A beta1,3-glucan recognition protein from an insect, Manduca sexta, agglutinates microorganisms and activates the phenoloxidase cascade.

Pattern recognition proteins function in innate immune responses by binding to molecules on the surface of invading pathogens and initiating host defense reactions. We report the purification and molecular cloning of a cDNA for a 53-kDa beta1,3-glucan-recognition protein from the tobacco hornworm, Manduca sexta. This protein is constitutively expressed in fat body and secreted into hemolymph. The protein contains a region with sequence similarity to several glucanases, but it lacks glucanase activity. It binds to the surface of and agglutinates yeast, as well as gram-negative and gram-positive bacteria. Beta1,3-glucan-recognition protein in the presence of laminarin, a soluble glucan, stimulated activation of prophenoloxidase in plasma, whereas laminarin alone did not. These results suggest that beta1,3-glucan-recognition protein serves as a pattern recognition molecule for beta1,3-glucan on the surface of fungal cell walls. After binding to beta1,3-glucan, the protein may interact with a serine protease, leading to the activation of the prophenoloxidase cascade, a pathway in insects for defense against microbial infection.

Pattern recognition proteins function in innate immune responses by binding to molecules on the surface of invading pathogens and initiating host defense reactions. We report the purification and molecular cloning of a cDNA for a 53-kDa ␤1,3-glucan-recognition protein from the tobacco hornworm, Manduca sexta. This protein is constitutively expressed in fat body and secreted into hemolymph. The protein contains a region with sequence similarity to several glucanases, but it lacks glucanase activity. It binds to the surface of and agglutinates yeast, as well as Gram-negative and Gram-positive bacteria. ␤1,3-Glucan-recognition protein in the presence of laminarin, a soluble glucan, stimulated activation of prophenoloxidase in plasma, whereas laminarin alone did not. These results suggest that ␤1,3glucan-recognition protein serves as a pattern recognition molecule for ␤1,3-glucan on the surface of fungal cell walls. After binding to ␤1,3-glucan, the protein may interact with a serine protease, leading to the activation of the prophenoloxidase cascade, a pathway in insects for defense against microbial infection.
Pattern recognition molecules serve as biosensors for detection of invading pathogens in the innate immune systems of vertebrate and invertebrate animals (1). They also play a crucial role in regulating the adaptive immune reactions carried out by vertebrate lymphocytes (2). Pattern recognition molecules bind to certain pathogen-associated molecular patterns that are not found in the host, such as lipopolysaccharide or peptidoglycan from bacterial cell walls and ␤1,3-glucan from fungal cell walls. Upon binding to the foreign invaders, pattern recognition proteins trigger defense pathways such as the complement system in vertebrates (2) and the prophenoloxidase (PPO) 1

activation pathway in insects and other arthropods (3).
The PPO activation pathway, like the complement system, involves a protease cascade. Determining the molecular mechanisms by which pattern recognition proteins differentiate nonself from self and transduce signals that stimulate defensive responses is a key to understanding the regulation of innate immune systems.
Some pattern recognition proteins from mammals, such as the cellular receptor CD14, the mannose-binding protein, the first component of complement (C1q), and the macrophage mannose receptor and scavenger receptor have been well characterized (4). Recent investigations of pattern recognition molecules from insects have identified a group of C-type lectins, which bind bacterial lipopolysaccharides and stimulate antibacterial responses of hemocytes or activate the phenoloxidase pathway (5)(6)(7)(8). A peptidoglycan recognition protein has also been discovered in the silkworm (9,10). This 19-kDa hemolymph protein is constitutively expressed in the fat body, epithelial cells, and hemocytes of naive silkworms and can be induced by bacterial challenge. Binding of peptidoglycan to peptidoglycan recognition protein can trigger the PPO defense pathway (9). A cDNA which encodes a similar bacteria-inducible peptidoglycan recognition protein has been cloned from the moth, Trichoplusia ni (11). Furthermore, proteins homologous to the insect peptidoglycan recognition proteins have been identified in human and mouse and are expressed in bone marrow and spleen (10,11), indicating that this peptidoglycan recognition protein is conserved from insects to humans.
Fungal infections can be recognized by proteins which bind to cell wall ␤1,3-glucans, a molecular pattern specific for fungi. A group of similar ␤1,3-glucan-binding proteins from crustaceans has been isolated, and a cDNA for one of them was cloned from the crayfish, Pacifastacus leniusculus (12)(13)(14)(15). The crayfish protein can induce spreading and degranulation of crayfish granular hemocytes when bound to ␤1,3-glucan (16). An insect ␤1,3-glucan-recognition protein has been purified from the silkworm, Bombyx mori (17). Binding of this 62-kDa protein with ␤1,3-glucan triggers the PPO pathway in plasma. Several ␤1,3glucan-binding proteins have been purified from cockroach species (18 -20). These proteins also enhance PPO activation triggered by soluble ␤1,3-glucan. However, so far no cDNA or amino acid sequence for an insect ␤1,3-glucan recognition protein has been reported.
In this paper, we describe the purification and cDNA cloning of a ␤1,3-glucan-recognition protein (GRP) from the tobacco hornworm, M. sexta. It is synthesized in the fat body and secreted into hemolymph. The deduced amino acid sequence of GRP shows similarity with bacterial and sea urchin glucanases (21), clotting factor G subunit ␣ from a horseshoe crab (22), earthworm CCF-1 (23), and also with a Gram-negative bacteria-binding protein from the silkworm (24). M. sexta GRP at physiological concentration caused aggregation of yeast and bacteria. Finally, we report that GRP can trigger the activation of the PPO pathway upon binding to soluble ␤1,3-glucan.

EXPERIMENTAL PROCEDURES
Insects-M. Sexta eggs were originally obtained from Carolina Biological Supply. Larvae were reared on an artificial diet as described by Dunn and Drake (25).
Affinity Purification of GRP for Antiserum Production-Hemolymph (100 ml) was collected from day 3 fifth instar M. sexta larvae. Cell-free hemolymph (plasma) was prepared by diluting the hemolymph 1:2 in anticoagulant buffer (4 mM NaCl, 40 mM KCl, 0.1% polyvinylpyrrolidone, 1.9 mM PIPES, 4.8 mM citric acid monohydrate, 13.6 mM sodium citrate, 4 mM EDTA, 5% sucrose, pH 6.8) and centrifuging at 12,000 ϫ g for 15 min to remove hemocytes. The plasma (100 ml) was incubated with 0.5 g of curdlan (an insoluble ␤1,3-glucan preparation, Sigma) pre-equilibrated with PBS (0.01 M phosphate buffer, 0.0027 M potassium chloride, and 0.137 M sodium chloride, pH 7.4, at 25°C) at room temperature for 20 min with mixing and then centrifuged (12,000 ϫ g, 5 min). The curdlan pellet was washed three times with 30 ml of PBS and then incubated with 5 ml of PBS and 1 ml of 6ϫ SDS sample buffer (0.35 M Tris, 10% (w/v) SDS, 30% (v/v) glycerol, 9.3% (v/v) ␤-mercaptoethanol, 0.012% bromphenol blue) at 95°C for 10 min. The sample was then centrifuged (12,000 ϫ g, 5 min) and the supernatant was analyzed by SDS-PAGE. The major band of 53 kDa was cut from the SDS-PAGE gel and used as an antigen for the production of a rabbit antiserum (Cocalico Biologicals, Reamstown, PA).
Western Blot Analysis-For immunoblot analysis of proteins after separation by SDS-PAGE, proteins were transferred to a nitrocellulose membrane. The membrane was blocked with 3% dry skim milk, and then incubated with rabbit antiserum to GRP (1:1000 dilution). Antibody binding was visualized by a color reaction catalyzed by alkaline phosphatase conjugated to goat anti-rabbit IgG (Bio-Rad). To estimate the concentration of GRP in hemolymph, 1-l samples of plasma were separated by SDS-PAGE followed by immunoblotting as described above. The immunoreactive bands were digitized using a Kodak DC120 digital camera, and band intensities were measured using one-dimensional Image Analysis Software (Kodak Digital Science). Lanes containing known concentrations of purified, native GRP were used to produce a standard curve, for determining GRP concentration in plasma samples. Plasma from larvae of several insect species including the Indian mealmoth Plodia interpunctella, the fruit fly Drosophila melanogaster, and the tobacco budworm Heliothis virescens (obtained from the Department of Entomology, Kansas State University) was collected and analyzed by SDS-PAGE and Western blotting, using antiserum to M. sexta GRP.
Purification of Native GRP-One hundred and fifty ml of hemolymph were collected as described above, and then fractionated by ammonium sulfate precipitation. The 20 -35% saturated ammonium sulfate fraction of the plasma proteins was dissolved in 50 ml of 50 mM Tris, 20 mM NaCl, pH 7.6, and dialyzed against the same buffer overnight at 4°C with several changes of buffer. Then the sample was centrifuged (10,000 ϫ g, 15 min) before it was applied to a Q-Sepharose Fast Flow (Amersham Pharmacia Biotech) anion exchange column (3 ϫ 10 cm). The column was washed with the same buffer until A 280 was less than 0.01 and then eluted at 2 ml/min with a gradient of 20 to 200 mM NaCl (200 ml total) in 50 mM Tris buffer, pH 7.6. Finally, the column was eluted with 100 ml of 200 mM NaCl in 50 mM Tris buffer, pH 7.6. Fractions (5 ml) were collected and analyzed for absorbance at 280 nm and by Western blotting, using antibody to GRP. Fractions containing GRP were pooled and passed through a concanavalin A (ConA) affinity column (3 ϫ 5 cm). The flow-through fractions, containing GRP, were pooled and applied to a hydroxyapatite column (3 ϫ 3 cm) pre-equilibrated with 10 mM sodium phosphate buffer, pH 6.8. The hydroxyapatite column was washed with the same buffer until A 280 was less than 0.01 and then eluted at 0.1 ml/min with a gradient of 10 to 100 mM sodium phosphate, pH 6.8 (200 ml total). Finally, the column was eluted with 100 ml of 500 mM sodium phosphate buffer, pH 6.8. Fractions (5 ml) containing GRP (detected by Western blot analysis) were pooled and analyzed by gel filtration HPLC using a Bio-Sil SEC250 column (300 ϫ 7.8 mm, Bio-Rad). The column was eluted with 50 mM sodium phosphate, pH 6.8, 150 mM NaCl at 1 ml/min, and fractions collected every 15 s were analyzed by SDS-PAGE and silver staining. The column was calibrated by separating a mixture of molecular mass standard proteins (Bio-Rad) containing thyroglobulin (670 kDa), bovine ␥-globulin (158 kDa), chicken ovalbumin (44 kDa), equine myoglobin (17 kDa), and vitamin B12 (1.35 kDa) to produce a standard curve for estimating the apparent mass of native GRP.
Amino-terminal and Internal Amino Acid Sequence Determinations-GRP (2 g) purified by affinity to curdlan was separated by SDS-PAGE, transferred to polyvinylidene difluoride membrane (0.2 m, Bio-Rad), and stained with Amido Black. The GRP band was cut from the membrane and subjected to automated Edman degradation in an Applied Biosystems Model 473 Pulse-liquid Sequencer. To generate a peptide fragment used to obtain an internal sequence, purified GRP (100 g) in 100 l of 10 mM sodium phosphate buffer was dried using a Speed-Vac and dissolved in 100 l of freshly prepared 70% formic acid (26). After incubation at 37°C for 48 h, the sample was dried using a Speed-Vac and dissolved in 20 l of water. The hydrolyzed peptides were separated by two-dimensional gel electrophoresis using the O'Farrell System (27), transferred to polyvinylidene difluoride membrane, and stained with Amido Black. One peptide spot with mass of 25 kDa was cut from the membrane and subjected to automated Edman degradation.
Glycosylation and Mass Analysis-GRP purified from plasma (0.07 g/l, 100 l) was mixed with 4 l of 0.5 M EDTA, 1 l of 10% SDS, 1 l of ␤-mercaptoethanol and incubated at 100°C for 3 min, then cooled to room temperature. Then 10 l of 10% Triton X-100 and 10 l of N-glycosidase F (0.2 units/l, Roche Molecular Biochemicals) were added, and the mixture was incubated at 37°C for 24 h. The digested GRP and the untreated GRP were analyzed by SDS-PAGE and by matrix-assisted laser desorption ionization mass spectrometry on a Lasermat 2000 instrument (Finnigan MAT). Protein samples were mixed with cyano-4-hydroxycinnamic acid (1 mg/ml) as the matrix, and bovine serum albumin was used as an internal standard.
cDNA Isolation and Sequencing-Primers for PCR amplification of a cDNA fragment encoding GRP were designed based on amino acid sequences of the amino terminus and an internal fragment (Fig. 3). . PCR was performed with denaturing at 94°C for 30 s, annealing at 50°C for 40 s, and extension at 72°C for 1 min for 40 cycles, using a DNA Thermal Cycler 480 (Perkin Elmer). The PCR product was used as a template to amplify again using primers P2 and P7 or P3 and P6 under the same conditions described above. A 700-base pair PCR product amplified using primers P2 and P7 was cloned into plasmid vector pGEM-T using a PCR cloning kit (Promega).
A M. sexta larval fat body cDNA library in Uni-Zap XR vector (Stratagene) (28) was screened using the 700-base pair cloned cDNA PCR product, labeled with [␣-32 P]dCTP (multiprime DNA labeling system, Amersham Pharmacia Biotech) as a probe. Plaques that hybridized to the probe were purified to homogeneity and subcloned by in vivo excission of pBluescipt plasmids. The positive clone with the longest cDNA insertion (1.6 kilobases) was sequenced from both strands by the DNA Sequencing Facility at Iowa State University, using vector primers and primers designed from previously determined sequences.
Northern Blot Analysis-Total RNA from day 3 fifth instar M. sexta larval fat body, hemocytes, or integument was extracted using the Rapid Total RNA Isolation Kit (5 Prime 3 3 Prime, Inc.). Samples of total RNA (20 g) from these tissues were resolved by electrophoresis on an agarose gel containing formaldehyde, transferred to a nitrocellulose membrane, and hybridized with 32 P-labeled GRP cDNA. To confirm equal mRNA loading in different samples, a duplicate blot was hybridized with a cDNA for M. sexta ribosomal protein S3 (29) as a control. To test whether GRP mRNA level is affected by microbial infection, M. sexta fifth instar day 3 larvae were injected with a mixture of killed E. coli (10 7 cells) and yeast (10 7 cells). Control larvae were not injected. Total RNA from fat body was extracted 24 h after injection and analyzed by Northern blotting as described above. Band intensities were measured using one-dimensional Image Analysis Software (Kodak Digital Science), and the ratio of GRP mRNA to ribosomal protein S3 mRNA band intensity was calculated.
Fat Body Culture-Fat body from day 2 fifth instar M. sexta larvae was dissected, washed with PBS, and then incubated at 28°C in 2 ml of EX-CELL 405 (JRH Biosciences) insect cell culture medium containing penicillin (100 units/ml) and streptomycin (100 g/ml) in wells of a 24-well plate with shaking at 100 rpm. At different times of incubation (0, 7, 11, 29, 40, and 64 h), 20 l of the medium was removed and stored at Ϫ20°C for later analysis. These samples were then analyzed for the presence of GRP by Western blotting using antibody to GRP as described above.
Computer Analysis of Sequence Data-Preliminary sequence editing and analysis was performed using IBI pustell programs. The amino acid sequence deduced from the cDNA was used to search the nonredundant peptide sequence data base and the expressed sequence tag data base (National Center for Biotechnology Information) with the blast program (30). The most similar protein sequences were retrieved and aligned with the GRP sequence using the pileup program from the GCG Sequence Analysis Software Package 7.3.1 (31).
Recombinant GRP-PCR was used to generate a GRP cDNA fragment encoding amino acid residues 1-468, which was cloned into the NcoI and HindIII sites of E. coli expression vector H6-pQE-60 (32) and used to transform E. coli strain XL1Blue. Expression of GRP in this vector yields a protein with an amino-terminal sequence of Met-(His) 6 -Ala-Met-Gly-Leu-Glu-Val. Maximum expression of the recombinant protein (0.1 mg/ml E. coli culture) was obtained from cultures in Luria-Bertani medium containing 100 g/ml ampicillin, with shaking for 20 h at 37°C. The bacteria from 100 ml of culture were centrifuged at 10,000 ϫ g for 5 min and then resuspended in 10 ml of 8 M urea, 100 mM sodium phosphate, pH 6.8, 200 mM sodium chloride, 10 mM ␤-mercaptoethanol (buffer B) to lyse the bacteria. After centrifugation at 12,000 ϫ g for 15 min, the supernatant was used for purification of recombinant GRP (H 6 GRP) by Ni 2ϩ -affinity chromatography.
Approximately 5 mg of recombinant H 6 GRP was mixed with 5 ml of Ni-agarose resin (Qiagen) for 12 h at room temperature. This mixture was then packed in a column and washed with 15 ml of buffer B. Then the column was washed with a 60-ml gradient from 8 M urea in Buffer B to 0 M urea in 100 mM sodium phosphate, pH 6.8, 200 mM NaCl, 2 mM glutathione, 0.02 mM oxidized glutathione, 10% glycerol (buffer C) at 1 ml/min. Finally, the column was eluted with 100 mM imidazole in buffer C. Three-ml fractions were collected and analyzed by SDS-PAGE. Fractions containing purified H 6 GRP were stored at 4°C. Immediately before use, samples were passed through a Sephadex G-25 column (PD-10, Amersham Pharmacia Biotech) to change the buffer to 50 mM Tris, pH 7.0, 50 mM NaCl, 10% glycerol.
Labeling H 6 GRP with Fluorescein Isothiocyanate (FITC)-Purified H 6 GRP (1.6 mg) in 4 ml of 0.1 M sodium carbonate, pH 9.0, was mixed with 14 mg of fluorescein isothiocyanate (Isomer I, on Celite, Sigma), stirred at room temperature for 4 h, and then centrifuged (10,000 ϫ g, 1 min). The FITC-H 6 GRP conjugate in the supernatant was separated from the unbound dye by gel filtration using a Sephadex G-25 column (PD-10, Amersham Pharmacia Biotech). The purity of the FITC-H 6 GRP was analyzed by SDS-PAGE and Coomassie Blue staining.
Binding of FITC-labeled H 6 GRP to Microbes-An aliquot of 50 l of FITC-H 6 GRP (0.3 mg/ml) in 50 mM Tris, pH 7.0, 50 mM NaCl, and 10% glycerol was mixed with Saccharomyces cerevisiae (5 ϫ 10 7 cells), E. coli (5 ϫ 10 7 cells), or Micrococcus lysodeikticus (5 ϫ 10 7 cells) at room temperature for 8 h and then at 4°C overnight. The cells were then washed with 1 ml of TBS (0.137 M NaCl, 0.003 M KCl, 0.025 M Tris-HCl, pH 7.6) and resuspended in 1 ml of TBS. Binding of FITC-H 6 GRP was measured by subjecting the washed cells to flow cytometry analysis using a Becton Dickinson FACScan flow cytometer. For each sample, fluorescence of 10,000 cells was determined. Untreated cells of each kind were used as controls to measure background fluorescence. To test the concentration dependence of FITC-H 6 GRP binding to S. cerevisiae or E. coli, aliquots of 20 l of FITC-H 6 GRP (0 to 0.3 mg/ml) in 50 mM Tris, pH 7.0, 50 mM NaCl were mixed with 20 l of S. cerevisiae or E. coli (5 ϫ 10 7 cells/ml) suspended in TBS. After incubation at room temperature for 30 min, the cells were washed with TBS, resuspended in 1 ml of TBS, and subjected to flow cytometry as described above.
Aggregation of Microbes by H 6 GRP-An aliquot of 10 l of H 6 GRP (0.1 mg/ml) in 50 mM Tris, pH 7.0, 50 mM NaCl, and 10% glycerol was incubated with 10 l of fluorescein-conjugated Staphylococcus aureus (2 ϫ 10 9 cells/ml), E. coli (K-12 strain, 2 ϫ 10 9 cells/ml), or S. cerevisiae (2 ϫ 10 8 cells/ml) in TBS (all from Molecular Probes) at room temperature for 30 min. Samples of the cells were then applied to microscope slides, and the degree of aggregation of the cells was observed using an Olympus BH-2 fluorescence microscope.
Activation of the PPO Pathway by H 6 GRP in the Presence of Laminarin-Twenty l of H 6 GRP (0.2 mg/ml in 50 mM Tris, pH 7.0, 50 mM NaCl, and 10% glycerol) was mixed with 20 l of fresh plasma from day 3 M. sexta larvae (diluted 1:2 in anticoagulant buffer) and 5 l of laminarin (10 mg/ml or 1 mg/ml in TBS) in wells of a 96-well microplate. After incubation for 20 min at room temperature, 20 l of dopamine (16 mM in distilled water) as a phenoloxidase substrate, was added to each well. Phenoloxidase activity was determined by measuring the absorbance at 492 nm at 1-min intervals using a PowerWave X 340 microplate scanning spectrophotometer (Bio-Tek Instruments, Inc.). Control samples lacked either H 6 GRP, laminarin, or both, which were replaced with appropriate buffers.

Identification of a ␤1,3-Glucan-binding Protein from M. sexta
Larval Hemolymph -To identify hemolymph proteins that can bind to ␤1,3-glucan, curdlan (an insoluble ␤1,3-glucan preparation) was used as an affinity matrix to purify proteins from M. sexta larval plasma. After incubating curdlan with plasma and washing the curdlan with buffer, the bound proteins were eluted by treatment with SDS and ␤-mercaptoethanol at 95°C for 10 min. When this sample was analyzed by SDS-PAGE, a major protein with a molecular mass of approximately 53 kDa was observed (Fig. 1A). This protein was later named ␤1,3glucan-recognition protein (GRP). GRP accounted for more than 50% of the total protein eluted from curdlan. Several other weaker bands were also observed, including a band at ϳ17 kDa. GRP bound so tightly to curdlan that it could not be eluted without denaturing in SDS. Treatments with acid (pH 1.5), high salt concentration (2 M NaCl), or 8 M urea did not elute GRP from curdlan. Approximately 3 mg of denatured GRP was obtained from 100 ml of hemolymph by this affinity method.
GRP eluted from curdlan was further purified by SDS- PAGE, and the 53-kDa band on the gel was excised and injected into a rabbit to raise an antiserum. This antiserum recognized a single protein band of 53 kDa in Western blot analysis of M. sexta plasma (Fig. 1B). The sequence of the amino-terminal 23 residues of this protein was determined by Edman degradation (Fig. 3). The amino-terminal sequence (19 residues) of the protein in the 17-kDa band that bound to curdlan also was determined and found to be identical to that of the 53-kDa GRP. The 17-kDa protein thus appears to be an amino-terminal fragment of GRP. When plasma proteins bound to curdlan were eluted by SDS treatment in the absence of ␤-mercaptoethanol, the 17-kDa amino-terminal fragment of GRP was still observed by SDS-PAGE analysis (data not shown), suggesting that this fragment bound to curdlan by itself rather than through a disulfide link to the carboxylterminal fragment of GRP.
Purification and Analysis of GRP-To obtain nondenatured GRP for functional studies, GRP was purified from fifth instar larval plasma by a series of chromatographic steps that did not involve binding to curdlan. The purification procedure consisted of ammonium sulfate fractionation, followed by column chromatographic separations using Q-Sepharose, ConA, and hydroxyapatite (Fig. 2). The 20 -35% ammonium sulfate fraction contained about 50% of the total GRP in plasma. This step removed most of the storage proteins, which are present at high concentration in the plasma. GRP bound to Q-Sepharose at pH 7.6 in a buffer containing 50 mM Tris and 20 mM NaCl. When the column was eluted with a sodium chloride gradient from 20 to 200 mM at pH 7.6, GRP eluted at the end of the gradient ( Fig.  2A). No additional GRP was eluted when the column was washed with 1 M sodium chloride. Fractions from the Q-Sepharose column containing GRP were combined and applied to a ConA column. GRP did not bind to the ConA column and was present in the flow-through fractions. However, other glycoproteins that did bind to ConA were separated from GRP in this step. Pooled fractions from the ConA column were applied to a hydroxyapatite column pre-equilibrated with 10 mM sodium phosphate buffer, pH 6.8. GRP eluted near the end of a sodium phosphate gradient from 10 to 100 mM.
When the fractions containing GRP were analyzed by SDS-PAGE, GRP appeared as a single band with an apparent molecular mass of 53 kDa (Fig. 1A, lane 2). Analysis by MALDI mass spectrometry gave a mass of 53,583 Da for GRP. Approximately 0.3 mg of pure GRP was obtained from 100 ml of plasma. When this purified GRP was analyzed by gel filtration HPLC, it eluted as a single peak, with an elution volume consistent with a native molecular mass of about 70 kDa. This result suggests that GRP exists as a monomer in solution.
The amino-terminal sequence of GRP isolated as described above was identical to that obtained from the protein eluted from curdlan. To obtain sequence information from an internal peptide fragment, GRP was treated with formic acid to hydrolyze between Asp and Pro residues. The resulting peptide fragments were separated by two-dimensional gel electrophoresis and transferred to a polyvinylidene difluoride membrane. From one of the peptide spots on the membrane with a mass of approximately 25 kDa, an amino acid sequence of 17 residues was determined by Edman degradation (Fig. 3).
cDNA Cloning and Deduced Protein Sequence-The aminoterminal sequence of GRP and the internal amino acid sequence of the fragment derived from formic acid hydrolysis were used to design degenerate oligonucleotide primers for PCR, to amplify a fragment of a GRP cDNA. PCR was carried out as described under "Experimental Procedures" using single-stranded cDNA synthesized from fat body mRNA as a template. A 700-base pair PCR product was obtained and cloned into a plasmid vector, and this cloned fragment was used as a hybridization probe to screen a larval fat body cDNA library. Forty-eight positive clones were isolated from a screen of approximately 3 ϫ 10 5 plaques. The GRP cDNA clone (pGRP9) containing the longest insert (approximately 1.6 kilobases) was sequenced.
The sequence of pGRP9 contains a 47-nucleotide 5Ј noncoding region, an open reading frame of 1461 nucleotides, and a 3Ј-untranslated sequence of 56 nucleotides (Fig. 3). The open reading frame encodes 487 amino acid residues. The first 19 amino acid residues make up a secretion signal peptide, and the amino-terminal sequence of the mature protein begins at residue 20. The internal amino acid sequence of 17 residues determined from a peptide obtained after hydrolysis of GRP with formic acid was consistent with the deduced sequence between residues 240 and 256 (Fig. 3). The calculated molecular mass of the 468-residue mature protein is 52,335 Da, which is 1,247 Da less than the mass determined by mass spectrometry. There are two putative N-linked glycosylation sites in the carboxyl-terminal region of the protein. After GRP was treated with N-glycosidase F, it migrated slightly faster than untreated GRP in SDS-PAGE analysis (Fig. 4), indicating that this glycosidase removed approximately 1 kDa of N-linked car-bohydrate from the protein. There are five Cys residues in the mature protein. Therefore, GRP must contain at least one free Cys that is not involved in a disulfide bond. The calculated pI of the mature GRP is 5.0. This is in agreement with the results of two-dimensional gel electrophoresis, in which the pI of GRP was estimated to be approximately 5.5 (data not shown).
Sequence Comparisons-Searching of the GenBank sequence data bases indicated that the GRP open reading frame has significant sequence similarity with bacterial ␤-glucanases and with several invertebrate proteins that contain glucanase- The deduced amino acid sequence is shown below the cDNA sequence. Amino acid residues in the mature protein are assigned positive numbers, and those in the signal peptide are assigned negative numbers. Two potential N-linked glycosylation sites are marked with a OE. A single underline represents sequences of the amino-terminal of mature GRP and the isolated peptide determined by Edman degradation. In the cDNA sequence, the polyadenylation sequence AATAAA is double underlined, and the termination codon TGA was marked with an asterisk. like regions. The carboxyl-terminal region (residues 174 -442) was similar to a group of ␤1,3-glucanases, with the highest degree of similarity among these to enzymes from a sea urchin and from Bacillus circulans (Fig. 5). The amino-terminal region was similar to several proteins from invertebrates, which also contain a carboxyl-terminal glucanase-like sequence. GRP was similar throughout its sequence to a Gram-negative bacteriabinding protein (GNBP) from the silkworm, B. mori (24) (42% identity), and to homologous putative GNBPs from the fall webworm, Hyphantria cunea (33) (39% identity), and a mosquito Anopheles gambiae (34) (38% identity). Partial amino acid sequences available from cDNAs corresponding to expressed sequence tags (EST) from B. mori and from Drosophila melanogaster also were quite similar to GRP. M. sexta GRP is 59% identical to the protein encoded by the B. mori EST (accession number AU004243) but only 37% identical to the corresponding amino-terminal region of B. mori GNBP, which suggests that this EST, rather than GNBP, is the silkworm ortholog of M. sexta GRP. In addition to these insect proteins, GRP is similar to sequences of coelomic cytolytic factor 1 (CCF-1), which is a glucan-binding protein from an earthworm, Eisenia foetida (23) (34% identity) and to a ␤1,3-glucanase from a sea urchin, Strongylocentrotus purpuratus (21) (34% identity) (Fig. 5). These proteins are similar to GRP in both the glucanase domain and the amino-terminal region, although the amino-terminal extension is much shorter in CCF-1 than in the sea urchin glucanase or the insect proteins.
The most conserved region of the glucanase-like domain of GRP and the similar invertebrate proteins is in a sequence that aligns with a region near the active site of bacterial ␤1,3glucanases and ␤1,3-1,4-glucanases (Fig. 5). However, GRP lacks four conserved residues that line the active site of such enzymes, and in particular has nonconservative replacements of two Glu residues that are believed to act as the catalytic residues responsible for cleaving ␤1,3or ␤1,4-glycosidic bonds in the bacterial glucanases (35). This observation is consistent with the fact that we have not detected any ␤1,3-glucanase activity associated with GRP (data not shown).
In addition to the sequences discussed above, BLAST searches using the GRP sequence identified a number of se- circulans ␤-1,3-glucanase A1 (P23903), the bacterial glucanse most similar to M. sexta GRP; Bm-GLUC, Bacillus macerans endo-1,3-1,4-␤glucanase (P23904), a related glucanase whose structure has been determined by x-ray crystallography (35). Residues conserved in all of the invertebrate sequences are marked with *, and residues identical in at least 4 of the invertebrate sequences are marked with ϩ. Residues corresponding to the active site of the B. macerans glucanase are marked with #. Positions in M. sexta GRP which are identical to at least one of the bacterial glucanases are underlined. Numbering corresponds to the M. sexta GRP sequence as shown in Fig. 3. In the sequences derived from expressed sequence tags, regions that have not yet been sequenced are shown with ϳ symbols. A marks the approximate site of proteolytic cleavage of GRP to produce the 17-kDa fragment shown in Fig. 1. quences with lower but still significant similarity to the conserved region near the glucanase active site. These include ␤1,3-glucanases from bacteria and plants, an expressed sequence tag from the solitary ascidian, Ciona intestinalis, and a ␤1,3-glucanase-like domain from the ␣-chain of coagulation factor G from the horseshoe crab, Tachypleus tridentatus (22) (Fig. 6). In all of these sequences, the putative glucanase active site residues are conserved, in contrast to GRP and the other proteins from lepidopteran insects. The glucanase domain, widespread in evolution, appears to have evolved in the insects to function as a glucan-binding protein and has lost its hydrolase activity and function.
Expression of GRP in Vivo-In Northern blot analysis, the GRP cDNA hybridized to a 1.5-kilobase nucleotide band in RNA samples from larval fat body, consistent with the size expected from the cloned cDNA. No band was detected in RNA from hemocytes or integument (Fig. 7A). These results indicate that fat body is the primary site of GRP synthesis, as is true for most insect hemolymph proteins. Furthermore, when fat body was cultured in vivo, GRP was first detected in the culture medium after 7 h and its concentration gradually increased up to 64 h in culture (data not shown). This result is also consistent with synthesis of GRP by fat body and secretion into hemolymph. The level of GRP mRNA in fat body did not increase significantly after larvae were injected with bacteria and yeast (Fig. 7B). This lack of inducibility distinguishes GRP expression from that of three similar proteins from insects, B. mori GNBP and homologous proteins from H. cunea and A. gambiae, which are synthesized in response to exposure to microbial elicitors (24,33,34).
Production of Recombinant GRP-To obtain sufficient amounts of GRP for functional studies, GRP was expressed as a recombinant protein in E. coli. Recombinant GRP (H 6 GRP), which contains an amino-terminal 6-histidine tag, was insoluble in the buffer extract obtained after lysis of E. coli by sonication. It could be dissolved in 8 M urea and was purified by affinity chromatography. After H 6 GRP bound to the Ni 2ϩ affinity column, it was renatured using a linear gradient of urea from 8 to 0 M, and then eluted by washing with 100 mM imidazole. This H 6 GRP preparation was soluble and contained a single band at 53 kDa when analyzed by SDS-PAGE and Coomassie Blue staining (Fig. 1A, lane 3). The renatured H 6 GRP remained stable at 200 g/ml and at 4°C in buffer C containing 10% glycerol for at least 1 month.
Binding of H 6 GRP to Microorganisms-To investigate potential functions of GRP, experiments were conducted to test whether the recombinant protein can bind to microorganisms. H 6 GRP labeled with fluorescein isothiocyanate (FITC-H 6 GRP) was incubated with S. cerevisiae, E. coli, and M. lysodeikticus as examples of yeast, Gram-negative, and Gram-positive bacteria. After the incubation period, the microbial particles were washed, and then the fluorescence due to bound FITC-H 6 GRP was measured by flow cytometry. We observed binding of FITC-H 6 GRP to yeast (whose cell walls contain ␤1,3-glucan) and to bacteria (Fig. 8A). When we normalized the fluorescence intensity per cell to obtain fluorescence intensity per m 2 of cell surface area, we found that GRP bound to all three types of microorganisms to about the same degree (Fig. 8B). The binding of FITC-H 6 GRP to yeast and to E. coli was concentrationdependent and saturable (Fig. 8, C and D), with a K d of 77 g/ml (1.5 M) for binding to yeast and a K d of 65 g/ml (1.2 M) for E. coli. The concentration of GRP in plasma is approximately 30 g/ml (determined by densitometry analysis of Western blot samples as described under "Experimental Procedures"). These results indicate that such binding would occur at physiological GRP concentrations. Because H 6 GRP was able to bind to bacteria and yeast, we tested whether H 6 GRP can aggregate these microorganisms. The presence of H 6 GRP at 50 g/ml caused significant aggregation of S. cerevisiae, E. coli, and S. aureus, whereas bovine serum albumin used as a control at the same concentration did not lead to aggregation of these organisms (Fig. 9). These results suggest that one function of GRP may be to aggregate invading microorganisms, leading to more efficient clearance by hemocytes.
Activation of the PPO Pathway by GRP in the Presence of Laminarin-Activation of PPO in insect plasma can be triggered by ␤1,3-glucan (17,36,37). We performed experiments to determine whether interaction of GRP with ␤1,3-glucan might participate in activation of this pathway. M. sexta larval plasma was diluted 1:2 in anticoagulant buffer, which decreased the endogenous GRP concentration from 30 to 10 g/ ml. This diluted plasma was incubated with different amounts of the ␤1,3-glucan laminarin and 100 g/ml H 6 GRP at room agarose gel containing formaldehyde. The RNA was transferred to a nitrocellulose membrane and then hybridized with 32 P-labeled M. sexta GRP cDNA. Approximately equal RNA loading of each lane was confirmed by probing a duplicate blot with 32 P-labeled cDNA for M. sexta ribosomal protein S3 (29). The approximate size of the mRNA was derived from RNA standards. B, RNA samples from fat body of fifth instar larvae injected with a mixture of E. coli and yeast 24 h earlier were analyzed by Northern blotting as described above. Control larvae were injected with PBS. The bands intensities were quantified using one-dimensional Image Analysis Software (Kodak Digital Science), and the ratio of GRP mRNA to ribosomal protein S3 mRNA was calculated. Bars represent the mean Ϯ S.E. (n ϭ 5). temperature for 20 min, and then the phenoloxidase activity in the plasma was determined (Fig. 10). Laminarin alone (1 g/l) did not trigger the PPO pathway within 20 min. However, when laminarin (1 or 0.1 g/l) combined with 100 g/ml H 6 GRP was added to plasma, PPO activity significantly increased. H 6 GRP alone stimulated a much smaller degree of PPO activation. These results suggest that GRP serves as a biosensor for ␤1,3-glucan in insect defense and that binding of GRP to fungi may lead to activation of a proteinase in the PPO activation cascade. DISCUSSION We have purified from hemolymph of the tobacco hornworm, M. sexta, a 53-kDa protein that binds very tightly to ␤1,3glucans. We have named this protein GRP, which is the name used for a very similar protein from hemolymph of silkworms (17). We first isolated GRP by taking advantage of its affinity to curdlan, an insoluble ␤1,3-glucan, but this protein bound so tightly to curdlan that it could only be eluted by treatment with SDS. We then developed a method to purify GRP that did not require denaturing conditions. GRP is present in hemolymph  at all developmental stages of M. sexta at approximately 30 g/ml (0.6 M). In immunoblot analysis using antiserum to M. sexta GRP, we have detected an immunoreactive band at ϳ50 kDa in plasma from other insect species, including two lepidopterans: P. interpunctella and H. virescens; and a dipteran: D. melanogaster, 2 suggesting that the presence of a similar GRP in plasma may be expected in other insects. M. sexta GRP has a sequence very similar to that of a ␤1,3-glucanrecognition protein isolated from hemolymph of the silkworm, B. mori (17), which is a 62-kDa protein that binds very strongly to curdlan and is involved in phenoloxidase activation. 3 We isolated a cDNA clone that encodes GRP from a larval fat body library. The amino acid sequence of GRP, deduced from the nucleotide sequence of the cloned cDNA, indicated that the protein is composed of two regions, which appear to represent amino-terminal and carboxyl-terminal domains. The carboxylterminal domain is similar in sequence to ␤1,3and ␤1,3-1,4glucanases from bacteria (38) and to a ␤1,3-glucanase from a sea urchin (21), but GRP had no detectable ␤1,3-glucanase activity. Two catalytic Glu residues, conserved in the active site of the bacterial glucanases (35) are replaced with Leu and Cys in Manduca GRP, which explains the absence of glucanase activity (Fig. 4). Thus, GRP may be an evolutionary descendent of a glucanase that has lost its catalytic activity but has maintained the ability to bind to ␤1,3-glucans. Clotting factor G from a horseshoe crab, Tachypleus tridentatus, also contains a region with similarity to these glucanases (Fig. 6), and it is possible that binding of ␤1,3-glucans to this domain triggers the activation of the enzyme's proteolytic subunit (22). Although no sequences similar to the glucanase-like domain of GRP appear to have been found so far in vertebrates, the discovery of a homologous sequence in an ascidian (a primitive chordate) (39) suggests that such glucan-binding proteins may exist in vertebrates as well.
A group of proteins from insects and from an earthworm are also similar in sequence to GRP. They contain the carboxylterminal glucanase-like domain and also a unique amino-terminal domain. These include a GNBP from hemolymph of the silkworm, B. mori. (24). GNBP-like sequences have also been identified in the fall webworm, H. cunea (33), and a mosquito, A. gambiae (34), although these two proteins are known so far only from cDNA sequences and have not been tested for function. A partial B. mori sequence encoded by an expressed sequence tag cDNA is more similar to M. sexta GRP than is B. mori, GNBP (Fig. 5), suggesting that this cDNA may encode the silkworm GRP.
A protein called CCF-1 from an earthworm, E. foetida, also has a sequence similar to that of GRP (23). CCF-1 binds to ␤1,3-glucan and bacterial lipopolysaccharide and thus may function in a manner similar to GRP. Although the glucanase active site residues are conserved in CCF-1 (Fig. 5), it also lacks apparent glucohydrolase activity (23). The catalytic Glu residues are also conserved in A. gambiae GNBP (Fig. 5), but whether it can function as a glucanase is unknown.
The insect proteins related to GRP as well as the earthworm CCF-1 and the sea urchin ␤1,3-glucanase contain an aminoterminal domain that lacks similarity to glucanases or to any other sequences currently in the Genbank data base. This amino-terminal domain (residues 1-173 in M. sexta GRP) has a molecular mass of approximately 17 kDa in M. sexta GRP and in the related invertebrate proteins, except for CCF-1 and the A. gambiae protein, in which this region is somewhat smaller (Fig. 5). We identified a 17-kDa polypeptide from M. sexta plasma that bound to curdlan and had an amino-terminal sequence identical to that of GRP (Figs. 1 and 5). This polypeptide appears to represent the amino-terminal domain of GRP and perhaps is a product of endogenous proteolytic activity in plasma. We observed that this species accumulated upon storage of plasma, which is consistent with the hypothesis that it is a degradation product of GRP and that the two domains are linked by a region that is sensitive to proteolysis. Although the function of this domain is not known, it appears that it can bind to curdlan independently from the carboxyl-terminal glucanase-like domain and thus may have its own glucan-binding site.
Northern blot analysis indicated that GRP mRNA is present in fat body (a tissue analogous to mammalian liver) and not in other tissues tested, including hemocytes. Cultured fat body released GRP into the medium, which is consistent with the hypothesis that fat body is the primary source of GRP in plasma. A significant difference between M. sexta GRP and the insect GNBPs is that synthesis of the GNBPs is strongly induced by microbial infection (24,33,34), whereas GRP is constitutively expressed and not induced as an acute phase response protein.
GRP bound to the surface of yeast cells, which contain ␤1,3glucans in their cell walls, and to Gram-negative and Grampositive bacteria. E. foetida CCF-1 also was shown to bind to ␤1,3-glucan and bacterial lipopolysaccharide (23). The ability of GRP to bind to microbial surfaces suggests that it may function as a pattern recognition molecule for detection of bacterial or fungal pathogens. Incubation of GRP with yeast or bacteria caused them to aggregate, which could improve the efficiency of clearance of these microorganisms by hemocytes through phagocytosis or nodule formation (40).
Activation of prophenoloxidase in hemolymph is a commonly observed response to microbial infection in insects and many other invertebrates (3). When GRP preincubated with the soluble ␤1,3-glucan, laminarin, was added to diluted plasma, the prophenoloxidase activation pathway, which involves a serine proteinase cascade (3) was triggered, and active phenoloxidase accumulated in the plasma. This activation occurred to a much lesser degree and took a significantly longer time when laminarin or GRP was added to plasma separately. This result suggests that a complex of GRP with laminarin stimulated activation of the phenoloxidase cascade. Similar results have been observed with B. mori GRP (17) and support the conclusion that these proteins act as pattern recognition proteins as they bind to microbial surfaces and stimulate a defensive response.
Proteins that bind to ␤1,3-glucans and stimulate phenoloxidase activation have been identified in other invertebrate species and appear to fall into several classes. A 100-kDa ␤1,3glucan-binding protein, which enhances activation of phenoloxidase has been isolated from a crayfish, P. leniusculus (12). Proteins very similar to the crayfish glucan-binding protein have also been identified in shrimp (13,14). These 100-kDa proteins from crustaceans appear to represent one family of invertebrate ␤1,3-glucan-binding proteins that function as pattern recognition molecules. Two different types of ␤1,3glucan-binding proteins have been isolated from hemolymph of cockroaches from the genus Blaberus. An approximately 90-kDa protein isolated from Blaberus craniifer, with subunits of 63 and 52 kDa, binds to laminarin and enhances activation of proPO from a hemocyte lysate (18). A different type of ␤1,3glucan-specific lectin isolated from Blaberus discoidalis is a hexamer of 80-and 82-kDa subunits, and appears to be a member of the hexamerin family of arthropod hemolymph proteins (20). Similar proteins have been isolated from plasma of a group of different cockroach species (19). These cockroach ␤1,3glucan-specific lectins, in combination with laminarin, are also able to activate the phenoloxidase system (20).
M. sexta GRP, B. mori GRP, and the earthworm protein CCF1 are members of another family of circulating proteins that bind to ␤1,3-glucans and lead to activation of phenoloxidase. The mechanism by which these pattern recognition molecules function to stimulate activation of the proteolytic cascade is not yet known. Perhaps binding to the polysaccharide produces a conformational change in GRP, which confers an ability to interact with a proteinase zymogen. Such an interaction may stimulate autoactivation of the proteinase, as is thought to occur in the lectin-mediated pathway for complement activation (41). Further study of these ␤1,3-glucan-recognition proteins in arthropods should lead to a better understanding of the function and evolution of pattern recognition molecules that bind to polysaccharides on the surface of microbial pathogens and stimulate innate immune responses.