivities deduced from the combination of overall strong d_N(i, i + 1) and weak d_NN(i, i + 1) indicate that the peptide is rich in -structures (26). Medium and long range NOEs were also assigned completely using NOESY spectra, and no characteristic NOEs for the helical structure were found as a result of the sequential assignments.

View larger version (10K): [in this window] [in a new window]   Fig. 1.   Summary of sequential NOE connectivities, coupling constant 3JHN, and the slowly exchanging backbone amino acid amide protons observed in tachystatin A. The sequential NOEs, dN, dNN, and dN, are indicated by bars, and their intensities are classified as strong, medium, or weak according to the heights of the bars. 3JHN values of more than 8 Hz are indicated by upward arrows. Slowly hydrogen-deuterium-exchanging amides are indicated by filled circles. These were still observed in a DQF-COSY spectrum recorded 24 h after dissolving in D2O.

The analysis of the one-dimensional spectra and DQF-COSY, which were recorded 24 h after dissolving the sample in D₂O, revealed the backbone amide protons of Phe⁹, Cys¹¹, Leu²⁷, Thr²⁸, Arg³⁰, Gly³⁹, Arg⁴⁰, Cys⁴¹, and Gln⁴² exchange slowly with a deuterium solvent, thus indicating that these amide protons form hydrogen bonds or are buried inside the molecule. The coupling constants between H and backbone NH, ³J_HN, the value of which is used to determine the restraint of dihedral angle , were estimated from the high resolution DQF-COSY spectrum. Generally, residues consisting of a -sheet produce a value of more than 8.0 Hz for ³J_HN, and those consisting of an -helix produce a value of less than 6.0 Hz (26). Coupling constants ³J_HN of more than 8.0 Hz were identified for 17 of the 44 residues in tachystatin A, and there were no peaks with a ³J_HN value of less than 6.0 Hz. Slowly exchanging amides and coupling constants ³J_HN are also summarized in Fig. 1. The above evidence regarding ³J_HN and exchange rates as well as the properties of sequential assignments may suggest that tachystatin A is rich in -structures.

It was possible to identify the secondary structure of tachystatin A as consisting of a triple-stranded antiparallel -sheet constructed by Phe⁹-Val¹², Thr²⁸-Arg³⁰, and Gly³⁹-Gln⁴² from the compatible combination of NOEs,  angle values estimated from ³J_HN, and hydrogen bonds predicted by slowly exchanging amide protons. As an important factor in the determination of the -sheet, three long range NOEs were observed between Hs, which are located between Asn¹⁰ and Arg⁴⁰, Val¹² and Tyr³⁸, and Cys²⁹ and Cys⁴¹. Fig. 2 shows a schematic diagram of this -sheet structure, including the observed NOEs and hydrogen bonds. From these analyses, it was found that the main element of the secondary structure of tachystatin A is a triple-stranded -sheet with a topology of +2x, 1 (27).

View larger version (23K): [in this window] [in a new window]   Fig. 2.   The -sheet schematic diagram in tachystatin A. Observed NOEs are shown by arrows, and slowly exchanging protons are circled. Dashed lines show the predicted hydrogen bonds. Two tight turns are also identified by NOE networks.

Determination of Disulfide Bonds and Global Structure-- Tachystatin A was treated with iodoacetamide in the presence of 8 M urea, under nonreducing conditions. Amino acid analysis revealed that no cysteine residues were S-alkylated, thereby indicating the presence of three disulfide bonds. The intact tachystatin A was digested with trypsin and thermolysin, and one disulfide-containing peptide (T-TL-15) was isolated, as described under "Experimental Procedures." By amino acid analysis, T-TL-15 was found to be composed of two chains, Phe⁹-Cys¹¹ and Leu²⁷-Arg³⁰, indicating the presence of a disulfide linkage of Cys¹¹-Cys²⁹. Peptides containing the remaining two disulfide bonds could not be isolated, because the -Cys²³-Cys²⁴- sequence was highly resistant to proteolytic digestion. These disulfide bonds connecting Cys²³, Cys²⁴, and the remaining two cysteine residues have been established by nuclear magnetic resonance, which we used to elucidate the conformational structure of tachystatin A.

We carried out the structural calculations using the program X-PLOR version 3.1 (23) to determine the three-dimensional solution structure of tachystatin A with the restraints derived from NMR experiments. This solution structure consisted of 274 intramolecular NOEs, 19 dihedral angles, and 22 hydrogen bonds. The 274 NOEs as distance restraints included 56 intraresidues (|i  j|= 0), 120 sequential (|i  j|= 1), 20 medium range (2 |i  j| 4), and 78 long range (|i  j| 5) NOEs.

In the first step, the structure calculations were performed without the restraints of dihedral angles, hydrogen bonds, or disulfide bonds and with only the distance constraints estimated from the NOE intensity. Restraints with large violations were removed or modified in this step. In the next step, a total of 22 hydrogen bond and three disulfide bond restraints were added for calculation. Hydrogen bond restraints were added for the pairs indicated in the analysis of secondary structure shown in Fig. 2. Although three pairs of disulfide bonds are composed of six cysteines, positions 4, 11, 23, 24, 29, and 41 in this molecule, only one pair, Cys¹¹-Cys²⁹, was determined by chemical methods. There are three possibilities in the other two pairs, Cys⁴-Cys²³ and Cys²⁴-Cys⁴¹, Cys⁴-Cys²⁴ and Cys²³-Cys⁴¹, and Cys⁴-Cys⁴¹ and Cys²³-Cys²⁴. In the NOESY spectra, the NOE cross-peaks that correspond to disulfide bonds were observed between the  protons of Cys⁴ and Cys²⁴ and of Cys²³ and Cys⁴¹. However, no NOE cross-peaks corresponding to the other patterns of disulfides were observed. These results strongly suggested that the reasonable disulfide bonds are Cys⁴-Cys²⁴ and Cys²³-Cys⁴¹.

In addition, to confirm the correct connections from the standpoint of energetic study, the calculations were performed with Cys¹¹-Cys²⁹ connectivity and with one of three patterns of disulfide bond connectivities, Cys⁴-Cys²³ and Cys²⁴-Cys⁴¹, Cys⁴-Cys²⁴ and Cys²³-Cys⁴¹, or Cys⁴-Cys⁴¹ and Cys²³-Cys²⁴. Well converged structures were then obtained in the case of Cys⁴-Cys²⁴ and Cys²³-Cys⁴¹ at 110 kcal/mol and 80 kcal/mol lower than the structures obtained with disulfide bonds Cys⁴-Cys²³ and Cys²⁴-Cys⁴¹ and Cys⁴-Cys⁴¹ and Cys²³-Cys²⁴, respectively, in terms of the average total molecular potential energy calculated by X-PLOR. The energy analysis of these sets of structures proved that the disulfide bonds Cys⁴-Cys²⁴ and Cys²³-Cys⁴¹ were the correct bonds, so these pairs were used for the final calculation step.

Finally, a family of 20 accepted three-dimensional structures was selected with the lowest potential energies that contained no experimental violations greater than 1.0 Å and 1° in the distance and torsion angle restraints, respectively. In a Ramachandran plot, 99.6% of the backbone dihedral angles of the 20 converged structures fall either in the -sheet region or in generally allowed regions. A summary of the energetic statistics for tachystatin A is shown in Table I. Pairwise r.m.s. of difference values for the 20 lowest energy structures of well defined regions of the 4-14, 21-32, and 37-42 residues were found to be 0.70 Å for backbone atoms and 1.63 Å for all heavy atoms when the structures were fitted in this region.

Fig. 3 represents the superposition of the backbone coordinates for 20 converged structures. The topology and constituting residues of the -sheet determined by calculations were found to correspond to those determined by the analysis of secondary structures mentioned above. Two tight turns, Leu⁶-Phe⁹ and Cys²⁴-Leu²⁷, were also identified in the peptide using the standard criterion that the distance between C(i) and C(i + 3) is less than 7 Å (28) and with the characteristic NOE connectivities. The distances between C atoms at positions i and i + 3 of the first (Leu⁶-Phe⁹) and second (Cys²⁴-Leu²⁷) -turns were 5.16 and 5.39 Å, respectively. Although the first turn had a nearly type II conformation from the viewpoint of the dihedral angles of constructing residues, there was no capability to form an intraturn hydrogen bond, because Phe⁹ was involved in the first strand, and the carbonyl formed a hydrogen bond with the amide of Cys⁴¹. This turn was therefore defined as a _E-type turn (29) (i.e. a type IV turn (miscellaneous type) in the classical nomenclature). In contrast, the second turn has a typical type II (_P-type) conformation with an intraturn hydrogen bond between Cys²⁴ and Leu²⁷.

View larger version (26K): [in this window] [in a new window]   Fig. 3.   The NMR structure of tachystatin A. A, superposition of the backbones of all 20 tachystatin A conformers. B, ribbon representation of the lowest energy model of tachystatin A. These images were generated by the program MOLMOL (25).

The loop regions with disordered conformation were constructed by Tyr¹⁶-Pro²² (loop 1) and Tyr³²-Thr³⁷ (loop 2). The average r.m.s. of difference values for these loops were 2.57 and 1.63 Å for the backbone atoms, respectively, when the structures were superposed as shown in Fig. 3. Both loops were rich in hydrophobic residues, and most of these residues were exposed to solvent. Because the hydrophobicity encourages interaction with the membrane, these loops might be an important factor in interactions with the bacterial membrane.

Comparison with the Ca²⁺ Channel Blocker and Evolutionary Considerations-- Whereas few structural similarities have been identified between tachystatin A and other invertebrate antimicrobial peptides, it has been found that tachystatin A shows a weak amino acid sequence homology (22%) with -aga IVA (10), a P-type calcium channel (11-12) blocker found in the venom of the funnel web spider (A. aperta), as shown in Fig. 4A. Despite the low amino acid sequence homology between tachystatin A and -aga IVA, tachystatin A has also been shown to share a remarkable structural similarity with -aga IVA, as shown in Fig. 4B. In tachystatin A, there are three -strands: Phe⁹-Val¹² (strand I), Thr²⁸-Arg³⁰ (strand II), and Gly³⁹-Gln⁴² (strand III). These -strands correspond to Gly¹⁰-Cys¹², Gly²⁴-Cys²⁷, and Cys³⁴-Lys³⁷ in -aga IVA, respectively. Regarding -turns, tachystatin A has two -turns that involve the residues Leu⁶-Phe⁹ (type IV) and Cys²⁴-Leu²⁷ (type II), whereas in the case of -aga IVA, there are three -turns. The first turn in tachystatin A corresponds to residues 7-10 (type II -turn) in -aga IVA, and the second turn corresponds to residues 20-23 (type IV), although there is no rigid turn corresponding to the remaining turn of -aga IVA (13-16 residues), and this site is as disordered as the N-terminal region in tachystatin A. The disulfide bonds of Cys⁴-Cys²⁴, Cys¹¹-Cys²⁹, and Cys²³-Cys⁴¹ in tachystatin A correspond to Cys⁴-Cys²⁰, Cys¹²-Cys²⁵, and Cys¹⁹-Cys³⁶ in -aga IVA, respectively (Fig. 4A), although the disulfide bond Cys²⁷-Cys³⁴ of -aga IVA is lacking in tachystatin A. Three identical disulfides might be a major factor in the similarity between the solution structures of tachystatin A and -aga IVA.

View larger version (17K): [in this window] [in a new window]   Fig. 4.   Structure comparisons of tachystatin A and -aga IVA. A, an amino acid sequence alignment of tachystatin A and -aga IVA. Consensus amino acids are represented in small boldface letters, and the conserved cysteines are indicated in large boldface letters. Lines show the identical pattern of disulfide bonds in tachystatin A and -aga IVA. B, structure comparison of tachystatin A and -aga IVA. The antiparallel -sheet with a topology of +2x, 1 is identified in both molecules. Disulfide bonds are shown by ball-stick models. The essential residues for chitin binding, Phe9 in tachystatin A and Tyr9 in -aga IVA, are also shown by a ball-stick model. The lowest energy molecule of 20 converged structures was used for tachystatin A, and the -aga IVA energy-minimized structure was extracted from the Brookhaven Protein Data Bank, entry code 1OAV (29). These images were generated by the program MOLMOL (25).

Kim et al. (30) have reported that the active site of -aga IVA as a calcium channel blocker is the highly hydrophobic C-terminal region with no secondary structures. They have also suggested that the globular structure of the molecule might be essential for the high affinity binding of calcium channel antagonists to channel receptor sites. Although tachystatin A has the same -sheet topology as -aga IVA, a region corresponding to the active site as a calcium channel antagonist of -aga IVA is lacking. A recent study on calcium channel blocking has clarified that tachystatin A does not have the properties of a calcium channel antagonist (9); this is a reasonable result in light of the structure-function relationship.

To determine whether -aga IVA has antimicrobial and chitin-binding activities, the corresponding experiments were carried out for each activity. Tachystatin A represents strong antimicrobial activity to a fungus (P. pastoris) (9). The 50% inhibitory concentration (IC₅₀) of -aga IVA for the fungus P. pastoris was found to be 7.8 µg/ml. Although this value is higher than that of tachystatin A (0.5 µg/ml) (9), the antifungal activity was represented clearly. In addition, the concentration of -aga IVA required for 50% binding to chitin was found to be 8.1 µM, which is nearly equivalent to that of tachystatin A (8.4 µM) (9). These results are summarized in Table II. These findings suggest that -aga IVA and tachystatin A have similar characteristics as antifungal/chitin-binding peptides and that a -sheet with a topology of +2x, 1 may be an essential factor for antifungal/chitin binding activity.

View this table: [in this window] [in a new window]   Table II Antifungal and chitin binding activities of tachystatin A and -aga IVA

	DISCUSSION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

We here determined the three-dimensional structure of the antimicrobial peptide tachystatin A by means of ¹H NMR measurements and distance geometry-simulated annealing calculations. The structure has an antiparallel triple-stranded -sheet with a topology of +2x, 1 as a dominant secondary structure and three disulfide linkages. Many previous reports of protein structures with this topology have indicated that the folding is a versatile scaffold for molecular stabilization. In the case of antimicrobial substances, the mammalian - and -defensins have a triple-stranded -sheet with +2x, 1 as a common structural feature (31-33). To our knowledge, the present report is the first to compare the structural homology of invertebrate and vertebrate antimicrobial peptides. Generally, the high net positive charge of defensins facilitates their electrostatic interaction with polyanionic surfaces of bacterial cells, but their detailed mechanism has not been characterized yet. Because of the structural similarity between tachystatin A and mammalian defensins, it could be considered that tachystatin A exerts antimicrobial activity in a manner similar to defensins. Tachystatin A is rich in arginine residues and is thus a positively charged peptide. All arginines involved in tachystatin A (Arg³, Arg¹⁴, Arg²⁵, Arg³⁰, Arg⁴⁰, and Arg⁴³) are favorably oriented for solvent accessibility. The molecular surface shown in Fig. 5 indicated that the regions with positive electrostatic potential are concentrated on one side of the -structure regions, with the exception of Arg¹⁴. Contrary to the assembly of charged residues, the opposite side of the -structures is occupied by a hydrophobic surface widely constructed with disordered loops, Tyr¹⁶-Pro²² (loop 1) and Tyr³²-Thr³⁷ (loop 2). This amphiphilic character is also found in mammalian defensins and may be favorable when the peptide comes into contact with bacteria and interacts with their membranes. A positive potential surface may play an important role in the interaction with the negatively charged membrane, and the exposed hydrophobic surface may accommodate the binding of tachystatin A to the membrane.

View larger version (32K): [in this window] [in a new window]   Fig. 5.   Surface of tachystatin A colored according to the electrostatic potential. On the -sheet surface (left), the area with positive potential (blue) distributes widely, whereas the neutral potential area occupies the opposite surface (right), where the hydrophobic loop (Ser15-Ile21) is exposed to the solvent. The positively charged residues are labeled. These images were generated by the program MOLMOL (25).

To date, many antimicrobial peptides have been found in invertebrates, and these can be classified into three categories on the basis of their structures: (a) linear or -helical peptides without disulfide bonds, (b) peptides with a high content of one or two kinds of amino acids, and (c) cysteine-rich peptides (34). Cysteines are rich in the antimicrobial peptides found in the Japanese horseshoe crab, including tachystatin A, and these peptides can be classified as cysteine-rich antimicrobial peptides (2). Noticeable structural motifs of cysteine-rich antimicrobial peptides isolated from invertebrates include the hairpin-like -sheet motif and the cysteine-stabilized - motif (35). Although tachystatin A belongs to the family of cysteine-rich peptides of invertebrate antimicrobial substances, no structural homology with tachystatin A has been identified in the cysteine-rich peptide family. Whereas few structural homologies have been identified between tachystatin A and other invertebrate antimicrobial peptides, it has been revealed that tachystatin A shows a significant structural similarity to -aga IVA, a calcium channel blocker found in the venom of the funnel web spider (9). Surprisingly, our results showed that -aga IVA also has the antifungal activity, although it was weaker than that of tachystatin A (Table II). Because -aga IVA does not form a clear amphiphilic conformation, its antifungal activity might appear to be weaker than that of tachystatin A.

The horseshoe crab is a close relative of spiders in evolutionary history. Thus, both tachystatin A and -aga IVA are considered to have evolved from the common ancestor of ancient arthropods as antimicrobial substances. It is assumed that -aga IVA is a more developed molecule than tachystatin A because -aga IVA affects the nervous system but tachystatin A does not, and it is thought that the ancestral antimicrobial peptide added the activity of the calcium channel antagonist to its function. In contrast, it could be considered that tachystatin A did not need to develop such a function because of its life circumstances, given that the horseshoe crab has not developed for a long time. It is thus considered that tachystatin A has conserved the original features of the antimicrobial peptide.

Tachystatin A has the chitin-binding property, which has also been observed in other proteins in small granules of T. tridentatus hemocytes. This study suggests that the molecular folding observed in tachystatin A marks it as belonging to a newly identified structural family of antimicrobial and chitin-binding peptides, because the structure of tachystatin A shows no structural homology with the known chitin-binding motif. As a characteristic motif of the peptides with both chitin-binding and antimicrobial activities, the hevein domain (36, 37) is widely known. Hevein (38) is a chitin-binding/antimicrobial peptide isolated from rubber tree (Hevea brasiliensis) latex, and its chitin-binding domain includes a cysteine-stabilized antiparallel -sheet and a helical turn. Although it has been proposed that this domain is involved in plant peptides, it has also been reported that some invertebrate chitinases involve this domain and conserve its structural motif (39-43). Recently, in the first report on the three-dimensional structure of an invertebrate chitin-binding short peptide (73 residues), Suetake et al. (14) determined the three-dimensional structure of tachycitin (6), which is an antimicrobial peptide with the chitin-binding property found in the hemocytes of the Japanese horseshoe crab. Tachycitin contains the hevein domain as a chitin-binding domain. Although tachystatin A and tachycitin are derived from the same source, no structural similarities have been found between them. The chitin affinity and antifungal activity of tachystatin A are stronger than those of tachycitin, although tachystatin A does not have the hevein domain. Thus, the structural motif of tachystatin A may be more favorable for the chitin binding and antifungal activity than those of the hevein domain. Furthermore, -aga IVA with a structural feature common to tachystatin A also shows an equivalent level of chitin-binding activity. Based on the results of the chitin-binding assay of tachystatin A and -aga IVA (Table II), both peptides are presumed to have a similar manner of binding to chitin.

There have been numerous reports that the exposed side chain of the aromatic residue plays an important role in the recognition of polysaccharides. Because only two aromatic residues are contained in -aga IVA, we can select the aromatic residues from both peptides, Tyr⁹ and Trp¹⁴ of -aga IVA and the corresponding Phe⁹ and Tyr¹⁶ of tachystatin A, as essential factors for the recognition of chitin, based on a comparison of the three-dimensional structures of tachystatin A and -aga IVA. Phe⁹ of tachystatin A and Tyr⁹ of -aga IVA are located on the -turn, and Tyr¹⁶ of tachystatin A and Trp¹⁴ of -aga IVA are located on the disordered loop region. In the case of the cellulose- and xylan-binding proteins, including the modular structure known as the carbohydrate-binding module, the aromatic ring of tryptophan on the -turn plays a key role in the hydrophobic interaction with the nonpolar face of a sugar ring (44-46). Furthermore, a previous study on the carbohydrate-binding module demonstrated that tryptophan residues form a planar surface ideally placed to interact with the flat surface of cellulose (47). In this study, the structures of tachystatin A and -aga IVA showed clearly that the side chains of Phe⁹ of tachystatin A and Tyr⁹ of -aga IVA form a planar surface (Fig. 4B). In addition, chitin polymer has a flat surface like that of cellulose. If tachystatin A and -aga IVA recognize chitin in a manner similar to that by which the carbohydrate-binding module recognizes carbohydrates, then Phe⁹ of tachystatin A and Tyr⁹ of -aga IVA would be essential residues for the binding to chitin.

Tachystatin A cannot bind to chitin oligomer but binds to chitin polymer, suggesting that an additional amino acid may be needed for strong binding to chitin. Although the aromatic residues in the disordered region Tyr¹⁶ of tachystatin A and Trp¹⁴ of -aga IVA could not be ruled out as being related to the binding of chitin in this study, these residues might also be important for the chitin-binding property. In the future, the chitin-binding mechanism of tachystatin A will be revealed in detail by experiments using the mutants altered from Phe⁹ and Tyr¹⁶ to nonaromatic residues. The causation between the antimicrobial and chitin binding activities of the newly identified antimicrobial/chitin-binding structure observed in tachystatin A will also be clarified.

	FOOTNOTES

^* This work was supported in part by the Center of Excellence Research Program of the Science and Technology Agency and grants-in-aid for scientific research from the Ministry of Education, Science, and Culture of Japan (10680625); by a grant from the Integrated Research Program for the Development of Insect Technology of the Ministry of Agriculture, Forestries, and Fisheries of Japan; and by the Program for Promotion of Basic Research Activities for Innovative Biosciences (Japan).

The assignment data have been deposited in the BioMagResBank (BMRB), a National Institutes of Health-funded bioinformatics resource (Department of Biochemistry, University of Wisconsin, Madison, WI; www.bmrb.wisc.edu) under accession code 5268.

^** To whom correspondence should be addressed: Division of Biological Sciences, Graduate School of Science, Hokkaido University, Sapporo 060-0810, Japan. Tel.: 81-11-706-2773; Fax: 81-11-706-2771; E-mail: nitta@sci.hokudai.ac.jp.

Published, JBC Papers in Press, April 16, 2002, DOI 10.1074/jbc.M111120200

	ABBREVIATIONS

The abbreviations used are: -aga IVA, -agatoxin IVA; r.m.s., root mean square; NOE, nuclear Overhauser effect; NOESY, NOE spectroscopy; DQF-COSY, double quantum-filtered correlation spectroscopy; TOCSY, total correlation spectroscopy; E. COSY, exclusive correlation spectroscopy; DANTE, delays alternating with nutation for tailored excitation.

	REFERENCES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

CiteULike

Complore

Connotea

Del.icio.us

Digg

Reddit

Technorati

This Article Abstract Full Text (PDF) All Versions of this Article: 277/26/23651    most recent M111120200v1 Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via Google Scholar Google Scholar Articles by Fujitani, N. Articles by Kawano, K. Search for Related Content PubMed PubMed Citation