Substrate and inhibitor studies of thermolysin-like neutral metalloendopeptidase from kidney membrane fractions. Comparison with bacterial thermolysin

The inhibitory constants of a series of synthetic N-carboxymethyl peptide inhibitors and the kinetic parameters (Km, kcat, and kcat/Km) of a series of model synthetic substrates were determined for the membrane-bound kidney metalloendopeptidase isolated from rabbit kidney and compared with those of bacterial thermolysin. The two enzymes show striking similarities with respect to structural requirements for substrate binding to the hydrophobic pocket at the S1' subsite of the active site. Both enzymes showed the highest reaction rates with substrates having leucine residues in this position while phenylalanine residues gave the lowest Km. The two enzymes were also inhibited by the same N-carboxymethyl peptide inhibitors. Although the mammalian enzyme was more susceptible to inhibition than its bacterial counterpart, structural variations in the inhibitor molecules affected the inhibitory constants for both enzymes in a similar manner. The two enzymes differed significantly, however, with respect to the effect of structural changes in the P1 and P2' positions of the substrate on the kinetic parameters of the reaction. The mammalian enzyme showed the highest reaction rates and specificity constants with substrates having the sequence -Phe-Gly-Phe- or -Phe-Ala-Phe- in positions P2, P1, and P1', respectively, while the sequence -Ala-Phe-Phe- was the most favored by the bacterial enzyme. The sequence -Gly-Gly-Phe- as found in enkephalins was not favored by either of the enzymes. Of the substrates having an aminobenzoate group in the P2' position, the mammalian enzyme favored those with the carboxyl group in the meta position while the bacterial enzyme favored those with the carboxyl group in the para position.(ABSTRACT TRUNCATED AT 250 WORDS)     

Purification, characterization, and partial sequence of the glutathione-dependent formaldehyde dehydrogenase from Escherichia coli: a class III alcohol dehydrogenase.

The glutathione-dependent formaldehyde dehydrogenase from Escherichia coli has been purified to homogeneity and characterized. It is a 83,000-kDa homodimer containing 4 g-atom of zinc per dimer with a specific activity of 60 units/mg toward S-(hydroxymethyl)glutathione and NAD+ as substrates. Its isoelectric point, 4.4, is consistent with both its amino acid composition and chromatographic behavior on DEAE HPLC. The N-terminus is unblocked, and 47 residues from the N-terminus were sequenced. A computer search of the Swiss-Prot protein sequence data bank shows that the N-terminal sequence, [sequence; see text], is homologous with the mammalian class III alcohol dehydrogenases with 27 identities when compared to the human enzyme. Like the human, rat, and rabbit enzymes, it has high formaldehyde dehydrogenase activity in the presence of glutathione and catalyzes the oxidation of normal alcohols (ethanol, octanol, 12-hydroxydodecanoate) in a reaction that is not GSH-dependent. In addition, hemithiolacetals other than those formed from GSH, including omega-thiol fatty acids, also are substrates. The wide distribution and high degree of similarity of this enzyme to the plant and animal alcohol dehydrogenases suggest that the E. coli enzyme is closely related to the ancestor of the plant and animal dimeric zinc alcohol dehydrogenases.     

Purification and characterization of a novel dipeptidyl peptidase from Dictyostelium discoideum

A dipeptidyl peptidase (DPP) was purified to homogeneity using lys-ala-beta-naphthylamide, the standard substrate for DPP II. The enzyme is a monomer with a Mr of 70kDa, pl 5.2, and Km 5.0 microM. Its terminal amino acid sequence was XXLLYAIQKRLF and was not identical to that of any known protein. Although initially considered to be a DPP II, the enzyme differed in some properties from classical DPP IIs. It had a pH optimum of 7.9, was not active on X-pro-naphthylamides, the usual substrates of mammalian DPP II, but was active on arg-arg- and asp-arg-naphthylamides, substrates acted on by the DPP III class of enzymes. This enzyme therefore combines properties typical of both DPP II and III and differs from all previously described DPPs. Activity on lys-ala-beta-naphthylamide was most abundant during aggregation and its activity is consistent with processing specific peptides during development.     

Purple acid phosphatases from bacteria: similarities to mammalian and plant enzymes

Mammalian and plant purple acid phosphatases have similar active site structures despite low sequence identity (<20%). Although no bacterial enzyme has been purified, a sequence database search revealed that genes that could encode potential purple acid phosphatases may be restricted to a small number of organisms (i.e. myco- and cyanobacteria). Analysis of their deduced amino acid sequences and predicted secondary structures indicates that the cyanobacterial enzyme is similar to both the mammalian and the recently discovered low-molecular-weight plant purple acid phosphatases, while the mycobacterial enzyme is homologous to the fungal and high-molecular-weight plant purple acid phosphatases. Homology models indicate that both bacterial proteins appear to be similar to mammalian purple acid phosphatases in the immediate vicinity of the active site. It is likely that these enzymes act as Fenton-type catalysts in order to prevent damage caused by reactive oxygen species generated by invaded host cells (M. tuberculosis) or by the light-harvesting complex (Synechocystis sp.).     

Molecular cloning and primary structure of rat testes metalloendopeptidase EC 3.4.24.15

The complete amino acid sequence of rat testes metalloendopeptidase (EC 3.4.24.15) was deduced from the nucleotide sequence of a cDNA clone isolated by screening a rat testes library with a polyclonal antibody raised against a homogeneous preparation of the rat testes enzyme. The correctness of the sequence was verified by N-terminal amino acid sequence analysis of the isolated enzyme and by partial amino acid sequence analysis of three tryptic peptides located near the N-terminus, the middle, and C-terminus of the native protein. The enzyme is composed of 645 amino acids with a molecular weight of 72,985. This value is close to that of the purified rat testes and brain enzyme as determined by polyacrylamide gel electrophoresis under denaturing and reducing conditions and by molecular sieving chromatography. The enzyme contains the putative active-site sequence -H-E-F-G-H- that is homologous to the sequence in the active site of thermolysin and several other related bacterial enzymes, as well as to active-site sequences of several mammalian zinc metallopeptidases. No amino acid sequence homology, beyond this active site, was found with thermolysin, a bacterial zinc metalloendopeptidase, nor with several mammalian zinc metallopeptidases. Northern blot hybridization analyses showed the presence of mRNA encoding the enzyme in rat testes, but not in other rat tissues in spite of the finding that enzyme activity is widely distributed in all tissues and that relatively high activities are present in rat brain and pituitary.     

Purification, characterization and cloning of isovaleryl-CoA dehydrogenase from higher plant mitochondria.

Between the different types of Acyl-CoA dehydrogenases (ACADs), those specific for branched chain acyl-CoA derivatives are involved in the catabolism of amino acids. In mammals, isovaleryl-CoA dehydrogenase (IVD), an enzyme of the leucine catabolic pathway, is a mitochondrial protein, as other acyl-CoA dehydrogenases involved in fatty acid beta-oxidation. In plants, fatty acid beta-oxidation takes place mainly in peroxisomes, and the cellular location of the enzymes involved in the catabolism of branched-chain amino acids had not been definitely assigned. Here, we describe that highly purified potato mitochondria have important IVD activity. The enzyme was partially purified and cDNAs from two different genes were obtained. The partially purified enzyme has enzymatic constant values with respect to isovaleryl-CoA comparable to those of the mammalian enzyme. It is not active towards straight-chain acyl-CoA substrates tested, but significant activity was also found with isobutyryl-CoA, implying an additional role of the enzyme in the catabolism of valine. The present study confirms recent reports that in plants IVD activity resides in mitochondria and opens the way to a more detailed study of amino-acid catabolism in plant development.     

Primary structure of two distinct rat pancreatic preproelastases determined by sequence analysis of the complete cloned messenger ribonucleic acid sequences

The mRNA sequences for two rat pancreatic elastolytic enzymes have been cloned by recombinant DNA technology and their nucleotide sequences determined. Rat elastase I mRNA is 1113 nucleotides in length, plus a poly(A) tail, and encodes a preproelastase of 266 amino acids. The amino acid sequence of the predicted active form of rat elastase I is 84% homologous to porcine elastase 1. Key amino acid residues involved in determining substrate specificity of porcine elastase 1 are retained in the rat enzyme. The activation peptide of the zymogen does not appear related to that of other mammalian pancreatic serine proteases. The mRNA for elastase I is localized in the rough endoplasmic reticulum of acinar cells, as expected for the site of synthesis of an exocrine secretory enzyme. Rat elastase II mRNA is 910 nucleotides in length, plus a poly(A) tail, and encodes a preproenzyme of 271 amino acids. The amino acid sequence is more closely related to porcine elastase 1 (58% sequence identity) than to the other pancreatic serine proteases (33-39% sequence identity). Predictions of substrate preference based upon key amino acid residues that define the substrate binding cleft are consistent with the broad specificity observed for mammalian pancreatic elastase 2. The activation peptide is similar to that of the chymotrypsinogens and retains an N-terminal cysteine available to form a disulfide link to an internal conserved cysteine residue.     

Multispecific aspartate and aromatic amino acid aminotransferases in Escherichia coli.

Two aminotransferases from Escherichia coli were purified to homogeneity by the criterion of gel electrophoresis. The first (enzyme A) is active on L-aspartic acid, L-tyrosine, L-phenylalanine, and L-tryptophan; the second (enzyme B) is active on the aromatic amiono acids. Enzyme A is identical in substrate specificity with transaminase A and is mainly an aspartate aminotransferase; enzyme B has never been described before and is an aromatic amino acid aminotransferase. The two enzymes are different in the Vmax and Km values with their common substrates and pyridoxal phosphate, in heat stability (enzyme A being heat-stable and enzyme B being heat-labile at 55 degrees) and in pH optima with the amino acid substrates. They are similar in their amino acid composition, each enzyme appears to consist of two subunits, and enzyme B may be converted to enzyme A by controlled proteolysis with subtilsin. The conversion was detected by the generation of new aspartate aminotransferase activity from enzyme B and was further verified by identification by acrylamide gel electrophoresis of the newly formed enzyme A. The two enzymes appear to be products of two genes different in a small, probably terminal, nucleotide sequence.     

Novel flavonol 3-sulfotransferase. Purification, kinetic properties, and partial amino acid sequence.

A flavonol sulfotransferase (EC 2.8.2.-), which catalyzes the transfer of the sulfate group from 3'-phosphoadenosine 5'-phosphosulfate to the 3-hydroxyl group of flavonol aglycones, has been purified to apparent homogeneity from Flaveria chloraefolia. The specific activity of flavonol 3-sulfotransferase was enriched 2000-fold, as compared with the homogenate, with a recovery of 9%. The molecular mass of the native and denatured enzyme was found to be 34.5 kDa, suggesting that the active from of the enzyme is a monomer. The enzyme exhibited expressed specificity for position 3 of flavonol aglycones, showed two activity optima at pH 6.0 and 8.5, did not require divalent cations, and was not inhibited by either EDTA or sulfhydryl group reagents. The results of substrate interaction kinetics and product inhibition are consistent with an Ordered Bi Bi mechanism where 3'-phosphoadenosine 5'-phosphosulfate is the first substrate to bind to the enzyme and 3'-phosphoadenosine 5'-phosphate is the final product to be released. The amino acid sequence of two peptides representing 17 and 33 amino acids showed no significant sequence similarity with the amino acid sequences reported for animal sulfotransferases. Antibodies raised against F. chloraefolia 3-sulfotransferase were found to cross-react with the 3'- and 4'-sulfotransferase activities of the same plant, suggesting that the three enzymes are structurally related.     

Site-directed mutagenesis of human membrane-associated ganglioside sialidase: identification of amino-acid residues contributing to substrate specificity.

Unlike microbial sialidases, mammalian sialidases possess strict substrate specificity, for example the human membrane-associated sialidase, which hydrolyzes only gangliosides. To cast light on the molecular basis of this narrow substrate preference, predicted active site amino-acid residues of the human membrane sialidase were altered by site-directed mutagenesis. When compared with the active site amino-acid residues proposed for Salmonella typhimurium sialidase, only five out of 13 residues were found to be different to the human enzyme, these being located upstream of the putative transmembrane region. Alteration of seven residues, including these five, was followed by transient expression of the mutant enzymes in COS-1 cells and characterization of their kinetic properties using various substrates. Substitution of glutamic acid (at position 51) by aspartic acid and of arginine (at position 114) by glutamine or alanine resulted in retention of good catalytic efficiency toward ganglioside substrates, whereas other substitutions caused a marked reduction. The mutant enzyme E51D exhibited an increase in hydrolytic activity towards GM2 as well as sialyllactose (which are poor substrates for the wild-type) with change to a lower Km and a higher Vmax. R114Q demonstrated a substrate specificity shift in the same direction as E51D, whereas R114A enhanced the preference for gangliosides GD3 and GD1a that are effectively hydrolyzed by the wild-type. The inhibition experiments using 2-deoxy-2,3-didehydro-N-acetylneuraminic acid were consistent with the results in the alteration of substrate specificity. The findings suggest that putative active-site residues of the human membrane sialidase contribute to its substrate specificity.     

Specificity of two isolated wheat carboxypeptidases

The substrate specificity of two purified carboxypeptidases from germinated wheat has been examined. Both enzymes were active on a wide variety of carbobenzoxy substituted peptides but inactive with unsubstituted dipeptides. Neither enzyme was active upon endoprotease or amidase substrates and only low levels of esterase activity were evident. In time course studies, both enzymes gave rapid non-specific sequential release of amino acids, including proline, from the carboxyterminal of proteins and polypeptides of known amino acid sequence.     

A novel acyl-CoA oxidase that can oxidize short-chain acyl-CoA in plant peroxisomes

Short-chain acyl-CoA oxidases are beta-oxidation enzymes that are active on short-chain acyl-CoAs and that appear to be present in higher plant peroxisomes and absent in mammalian peroxisomes. Therefore, plant peroxisomes are capable of performing complete beta-oxidation of acyl-CoA chains, whereas mammalian peroxisomes can perform beta-oxidation of only those acyl-CoA chains that are larger than octanoyl-CoA (C8). In this report, we have shown that a novel acyl-CoA oxidase can oxidize short-chain acyl-CoA in plant peroxisomes. A peroxisomal short-chain acyl-CoA oxidase from Arabidopsis was purified following the expression of the Arabidopsis cDNA in a baculovirus expression system. The purified enzyme was active on butyryl-CoA (C4), hexanoyl-CoA (C6), and octanoyl-CoA (C8). Cell fractionation and immunocytochemical analysis revealed that the short-chain acyl-CoA oxidase is localized in peroxisomes. The expression pattern of the short-chain acyl-CoA oxidase was similar to that of peroxisomal 3-ketoacyl-CoA thiolase, a marker enzyme of fatty acid beta-oxidation, during post-germinative growth. Although the molecular structure and amino acid sequence of the enzyme are similar to those of mammalian mitochondrial acyl-CoA dehydrogenase, the purified enzyme has no activity as acyl-CoA dehydrogenase. These results indicate that the short-chain acyl-CoA oxidases function in fatty acid beta-oxidation in plant peroxisomes, and that by the cooperative action of long- and short-chain acyl-CoA oxidases, plant peroxisomes are capable of performing the complete beta-oxidation of acyl-CoA.     

Mesocarp localization of a bi-functional resveratrol/hydroxycinnamic acid glucosyltransferase of Concord grape (Vitis labrusca)

Resveratrol is a stilbene with well-known health-promoting effects in humans that is produced constitutively or accumulates as a phytoalexin in several plant species including grape (Vitis sp.). Grape berries accumulate stilbenes in the exocarp as cis- and trans-isomers of resveratrol, together with their respective 3-O-monoglucosides. An enzyme glucosylating cis- and trans-resveratrol was purified to apparent homogeneity from Concord (Vitis labrusca) grape berries, and peptide sequencing associated it to an uncharacterized Vitis vinifera full-length clone (TC38971, tigr database). A corresponding gene from Vitis labrusca (VLRSgt) had 98% sequence identity to clone TC38971 and 92% sequence identity to a Vitis viniferap-hydroxybenzoic acid glucosyltransferase that produces glucose esters. The recombinant enzyme was active over a broad pH range (5.5-10), producing glucosides of stilbenes, flavonoids and coumarins at higher pH and glucose esters of several hydroxybenzoic and hydroxycinnamic acids at low pH. Vitis labrusca grape berries accumulated both stilbene glucosides and hydroxycinnamic acid glucose esters, consistent with the bi-functional role of VLRSgt in stilbene and hydroxycinnamic acid modification. While phylogenetic analysis of VLRSgt and other functionally characterized glucosyltransferases places it with other glucose ester-producing enzymes, the present results indicate broader biochemical activities for this class of enzymes.     

Cloning and sequencing of a cDNA encoding 2,3-bisphosphoglycerate-independent phosphoglycerate mutase from maize. Possible relationship to the alkaline phosphatase family.

The primary sequence of maize 2,3-bisphosphoglycerate-independent phosphoglycerate mutase was deduced from cDNAs isolated from maize cDNA libraries by screening with specific antibodies to the cofactor-independent enzyme and from a maize genomic clone. The genomic clone provided the 5'-nucleotide sequence encoding the N-terminal amino acids which could not be obtained from the cDNA. Confirmation that the nucleotide sequence was for the cofactor-independent phosphoglycerate mutase was obtained by sequencing the peptides generated from cyanogen bromide cleavage of the purified protein. This is the first report of the amino acid sequence of a 2,3-bisphosphoglycerate cofactor-independent phosphoglycerate mutase, which consists of 559 amino acids and is twice the molecular size of the mammalian cofactor-dependent enzyme subunit. Analysis of the cofactor-independent phosphoglycerate mutase amino acid sequence revealed no identity with the cofactor-dependent mutase types. Northern blot analysis confirmed this difference since the maize cofactor-independent phosphoglycerate mutase cDNA did not hybridize with mRNA of the cofactor-dependent mutase. The lack of amino acid identity between cofactor-dependent and -independent enzymes is consistent with their different catalytic mechanisms and suggests that both enzymes are unrelated evolutionarily and arose from two independent ancestral genes. However, a constellation of residues which are involved in metal ion binding in various alkaline phosphatases is conserved in the maize cofactor-independent phosphoglycerate mutase, which suggests that the enzyme is a member of the alkaline phosphatase family of enzymes.     

The role of the non-conserved residue at position 104 of class A beta-lactamases in susceptibility to mechanism-based inhibitors

The role of the non-conserved amino acid residue at position 104 of the class A beta-lactamases, which comprises a highly conserved sequence of amino acids at the active sites of these enzymes, in both the hydrolysis of beta-lactam substrates and inactivation by mechanism-based inhibitors was investigated. Site-directed mutagenesis was performed on the penPC gene encoding the Bacillus cereus 569/H beta-lactamase I to replace Asp104 with the corresponding Staphylococcus aureus PC1 residue Ala104. Kinetic data obtained with the purified Asp104Ala B. cereus 569/H beta-lactamase I was compared to that obtained from the wild-type B. cereus and S. aureus enzymes. Replacement of amino acid residue 104 had little effect on the Michaelis parameters for the hydrolysis of both S- and A-type penicillins. Relative to wild-type enzyme, the Asp104Ala beta-lactamase I had 2-fold higher Km values for benzylpenicillin and methicillin, but negligible difference in Km for ampicillin and oxacillin. However, kcat values were also slightly increased resulting in little change in catalytic efficiency, kcat/Km. In contrast, the Asp104Ala beta-lactamase I became more like the S. aureus enzyme in its response to the mechanism-based inhibitors clavulanic acid and 6-beta-(trifluoromethane sulfonyl)amido-penicillanic acid sulfone with respect to both response to the inhibitors and subsequent enzymatic properties. Based on the known three-dimensional structures of the Bacillus licheniformis 749/C, Escherichia coli TEM and S. aureus PC1 beta-lactamases, a model for the role of the non-conserved residue at position 104 in the process of inactivation by mechanism-based inhibitors is proposed.     

Human NAD(+)-dependent mitochondrial malic enzyme. cDNA cloning, primary structure, and expression in Escherichia coli

Mitochondrial NAD(+)-dependent malic enzyme (EC 1.1.1.40) is expressed in rapidly proliferating cells and tumor cells, where it is probably linked to the conversion of amino acid carbon to pyruvate. In this paper, we report the cDNA cloning, amino acid sequence, and expression in Escherichia coli of functional human NAD(+)-dependent mitochondrial malic enzyme. The cDNA is 1,923 base pairs long and contains an open reading frame coding for a 584-amino acid protein. The molecular mass is 65.4 kDa for the unprocessed precursor protein. Comparison of the amino acid sequence of the human protein with the published NADP(+)-dependent mammalian cytosolic or plant chloroplast malic enzymes reveals highly conserved regions interrupted with long stretches of amino acids without significant homology. Expression of the processed protein in E. coli yielded an enzyme with the same kinetic and allosteric properties as malic enzyme purified from human cells.     

Aromatic amino acid catabolism in trypanosomatids

Trypanosomatids cause important human diseases, like sleeping sickness, Chagas disease, and the leishmaniases. Unlike in the mammalian host, the metabolism of aromatic amino acids is a very simple pathway in these parasites. Trypanosoma brucei and Trypanosoma cruzi transaminate the three aromatic amino acids, the resulting 2-oxo acids being reduced to the corresponding lactate derivatives and excreted. In T. cruzi, two enzymes are involved in this process: a tyrosine aminotransferase (TAT), which despite a high sequence similarity with the mammalian enzyme, has a different substrate specificity; and an aromatic L-2-hydroxyacid dehydrogenase (AHADH), which belongs to the subfamily of the cytosolic malate dehydrogenases (MDHs), yet has no MDH activity. In T. cruzi AHADH the substitution of Ala102 for Arg enables AHADH to reduce oxaloacetate. In the members of the 2-hydroxyacid dehydrogenases family, the residue at this position is known to be responsible for substrate specificity. T. cruzi does not possess a cytosolic MDH but contains a mitochondrial and a glycosomal MDH; by contrast T. brucei and Leishmania spp. possess a cytosolic MDH in addition to glycosomal and mitochondrial isozymes. Although Leishmania mexicana also transaminates aromatic amino acids through a broad specificity aminotransferase, the latter presents low sequence similarity with TATs, and this parasite does not seem to have an enzyme equivalent to T. cruzi AHADH. Therefore, these closely related primitive eukaryotes have developed aromatic amino acid catabolism systems using different enzymes and probably for different metabolic purposes.     

Identification and characterization of mammalian mitochondrial tRNA nucleotidyltransferases

The CCA-adding enzyme (ATP:tRNA adenylyltransferase or CTP:tRNA cytidylyltransferase (EC )) generates the conserved CCA sequence responsible for the attachment of amino acid at the 3' terminus of tRNA molecules. It was shown that enzymes from various organisms strictly recognize the elbow region of tRNA formed by the conserved D- and T-loops. However, most of the mammalian mitochondrial (mt) tRNAs lack consensus sequences in both D- and T-loops. To characterize the mammalian mt CCA-adding enzymes, we have partially purified the enzyme from bovine liver mitochondria and determined cDNA sequences from human and mouse dbESTs by mass spectrometric analysis. The identified sequences contained typical amino-terminal peptides for mitochondrial protein import and had characteristics of the class II nucleotidyltransferase superfamily that includes eukaryotic and eubacterial CCA-adding enzymes. The human recombinant enzyme was overexpressed in Escherichia coli, and its CCA-adding activity was characterized using several mt tRNAs as substrates. The results clearly show that the human mt CCA-adding enzyme can efficiently repair mt tRNAs that are poor substrates for the E. coli enzyme although both enzymes work equally well on cytoplasmic tRNAs. This suggests that the mammalian mt enzymes have evolved so as to recognize mt tRNAs with unusual structures.     

Purification, molecular cloning, and properties of a beta-glycosidase isolated from midgut lumen of Tenebrio molitor (Coleoptera) larvae

Two beta-glycosidases (M(r) 59k) were purified from midgut contents of larvae of the yellow mealworm, Tenebrio molitor (Coleoptera: Tenebrionidae). The two enzymes (betaGly1 and betaGly2) have identical kinetic properties, but differ in hydrophobicity. The two glycosidases were cloned and their sequences differ by only four amino acids. The T. molitor glycosidases are family 1 glycoside hydrolases and have the E379 (nucleophile) and E169 (proton donor) as catalytic amino acids based on sequence alignments. The enzymes share high homology and similarity with other insect, mammalian and plant beta-glycosidases. The two enzymes may hydrolyze several substrates, such as disaccharides, arylglucosides, natural occurring plant glucosides, alkylglucosides, oligocellodextrins and the polymer laminarin. The enzymes have only one catalytic site, as inferred from experiments of competition between substrates and sequence alignments. The observed inhibition by high concentrations of the plant glucoside amygdalin, used as substrate, is an artifact generated by transglucosylation. The active site of each purified beta-glycosidase has four subsites, of which subsites +1 and +2 bind glucose with more affinity. Subsite +2 has more affinity for hydrophobic groups, binding with increasing affinities: glucose, mandelonitrile and nitrophenyl moieties. Subsite +3 has more affinity for glucose than butylene moieties. The intrinsic catalytic constant calculated for hydrolysis of the glucose beta-1,4-glucosidic bond is 21.2 s(-1) x M(-1). The putative physiological role of these enzymes is the digestion of di- and oligosaccharides derived from hemicelluloses.