Self-association of an Insect β-1,3-Glucan Recognition Protein Upon Binding Laminarin Stimulates Prophenoloxidase Activation as an Innate Immune Response*

Background: The N-terminal domain of an insect β-glucan recognition protein (N-βGRP) stimulates innate immune responses. Results: N-βGRP forms soluble and insoluble protein-carbohydrate complexes involving specific protein-protein interactions, which activates the prophenoloxidase pathway. Conclusion: Assembly of βGRP oligomers stimulated by binding to microbial polysaccharide triggers phenoloxidase activation. Significance: βGRP interactions stimulate an innate immune response by a novel mechanism. Insect β-glucan recognition protein (βGRP), a pathogen recognition receptor for innate immune responses, detects β-1,3-glucan on fungal surfaces via its N-terminal carbohydrate-binding domain (N-βGRP) and triggers serine protease cascades for the activation of prophenoloxidase (pro-PO) or Toll pathways. Using biophysical and biochemical methods, we characterized the interaction of the N-terminal domain from Manduca sexta βGRP2 (N-βGRP2) with laminarin, a soluble form of β-1,3-glucan. We found that carbohydrate binding by N-βGRP2 induces the formation of two types of protein-carbohydrate complexes, depending on the molar ratio of carbohydrate to protein ([C]/[P]). Precipitation, analytical ultracentrifugation, and chemical cross-linking experiments have shown that an insoluble aggregate forms when the molar ratio of carbohydrate to protein is low ([C]/[P] ∼ 1). In contrast, a soluble complex, containing at least five N-βGRP2 molecules forms at a higher molar ratio of carbohydrate/protein ([C]/[P] >5). A hypothesis that this complex is assembled partly due to protein-protein interactions was supported by chemical cross-linking experiments combined with LC-MS/MS spectrometry analysis, which permitted identification of a specific intermolecular cross-link site between N-βGRP molecules in the soluble complex. The pro-PO activation in naive plasma was strongly stimulated by addition of the insoluble aggregates of N-βGRP2. The soluble complex with laminarin formed in the plasma also stimulated pro-PO activation, but at a lower level. Taken together, these results provide experimental evidence for novel mechanisms in which associations of βGRP with microbial polysaccharide promotes assembly of βGRP oligomers, which may form a platform needed to trigger the pro-PO pathway activation cascade.

Insects rely on innate immune responses to fight against a broad spectrum of pathogens. These responses can be triggered by the recognition of pathogen-associated molecular patterns by pathogen recognition receptors. Recognition signals initiated by such interactions are amplified by extracellular serine protease cascades leading to at least two types of responses: prophenoloxidase (pro-PO) 3 activation and subsequent melanization of pathogens, and activation of the Toll ligand, which leads to synthesis of antimicrobial peptides (1,2).
In a recent study, we demonstrated that the binding of laminarin to N-␤GRP from Plodia interpunctella (Pi-N-␤GRP) induces formation of a protein-carbohydrate macrocomplex containing multiple Pi-N-␤GRP molecules and suggested that this complex is an initiation signal for activation of serine protease cascades that promote immune responses (12). Here, to better understand biological and biochemical aspects of such complex formation of ␤GRP, we studied the molecular interaction between a soluble ␤-1,3-glucan and N-␤GRP from Manduca sexta, one of the most common model insects for invertebrate biochemistry, molecular biology, and innate immunity. In the past, we have identified two ␤GRPs from M. sexta (Ms-␤GRP1 and -␤GRP2) (4,8). Another pathogen recognition receptor (microbial-binding protein), a new member of ␤-1,3glucanase-related protein superfamily to which the ␤GRPs belong, has also been identified from M. sexta (15). Whereas Ms-␤GRP1 is constitutively expressed in fat body and secreted into hemolymph (4), the transcription of Ms-␤GRP2 is up-regulated after immune challenges with yeast or bacteria (8), suggesting that Ms-␤GRP2 is involved in an acute immune response against microbial infection. Because of this biological significance, we here studied the N-terminal domain from Ms-␤GRP2 (N-␤GRP2) and demonstrated that N-␤GRP2 forms two types of high molecular mass complexes with laminarin. A soluble complex of N-␤GRP2 and laminarin formed when the molar ratio of the carbohydrate to protein was high ([C]/[P] Ͼ5). However, when this ratio was low ([C]/[P] ϳ 1), N-␤GRP2 and laminarin associated to form insoluble and irreversible aggregates. We characterized the structural properties of the soluble complex using size exclusion chromatography and chemical cross-linking experiments combined with LC-MS/MS analysis. The effects of the soluble and insoluble macromolecular associations on activation of the pro-PO pathway were also studied to gain insight into the biological relevance of the distinct complexes formed by N-␤GRP2 with longer ␤-1,3glucan molecules.
Expression and Purification of N-␤GRP2-A DNA sequence encoding the amino-terminal 109 residues of ␤GRP2 from M. sexta (N-␤GRP2) (8) was inserted via NcoI/XhoI sites into plasmid pET28a, for expression of recombinant N-␤GRP2 with a carboxyl-terminal His 6 tag. Escherichia coli strain BL21(DE3) cells transformed with this plasmid vector was cultured in LB media or with M9 minimal media supplemented with 1 g/liter of 15 NH 4 Cl and 2 g/liter of D-[ 13 C]glucose (Cambridge Isotope Laboratories) for 15 N and 13 C labeling, respectively. N-␤GRP2 was expressed by induction with 1 mM isopropyl 1-thio-␤-Dgalactopyranoside for 4 h at 37°C. After lysis of the bacteria, soluble N-␤GRP2 was purified using Ni 2ϩ affinity chromatography, followed by size exclusion chromatography on a Superdex 75 column (GE healthcare). Purity of N-␤GRP2 was confirmed by SDS-PAGE, and the amino acid sequence was confirmed by mass spectrometry.
NMR Spectroscopy and Structural Modeling-All NMR spectra were collected at 25°C with a Varian VNMR 500 MHz equipped with 5-mm cryogenic triple resonance probes. For backbone resonance assignments, two-dimensional 1 H-15 N heteronuclear single quantum coherence spectrum (HSQC) and three-dimensional HNCA, HN(CO)CA, HNCACB, and HNCO were recorded on 1.5 mM 13 C/ 15 N-labeled N-␤GRP2 prepared in 20 mM sodium phosphate buffer, pH 7.0, with 3 mM NaN 3 in 90% H 2 O, 10% D 2 O. NMR spectra were processed using NMRPipe (17), and analyzed with Sparky (T. D. Goddard and D. G. Kneller, Sparky, University of California, San Francisco, CA) and CARA (18). The secondary structure of N-␤GRP2 was predicted by the program TALOSϩ (19) using the chemical shift of assigned 13 C␣, 13 C␤, and 13 CЈ resonances. For titration experiments, a series of two-dimensional 1 H-15 N HSQC spectra of 0.3 mM 15 N-labeled N-␤GRP2 were collected in 50 mM sodium phosphate, pH 6.5, in the presence and absence of 200 mM NaCl. Laminarin or laminarihexaose prepared in the same buffer were titrated into the protein sample. The final molar ratio of carbohydrate to protein was 1 for laminarin and 10 for laminarihexaose.
Three-dimensional structural models of N-␤GRP2 were generated using the homology modeling tool, I-TASSER (20,21). The N-␤GRP2 amino acid sequence including the C-terminal His tag was used as an input for processing by the I-TASSER algorithm.
Protein Precipitation Analysis-Various concentrations of N-␤GRP2 (200, 95, 47, and 4 M) and laminarin (ϳ1 mM) were incubated in microcentrifuge tubes for 2 h at room temperature. After centrifugation to pellet precipitates, the protein concentration in the supernatant was determined by absorbance measurement, using a molar extinction coefficient of 30,940 M Ϫ1 cm Ϫ1 at 280 nm for N-␤GRP2. Analytical Ultracentrifugation-Sedimentation velocity experiments were conducted with an Optima XL-I ultracentrifuge (Beckman Coulter, Inc.) using an An-60 Ti rotor at 20°C with 50 mM Tris-HCl, pH 7.3, containing 50 mM NaCl (22). Sedimentation was monitored by absorbance or interference optics using double-sector aluminum cells with a final loading of 400 l/sector. Sedimentation was performed at 49,000 ϫ g with scans made at 5-min intervals. Data were analyzed using DCDTϩ software version 1.16. Sedimentation coefficients were calculated using g(s*) and dc/dt fitting functions in DCDTϩ software. Buffer density and viscosity were calculated by SEDNTERP version 1.08.
Size Exclusion Chromatography-Size exclusion chromatography was performed on a Sephacryl S-100 column (GE Healthcare) equilibrated with 50 mM sodium phosphate, pH 6.5, with 200 mM NaCl to investigate formation of complexes between N-␤GRP2 and laminarin. Gel filtration standards (Bio-Rad) containing thyroglobulin (670 kDa), ␥-globulin (158 kDa), ovalbumin (44 kDa), myoglobin (17 kDa), and vitamin B12 (1.3 kDa) were used to determine the molecular mass of proteins. N-␤GRP2 was incubated with a 20-fold molar excess of laminarin in the same buffer for 5 min and applied to the column and eluted at a flow rate of 1 ml/min.
Purification of a Soluble Complex Containing N-␤GRP2 and Laminarin-A soluble complex of N-␤GRP2 and laminarin was purified from the size exclusion chromatography described above. Centrifugal filter (Amicon Ultra-15) was used to concentrate the soluble complex and exchange a desired buffer as indicated in the following experiments.
Chemical Cross-linking and MS Analysis-To identify interacting regions in adjacent N-␤GRP2 molecules within the N-␤GRP2-laminarin soluble complex, the cross-linking reagent DTSSP was used to cross-link amino groups. The ⑀-amino groups from nine lysine residues as well as the N-terminal ␣amino group of N-␤GRP2 were considered as potential crosslinking sites for DTSSP. The purified soluble complex of N-␤GRP2 and laminarin was prepared in 50 mM sodium phosphate buffer, pH 7.0, with 200 mM NaCl. DTSSP was added at a molar ratio of 100:1 (DTSSP to the N-␤GRP2 soluble complex) with a final protein concentration of 5 M. After 30 min of incubation at 25°C, the cross-linking reaction was quenched by addition of Tris-HCl, pH 7.5, to a final concentration of 20 mM. Cross-linked samples were separated by SDS-PAGE under non-reducing conditions and visualized by silver staining using SilverQuest Silver Staining Kit (Invitrogen). Protein bands of interest were excised from the gel, and in-gel trypsin digestion was conducted following the manufacturer's protocol (Sigma) and the method described by Shevchenko et al. (23).
Digested peptide samples were analyzed using LC-MS/MS at the Nevada Proteomics Center (University of Nevada, Reno, NV). The peptides were separated and analyzed using a Michrom Paradigm Multi-dimensional Liquid Chromatography instrument (Michrom Bioresources) coupled with a Thermo LTQ Orbitrap XL mass spectrometer (Thermo Fisher Scientific). Peptide samples dissolved in 100 l of 0.1% formic acid were loaded onto a ZORBAX 300SB-C 18 5 (5 ϫ 0.3 mm) trap column (Agilent Technologies), eluted from the trap, and then separated with a reverse phase Michrom Magic C 18 AQ column (3 m, 200 Å, 0.2 ϫ 150 mm) by a gradient elution using solvent A (0.1% formic acid) and solvent B (0.1% formic acid in acetonitrile) at a flow rate of 2 l/min. The gradient was set from 5 to 40% solvent B for 90 min, increased to 80% solvent B in 10 s, and held at 80% solvent B for 1 min. MS spectra were recorded over the mass range of m/z 400 -1600 with resolution of 60,000. The three most intense ions were isolated for frag-mentation in the linear ion trap using collision-induced dissociation with a minimal signal of 500 and collision energy of 35.0 or using electron transferring dissociation with a minimal signal of 1000 and collision energy of 35.0. Dynamic exclusion was implemented with 2 repeat counts, repeat duration of 15 s, and exclusion duration of 90 s.
Intra-and intermolecular cross-linked peptides were identified by analyzing MS and MS/MS data with StavroX software (24). Potential cross-links were evaluated manually. The criteria for selection of a possible cross-linked peptide include a maximum mass deviation of 50 ppm between theoretical and experimental mass, the presence of sequential ions for a predicted cross-linked peptide. Cross-linked peptides identified in the N-␤GRP2 monomer band and those with the measured distance between ⑀-amino groups of cross-linked lysine residues up to 12 Å apart in the structural model of the N-␤GRP2 monomer are assumed to be intra-molecular cross-links. Cross-linked peptides involving the N-terminal ␣-amino group, however, are excluded from this criterion due to possible flexibility of the N-terminal region. Cross-linked peptides identified only in the N-␤GRP2 dimer band and those with the measured distance between ⑀-amino groups of cross-linked lysine residues more than 12 Å apart in the structural model of N-␤GRP2 monomer are defined as intermolecular cross-links. Oxidation of Met, deamidation of Gln, and Asn were added as potential modifications.
Pro-PO Activation Assay-Phenoloxidase activation assay was conducted using a method modified from Fabrick et al. (7) and Dai et al. (12). Hemolymph was collected from the day 2 fifth instar larvae of M. sexta. After hemocytes were removed by centrifugation at 8000 ϫ g for 20 min at 4°C, the plasma sample (10 l) was incubated with 10 g of N-␤GRP2, the soluble N-␤GRP2⅐laminarin complex, or insoluble N-␤GRP2⅐laminarin aggregate in wells of a 96-well microplate. The volume of each sample well was brought to 100 l with 50 mM sodium phosphate buffer, pH 6.5. After incubation for 1 h at room temperature, 100 l of 5 mM dopamine hydrochloride was added. Phenoloxidase activity was determined by measuring the absorbance at 470 nm at 30-s intervals for 20 min. PO activity was presented as the slope of initial linear region (mOD/min). The assay was conducted with more than three technical replicates, and statistical analysis was performed using Prism 5 (GraphPad Software).

RESULTS
Structural Analysis of N-␤GRP2-Recombinant N-␤GRP2 expressed in E. coli was purified to homogeneity using Ni 2ϩ affinity chromatography and size exclusion chromatography. The protein eluted as a monomer, based on the comparison of its elution volume with those of molecular weight standards (Fig. 1A). The two-dimensional 15 N-1 H HSQC spectrum for 15 N-labeled N-␤GRP2 showed excellent chemical shift dispersion (Fig. 1B), and the backbone chemical shifts of 1 H, 15 N, and 13 C nuclei were assigned using a series of the multidimensional heteronuclear NMR methods; 94% of the backbone amide resonance of 109 non-proline residues could be assigned. The secondary structures of N-␤GRP2 were predicted using the TALOS program (19) and the chemical shift assignments of 13 C␣, 13 C␤, and 13 CЈ resonances (Fig. 1C), which agreed well with those of Pi-N-␤GRP determined by solution NMR spectroscopy (12). A three-dimensional structural model of N-␤GRP2 predicted using the I-TASSER server (C-score ϭ 0.83 Ϯ 0.08, expected root mean square deviation ϭ 2.7 Ϯ 2.0) (20, 21) aligned very well with N-␤GRP from other insects, with root mean square deviations of the backbone C␣ atoms between N-␤GRP2 and Pi-N-␤GRP (Protein Data Bank 2KHA (12)) being 0.38 Å (Fig. 1D).
Binding to Laminarin Induces the Association of N-␤GRP2 Molecules-We carried out a series of NMR chemical shift titration experiments to study interactions between N-␤GRP2 and laminarin. Laminarin is a water-soluble ␤-1,3-glucan with ␤-1,6-glucan branches. The ␤-1,3 to ␤-1,6 cross-linkage ratio varies depending on the biological source. Laminarin from L. digitata used in the present study has a cross-linkage ratio of 7 (16). Solubility of laminarin depends on the degree of ␤-1,6 branching, with more branched laminarin being more soluble. To calculate molarities of laminarin solutions, we adopted a value of 5.5 kDa for the molecular mass, as reported by the supplier (Sigma). Titration of laminarin into 15 N-labeled N-␤GRP2 solution did not result in chemical shift changes, but significant decreases of resonance peak intensities occurred (supplemental Fig. S2). On average, the resonance intensities decreased by more than 90% when the molar ratio of the carbohydrate to protein ([C]/[P]) was 0.5 in the absence of NaCl (supplemental Fig. S2C). There were no site-specific changes in resonance intensity, and almost all of the peaks, excluding the backbone NH group of C-terminal residues and side chain NH 2 groups, disappeared from the spectrum when the laminarin was added at a 1:1 [C]/[P] ratio, as observed in our previous NMR study for Pi-N-␤GRP (12). The presence of 0.2 M NaCl suppressed the decrease in resonance intensities to some extent (supplemental Fig. S2, B and C), but the addition of 0.3 mM laminarin resulted in peak disappearance similar to that observed in the absence of NaCl. Such peak disappearance can be caused by NMR line broadening due to large protein complex formations and/or conformational exchanges on millisecond time scale. However, we observed precipitation of N-␤GRP2 during the titration experiments, which has not been reported for other NMR studies of insect ␤GRPs (12,13) and this made the chemical shift titration approach inapplicable. Titration of N-␤GRP2 with laminarihexaose, a shorter ␤-1,3glucan, also resulted in no chemical shift perturbation, but resonance intensities were decreased when the molar ratio of oligosaccharide to protein was greater than 5 (data not shown). This result suggests weaker interaction of N-␤GRP2 with laminarihexaose than with laminarin, consistent with other studies (11,12).
When the laminarin concentration was lower than or comparable with the protein concentration (0.2 mM), insoluble aggregation was visually confirmed (Fig. 2A). Maximum precipitation was observed when the molar ratio of the carbohydrate to protein was less than 1 (Fig. 2B). However, no precipitation was observed when N-␤GRP2 was incubated with a molar excess of laminarin, at the higher [C]/[P] ratios (Fig. 2). Similar precipitation behavior of N-␤GRP2 with laminarin was observed at different initial protein concentrations (4, 47, and 95 M) (data not shown). The insoluble aggregates could not be re-solubilized by dilution or further addition of laminarin, laminarihexaose, or laminaritetraose, indicating that this aggregation occurs irreversibly. Microscopic observation showed that the insoluble aggregates of N-␤GRP2 with laminarin formed evenly dispersed particles with a diameter of ϳ0.5 m (Fig. 2C).
We recently demonstrated using solution NMR and analytical ultracentrifugation methods that Pi-N-␤GRP forms a soluble, high molecular mass complex upon binding to laminarin (12). To directly compare the effects of laminarin on N-␤GRP2 and Pi-N-␤GRP, sedimentation velocity analysis was performed (Fig. 3). N-␤GRP2 (13.8 kDa) in the absence of laminarin had a sedimentation coefficient of 1.9 s. A mixture of 60 M N-␤GRP2 with 79 M laminarin resulted in a broadening of the sedimentation peak, a decrease of signal height, and an increase of sedimentation coefficient to 5.5 s. Decrease of the signal height was due to precipitation or high order of aggregation. A further increase of laminarin concentration led to a slight increase of signal height and gave s values of 4.5 s when the laminarin concentration was 3.9 mM. Analysis of this sedimentation profile yielded an average molecular mass of 70 kDa. A very similar sedimentation profile was observed for Pi-N-␤GRP when the laminarin concentration was much higher than the protein concentration.
The results of these analytical ultracentrifuge and precipitation experiments are consistent with a model in which N-␤GRP2 forms two types of macromolecular complexes with laminarin, depending on the molar ratio of the carbohydrate to protein. Aggregation or assembly to form insoluble particles of ϳ0.5 m diameter occurs at the lower molar ratio ([C]/[P] Ͻ 1), and a soluble complex (s ϭ 4.5) forms at the higher molar ratio ([C]/[P] Ͼ 5).

Characterization of the Soluble Protein-Carbohydrate
Complex-To structurally characterize the soluble N-␤GRP2⅐ laminarin complex, we first purified this complex using size exclusion chromatography (Fig. 4). After N-␤GRP2 was incubated with a 20-fold molar excess of laminarin, the soluble protein-carbohydrate complex eluted as a single and relatively sharp peak at 57 ml, which corresponds to molecular mass of ϳ75 kDa, based on a calibration curve obtained with molecular mass standard proteins. In contrast, N-␤GRP2 in the absence of laminarin eluted at 77 ml, consistent with its existence as a monomer. This result suggests that association of the N-␤GRP2 and laminarin leads to formation of a soluble macrocomplex with defined size and that the complex is stable due to strong protein-protein and protein-carbohydrate interactions.
Chemical cross-linking of noncovalent protein complexes has been used to determine the stoichiometry of constituent monomers, and mass spectrometry analysis of the cross-linked species can identify interacting regions in the protein complexes (25)(26)(27)(28)(29). We have employed this combined approach to gain structural insights into molecular organization of the soluble N-␤GRP2⅐laminarin macrocomplex. DTSSP is a crosslinking agent with a spacer arm of 12 Å that can be cleaved by reducing agents, and reacts with primary amine groups in proteins (25,26). N-␤GRP2 has nine ⑀-amino groups from lysine residues as well as the N-terminal ␣-amino group that could serve as targets for cross-linking by DTSSP if the residues are near 12 Å apart in the protein monomer or complex. Incubation of the soluble complex of N-␤GRP2 and laminarin with 0.5 mM DTSSP yielded at least 4 higher molecular mass bands (Fig.  5, lane 5, b-e) in addition to the monomer species (13.8 kDa, a in Fig. 5). These bands correspond to molecular masses of 27, 44.6, 62.7, and 78.1 kDa, each of which approximately equals molecular masses expected for oligomers (from dimer to pentamer) of N-␤GRP2. Most of these higher molecular mass species were reduced to monomer in the presence of 20 mM DTT (Fig. 5, lane 4). Although a band corresponding to the N-␤GRP2 dimer was also observed when the N-␤GRP2 monomer was treated with DTSSP in the absence of laminarin (Fig. 5,  lanes 3), this band is of substantially weaker intensity than that for the soluble complex, probably suggesting a nonspecific inter-monomer cross-linking. This cross-linking experiment provides further evidence for the soluble complex of N-␤GRP2 and laminarin, and demonstrates that the complex encompasses at least five N-␤GRP2 molecules.
To identify the interacting region of N-␤GRP2 molecules within the soluble complex, we digested the cross-linked dimer with trypsin and analyzed the peptide fragments with LC-MS/MS (Table 1). We determined that in the dimer, lysine 61 in the peptide (AK 61 QGR) is cross-linked to lysine 79 in the peptide (LGDK 79 IYFWTYVIK). The product ion spectrum of the identified intermolecular cross-link (Fig. 6) confirmed the sequence of the peptide and mapping of the cross-link sites between AK 61 QGR and LGDK 79 IYFWTYVIK. Lysine 61 is located on the concave surface of N-␤GRP2, whereas lysine 79 is on the convex surface, and the measured distance between their ⑀-amino groups within the structural model of the N-␤GRP2 monomer is 29.1 Å, significantly longer than the range of an intra-molecular cross-link with DTSSP. This dipeptide was not found in digested fragments of the cross-linked monomer. This result demonstrates that in the dimer, Lys 61 in one N-␤GRP2 is within ϳ12 Å from Lys 79 in an adjacent N-␤GRP2 molecule.
Effects of the N-␤GRP2 Complexes with Laminarin on Pro-PO Activation-To gain insights into biological implications of the soluble and insoluble associations of N-␤GRP2 with laminarin, we investigated their effects on the activation of pro-PO in plasma from M. sexta larvae. PO activity increased in plasma supplemented with purified N-␤GRP2, and was further enhanced when both N-␤GRP2 and laminarin were incubated with plasma sample (Fig. 7A), which is consistent with earlier studies (7,8,12). At this condition, the molar ratio of laminarin to N-␤GRP2 is more than 20, which is high enough to form the soluble association we observed with purified components in the experiments described above. However, addition of the purified soluble complex to plasma did not stimulate significant PO activation. On the other hand, incubation of the insoluble complex with plasma led to a strong enhancement of PO activity, suggesting that the insoluble aggregates, formed when the molar ratio of protein to carbohydrate is high, is a potent activator of the pro-PO activation pathway. As the insoluble complex could not be resolubilized and was extensively washed with 50 mM sodium phosphate buffer before addition to plasma, it is not likely that the enhancement of PO activity was derived from the soluble complex and/or laminarin. Addition of starch (mixture of amylose and amylopectin) as a control material containing insoluble polysaccharide resulted in no pro-PO activation (Fig. 7A), indicating that the pro-PO activation by the insoluble complex of N-␤GRP2 with laminarin is not simply due to the recognition of an insoluble phase in plasma.
We further tested whether the molecular assembly of N-␤GRP2 with laminarin to form a larger complex occurs under the conditions of pro-PO activation. Plasma was supplemented with N-␤GRP2 or the N-␤GRP2⅐laminarin soluble complex, and after a 1-h incubation, the proteins in those plasma samples were separated by size exclusion chromatography. The resulting fractions were analyzed by SDS-PAGE and immunoblotting using the anti-␤GRP2 antibody. N-␤GRP2 eluted as a monomer in the absence of laminarin (Fig. 7B, left), but eluted much earlier when the protein was incubated with plasma in the presence of laminarin (Fig. 7B, middle), indicating that N-␤GRP2 was associated to form a high molecular mass complex. This elution profile of N-␤GRP2 was similar to that observed after the pre-formed soluble complex was incubated with the plasma (Fig. 7B, right). These results indicate that adding recombinant N-␤GRP2 and laminarin to plasma results in formation of complexes as observed in the absence of plasma.
We also detected a 54-kDa plasma protein with the anti-␤GRP2 antibodies (Fig. 7B). Our previous studies showed that ␤GRP2 is absent in the naive larvae and ␤GRP1 is constitutively present in hemolymph of the feeding stage larvae (4,8). Due to their sequence similarity (57% identity), the antibody to ␤GRP1 cross-reacts with ␤GRP2 (8). Based on these results, the 54-kDa protein was deduced to be ␤GRP1, and this was further confirmed with immunoblotting analysis using anti-␤GRP1 antibody (data not shown). The immunoblot analysis showed that ␤GRP1 eluted earlier in the presence of N-␤GRP2 and laminarin and of the soluble complex (Fig. 7B, middle and right) when compared with it in the presence of N-␤GRP2 (Fig. 7B,  left), which indicates a higher molecular mass complex formation of natural ␤GRP1 in the presence of the soluble complex of N-␤GRP2-laminarin.

DISCUSSION
Insoluble ␤-1,3-glucan, a major surface component of fungal cell walls, is detected by pattern recognition proteins to initiate

Complexes of Insect Glucan Recognition Protein and ␤-1,3-Glucan
OCTOBER 10, 2014 • VOLUME 289 • NUMBER 41 innate immune responses to fungal infections in insects. In this study, we investigated the interactions between the amino-terminal carbohydrate-binding domain of the hemolymph plasma protein Ms-␤GRP2 (N-␤GRP2) and a soluble ␤-1,3-glucan, laminarin, by means of biophysical and biochemical methods. We identified two types of association between N-␤GRP2 and laminarin. When the concentration of N-␤GRP2 was greater than that of laminarin, insoluble complexes of N-␤GRP2 and laminarin were formed, which appeared as fairly uniform particles when viewed by light microscopy. This preparation strongly stimulated the activation of pro-PO when added to plasma. However, when the concentration of laminarin was greater than that of N-␤GRP2, we detected the formation of a soluble protein-carbohydrate complex (ϳ75 kDa), as previously observed for Pi-N-␤GRP (12). This soluble complex was less effective in activation of pro-PO in plasma.
Macromolecular assembly of carbohydrate-binding proteins and their carbohydrate ligands can serve as effective and specific platforms to drive and control concomitant signaling events (30 -33). Drosophila PGRP-LCx, another insect pathogen recognition receptor, forms a homodimer or heterodimer with PGRP-LCa, and their cytoplasmic domains recruit the adaptor molecule Imd, leading to downstream signaling in the immune deficiency pathway (32). Clustering of PGRP-LCx with DAP-type peptidoglycans on the bacterial surface was suggested based on the crystal structure of PGRP-LE complexed with a monomeric peptidoglycan (33). Many lectins interact with multivalent polysaccharides or glycoproteins to form highly organized, multidimensional cross-linked clusters (lattice-like structures), many of which form insoluble precipitates (34 -36).
For insect ␤GRP, Kanagawa et al. (14) suggested from the crystal structure of N-␤GRP complexed with triple helical laminarihexaoses that an unusually large carbohydrate binding surface on N-␤GRP would enhance its affinity to carbohydrate. We propose that N-␤GRP2 may also employ another strategy in which protein-protein interactions become established as a carbohydrate-protein complex is formed. This idea is supported by the formation of N-␤GRP2 multimers only in the presence of laminarin. Such protein-protein interactions might increase overall avidity of carbohydrate binding, as observed for many lectins (31). Formation of oligomeric glucan-protein complexes has also been suggested for other ␤-glucan recognition proteins involved in immune responses both in vertebrates (37) and invertebrates (38).

TABLE 1 DTSSP cross-linked peptides for the N-␤GRP2⅐laminarin complex
Purified soluble complex of N-␤GRP2 and laminarin were incubated at 25°C with DTSSP at a molar ratio of the cross-linker to protein of 100:1, followed by SDS-PAGE with silver staining under non-reducing condition to separate cross-linked products (Fig. 6). Protein bands corresponding to the N-␤GRP2 monomer and dimer were excised and in-gel digested with trypsin. Digested peptide fragments were then analyzed using LC-MS/MS, and intra-and intermolecular cross-linked peptides were identified with the software StavroX (24).  In the crystal structure of Plodia and Bombyx N-␤GRPs complexed with laminarihexaose (14), protein molecules pack in a side-by-side fashion through electrostatic interactions in a loop that includes Asp 39 . We previously conducted site-directed mutagenesis of the corresponding aspartic acid residue (Asp 45 in Ref. 12) and Glu 46 in Ms-N-␤GRP2) located in the contact "loop" region between the adjacent N-␤GRP molecules in the crystals (Fig. 8), and demonstrated that the electrostatic interaction involving Asp 45 contributes to self-association of the N-␤GRP⅐laminarin complex (12). To characterize proteinprotein interactions that accompany formation of the N-␤GRP2⅐laminarin complex, we applied chemical cross-linking combined with mass spectrometry analysis and identified an inter-molecular cross-link formed between Lys 61 and Lys 79 in a cross-linked N-␤GRP2 dimer formed in the presence of laminarin. Consistent with this finding, the distance between the ⑀-amino groups of lysine residues corresponding to Lys 61 and Lys 79 in adjacent proteins in the crystal structure of N-␤GRP complexed with laminarihexaose is close enough to form a cross-linking bridge: 8.1 and 12.9 Å for Plodia N-␤GRP (Fig. 8) and 13.9 and 17.6 Å for Bombyx N-␤GRP. These results suggest that protein-protein interactions formed in the complex of N-␤GRP2 and laminarin in our experiment may be similar to those observed in the crystals of N-␤GRP formed with laminarihexaose.

Theoretical mass of cross-linked peptide
Laminarin is a ␤-1,3-linked glucose polymer with ␤-1,6 cross-link branches and contains 30 glucose residues on average, based on its average molecular mass (5.5 kDa). In contrast to insoluble linear polysaccharides such as curdlan and cellulose, laminarin is soluble in water, because its ␤-1,6 branches provide an extra degree of freedom by the rotation about the C-5 and C-6 bonds of the glucose ring (39). The crystal structure of N-␤GRP complexed with laminarihexaoses arranged in a triple helical form has shown that N-␤GRP interacts with ␤-1,3-glucans through six glucose residues, two from each laminarihexaose chain (14). Our chemical cross-linking exper- FIGURE 7. Activation of the prophenoloxidase pathway by N-␤GRP2 and its multimeric complexes formed with laminarin. A, pro-PO activation assay. Ten microliters of plasma from day 2 5th instar larvae of M. sexta was incubated with 10 g of N-␤GRP2 with and without 100 g of laminarin, soluble complex, or insoluble aggregate, in wells of a 96-well plate for 1 h at 25°C. Phenoloxidase activity was determined using dopamine as a substrate by measuring the absorbance at 470 nm. The control contained no carbohydrate or protein, but 50 mM sodium phosphate buffer. BSA and starch were also used as negative controls. Values represent the mean slope (mOD/min) Ϯ S.D. from triplicates. B, analysis of the complex formation in plasma by size exclusion chromatography combined with immunoblotting. The chromatography was performed for plasma with addition of N-␤GRP2 (left), N-␤GRP2 with laminarin (middle), and the purified soluble complex of N-␤GRP2 and laminarin (right). Shown above each chromatograph is an immunoblot analysis of eluted fractions using anti-␤GRP2 antibody.
iment suggests that the mode of interaction between N-␤GRP2 molecules bound to laminarin is consistent with that observed in the crystals of N-␤GRP formed with laminarihexaose, which may indicate that binding of N-␤GRP2 to laminarin chains promotes their association as a partial triple helical structure (although the ␤-1,6 branches would interfere with the helical structure formation to some extent). When the molar ratio of protein to carbohydrate is high (at the low [C]/[P] ratio), we predict that multiple N-␤GRP2 molecules interacting with a helical form of laminarins assemble through protein-protein interaction to form a larger complex structure. We speculate that association of ␤-glucan chains to form triple helical structures, promoted by protein-protein interactions, leads to production of large complexes in which a significant loss of the motional freedom of glucan chains results in formation of the insoluble aggregates observed in this study. Hydrophobic association and hydrogen bond interactions between polysaccharide chains likely contribute to stabilize the association of polysaccharides in an insoluble form as occurs in curdlan and cellulose (40). The insoluble complex of N-␤GRP2 and laminarin may mimic assembly of ␤GRP2 upon binding to ␤-1,3-glucan on a fungal surface, and serves as a large and effective platform for downstream signaling events by recruiting serine proteases and prophenoloxidase, increasing their local concentrations. To our knowledge, this study represents the first report demonstrating that the insoluble complex of N-␤GRP2 and ␤-1,3glucan is a powerful activator for the pro-PO activation pathway in insect hemolymph (Fig. 7).
In the presence of excess laminarin (at a higher [C]/[P] ratio), on the other hand, the N-␤GRP2 molecules bound to laminarins may be more dispersed, such that there are fewer proteinprotein contacts, resulting in formation of smaller complexes, in which the laminarin chains retain sufficient flexibility to remain soluble. In our previous study (12), analytical ultracentrifuge results indicated that a Pi-N-␤GRP⅐laminarin complex likely contained six protein and three laminarin molecules, consistent with our current observation of stable soluble complexes of ϳ75 kDa, estimated by analytical ultracentrifuge and size exclusion chromatography. Such a soluble complex is possibly not large enough to serve as an effective platform for activation of downstream serine protease cascades.
Previous studies of a crayfish immune response demonstrated that activation of pro-PO in plasma by laminarin reached a maximum at relatively low laminarin concentration (ϳ1 g/ml) and decreased at higher laminarin concentrations (41). In addition, when the concentration of laminarin was fixed, activation of crayfish pro-PO was greater at higher molar ratios of the glucan-binding protein to laminarin (42). We also previously observed that pro-PO activation by laminarin in M. sexta plasma can decrease as the laminarin concentration increases past an optimum level (43). In this study, the soluble complex of N-␤GRP2 with laminarin formed in the plasma at a higher laminarin concentration (a higher [C]/[P] ratio) stimulated pro-PO activation, but at a significantly lower level compared with the insoluble complex (Fig. 7A). Thus, our result showing formation of the less active soluble complex may help to explain why high concentrations of laminarin lead to the inhibition or a lower level stimulation of pro-PO activation in crayfish (41,42) and M. sexta (43).
It is likely that Toll activation can also be stimulated in a similar manner by formation of ␤GRP-glucan complexes resulting in activation of connected serine protease cascades for the Toll and pro-PO activation pathways (44 -46). However, the molecular mechanism by which the downstream serine protease cascades are activated upon the ␤-1,3-glucan recognition by ␤GRP still remains to be established. This is partly because the molecular function of the C-terminal glucanaselike domain of ␤GRP and the unique modular domains of the initiating serine proteases (47,48), and their interactions have not been characterized in detail. Intriguingly, within the higher order structure of N-␤GRP2 observed in the crystal structure (Fig. 8), the carboxyl terminus of N-␤GRP2 molecules positions away both from the carbohydrate-binding site and the proteinprotein contact site. It is possible that the C-terminal glucanase-like domain can interact with the initiating proteases without interfering with protein complex organization mediated by N-␤GRP. However, further biochemical and biophysi-cal studies on full-length ␤GRP and the initiating serine proteases are necessary to better understand the molecular basis underlying this initial step for insect innate immune responses. The questions remaining to be answered include whether fulllength ␤GRP associates to form protein complexes observed here, whether the C-terminal domain may mediate protein interactions with the initiating protease, and which modular domains in the initiating protease mediate the interaction with ␤GRP resulting in protease autoactivation.