Structure of the Insect Cytokine Peptide Plasmatocyte-spreading Peptide 1 from Pseudoplusia includens *

The structure of the recently identified plasmatocyte spreading peptide from the moth Pseudoplusia includens (PSP1) has been determined by NMR spectroscopy. This novel insect cytokine consists of 23 amino acid residues and a single disulfide bond. Torsion angle dynamics calculations utilizing a total of 337 distance constraints yielded an ensemble of 30 structures with an average backbone root mean square deviation for residues 7–22 of 0.18 Å from the mean structure. The structure consists of a disordered N-terminal region and a well defined core that is stabilized by numerous hydrophobic interactions and a short β-hairpin. Structural comparisons confirm that PSP1 adopts an epidermal growth factor (EGF)-like fold with close similarity to the C-terminal subdomain of EGF-like module 5 of human thrombomodulin. The combination of the three-dimensional structure of PSP1 and the extensive literature on EGF-receptor interactions should accelerate the process of identifying the specific residues responsible for receptor binding activity of this family of immunoregulatory peptides.

The insect immune response to parasites and pathogens involves the action of different classes of blood cells (hemocytes), which adhere to and spread across the surface of foreign targets. In phylogenetically advanced insects like the Lepidop-tera (moths and butterflies), plasmatocytes and granular cells are the two most important classes of hemocytes involved in cellular defense responses like encapsulation and clotting (3). However, relatively little is known about the factors mediating the movement and action of these immune cells (3). The identification of a peptide that induces the adhesion and spreading of plasmatocytes on foreign surfaces was recently reported (1). This plasmatocyte-spreading peptide (PSP1) 1 of Pseudoplusia includens is expressed as a preproprotein of 142 residues (2), but the mature peptide consists of only the C-terminal 23 amino acids containing one disulfide bond. Although the existence of cytokine-like factors to regulate hemocyte activity has been proposed previously (3), PSP1 is among the first soluble mediators of the insect cellular immune response to be purified and functionally characterized.
The amino acid sequence of PSP1 shows no significant similarity to vertebrate cytokines, but is related to two classes of insect peptides previously identified (1). As shown in Fig. 1, the growth-blocking peptide (GBP) of Pseudaletia separata and the paralytic peptides from Manduca sexta, Heliothis virescens, and Spodoptera exigua have high (Ͼ70%) sequence identity with PSP1, including the two cysteine residues and a highly conserved glycine. High sequence variabilty within this group is found only at positions 4 and 8. Some similarity between the sequences of GBP and human epidermal growth factor has been recently suggested (4), and this would extend to PSP1 and the paralytic peptides.
Chemically synthesized PSP1 was shown to induce an identical plasmatocyte spreading and adhesion response to the peptide purified from the plasma of P. includens larvae and is therefore suitable for detailed structure-function analysis (1). Here we report the three-dimensional structure of this novel insect cytokine determined by NMR spectroscopy and torsion angle dynamics calculations. The structure of PSP1 shows clear homology to the C-terminal subdomain of the EGF domain family. The backbone conformation of PSP1 is very similar to that of the fifth EGF-like domain of the human anticoagulant protein thrombomodulin (hTM5), despite sharing sequence identity at only four positions, including the two cysteines of the conserved disulfide.

EXPERIMENTAL PROCEDURES
Sample Preparation-PSP1 was synthesized using tert-butyloxycarbonyl chemistry and purified by reversed-phase high performance liquid chromatography. Its identity was confirmed by matrix-assisted laser desorption/ionization time of flight mass spectrometry and NMR. The NMR sample consisted of 2 mg of pure lyophilized peptide, dissolved in 500 l of a buffer containing 90% H 2 O/10% D 2 O, 20 mM sodium phosphate, pH 6.0.
NMR Spectroscopy-All NMR spectra were recorded at 10°C on Bruker DMX750 and DMX500 spectrometers equipped with tripleresonance ( 1 H/ 13 C/ 15 N) probes and with ZϪ or three-axis pulsed field gradient capabilities. Quadrature detection in the indirectly detected dimensions was obtained with the States-TPPI method (5). Water suppression was achieved using a watergate sequence with a 3-9-19 selective inversion pulse (6). A two-dimensional total correlation spectroscopy spectrum with an isotropic mixing period of 67 ms was acquired at 499.84 MHz, with spectral widths of 6009. 6 (7). Time domain data in both dimensions were apodized with a squared cosine bell prior to zero filling to the final matrix size, Fourier transformation, and phase correction. The initial value for the incremented delay was set in a manner that allowed predictable phasing in each dimension and minimized roll and offset of the base line (8). All 1 H dimensions were referenced to internal 2,2-dimethyl-2-silapentane-5-sulfonate.
Resonance Assignments and Structure Calculations-Sequential resonance assignments for the backbone were obtained manually from analysis of the total correlation spectroscopy and NOESY spectra (9). Subsequent assignment of side chains and NOEs for structure calculations were obtained with the programs Garant (10) and XEASY (11).
Structures were calculated with the torsion angle dynamics approach of the program DYANA 1.5 (12) using the anneal.dya macro with 4000 steps/structure in the early stages of refinement and 20,000 steps in the final rounds, followed by 1000 steps of conjugate gradient minimization. Distance constraint upper bounds were generated with the CALIBA function of DYANA. For the initial round of structure calculations, sequential and intraresidue NOEs assigned by Garant, as well as a small number of manually assigned additional medium and long range NOEs were used as input restraints. Each round started with 60 randomized conformers, from which the 30 with the lowest target function were used to analyze constraint violations and assign additional NOE constraints (in Garant or XEASY) for the following round of refinement. This process was repeated until all peaks in the spectrum had been assigned and all consistent violations were eliminated. In the later stages of refinement, the GLOMSA routine of DYANA was applied to the set of restraints and ensemble of structures to obtain a small number of stereospecific assignments based on differences in NOE intensities for diastereotopic groups. From the total of 60 structures calculated in the final round of refinement, the 30 conformers with the lowest target function were considered for analysis. The mean structure calculated in MOLMOL (13) from this ensemble of 30 structures was minimized in DYANA with 1000 steps of conjugate gradient minimization and reported as the representative PSP1 structure. All analyses of root mean square deviation, secondary structure, and hydrogen bonds were performed with MOLMOL.

RESULTS AND DISCUSSION
Structure Determination-Two-dimensional 1 H NMR spectroscopy was used to obtain complete resonance assignments for chemically synthesized plasmatocyte-spreading peptide 1 (PSP1). 2 As shown in Fig. 2, the chemical shift dispersion and large number of NOEs observed indicate the presence of well defined secondary and tertiary structure. A total of 699 crosspeaks were assigned in the two-dimensional NOESY spectrum of PSP1, from which 333 nontrivial NOE distance constraints were produced after discarding duplicate and structurally re-dundant NOEs. Two additional pairs of upper and lower bound distance constraints were included between the S ␥ and C ␤ atoms of Cys 7 and Cys 19 to produce normal disulfide bond geometry (2.0 Å Ͻ d SϪS Ͻ 2.1 Å, 3.72 Å Ͻ d C␤ϪC␤ Ͻ 3.99 Å). A summary of all 337 distance constraints by class is given in Table I, and the distribution of constraints by residue is shown in Fig. 3. Stereospecific assignments were obtained from analysis of structures and NOE intensities for the H ␣ protons of Gly 17 , the H ␤ protons of Tyr 11 and Pro 21 , and the H ␥ and H ␦ protons of Pro 21 .
The ensemble of 30 PSP1 conformers, which resulted from the final stage of refinement, is shown in Fig. 4. The low average target function (0.12 Ϯ 0.02 Å 2 ) indicates good agreement between the ensemble of calculated conformers and the experimental constraints. Superposition of the well ordered residues (7-22) of this ensemble onto the corresponding residues of the mean structure produced an average RMSD to the mean of 0.18 Å for the backbone atoms and 0.84 Å for all nonhydrogen atoms. After minimization of the mean structure in DYANA to a final target function of 0.13, the average RMSD for residues 7-22 became 0.23 Å for the backbone and 0.98 Å for the heavy atoms. A summary of residual violations and RMSDs is given in Table I.
The tertiary structure of PSP1 is stabilized by a combination of the covalent disulfide linkage, hydrogen bonding within the ␤-hairpin structure, and hydrophobic side chain packing. All 30 torsion angle dynamics conformers were consistent with a pair of hydrogen bonds involving the backbone amide and carbonyl groups of Met 12 and Lys 20 . Evidence for two additional hydrogen bonds (Asp 16 H N -Thr 14 O ␥1 , Thr 14 H N -Arg 18 O) was observed in a subset of the family of structures. Within the ordered portion of the molecule, a subset of side chains forms a well defined hydrophobic core centered around the side chain of Tyr 11 . This residue, conserved throughout the family of EGFlike domains as a phenylalanine or tyrosine (14), has NOE contacts to the side chains of Cys 7 , Leu 8 , Ala 9 , Cys 19 , Pro 21 , and Phe 23 . The opposite side of the structure is largely composed of exposed and disordered charged residues, including Arg 13 , Asp 16 , Arg 18 , and Lys 20 .
As seen in Fig. 1, the sequences of PSP1, GBP from Pseudaletia separata, and a paralytic peptide from M. sexta differ in FIG. 1. Comparison of the sequences of PSP1 from P. includens, GBP from P. separata, paralytic peptide from M. sexta  (PP1), and the C-terminal subdomains of human epidermal growth factor (hEGF) and the hTM5. Highlighted in boldface type are the cysteine, glycine, and tyrosine residues conserved in the EGFlike domain family and the corresponding residues of the insect peptides. very few positions, and all may be expected to adopt the same three-dimensional structures. Although the structures of these related peptides have not been reported, preliminary analysis of the NOESY spectrum of GBP indicates that its structure will be very similar to that reported here for PSP1. 3 Consistent with this result are functional assays using both PSP1 and GBP that show no significant differences in activity. 4 Comparison with Epidermal Growth Factor Modules-EGFlike domains are peptides of approximately 40 residues in length, containing six highly conserved cysteine residues. Three loops, designated A, B, and C, are defined by the pattern of disulfide bonding. EGF modules are alternatively described in terms of two subdomains, each containing a ␤-hairpin structure. The N-terminal EGF-like subdomain consists of the first two disulfide-linked loops, and the smaller C-terminal subdomain consists of the third disulfide and the C loop.
The structures of human epidermal growth factor (hEGF) (15) and the hTM5 (16), which have been determined by NMR spectroscopy, share the general structural features of the family of EGF-like domains. The variability of the EGF family is illustrated by the differences between these two structures, including alternative pairing for the first two disulfides and a longer C loop in hTM5 than hEGF (16). Sequence identity between PSP1 and these representatives of the family of EGFlike domains is limited to the essential cysteine, glycine, and tyrosine residues of the C-terminal EGF-like subdomain (Fig.  1), but PSP1 has an extended C loop as found in hTM5. The backbone conformations of these three peptides are compared in Fig. 5. Except for the displacement of the ␤-turn in hEGF, resulting from its shorter C loop, the structures of all three molecules are quite similar. PSP1 and hTM5 align to a remarkably high degree, with an average backbone atomic RMSD of ϳ1 Å for the 13 residues within the disulfide loop.
Implications for Receptor Binding-PSP1 was isolated as the peptide responsible for spreading of insect cell plasmatocytes on foreign surfaces, an important process in cellular defense. Hence, it will be of interest to determine the cellular receptor for this peptide. Given the striking similarity between the structure of PSP1 and the C-terminal subdomain of EGF-like modules, it is tempting to speculate that the ligand-receptor interactions of PSP1 may be analogous to those of the array of EGF domain-receptor complexes that have been investigated (17).   ing all side chains. The charged residues Arg 13 , Asp 16 , Arg 18 , and Lys 20 are clearly clustered on one side of the molecule, and the patch of hydrophobic residues surrounding the Tyr 11 side chain are located on the opposing side. No consensus receptorbinding site has been identified yet for EGF receptor ligands, despite intense efforts (17) stemming from the relevance of this system to the design of potential drugs for treating human cancers. However, the studies have determined that an argi-nine conserved in EGF receptor binding domains (Arg 41 in hEGF) is essential for binding. Interestingly, Arg 18 is the analogous residue in PSP1.
Significant effort has been spent investigating the details of thrombin recognition by thrombomodulin, including alaninescanning mutagenesis, NMR spectroscopy, and x-ray crystallography of the complexes between small peptide fragments of hTM5 and thrombin (18 -20). Although the distinction between residues critical for proper folding and those directly involved in binding may not be entirely clear from mutagenesis alone, a structural interpretation clearly implicates the C loop of hTM5, including the many acidic side chains on either side of Cys 421 (Fig. 1). None of these residues has a direct correlation, but the cluster of charged side chains occurs in the same vicinity of the PSP1 structure. In the absence of other information on the activity of PSP1, this concentration of acidic and basic residues is an obvious starting point for investigations of the receptorbinding determinants of this new class of insect cytokine peptides.