Investigations of the Structures and Inhibitory Properties of Intestinal Maltase Glucoamylase and Sucrase Isomaltase

Two enzyme complexes are largely responsible for the postamylase metabolism of starch limit dextrins into monomeric glucose in the human small intestine, maltase-glucoamylase (MGAM) and sucrase-isomaltase (SI) (1). MGAM and SI each consist of 2 active enzyme domains or modules, with related structures but somewhat different enzymatic characteristics. Because there is some overlap in their substrate tolerances, the nomenclature used here is structurally defined: ntMGAM and ntSI for the respective N-terminal enzyme modules, and ctMGAM and ctSI for the C-terminal units. All 4 MGAM and SI enzyme modules are classified into glycoside hydrolase family GH31 (Carbohydrate Active Enzymes Database, http://http://www.cazy.org) (2). This family is characterized by a number of enzymes with activities toward starch-derived structures. Interestingly, several GH31 enzymes are present in the proteome of gut-resident microbes, such as Bacteroides thetaiotamicron, implicating a role in the degradation of resistant starch structures by the bacterial flora (3,4). From amino acid sequence analysis, it is evident that MGAM and SI evolved from a single GH31 precursor, initially duplicating into primordial N-terminal and C-terminal domains, and subsequently into the 2 extant enzymes (5). The 4 modules have retained some overlap in substrate specificity; however, their efficiencies toward different substrates have evolved considerably, as reflected in their activity-based names and their various substrate profiles. This observation led us to propose the hypothesis that competitive inhibitors could be developed with selectivity toward each of the modules. Our goal in deriving such inhibitors is to investigate the respective roles of each enzyme module in starch digestion under differing physiological and dietary conditions. Such an approach would lead to a more complete picture of the human activities involved in starch digestion. The experimental approach involves an iterative series of structural and chemical synthesis studies. The structural work is based on the crystallographic structural determination of each of the 4 modules, individually expressed by recombinant DNA technology in host cells. Mouse MGAM was cloned by Nichols et al (1) and human SI (6,7). Two expression systems have been used for large-scale expression. The ntMGAM and ntSI modules were initially expressed by stable transfection of Drosophila S2 cells (8), and ctMGAM and ctSI were produced by baculovirus infection of insect cells (9). The first atomic structural results were derived for the mouse ntMGAM module, as described in detail by Sim et al (10). This structure was determined of the enzyme itself, in addition to complexes with the glucosidase inhibitors acarbose (1) (the boldface numbers correspond to the numbers in Fig. 1) and salacinol (2). Acarbose (1), a commercial compound used as an antidiabetic, is a pseudotetrasaccharide derivative with broad specificity. It is a relatively poor inhibitor of ntMGAM, but a much better one against ctMGAM. Salacinol (2), a natural product derived from Salacia plants (11), is a much more effective inhibitor of the MGAM and SI modules.FIGURE 1: Compounds studied in inhibitor analysis.The initial ntMGAM structure resulted in a hypothesis for the selective activity of acarbose towards ctMGAM. All of the MGAM/SI modules are exo-acting glycosidases that cleave a glucose monosaccharide from the nonreducing end of the polysaccharide substrate. The substrate-binding cleft of ntMGAM consists of 2 well-defined sugar subsites on the reducing end of the cleaved product; however, ctMGAM has an inserted sequence of 21 amino acid residues as compared with ntMGAM. This insert includes several side chains common to sugar-binding sites, including aromatic side chains. The ntMGAM model proposed that this insert formed a loop extending the number of subsites in ctMGAM, which is consistent with the observation that ctMGAM was more active than ntMGAM on longer polysaccharides and more effectively inhibited by acarbose (10). This proposal has subsequently been supported by a crystal structure of ctMGAM (21). The second of the series of crystal structures was human ntSI (12). This structure had 4 independent molecules in the crystallographic cell. Interestingly, as the enzyme was produced in insect cells, it was modified by N-glycosylation. The glycan from one of the glycosylation sites extends across an intermolecular contact into the vicinity of the substrate-binding region of a neighboring molecule. On comparison with the position of acarbose in the ntMGAM structure, it is evident that the glycan does not adopt the active substrate-binding conformation, but is nonetheless indicative of sugar-binding propensity. Because ntSI, unlike ntMGAM, is active against isomaltose in addition to maltose, a docking analysis was performed to investigate this property. ntSI was observed to have a more closed, constricted catalytic center than ntMGAM. The interpretation of this analysis was that ntSI, because of tighter active site, is able to distort isomaltose into the transition state–like conformation required to progress along the reaction coordinate, whereas the more open ntMGAM site cannot. Further support of this hypothesis will require further crystallographic studies in complex with inhibitors or substrate analogues. To further investigate the active site of the N- and C-terminal domains and improve the understanding of the functional requirement of each enzyme in the human digestive system, the effectiveness of various synthetic inhibitors were tested with each enzyme subunit, including 2 splice forms of ctMGAM (ctMGAM-N2 and ctMGAM-N20, (9)) (Table 1). Two, acarbose (1) (13) and miglitol (3) (14), are used as α-glucosidase inhibitors in the clinical treatment of noninsulin-dependent type II diabetes mellitus (15). Salacinol (2), its selenium analog, blintol (4), kotalanol (5), de-O-sulfonated kotalanol (6), 3′-O-β-maltosyl-de-O-sulfonated ponkoranol (7), 5′-O-β-maltosyl-de-O-sulfonated ponkoranol (8) (Fig. 1), are either natural products found in the extract of the Salacia species plant native to Sri Lanka and southern India and used in the ayurvedic medical system to treat diabetes (16), or synthetic analogues thereof (17). We propose a 21-amino acid extension in the C-terminal domains as compared to the N-terminal domains, providing an extended binding cleft and capable of accommodating larger substrates.TABLE 1: Ki (μmol/L) values of inhibitor analysis of compounds 1–8The inhibition profile of acarbose (1), a relatively large inhibitor, supports our hypothesis of an extended binding cleft in the C-terminal domains as it preferentially inhibits the C-terminal enzymes (Ki values: ctSI = 0.246 ± 0.005 μmol/L, ctMGAM-N2 = 0.009 ± 0.002 μmol/L, ctMGAM-N20 = 0.028 ± 0.005 μmol/L) (9) and is a poor inhibitor of the N-terminal enzymes (Ki values: ntMGAM = 62 ± 13 μmol/L, ntSI = 14 ± 1 μmol/L) (8). Interestingly, 3′-O-β-maltosyl-de-O-sulfonated ponkoranol (7), also a large inhibitor, demonstrates some selectivity for ntMGAM (Ki = 0.039 ± 0.025) compared to the C-terminal enzymes (Ki values: ct-MGAM-N2 = no inhibition, ct-MGAM-N20 = 0.655 ± 0.063, ctSI = 0.062 ± 0.05). When the maltosyl unit is transferred to the 5′ hydroxyl group (8), the inhibition of ntMGAM is further improved (Ki = 0.008 ± 0.002), whereas there is little differentiation in the inhibitory properties with respect to the other catalytic subunits (Ki values: ctMGAM-N2 = 0.077 ± 0.015, ctMGAM-N20 = 0.067 ± 0.012, ctSI = 0.045 ± 0.001, ntSI = 0.019 ± 0.008) (17). Miglitol (3) is a high macromolar inhibitor of the α-glucosidase enzymes studied, showing little differentiation with respect to inhibitory properties of the 5 catalytic subunits. Salacinol (2), a smaller inhibitor in comparison to acarbose, is a submicromolar inhibitor of ntSI and ntMGAM (Ki values: 0.277 ± 0.068 and 0.19 ± 0.02 μmol/L, respectively). Salacinol (2) also shows a small amount of selectivity between the 2 splice forms of ctMGAM (Ki values: ctMGAM-N2 = 0.213 ± 0.018 μmol/L, ctMGAM-N20 = 0.058 ± 0.003 μmol/L). When the sulfonium ion of salacinol is replaced with a selenonium ion, as seen in the inhibitor blintol (4), there is improved inhibition of ctMGAM-N2. The inhibitory effects of blintol on the other inhibitors are largely unchanged when compared with salacinol. Kotalanol (5) shows some selectivity for inhibiting C-terminal enzymes in comparison to the N-terminal enzymes (4- to 6-fold) and de-O-sulfonated kotolanol (6) shows similar inhibitory properties of all of the enzymatic subunits (9). In summary, there is a general preference for larger inhibitors for the theorized extended binding cleft of the C-terminal active sites, whereas smaller inhibitors are generally better suited to the shorter active site of the N-terminal catalytic domains. Furthermore, the compounds that have been synthesized and studied have been shown to be capable of selectivity between the catalytic subunits of SI and MGAM, which will further our understanding of the final stages of starch digestion in the small intestine and lead toward the development of oral agents for the treatment of type 2 diabetes mellitus. In addition to inhibitor studies, mutations have been identified in the isomaltase subunit of SI (ntSI) associated with patients with congenital sucrase-isomaltase deficiency (CSID). These mutations have been shown as affecting enzyme activity, trafficking of the enzymes to the intestinal lumen, or both. The mutations were modeled into the structure of ntSI (PDB: 3LPO) (12) in an attempt to better understand how they alter the enzyme's characteristics. There are 2 mutations in which enzymatic activity is maintained: L340P and Q117R; and 2 mutations in which enzymatic activity is either completely or partially abolished: C635R and L620P (7,18–20). Figure 2 illustrates the location of the mutation sites with respect to active site. The modeling results presented here provide possible explanations for these observations.FIGURE 2: ntSI mutations. ntSI model illustrating mutation sites and active site residues. Purple: Catalytic nucleophile and acid/base catalyst; orange: substrate-binding residues; red: mutation site.L340P The first mutation, in which a leucine is replaced with a proline at position 340 (corresponding to residue number 308 in PDB file 3LPO), results in an active protein. It is cleaved in the endoplasmic reticulum and secreted, but it is not accessible in the intestinal lumen to cleave sugars, presumably because of a trafficking defect (7,18–20). The mutation is 28.5 Å away from the catalytic nucleophile. Proline is added in place of a leucine in an α-helix at the surface of the protein monomer, which likely disrupts the secondary structure on the surface of the protein (Fig. 3A). We propose the failure of the protein to be properly trafficked to the intestinal lumen stems from the change in secondary structure as a result of the mutation, potentially leading to improper glycosylation. There is a predicted glycosylation site (ASN 119 in PDB file 3LPO; Fig. 4A) 12.6 Å from the mutation. The distortion of the mutated α-helix likely disorders the β-sheet containing ASN 119 (Figs. 3A, 4A). An additional consideration is that the mutation is on the surface of the protein at a location at which 2 monomers of the protein interact in the crystal form. For proper trafficking, the N- and C-terminal enzymatic subunits of SI may interact with one another, and this mutation may prevent this interaction from occurring properly. Disruption of any of these properties may lead to trafficking defects.FIGURE 3: L340P and Q117R mutations. A, L340P mutation and ntSI active site. B, Q117R mutation and ntSI active site. Purple: catalytic nucleophile and acid/base catalyst; orange: substrate-binding residues; yellow: original residue; red: mutated residue.FIGURE 4: Glycosylation sites in vicinity of L340P and Q117R. A, L340P and predicted glycosylation site ASN 119. B, Q117R and predicted glycosylation site ASN 119. Yellow: original residue; red: mutated residue; blue: ASN 1139.Q117R A second mutation, in which a glutamine is replaced with an asparagine at position 117 (corresponding to residue 84 in PDB 3LPO; Fig. 3B), leads to a protein that is active from an enzymatic standpoint, but is not active in patients with CSID. There is no transport to the small intestinal lumen and therefore no functionality of the enzyme in that location. The mutation is on the surface of the protein 52.2 Å from the catalytic nucleophile in the active site. Because of the distance from the active site, there is likely no direct effect on the active site conformation, explaining the enzymatic functionality of the protein. We propose an effect on trafficking that is likely caused by a change in glycosylation at the surface of the protein as a result of the mutation, causing the signal pathway in which the enzyme is transported to the intestinal lumen to fail. Similar to the mutation L340P, there is a predicted N-glycosylation site, ASN 119, located in the same β-sheet as the mutated arginine 117 (Fig. 4B). C635R The third mutation, in which a cysteine is replaced with an arginine at position 635 (corresponding to residue 602 in the PDB file 3LPO, Fig. 5A), results in a protein with only partial α-glucosidase activity. The mutation is 17.4 Å from the catalytic center and results in the disruption of a disulfide bond between the mutated residue and residue 613. The likely result is a significant disruption of secondary structure, disordering the active site and affecting function.FIGURE 5: C635R and L620P mutations. A, C635R mutation and ntSI active site. B, L620P mutation and ntSI active site. Purple: catalytic nucleophile and acid/base catalyst; orange: substrate-binding residues; yellow: original residue; red: mutated residue.L620P The final mutation, in which a leucine is replaced by a proline at position 620 (corresponding to residue 587 in the PDB file 3LPO; Fig. 5B), results in an inactive protein. Proline is added in the middle of an α-helix and likely disrupts the helix and associated secondary structure. The mutation is 21.2 Å away from the catalytic nucleophile, but it is in the interior of the protein where proline is found only infrequently because this residue typically lies on the surface of globular proteins. The disorder of the active site caused by this mutation likely affects function and causes the inactivity of the protein. In addition to the specifics of inhibitor-enzyme interactions, there are some important questions that can be addressed by future structural analyses. The roles of portions of the enzyme module structures that are distal to the active sites have not been investigated. In particular, the structure identified a β-trefoil region, a fold that has been associated with both carbohydrate-binding modules of microbial glycoside hydrolases and protein–protein interactions. Furthermore, although studies of individual modules have been critical in studying their characteristics, MGAM and SI both exist physiologically as dienzyme complexes. How do the modules interact within the intact complex? To what extent is this interaction important for their activities? Finally, what is the basis for the altered activity or trafficking of mutated SI domains in patients with CSID? Structural studies of intact MGAM and SI enzymes can contribute to answering these questions. Acknowledgment The authors thank Lyann Sim for previous structural studies and enzyme purification.