Innate Immunity in a Pyralid Moth

Invertebrates, like vertebrates, utilize pattern recognition proteins for detection of microbes and subsequent activation of innate immune responses. We report structural and functional properties of two domains from a β-1,3-glucan recognition protein present in the hemolymph of a pyralid moth, Plodia interpunctella. A recombinant protein corresponding to the first 181 amino-terminal residues bound to β-1,3-glucan, lipopolysaccharide, and lipoteichoic acid, polysaccharides found on cell surfaces of microorganisms, and also activated the prophenoloxidase-activating system, an immune response pathway in insects. The amino-terminal domain consists primarily of an α-helical secondary structure with a minor β-structure. This domain was thermally stable and resisted proteolytic degradation. The 290 residue carboxyl-terminal domain, which is similar in sequence to glucanases, had less affinity for the polysaccharides, did not activate the prophenoloxidase cascade, had a more complicated CD spectrum, and was heat-labile and susceptible to proteinase digestion. The carboxyl-terminal domain bound to laminarin, a β-1,3-glucan with β-1,6 branches, but not to curdlan, a β-1,3-glucan that lacks branching. These results indicate that the two domains of Plodia β-1,3-glucan recognition protein, separated by a putative linker region, bind microbial polysaccharides with differing specificities and that the amino-terminal domain, which is unique to this class of pattern recognition receptors from invertebrates, is responsible for stimulating prophenoloxidase activation.

The Plodia interpunctella ␤GRP (Pi␤GRP) was originally identified as a soluble 53-kDa plasma protein that binds ␤-1,3glucan, lipopolysaccharide, and lipoteichoic acid and subsequently activates the PPO cascade (12). A cDNA clone encoding Pi␤GRP was isolated and found to belong to a family of PRRs identified from Manduca sexta (␤GRPs 1 and 2), Bombyx mori (␤GRP and GNBP), Drosophila melanogaster (GNBPs 1-3), Hyphantria cunea (GNBP), and Eisenia foetida (coelomic cytolytic factor). Pi␤GRP is expressed in fat body and is constitutively present in the hemolymph. In this study, we describe structural and functional properties of recombinant proteins corresponding to putative amino-and carboxyl-terminal domains from Pi␤GRP.

Expression of Recombinant ␤GRP and Deletion Constructs-
The fulllength 471-residue ␤GRP (without secretion signal peptide) was expressed using an Invitrogen pTrcHis2-TOPO® expression vector in Escherichia coli and purified as outlined previously (12). Two aminoterminal deletions mutants composed of 118 and 181 residues that corresponded to amino-terminal domains reported in B. mori and M. sexta, respectively (8,9), were constructed (Fig. 1A). The 118 and 181 residue recombinant proteins are referred to as 118N and 181N, respectively. A carboxyl-terminal domain (290C) consisting of 290 residues was produced similarly. All of the recombinant proteins were expressed as fusion proteins with a c-myc and a six-histidine tag at their carboxyl terminus comprising 32 additional residues.
Circular Dichroism Spectroscopy-Circular dichroism (CD) spectroscopy was conducted by using a Jasco J-720 spectropolarimeter. All of the proteins were diluted to 3-4 M in 10 mM Tris-HCl, pH 8.0, containing 0.1% Triton X-100 and scanned from 260 to 180 nm at 20-nm min Ϫ1 . The mean molar residue ellipticity () values were calculated by inputting the raw CD data into the equation () ϭ /(10 ϫ l ϫ C r ), where l is the path length in centimeters and C r is the molar concentration of the protein multiplied by number of residues.
Proteolysis and Thermal Stability of Recombinant Proteins-Recombinant P. interpunctella ␤GRP (r␤GRP), 118N, 181N, and 290C were digested with equal amounts (2 g) of bovine trypsin or chymotrypsin in 20 mM Tris-HCl, pH 8.5, for 1 h at 25°C. Following incubation, samples were heat-treated in SDS sample buffer to inactivate the proteinase and separated by SDS-PAGE. In a second test, the proteolysis of r␤GRP through time was observed by incubating 7.2 g of r␤GRP with 2.9 g of trypsin or chymotrypsin at 25°C in 20 mM Tris-HCl, pH 8.5, in a total volume of 120 l. At appropriate time intervals (0, 0.01, 0.5, 1, 3, 6, and 24 h), a 20-l aliquot was removed, added to SDS sample buffer, and heated in a 70°C water bath for 10 min to inactivate the proteinases. In both experiments, proteins were transferred to nitrocellulose membranes after SDS-PAGE and immunoblot analysis was conducted using rabbit antiserum against M. sexta ␤GRP2.
Heat stability was determined by incubating an aliquot of each recombinant protein diluted to 0.12 mg ml Ϫ1 in 10 mM Tris-HCl, pH 8, containing 0.1% Triton X-100 in a boiling water bath for 10 min. Heat-treated samples were centrifuged at 10,000 ϫ g for 5 min at 4°C, and supernatants containing heat-stable protein were collected. Precipitates were resuspended and centrifuged. After an additional wash, pellets containing heat-labile proteins were resuspended in 125 l of the Tris buffer and analyzed as above. Curdlan binding activity of heat-treated recombinant Pi␤GRP and domains was determined as described below.
Aggregation of Microorganisms-Recombinant proteins were tested for agglutination of fluorescein-labeled E. coli (K12 strain), Staphylococcus aureus (Wood strain without protein A), and Saccharomyces cerevisiae (12). Samples containing BSA or buffer alone were also analyzed. Multiple images obtained with a SPOT digital camera from Diagnostic Instruments (Sterling Heights, MI) were used to determine the degree of aggregation.
Binding of Recombinant ␤GRP and Domains to Curdlan-Curdlan (an insoluble ␤-1,3-glucan, Sigma) was used as an affinity matrix to assess ␤-1,3-glucan binding of recombinant proteins. Proteins (10 g of r␤GRP, 15 g of 118N, 12 g of 181N, and 20 g of 290C) were incubated with 10 mg of curdlan equilibrated in phosphate-buffered saline (8.1 mM Na 2 HPO 4 , 1.5 mM KH 2 PO 4 , 2.7 mM KCl, and 137 mM NaCl, pH 7.4) for 0.5 h on ice with occasional mixing. The curdlanprotein mixture was centrifuged at 10,000 ϫ g for 5 min at 4°C, and the pellet was washed. Bound proteins were eluted from the curdlan precipitate by the addition of 50 l of 1 ϫ SDS-NuPAGE sample buffer (Novex, San Diego, CA) containing 50 mM dithiothreitol followed by heating for 10 min at 95°C. The slurry was centrifuged at 10,000 ϫ g for 5 min at 4°C, and equal volumes of supernatants were analyzed by SDS-PAGE with Coomassie Blue protein staining and by immunoblotting.
Competition for binding curdlan by r␤GRP in the presence of laminarin was tested by incubating 30 g of recombinant protein with or without laminarin (0.5 mg) prior to the addition to curdlan (10 mg in 0.5 ml of phosphate-buffered saline). Following incubation for 30 min on ice with occasional mixing, curdlan was precipitated by centrifugation and the precipitates were washed. Protein was eluted from curdlan by heat treatment in SDS sample buffer, and equivalent amounts of supernatants were analyzed.
Binding of Recombinant ␤GRP and Domains to Laminarin-Laminarin (a soluble ␤-1,3-glucan, Sigma) was biotinylated by following the protocols modified from Takaki et al. (39) and Shinohara et al. (40). 10 tubes, each containing 26 nmol of laminarin and 52 nmol of biotinbiotinamidocaproyl hydrazide-hydrazide in 30% CH 3 CN in a total of 10 l, were incubated at 90°C for 2 h. Samples were pooled and incubated with 0.5 ml of UltraLink immobilized streptavidin in binding buffer (0.1 M NaH 2 PO 4 /Na 2 HPO 4 and 150 mM NaCl, pH 7) at 4°C for 1 h. The resin was centrifuged at 2,500 ϫ g for 1 min, washed, and then suspended in 0.5 ml of buffer. Each recombinant protein sample (40 g) was added to the laminarin-affinity resin (50 l) and incubated at 4°C for 0.5 h. The resin was washed three times, resuspended in 50 l of 1ϫ SDS-PAGE sample buffer, and placed into a boiling water bath for 10 min. The eluted proteins were analyzed by SDS-PAGE with Coomassie Blue protein staining and immunoblot analysis. The specificity of the binding to laminarin (and not the affinity resin itself) was tested by preparing immobilized-streptavidin resin with biotin-biotinamidocaproyl hydrazide-hydrazide but without laminarin as the ligand and assaying as above.
Affinity Constant Determination via Surface Plasmon Resonance (SPR)-A BIAcore® 3000 SPR biosensor (BIAcore AB, Uppsala, Sweden) was used to measure kinetic interactions between laminarin and the recombinant proteins. Laminarin (195 nmol) was biotinylated and injected directly onto a streptavidin-coated sensor chip SA (BIAcore AB). As a control, biotin-biotinamidocaproyl hydrazide-hydrazide was similarly injected onto a different flow cell of the SA sensor chip to account for nonspecific interactions. A series of concentrations of r␤GRP, 118N, 181N, and 290C ranging from 0.5 nM to 2 M prepared in 10 mM HBS-P (10 mM HEPES, pH 7.4 with 150 mM NaCl and 0.005% Tween 20) was injected over the surface of the sensor chip at a flow rate of 15 l min Ϫ1 . Proteins were injected at the same rate onto both the control (immobilized biotin) and test (immobilized laminarin) flow cells at 25°C over 6 min, and the interaction was monitored as the change in SPR response. After sample injections, flow cells were washed with HBS-P buffer and dissociation of analyte (protein) from the immobilized ligand was monitored continuously over 15 min at a flow rate of 15 l min Ϫ1 . At the end of each cycle, the regeneration of the chip was accomplished with two successive 30-s washes of 50 mM H 3 PO 4 at a flow rate of 30 l min Ϫ1 . The chip was equilibrated with HBS-P prior to the next analysis. Both association rate constants (k a ) and dissociation rate constants (k d ) for the laminarin:protein interactions were obtained by fitting background-corrected SPR binding data to the 1:1 Langmuir binding model within the BIAevaluation software version 3.1. The association constant (K A ) was subsequently determined as k a /k d . The statistical analysis of association constants for each protein was performed by using the SAS PROC ANOVA at ␣ ϭ 0.05 (41).
Binding of Recombinant ␤GRP and Domains to Lipopolysaccharide and Lipoteichoic Acid-Binding to LPS and LTA was determined by using microplate binding assays modified from Tobias et al. (42) and Yu

FIG. 1. Expression of P. interpunctella ␤GRP and domains in E. coli.
A, ␤GRP truncation constructs expressed in E. coli as fusion proteins in-frame with a c-myc and a six-histidine tag at the carboxyl terminus of each recombinant protein. Each recombinant protein contains an additional 32 residues: three residues at the amino terminus (Met-Ala-Leu) and 29 residues at the carboxyl terminus (comprising the c-myc tag, the six-histidine tag, and linker regions). The approximate locations of amino-terminal carbohydrate recognition domain, putative linker region, and carboxyl-terminal glucanase-like domain are shown. r␤GRP comprised the entire mature polypeptide (471 amino acid residues) excluding the putative secretion signal peptide. 118N and 181N consisted of the first 118 and 181 amino-terminal residues of the mature ␤GRP, respectively. The carboxyl-terminal construct, 290C, consisted of the carboxyl-terminal 290 amino acid residues of ␤GRP beginning with Val-182. B, recombinant proteins (r␤GRP, 118N, 181N, and 290C) expressed in E. coli. Proteins were separated by SDS-PAGE and analyzed by Coomassie Blue staining (lanes 1) and immunoblot analysis (lanes 2). Protein molecular weight standards are shown in lane M.
and Kanost (22,27). LPS (from E. coli 0111:B4, smooth strain LPS, Sigma) or LTA (from S. aureus) suspensions (40 g ml Ϫ1 ) were used to coat wells of a microplate (Continental Lab Products, San Diego, CA). After blocking, r␤BRP, 118N, 181N, and 290C at concentrations of 0, 0.024, 0.047, 0.094, 0.19, 0.28, 0.47, and 0.94 M in TB (50 mM Tris-HCl, pH 8, with 50 mM NaCl) containing 0.1 mg ml Ϫ1 BSA were added to LPS-or LTA-coated wells and incubated for 3 h at 25°C. After washing and incubation with rabbit antiserum made against M. sexta ␤GRP2, wells were treated with secondary antibody (goat anti-rabbit IgG conjugated with alkaline phosphatase, Bio-Rad). Alkaline phosphatase activity against p-nitrophenyl phosphate, which is proportional to LPS or LTA binding activity, was detected at 405 nm. To validate the specificity of binding to LPS or LTA, 5 g ml Ϫ1 r␤GRP or 181N was preincubated with 0, 0.1, 1, 10, 50, 100, 250, or 500 g ml Ϫ1 free competitor (LPS or LTA). 50 l of the solution was added to wells precoated with LPS or LTA and assayed.
Activation of the PPO Pathway in the Presence of Laminarin-A method described previously by Ma and Kanost (8) and Fabrick et al. (12) was used to test ability of the recombinant proteins to activate M. sexta PPO in the presence of soluble ␤-1,3-glucan (laminarin). Protein samples (0.5 M) were incubated with 150 g of laminarin and mixed with 10 l of M. sexta plasma. Phenoloxidase activity was determined by measuring the absorbance at 490 nm. Values represent the means Ϯ S.E. of data from three sets of triplicate measurements on a single pooled plasma sample collected from five M. sexta 5 th Instar (day 3) larvae. PPO activation was analyzed for significance by using the SAS PROC Student's t test at ␣ ϭ 0.05 (41).

Production of Recombinant Full-length and Truncated
␤GRP-Pi␤GRP is 471 amino acids long with an amino-terminal region of ϳ120 residues, a putative linker region, and a carboxyl-terminal region (ϳ220 residues) that is similar in sequence to ␤-1,3-glucanases (12). The amino-terminal portion had no significant sequence similarity to any proteins other than the amino terminus of proteins that belong to the ␤GRP/ GNBP family. The structural and functional properties of the P. interpunctella ␤GRP were analyzed by producing a fulllength r␤GRP, two amino-terminal truncation mutants (118N and 181N) that lack the glucanase-like domain, and a carboxylterminal truncation mutant (290C) that includes only the glucanase-like domain (Fig. 1A). Purified recombinant proteins were obtained in milligram quantities from 0.5-1.0 liters of bacterial cultures. SDS-PAGE analysis with Coomassie Blue staining indicated that each recombinant protein solution contained a single major protein band (Fig. 1B). The molecular masses of the recombinant proteins (as estimated by SDS-PAGE) were 54-, 20-, 24-, and 36-kDa for r␤GRP, 118N, 181N, and 290C, respectively.
Structural Properties-r␤GRP and its domains appear to be folded as indicated by CD spectroscopy (Fig. 2) and possess binding and biological activities. CD spectroscopy of 118N and 181N indicates that these proteins contain primarily an ␣-helical secondary structure with a minor ␤-structure. Both proteins exhibit CD spectra reminiscent of a protein containing primarily ␣-helix with a positive absorbance peak near 192 nm and a negative peak near 208 nm. Proteins containing only ␣-helix also possess a second negative absorption peak near 222 nm, which is not clearly evident in the spectra obtained from 118N and 181N, indicating that the recombinant proteins may possess other forms of secondary structure (i.e. ␤-sheet or . r␤GRP was also treated with proteinase for different time periods, separated on SDS-PAGE gels, and transferred to nitrocellulose membranes, and the blots were probed with anti-␤GRP sera. Immunoblots C and D correspond to r␤GRP treated with trypsin and chymotrypsin, respectively. Lanes 1-7 for immunoblots C and D correspond to proteinasetreated r␤GRP at 0, 0.01, 0.5, 1, 3, 6, and 24 h, respectively. r␤GRP 24-h controls without proteinase are shown in lane 8. Trypsin and chymotrypsin controls (without r␤GRP) are shown in lanes T and C, respectively. Lane M contains protein molecular weight standards. turns). The CD spectra of r␤GRP and 290C were complicated and could not be easily interpreted, indicating that these proteins contained a mixture of secondary structure. We believe that the proteins were folded correctly because we could detect various functional activities in both proteins.
The amino-terminal domain of ␤GRP was resistant to both proteolysis and thermal denaturation. The treatment of r␤GRP or 181N with trypsin or chymotrypsin for 1 h produced a 20-kDa protein fragment that cross-reacted with antisera made against M. sexta ␤GRP2 (Fig. 3, A and B). Amino acid sequencing of this fragment revealed a single polypeptide with sequence (AQQYVV) corresponding to very near the amino-terminal end (residues 4 -9) of P. interpunctella ␤GRP (12). The 20-kDa fragment resisted degradation by trypsin for at least 24 h and by chymotrypsin for up to 6 h (Fig. 3, C and D). 118N did not resist degradation and was hydrolyzed to smaller peptides by both proteinases within 1 h. Both trypsin and chymotrypsin completely hydrolyzed 290C within 1 h of treatment.
Heat treatment and recovery of recombinant proteins indicated that the amino-terminal domain of ␤GRP is highly resistant to thermal denaturation (Fig. 4, A and B). Recombinant proteins 118N and 181N both remained in solution after treatment at 100°C for 10 min, suggesting a highly stable conformation. In addition, 118N and 181N retained their ability to bind to the ␤-1,3-glucan, curdlan, following heat-treatment (Fig. 4C). r␤GRP and 290C were found primarily in the heatdenatured pellet, indicating that the carboxyl-terminal domain of ␤GRP is heat-labile.
Aggregation of Microbes-Aggregation of S. cerevisiae, S. aureus, and E. coli was observed in the presence of physio-logical concentrations of r␤GRP (Table I). Incubation with r␤GRP resulted in large multicellular aggregates of each microorganism with very few individual cells visible. In contrast, only a few large aggregates of yeast or E. coli were observed in the presence of 118N and 181N. 290C did not cause significant aggregation of any of the microbes. We previously showed that the agglutination of yeast by r␤GRP was concentration-dependent and reversible in the presence of soluble ␤-1,3-glucan (12). Thus, the full-length r␤GRP is required for substantial microbial aggregation.
We further characterized the binding of the 1,3-glucan recognition protein domains to laminarin by surface plasmon resonance analysis. A single flow cell on a streptavidin-coated sensor chip SA was coated with either biotin (control) or biotinylated-laminarin and subsequently utilized for all of the analyses. Immobilization of biotin onto the control flow cell resulted in a net change of ϳ40 response units. Injection of biotinylated laminarin resulted in a net change of ϳ600 response units that appeared stable, indicating the immobilization of a fixed amount of laminarin. A typical sensorgram for the injection of r␤GRP over the control and bound-laminarin flow cell is shown in Fig. 5D. The binding constants of the recombinant proteins are summarized in Table II. All four recombinant proteins bound to the chip containing immobilized laminarin but did not interact with a CM5 dextran chip surface (data not shown), which consists of only the dextran linker. Corrections for nonspecific binding were made by the subtraction of data obtained from the biotin-control flow cell for each protein. r␤GRP had a higher affinity for laminarin (K A of 3.8 ϫ 10 8 M Ϫ1 ) than the truncated mutants 118N, 181N, or 290C, although the differences were not statistically significant by ANOVA at ␣ ϭ 0.05.
The binding of r␤GRP 2 and 181N to LPS or LTA immobilized in wells of microplates was concentration-dependent and saturable (Fig. 6) Ϫ as free LPS and LTA effectively reduced the binding of the protein to the bound ligand. No significant binding was observed for 118N or 290C with LPS or LTA (data not shown). However, it is difficult to compare binding between the proteins because the antibody generated against full-length M. sexta ␤GRP2 does not recognize the proteins equally. Note that in Fig. 5A, unequal amounts of protein were loaded to achieve similar levels of detection on immunoblots.
Activation of Prophenoloxidase-r␤GRP, 118N, and 181N stimulated PPO activation when added to M. sexta plasma, but a significantly greater (ϳ2.5-fold) stimulation of PPO activation occurred when 1 mg ml Ϫ1 laminarin was included in the reaction mixture with these proteins (Fig. 7). 290C did not activate PPO with or without laminarin. Laminarin alone or with bovine serum albumin did not result in significant PPO activation (although a longer incubation time with laminarin alone did stimulate PPO activation). No additive or synergistic effects were observed in samples that contained both 181N and 290C.

DISCUSSION
PRRs play an essential role in the activation of invertebrate innate immune pathways including those regulated by proteinase cascades such as the horseshoe crab-clotting reaction, the  Toll pathway in Drosophila, the prophenoloxidase-activating system in insects and other arthropods, and the processing of a cytokine that causes the spreading of insect hemocytes (43)(44)(45)(46)(47). These signaling pathways are ancient in origin and conserved in many organisms. The horseshoe crab blood-clotting pathway is controlled by a proteinase cascade in which the PRRs possess both carbohydrate recognition domains, which bind to PAMPs, and proteinase domains that are activated autocatalytically (50). The PPO pathway involves the proteolytic activation of prophenoloxidase to phenoloxidase, which results in the production of melanin or other compounds that function in protective and/or signaling capacities. Various PAMPs, including ␤-1,3-glucan can trigger the activation of insect PPO (48 -55). PRRs including ␤GRPs, peptidoglycan recognition proteins, and C-type lectins have been shown to stimulate PPO activation (1). In Drosophila, the antifungal peptide, drosomycin, is up-regulated by the action of the Toll pathway and requires the extracellular proteinase processing of spaetzle, a ligand for Toll. A peptidoglycan recognition proteins has been shown recently to activate this pathway in the presence of Gram-positive bacteria (but not fungi) (56). The mechanism by which PRRs, bound to PAMPs, activate proteinases remains unknown.
Ochiai and Ashida (9) and Ma and Kanost (8) found that polypeptides corresponding to the amino-terminal domain of two lepidopteran ␤GRPs bound to curdlan (an insoluble ␤-1,3glucan), whereas the carboxyl-terminal domain did not. Similarly, we demonstrated that a recombinant protein corresponding to the amino-terminal domain of P. interpunctella ␤GRP (the first 118 residues) could bind to curdlan, whereas no binding was detected with the carboxyl-terminal glucanase-like domain. However, the glucanase-like domain of P. interpunctella ␤GRP has an affinity to laminarin (a soluble ␤-1,3-glucan) as determined by affinity precipitation and SPR. Although laminarin and curdlan both contain ␤-(133)-linked D-glucopyranose backbones, the two polysaccharides differ. Laminarin contains ␤-(136) side chain branching (one branch approxi-mately every 20 glucose units) and has an average molecular mass of 7.7ϫ10 3 Da (57). In contrast, curdlan is a linear polymer of ␤-1,3-glucan with no side chain substitutions and is generally a heterogeneous mixture of polymer sizes (average degree of polymerization of 450) (58). It is likely that these structural differences allow for binding by the carboxyl-terminal domain of ␤GRP to laminarin but not curdlan. Laminarin and curdlan apparently compete for a common binding site or sites on r␤GRP, because excess laminarin decreased the amount of protein co-precipitated with curdlan.
Some insect PRRs function to opsonize microorganisms within the hemocoel, thereby activating or amplifying cellular immune responses (1,59). Aggregation of microorganisms in vivo may enhance immune signaling pathways or provide superior targets for cellular immunity. In this study, we showed that a full-length r␤GRP was necessary for agglutination of microbes, whereas deletion mutants of the protein were not effective. Aggregation of cells requires that agglutinins have at least two binding sites or can form self-oligomers. Because the truncated protein domains did not agglutinate microbes, they most probably have only one binding site. Although r␤GRP and/or its domains bound to surface compounds of various types of microorganisms and caused aggregation of these microbes, we could not demonstrate any direct anti-microbial activity associated with these proteins (data not shown).
The amino-terminal domain for P. interpunctella ␤GRP has several unique physical characteristics. The first 118 residues of Pi␤GRP were sufficient for binding ␤-1,3-glucan and for the activation of PPO, whereas the glucanase-like domain has less affinity toward some PAMPs and could not activate PPO. Furthermore, the amino-terminal domain is highly resistant to degradation by proteolysis and is resistant to heat denaturation. The folded structure of this domain must be extremely stable. Although 181N does possess a pair of cysteines (Cys-143 and Cys-157) that might participate in a disulfide bond, 118N lacks cysteine residues; thus, a disulfide linkage is not responsible for the thermal stability. Whereas no tertiary structure is available for proteins similar to ␤GRPs, we found that the amino-terminal recombinant truncations of Pi␤GRP have a predominantly ␣-helical secondary structure. Further studies of these domains should reveal structural features required for their stability as well as what motifs are important for binding carbohydrates and mediating immunological functions.