Porcine Pancreatic α-Amylase Shows Binding Activity towardN-Linked Oligosaccharides of Glycoproteins*

Porcine pancreatic α-amylase was shown by interaction analyses using a resonance mirror detector and α-amylase-immobilized Sepharose to bind with glycoproteins possessingN-glycans but not O-linked mucin-type glycans. Direct binding of three types of N-glycans to the α-amylase was demonstrated by surface plasmon resonance. Binding with biotin-polymer sugar probes revealed that the α-amylase has affinity to α-mannose, α-N-acetylneuraminic acid, and β-N-acetyllactosamine, which are components ofN-glycans. The binding of glycoproteins or carbohydrates enhanced the enzyme activity, indicating that the recognition site forN-glycans is different from its catalytic site. The binding activity was unique to porcine pancreatic α-amylase and was not observed for α-amylase from saliva, wheat, and fungus.

Porcine pancreatic ␣-amylase was shown by interaction analyses using a resonance mirror detector and ␣-amylase-immobilized Sepharose to bind with glycoproteins possessing N-glycans but not O-linked mucintype glycans. Direct binding of three types of N-glycans to the ␣-amylase was demonstrated by surface plasmon resonance. Binding with biotin-polymer sugar probes revealed that the ␣-amylase has affinity to ␣-mannose, ␣-N-acetylneuraminic acid, and ␤-N-acetyllactosamine, which are components of N-glycans. The binding of glycoproteins or carbohydrates enhanced the enzyme activity, indicating that the recognition site for N-glycans is different from its catalytic site. The binding activity was unique to porcine pancreatic ␣-amylase and was not observed for ␣-amylase from saliva, wheat, and fungus.
␣-Amylase (EC 3.2.1.1) is a well known starch hydrolase discovered in 1830 that has important roles in energy acquisition in animals, plants, and microbes. It is an endo-type enzyme that typically cleaves ␣1,4-glucose linkages in starch which is further processed by the exoenzymes. Because in vivo modulation of ␣-amylase activity in pancreatic secretions is important in nutrition and for developing therapies for diabetes mellitus, interaction between porcine pancreatic ␣-amylase (PPA) 1 and its inhibitors, including antinutrients, has been investigated (1)(2)(3).
Legumes contain large amounts of antinutrients such as lectins, tannins, phytates, and enzyme inhibitors that may be responsible for lowering the rate of starch digestion and decreasing the blood glucose response. The mechanisms of the antinutrient effect of lectins on pancreatic ␣-amylase activity have been studied by several groups (4 -7). Because cDNA of the ␣-amylase inhibitor from kidney bean, Phaseolus vulgaris, exhibits homology with a lectin from the same source (PHA) (8) and because PHA inhibits the growth of insect larvae in vivo (9), PHA together with arcelin and ␣-amylase inhibitor are classified as members of a family of defense proteins (10). Plant ␣-amylase inhibitors inhibit only animal ␣-amylase, exhibiting species specificity toward insect larvae but not its plant homologue. However, purified legume lectins of various kidney beans were found instead to increase the PPA activity in vitro (11), and factors other than direct interactions of lectins with pancreatic ␣-amylase are considered to lower the rate of starch digestion (12). Recently, the antinutrient effect of dietary lectins has been attributed to the fact that lectins bind to glycoproteins in the intestinal mucosa of mammals and insects, and it is hypothesized that lectin toxicity results from this initial binding, which interferes with immune function, gut bacterial population, or metabolic or hormonal reactions (13).
Most plant lectins and ␣-amylase inhibitors are glycoproteins. Because PPA possesses three potential glycosylation sites, we studied the possibility that carbohydrate-dependent interaction between ␣-amylase and various glycoproteins occurs in the physiological processes. PPA specifically recognizes N-linked glycans. The novel carbohydrate binding activity of PPA may provide new insight into the secretion, targeting, and modulation mechanism of this long known enzyme, and the antinutrient activities of plant lectins in the digestive tract.
Affinity Chromatography on PPA-Sepharose-PPA-Sepharose was prepared by coupling 15 mg of PPA with formyl-Sepharose (2 g) (15) in 20 ml of 0.15 M NaCl, 15 mM phosphate buffer (pH 7.4) containing 60 mg of NaCNBH 3 at 4°C for 2 days. Excess formyl groups were blocked with 40 ml of 1 M Tris-HCl (pH 7.4) containing 124 mg of NaCNBH 3 at room temperature for 1 h. About 5 mg of protein per ml of gel was immobilized. A PPA-Sepharose column (0.75 ϫ 6 cm) was equilibrated with 15 mM sodium succinate buffer containing 20 mM CaCl 2 and 0.5 M NaCl (pH 5.6). 200 l of each glycoprotein solution (1 mg/ml) was applied to the column and washed with the same buffer. The bound glycoproteins were eluted with 0.2 M methyl ␣-D-mannoside, methyl ␤-D-galactoside in the same buffer, or 0.1 M citrate buffer (pH 3.0) and monitored by absorbance at 280 nm.

Quantitative Interaction Analyses between PPA and Various Glycoproteins by Resonance Mirror Detector (IAsys)-IAsys (Affinity Sensors,
Cambridge, UK), a biosensor based on optical resonance, was used for kinetic analyses of interactions between PPA and glycoproteins. Immobilization was performed with 200 l of PPA solution (1 mg/ml) in 10 mM sodium acetate buffer (pH 5.5) containing 0.2 M methyl ␣-D-mannoside by adding it to a CM-dextran-coated cuvette that had been activated with N-hydroxysuccinimide and N-ethyl-NЈ-(dimethylaminopropyl)carbodiimide, and reacting it at room temperature for 30 min. After immobilization of PPA, excess reactive groups were blocked by 1 M ethanolamine (pH 8.0).
To measure binding curves, various concentrations of glycoproteins in 10 mM sodium acetate buffer (pH 5.5) containing 150 mM NaCl and 5 mM CaCl 2 were added to the cuvette, and after 200 s the cuvette was washed with the same buffer. Glycoproteins that bound to the cuvette were eluted with 0.1 M methyl ␣-D-mannoside to regenerate the cuvette or 3 M NaCl in the same buffer or diluted NaOH (pH 10) for extensive washing. Binding constants were calculated from their binding curves using connected software, according to the equation, dR/dt ϭ k a where R is response (arc seconds), [S] 0 is initial concentration of ligand, and R max is maximum response that will be seen at concentrations of ligand high enough to saturate the binding site of ␣-amylase. On introducing k on ϭ k a [S] 0 ϩ k d , k on is obtained from the slope of dR/dt versus R plot. The association constant, K a , is calculated according to the equation, K a ϭ k a /k d , from the slope and the intercept of k on versus [S] 0 under various concentrations of ligands.
Interaction Analyses between PPA and PA-oligosaccharides by Surface Plasmon Resonance (BIAcore)-For binding studies between PPA and PA-oligosaccharides, a BIAcore 2000 SPR apparatus (BIACORE AB, Uppsala, Sweden) was used. After equilibration with HEPESbuffered saline on a CM5 sensor chip, the surface of the sensor chip was activated with an amine coupling kit, and PPA (3 mg/ml) in 10 mM sodium acetate buffer ( Binding Studies with BP Probes-␣-Amylases of various origins were immobilized overnight in TBS (10000 -9.8 ng/100 l) on each well of a microtiter plate (Immulon 1, Dynatech Laboratories, Chantilly, VA) at 4°C. All other procedures were performed at room temperature using TBS containing 5 mM CaCl 2 , according to the procedure described (16). The incubation time with sugar-BP probes was 4 h. In the case of inhibition assays, the concentration of PPA added to each well was fixed at 100 g/ml. Various concentrations of saccharides were preincubated with immobilized PPA in the wells for 1 h, and 50 l of Man-BP probe (10 g/ml) was added, and they were incubated overnight at 4°C.
Measurement of Enzyme Activity-Enzyme activity was measured according to Bernfeld (17). PPA solution in PBS (0.1 ml) was preincubated at 25°C for 5 min, and various concentrations of glycoproteins (10 l) were added. After incubation for 1 h, 0.2 ml of 1% starch in 20 mM phosphate buffer (pH 6.9) containing 0.006 M NaCl was added, and the mixture was reacted at room temperature for 30 min. To stop the reaction, 0.2 ml of 3,5-dinitrosalicylic acid was added, and it was boiled in a waterbath for 5 min to develop the color. The reaction mixture was cooled to room temperature, diluted with 2 ml of water, and measured at 540 nm. Maltose was used as a standard for reducing sugar.
Carbohydrate Analyses of PPA-Neutral carbohydrate analysis was carried out from blotted polyvinylidene difluoride membranes according to the method described previously (18). Lectin reactivity was examined on the membrane according to the method described previously (16) using biotinyl lectins, concanavalin A, Psathyrella velutina lectin, and P. vulgaris leukoagglutinin. Fig. 1 shows the affinity of glycoproteins on a PPAimmobilized Sepharose chromatography column. As shown in Fig. 1A, fetuin bound to the column and was dissociated from it by pulse elution with 0.2 M Me ␣-D-Man (arrows, b and d) but not with Me ␤-D-Gal (arrows, a and c). As shown in Fig. 1, B and C, the bound ovalbumin and RNase B, respectively, were eluted with methyl ␣-D-mannoside but not with methyl ␤-D-galactoside (data not shown). On the contrary, ␣-amylase inhibitor bound to the column and eluted with 0.1 M sodium citrate buffer (pH 3.0) as reported previously (8), but it was not eluted with 0.2 M methyl ␣-D-mannoside, in contrast to other glycoproteins tested here (data not shown).

Affinity Chromatography of Glycoproteins on a PPA-Sepharose-
Interaction between PPA and Various Glycoproteins Analyzed by IAsys-For immobilization of PPA in the cuvette, methyl ␣-D-mannoside was necessary to protect the carbohydrate-binding site; otherwise, fetuin did not bind to the PPAimmobilized cuvette (data not shown). Changes of the resonance angle caused by the immobilization or interaction were measured in arc seconds, a response unit of IAsys (163 arc s ϭ 1 ng/mm 2 ), and plotted versus time in the binding curves. The total amount of immobilized PPA was about 2,000 arc s.
The optimum condition for interaction analyses was examined using fetuin. As shown in Fig. 2A, fetuin bound to PPA-immobilized cuvettes at pH 4.0 -7.4 with optimal binding at pH 4.5. Between pH 4.5 and 5.0, however, binding occurred too quickly to produce a biphasic curve, and pH 5.5 was adapted for quantitative interaction analyses because it is closer to the pH of pancreatic fluid. As shown in Fig. 2B, transferrin and fetuin bound to the PPA cuvette at pH 5.5, but BSM and BSA did not. Asialofetuin, CTA, and ovalbumin also bound to the PPA cuvette (data not shown). The bound glycoproteins were eluted with 0.2 M methyl ␣-D-mannoside, but when the elution was incomplete, 3 M NaCl was used to complete it. When 10 mM NaOH was used to wash the cuvette, the addition of Ca 2ϩ was necessary to restore the binding activity of PPA. CTA, a galactose/N-acetylgalactosamine-specific lectin, was not eluted from the cuvette with 0.2 M D-galactose, indicating that the binding was not caused by the carbohydrate recognition of CTA.
Quantitative Parameters for Interaction between PPA and Glycoproteins-Binding constants were measured for each glycoprotein from the plots of k on versus [S] 0 from the binding curves at various ligand concentrations, as shown in Fig. 2C. Association rate constants (k a ), dissociation rate constants (k d ), and affinity constants (K a ) were calculated as described under "Experimental Procedures" and are summarized in Table I. Transferrin bound best among the samples, with a k a of 2.3 ϫ 10 5 M Ϫ1 s Ϫ1 and a K a of over 2.3 ϫ 10 7 M Ϫ1 . Desialylation of fetuin markedly decreased k a and consequently K a to 1/30 of fetuin, suggesting that sialyl residues contribute to the binding. On the other hand, the k d values of all the samples tested were close. All the bound glycoproteins have N-linked oligosaccharides of various types, i.e. complex types (transferrin and fetuin), plant complex types with ␤-1,2linked xylose and ␣-1,3-linked fucose (CTA), oligomannose, and hybrid types (ovalbumin) as summarized in Table I. On the contrary, of the unbound glycoproteins, BSM possesses only O-linked oligosaccharides, and BSA does not contain any glycans. These results suggest that a common trimannosyl moiety of N-glycans may play an essential role in the interaction of glycoproteins with PPA.
Interaction between PPA and RNase B by IAsys-Except RNase B, the glycoproteins examined here bound to PPA optimally at a pH lower than 5, whereas RNase B bound optimally at pH 7-7.5 as well as at pH 5 (data not shown). As shown in Fig. 2D, RNase B bound to PPA better than its unglycosylated isoform RNase A, especially at pH 5.5, suggesting that a high mannose type N-glycan is involved in the interaction. Because RNase B and A adsorbed significantly to underivatized CMdextran cuvettes, probably due to electrostatic interaction, quantitative parameters could not be obtained.
Interaction between PPA and PA-oligosaccharides by BIAcore-The total amounts of immobilized PPA and BSA were 24,381 and 18,392 BIAcore resonance units (1000 resonance units ϭ 1 ng/mm 2 ), respectively. The changes of resonance units induced by binding of analytes to PPA-immobilized flow cell were corrected for bulk effect by subtracting the changes on the BSA-immobilized reference cell. As shown in Fig. 3, the binding and dissociation occurred rapidly at the start and end of the injection of PA-oligosaccharides, demonstrating the specific binding of the PA-oligosaccharides to PPA with quick association and dissociation rates. From the changes in response units and the molecular weight of each PA-oligosaccharide, the bound amounts were calculated to be 0.22, 0.18, and 0.044 pmol/mm 2 for NeuAc 2 Gal 2 Man 3 GlcNAc 4 -PA, NeuAc 3 -Gal 3 Man 3 GlcNAc 5 -PA, and Man 8 GlcNAc 2 -PA, respectively. The differential binding of PPA to PA-oligosaccharides clearly indicates its relative binding affinity toward oligosaccharides; PPA binds best with NeuAc 2 Gal 2 Man 3 GlcNAc 4 -PA, and then NeuAc 3 Gal 3 Man 3 GlcNAc 5 -PA, and to a lesser extent Man 8 -GlcNAc 2 -PA. Although absolute K a values could not be obtained because the amount of the samples was limited, the oligosaccharide specificity shown here agreed well with the binding affinities of PPA for glycoproteins that possess these oligosaccharides, that is transferrin, fetuin, and ovalbumin (Table I).
In the inhibition test, 50 mM mannan inhibited the interaction between PPA and the ␣-mannose-BP probe by about 75%, but methyl ␣-D-mannoside, methyl ␤-D-galactoside, or methyl ␣-D-glucoside did not, even at the concentration of 0.1 M (data not shown). From these results, the glycoprotein binding is considered to be due to the affinity of PPA for multiple carbohydrate residues that are components of N-linked glycans.
␣-Mannose-BP Binding Activity of ␣-Amylases of Different Origins-As shown in Fig. 4B, PPA from three manufacturers all exhibited the mannose-BP binding activity. In contrast, ␣-amylases of barley, Bacillus, and human saliva origins did not exhibit obvious binding activity toward the mannose-BP probe.
Carbohydrate Analyses of PPA-Carbohydrate analyses indicated that PPA contained 0.04 mol each of D-mannose and N-acetyl-D-glucosamine per 1 mol of PPA, and a trace amount of D-galactose, which is lower than the reported carbohydrate contents for PPA (20). L-Fucose, N-acetyl-D-galactosamine and D-xylose were not detected. The results indicate that although it has three potential glycosylation sites, most parts of PPA are unglycosylated, and a very small amount (Ͻ1% of potential sites) is N-glycosylated. PPA electroblotted onto the membrane after SDS-PAGE was very slightly stained with concanavalin A and only weakly stained with P. velutina lectin and P. vulgaris leukoagglutinin, suggesting that the glycosylated molecules contain sialylated tri-or tetraantennary branched N-glycans.
Effect of Glycoproteins on an Enzyme Activity of PPA-As shown in Fig. 5, glycoproteins that interacted with PPA did not inhibit the enzyme activity but rather enhanced it to various degrees with the exception of the wheat PPA inhibitor only. When transferrin was added at 5-, 30-, and 100-fold excess in molar ratio to PPA, it enhanced the enzyme activity by 125, 134, and 145%, respectively (Fig. 5A). Fetuin similarly enhanced the enzyme activity by 110% at 30-fold excess in molar TABLE I Binding parameters for the interaction between PPA and glycoproteins F, GlcNAc; }, Man; o, Gal; E, NeuAc␣ 236; ‚, NeuAc ␣233; Ⅺ, Fuc; r, Xyl. Interactions between PPA and glycoproteins are measured in 10 mM acetate buffer (pH 5.5) containing 150 mM NaCl and 5 mM CaCl 2 using IAsys. k a shows association rate constant; k d , dissociation rate constant; and Ka, association constant. K a ϭ k a /k d was calculated from the slope and the intercept of the k on versus [S] 0 in Fig. 2 C as described in the text. ratio to PPA, whereas ovalbumin and BSA did not affect the enzyme activity, and wheat PPA inhibitor inhibited it to 80% at the same concentration (Fig. 5B). DISCUSSION This paper demonstrates that PPA binds to glycoproteins by carbohydrate-specific interaction. The affinity chromatography and quantitative analyses with IAsys indicated that immobilized PPA interacted with glycoproteins possessing N-glycans at a K a of 10 5-7 M Ϫ1 at acidic pH, whereas mucin did not bind at all. The binding studies of PA-oligosaccharides to PPA with BIAcore indicated that sialylated complex-type oligosaccharides bound better than high mannose oligosaccharides at pH 5.5. Microplate assay using sugar-BP probes indicated that the interaction is due to the affinity of PPA for component saccharide residues of the N-linked complex type, i.e. for ␣-mannose, ␣-NeuAc, and for ␤ϪLacNAc. Supportingly, the glycoproteins that bound to immobilized PPA were most effectively eluted with methyl ␣-mannoside among the saccharides tested. Because the sugar-BP probe is a multivalent probe exhibiting an affinity constant higher than free sugar by 10 2 -10 5 M Ϫ1 (21,22), it was successfully used to reveal the binding specificity of PPA. The affinity of PPA toward free mono-or disaccharide would not be enough to elute completely the bound glycopro-teins from a PPA-immobilized cuvette.
The glycoprotein binding was inhibited in the presence of EDTA, although once bound, glycoproteins were hardly eluted with EDTA. When the PPA-immobilized cuvette was washed with 10 mM NaOH, the carbohydrate binding of PPA was inactivated, but equilibrating the cuvette with the buffer containing Ca 2ϩ restored the binding activity of PPA. These observations indicate that the carbohydrate recognition of PPA is Ca 2ϩdependent and that the bound glycoprotein might prevent EDTA from accessing Ca 2ϩ located near the lectin site of PPA. The interaction of glycoproteins with PPA did not inhibit its enzyme activity but instead enhanced it to various degrees. On the other hand, the N-glycan binding activity was completely lost in diisopropyl fluorophosphate-treated PPA that had been modified at serine residues to inactivate protease activity while maintaining the amylase activity (data not shown). These observations suggest that the N-glycan recognition is exhibited at a site different from the catalytic subsites of PPA.
The Man-BP binding activity is unique to PPA and was not observed for ␣-amylase isolated from barley, B. subtilis, or human saliva, indicating that the N-glycan binding activity has been acquired during evolution to adapt PPA to the pancreasspecific environment. Because mature PPA is almost unglycosylated, the receptor glycan for PPA may be present on other glycoconjugates.
The N-glycan-binding site may play a role in targeting PPA to intestinal membrane surfaces after secretion because the surface epithelium of the gut is extensively glycosylated and PPA is localized at the luminal surface. In this case, an oligomannosyl N-glycan, such as that of RNase B, may be a first candidate for the receptor glycan due to the slightly alkaline to neutral pH of pancreatic fluid and the intestinal lumen. The epithelium of the small intestine is organized into crypts and villi; the less differentiated crypt cells usually contain oligomannosyl glycans on the membrane glycoproteins, and upon differentiation of the cells, the glycosylation changes to express complex glycans on the fully mature cells of the villi (23)(24)(25). Binding to the membranous glycans on the epithelium would concomitantly protect PPA from proteolysis, stabilize it to extend its life, and/or activate it at the intestinal surface. Furthermore, the binding of PPA to intestinal surface glycans would make the product spatially available as a substrate for the exo-type enzymes that are naturally anchored to the intestinal brush border membranes, e.g. maltase-glucoamylase or sucrase-isomaltase complexes (26), to metabolize starch efficiently. Like many proteins anchored to the brush-border membrane, human maltase-glucoamylase is a glycoprotein with 32-38% carbohydrates that possesses unsialylated complex N-glycans at 19 potential glycosylation sites (27). The binding of PPA to complex glycans is rather weak at neutral pH, but the multivalent glycans on one molecule usually increase the affinity toward carbohydrate-binding proteins. Therefore, maltase-glucoamylase as well as other highly glycosylated membrane enzymes or receptors may also bind pancreatic ␣-amylase. When the mature epithelial cells are shed from the villus tips into the lumen as part of normal cell turnover, the cellular material is digested, but the liberated pancreatic ␣-amylase might bind to the next receptor. In this context, the antinutrient activity of plant lectins is primarily explained by the predominant binding of lectins to glycan receptors at the intestinal surface and blocking them from PPA.
Alternatively or compatibly, the carbohydrate binding activity of PPA found here might be involved in the formation of zymogen granules in the exocrine pancreas. PPA in zymogen aggregates may sort the aggregates into zymogen granules via binding to the N-glycans of the intragranular submembranous A, dose dependence of transferrin on PPA activity. Human transferrin was added at various molar ratios to PPA, and the enzyme activity was measured against starch as described in the text. Relative activity is expressed as %, taking the control as 100%. B, effect of various glycoproteins on the catalytic activity. Each glycoprotein was added at a molar ratio of 30:1 to PPA, and the enzyme activity was measured. OE, transferrin; F, fetuin; Ⅺ, ovalbumin; E, ␣-mylase inhibitor; f, control. glycoprotein matrix and/or to the Man 3 -GlcNAc portion of glycosylphosphatidylinositol-anchored proteins located on rafts of the trans-Golgi network or granular membrane for envelopment. This hypothesis is consistent with the report that glycosylphosphatidylinositol-anchored proteins and rafts play an important role in the granule formation and regulated apical secretion of zymogen in rat pancreas and that the inhibition of raft assembly results in missorting of pancreatic amylase to constitutive secretion (28). The binding characteristics of PPA found in this study, i.e. that PPA exhibits the highest binding to various types of N-glycan at around pH 5 but not at alkaline pH in the presence of Ca 2ϩ , is consistent with the fact that the association events take place Ca 2ϩ -dependently at mildly acidic pH (29,30) and that PPA would dissociate from the granular membrane at the alkaline pH of pancreatic fluid. The carbohydrate binding activity found for PPA was not observed for salivary ␣-amylase in this study, and that may indicate the lack of universality of the aggregation-sorting pathway among secretory tissues. Carbohydrate-binding specificity might be present but different in salivary ␣-amylase because component glycoproteins are not common to the luminal aspect of all secretory granule membranes (29). The function of carbohydrate-specific binding of PPA is still unknown and needs to be elucidated.