Directed evolution of a remarkably efficient Kdnase from a bacterial neuraminidase.

N-acetylneuraminic acid (5-acetamido-3,5-dideoxy-d-glycero-d-galacto-non-2-ulosonic acid), which is the principal sialic acid family member of the non-2-ulosonic acids and their various derivatives, is often found at the terminal position on the glycan chains that adorn all vertebrate cells. This terminal position combined with subtle variations in structure and linkage to the underlying glycan chains between humans and other mammals points to the importance of this diverse group of nine-carbon sugars as indicators of the unique aspects of human evolution and is relevant to understanding an array of human conditions. Enzymes that catalyze the removal N-acetylneuraminic acid from glycoconjugates are called neuraminidases. However, despite their documented role in numerous diseases, due to the promiscuous activity of many neuraminidases, our knowledge of the functions and metabolism of many sialic acids and the effect of the attachment to cellular glycans is limited. To this end, through a concerted effort of generation of random and site-directed mutagenesis libraries, subsequent screens and positive and negative evolutionary selection protocols, we succeeded in identifying three enzyme variants of the neuraminidase from the soil bacterium Micromonospora viridifaciens with markedly altered specificity for the hydrolysis of natural Kdn (3-deoxy-d-glycero-d-galacto-non-2-ulosonic acid) glycosidic linkages compared to those of N-acetylneuraminic acid. These variants catalyze the hydrolysis of Kdn-containing disaccharides with catalytic efficiencies (second-order rate constants: kcat/Km) of greater than 105 M-1 s-1; the best variant displayed an efficiency of >106 M-1 s-1 at its optimal pH.

Chemical insight into the emergence of influenza virus strains that are resistant to Relenza.

A reagent panel containing ten 4-substituted 4-nitrophenyl α-D-sialosides and a second panel of the corresponding sialic acid glycals were synthesized and used to probe the inhibition mechanism for two neuraminidases, the N2 enzyme from influenza type A virus and the enzyme from Micromonospora viridifaciens. For the viral enzyme the logarithm of the inhibition constant (Ki) correlated with neither the logarithm of the catalytic efficiency (kcat/Km) nor catalytic proficiency (kcat/Km kun). These linear free energy relationship data support the notion that these inhibitors, which include the therapeutic agent Relenza, are not transition state mimics for the enzyme-catalyzed hydrolysis reaction. Moreover, for the influenza enzyme, a correlation (slope, 0.80 ± 0.08) is observed between the logarithms of the inhibition (Ki) and Michaelis (Km) constants. We conclude that the free energy for Relenza binding to the influenza enzyme mimics the enzyme-substrate interactions at the Michaelis complex. Thus, an influenza mutational response to a 4-substituted sialic acid glycal inhibitor can weaken the interactions between the inhibitor and the viral neuraminidase without a concomitant decrease in free energy of binding for the substrate at the enzyme-catalyzed hydrolysis transition state. The current findings make it clear that new structural motifs and/or substitution patterns need to be developed in the search for a bona fide influenza viral neuraminidase transition state analogue inhibitor.

Bacterial and viral sialidases: contribution of the conserved active site glutamate to catalysis.

Mutagenesis of the conserved glutamic acid of influenza type A (E277) and Micromonospora viridifaciens (E260) sialidases was performed to probe the contribution of this strictly conserved residue to catalysis. Kinetic studies of the E260D and E260C M. viridifaciens mutant enzymes reveal that the overall mechanism of action has not changed. That is, the mutants are retaining sialidases in which glycosylation and deglycosylation are rate-limiting for k(cat)/K(m) and k(cat), respectively. The solvent kinetic isotope effect and proton inventory on k(cat) for the E260C mutant sialidase provide strong evidence that the newly installed cysteine residue provides little catalytic acceleration. The results are consistent with the conserved aspartic acid residue (D92) becoming the key general acid/base residue in the catalytic cycle. In addition, the E277D mutant influenza type A sialidase is catalytically active toward 4-nitrophenyl α-D-sialoside, although no measurable hydrolysis of natural substrates was observed. Thus, mutating the glutamate residue (E277) to an aspartate increases the activation free energy of hydrolysis for natural substrates by >22 kJ/mol.

Brønsted analysis of an enzyme-catalyzed pseudo-deglycosylation reaction: mechanism of desialylation in sialidases.

The Micromonospora viridifaciens Y370G inverting mutant sialidase has been found to possess beta-sialidase activity with various fluoro-substituted phenyl beta-sialosides. A reagent panel of seven mono- and difluorophenyl beta-d-sialosides was synthesized, and these compounds were used, in conjunction with the parent phenyl beta-d-sialoside, to probe the mechanism of M. viridifaciens Y370G mutant sialidase-catalyzed hydrolyses. These hydrolysis reactions mimic the deglycosylation reaction step of the crucial tyrosinyl enzyme-bound intermediate that is formed during the corresponding wild-type sialidase reactions. The derived Brønsted parameter (beta(lg)) on k(cat)/K(m) is -0.46 +/- 0.02 for the four substrates that display significant activity, and these span a range of leaving group abilities (as judged by the pK(a) of their conjugate acids being between 7.09 and 9.87). The 4-fluoro, 2,3- and 2,5-difluorosubstrates display a diminished activity, whereas the 3,5-difluoro compound undergoes catalyzed hydrolysis exceedingly slowly. These observations, taken with solvent deuterium kinetic isotope effects (k(H(2))(O)/k(D(2))(O)) on the catalyzed hydrolysis of the 2-fluorophenyl substrate of 0.88 +/- 0.24 (k(cat)/K(m)) and 1.16 +/- 0.12 (k(cat)) and the poor inhibition shown by phenol (IC(50) > 1 mM), are consistent with glycosidic C-O cleavage being rate determining for both k(cat)/K(m) and k(cat) with little or no protonation of the departing aryloxide leaving group. The kinetic data reported herein are consistent with rate-limiting glycoside hydrolysis occurring via two distinct transition states that incorporates a nonproductive binding component for the tighter binding substrates.

Contribution of the active site aspartic acid to catalysis in the bacterial neuraminidase from Micromonospora viridifaciens.

A recombinant D92G mutant sialidase from Micromonospora viridifaciens has been cloned, expressed and purified. Kinetic studies reveal that the replacement of the conserved aspartic acid with glycine results in a catalytically competent retaining sialidase that possesses significant activity against activated substrates. The contribution of this aspartate residue to the free energy of hydrolysis for natural substrates is greater than 19 kJ/mol. The three dimensional structure of the D92G mutant shows that the removal of aspartic acid 92 causes no significant re-arrangement of the active site, and that an ordered water molecule substitutes for the carboxylate group of D92.

Structure and role of sialic acids on the surface of Aspergillus fumigatus conidiospores.

Aspergillus fumigatus is an opportunistic fungal pathogen that causes a life-threatening invasive fungal disease (invasive aspergillosis, IA) in immunocompromised individuals. The first step of pathogenesis is thought to be the attachment of conidia to proteins in lung tissue. Previous studies in our laboratory have shown that conidia adhere to basal lamina proteins via negatively charged sugars on their surface, presumably sialic acids. Sialic acids are a family of more than 50 substituted derivatives of a nine-carbon monosaccharide, neuraminic acid. The purpose of this study was 2-fold: (1) to determine the structure of sialic acids and the glycan acceptor on A. fumigatus oligosaccharides and (2) to determine the effect on the removal of sialic acids from conidia on conidial binding to the extracellular matrix protein fibronectin and phagocytosis of conidia by cultured macrophages and type 2 pneumocytes. Surface sialic acids were removed using Micromonospora viridifaciens sialidase or using acetic acid, mild acid hydrolysis. Lectin binding studies revealed that the majority of conidial sialic acids are alpha2,6-linked to a galactose residue. High-pressure liquid chromatography of derivatized sialic acids released from conidia revealed that unsubstituted N-acetylneuraminic acid is the predominant sialic acid on the surface of conidia. Enzymatic removal of sialic acid significantly decreased the binding of conidia to fibronectin by greater than 65% when compared with sham-treated controls. In addition, removal of sialic acids decreased conidial uptake by cultured murine macrophages and Type 2 pneumocytes by 33% and 53%, respectively. Hence, sialylated molecules on A. fumigatus conidia are ligands for both professional and nonprofessional phagocytes.

Mechanistic requirements for the efficient enzyme-catalyzed hydrolysis of thiosialosides.

The sialidase from Micromonospora viridifaciens has been found to catalyze the hydrolysis of aryl 2-thio-alpha-D-sialosides with remarkable efficiency: the first- and second-order rate constants, kcat and kcat/Km, for the enzyme-catalyzed hydrolysis of PNP-S-NeuAc are 196 +/- 5 s(-1) and (6.7 +/- 0.7) x 10(5) M(-1) s(-1), respectively. A reagent panel of eight aryl 2-thio-alpha-D-sialosides was synthesized and used to probe the mechanism for the M. viridifaciens sialidase-catalyzed hydrolysis reaction. In the case of the wild-type enzyme, the derived Brønsted parameters (beta(lg)) on kcat and kcat/Km are -0.83 +/- 0.11 and -1.27 +/- 0.17 for substrates with thiophenoxide leaving groups of pKa values > or = 4.5. For the general-acid mutant, D92G, the derived beta(lg) value on kcat for the same set of leaving groups is -0.82 +/- 0.12. When the conjugate acid of the departing thiophenol was < or = 4.5, the derived Brønsted slopes for both the wild-type and the D92G mutant sialidase were close to zero. In contrast, the nucleophilic mutant, Y370G, did not display a similar break in the Brønsted plots, and the corresponding values for beta(lg), for the three most reactive aryl 2-thiosialosides, on kcat and kcat/Km are -0.76 +/- 0.28 and -0.84 +/- 0.04, respectively. Thus, for the Y370G enzyme glycosidic C-S bond cleavage is rate-determining for both kcat and kcat/Km, whereas, for both the wild-type and D92G mutant enzymes, the presented data are consistent with a change in rate-determining step from glycosidic C-S bond cleavage for substrates in which the pKa of the conjugate acid of the leaving group is > or = 4.5, to either deglycosylation (kcat) or a conformational change that occurs prior to C-S bond cleavage (kcat/Km) for the most activated leaving groups. Thus, the enzyme-catalyzed hydrolysis of 2-thiosialosides is strongly catalyzed by the nucleophilic tyrosine residue, yet the C-S bond cleavage does not require the conserved aspartic acid residue (D92) to act as a general-acid catalyst.

Natural sialoside analogues for the determination of enzymatic rate constants.

Two isomeric 4-methylumbelliferyl-alpha-D-N-acetylneuraminylgalactopyranosides (1 and 2) were synthesised. These compounds contain either the natural alpha-2,3 or alpha-2,6 sialyl-galactosyl linkages, as well as an attached 4-methylumbelliferone for convenient detection of their hydrolyses. These compounds were designed as natural sialoside analogues to be used in a continuous assay of sialidase activity, where the sialidase-catalysed reaction is coupled with an exo-beta-galactosidase-catalysed hydrolysis of the released galactoside to give free 4-methylumbelliferone. The kinetic parameters for 1 and 2 were measured using the wild-type and nucleophilic mutant Y370G recombinant sialidase from Micromonospora viridifaciens. Kinetic parameters for these analogues measured using the new continuous assay were in good agreement with the parameters for the natural substrate, 3'-sialyl lactose. Given the selection of commercially available exo-beta-galactosidases that possess a variety of pH optima, this new method was used to characterise the full pH profile of the wild-type sialidase with the natural sialoside analogue 1. Thus, use of these new substrates 1 and 2 in a continuous assay mode, which can be detected by UV/Vis or fluorescence spectroscopy, makes characterisation of sialidase activity with natural sialoside linkages much more facile.

The hydrolase and transferase activity of an inverting mutant sialidase using non-natural beta-sialoside substrates.

The Y370G inverting mutant sialidase from Micromonospora viridifaciens possesses beta-sialidase activity with phenyl beta-sialoside (Ph-betaNeuAc) to give alpha-sialic acid as the first formed product. The derived catalytic rate constants for k(cat) and k(cat)/K(m) are 13.3 +/- 0.3 and (2.9 +/- 0.3) x 10(5) M(-)(1) s(-)(1), respectively. This enzyme is highly specific for the phenyl substrate, with substituted phenyl and thiophenyl leaving groups having k(cat) values that are at least 1000-fold lower. In addition, the Y370G mutant can transfer the sialic acid moiety from Ph-betaNeuAc to lactose in yields of up to 13%. Greater than 90% of the sialyl-lactose product formed in the coupling reactions is the alpha-2,6-isomer. A library encoding 6 x 10(5) different sialidases was constructed by mutating Y370, E260, T309, N310, and N311, residues that include and are proximal the catalytic tyrosine residue. A total of 2628 individuals were screened for hydrolytic activity against 4-nitrophenyl 2-thio-beta-sialoside and 4-methylumbelliferyl beta-sialoside. However, none of the mutants screened possessed a significant activity against either of the beta-sialosides.

Two nucleophilic mutants of the Micromonospora viridifaciens sialidase operate with retention of configuration by two different mechanisms.

Mutants of the Micromonospora viridifaciens sialidase, Y370E and Y370F, are catalytically active retaining enzymes that operate by different mechanisms. Previous substitutions with smaller amino acids, including Y370D, yielded inverting sialidases. At least one water molecule can fit into the active-site cavity of this mutant and act as a nucleophile from the face opposite the leaving group (Biochemistry 2003, 42, 12 682). Thus, addition of a CH(2) unit (Asp versus Glu) changes the mechanism from inversion back to retention of configuration. Based on Brønsted beta(lg) values, it is proposed that the Y370E mutant reacts by a double-displacement mechanism (beta(lg) on k(cat)/K(m) -0.36+/-0.04) with Glu370 acting as the nucleophile. However, the Y370F mutant (beta(lg) on k(cat)/K(m) -0.79+/-0.12) reacts via a dissociative transition state. The crystal structure of the Y370F mutant complexed with 2-deoxy-2,3-dehydro-N-acetylneuraminic acid shows no significant active-site perturbation relative to the wild-type enzyme.

Galactose recognition by the carbohydrate-binding module of a bacterial sialidase.

Glycoside hydrolases often possess carbohydrate-binding modules (CBMs) in addition to their catalytic domains, which help target the enzymes to appropriate substrates and thereby increase their catalytic efficiency. Sialidases hydrolyse the release of sialic acid from a variety of glycoconjugates and play significant roles in the pathogenesis of a number of important diseases. The sialidase from Micromonospora viridifaciens has a CBM which recognizes galactose. The CBM is linked to the catalytic domain by an immunoglobulin-like domain, resulting in the galactose binding site sitting above the catalytic site, suggesting an interplay between the two sites. By studying nine crystallographically independent structures of the M. viridifaciens sialidase, the relative flexibility of the three domains was analysed. A detailed study is also presented of the recognition of galactose and lactose by the M. viridifaciens CBM. The striking structure of this sialidase suggests a role for the CBM in binding to galactose residues unmasked by the adjacent catalytic site.

Structure and mechanism of action of an inverting mutant sialidase.

Mutagenesis of the conserved tyrosine (Y370) of the Micromonospora viridifaciens sialidase to small amino acids changes the mechanism of catalysis from retention of anomeric configuration to inversion [Watson, J. N., et al. (2003) Biochemistry 42, 12682-12690]. For the Y370G mutant enzyme-catalyzed hydrolysis of a series of aryl sialosides and 3'-sialyllactose, the derived Brønsted parameters (beta(lg)) on k(cat) and k(cat)/K(m) are -0.63 +/- 0.05 and -0.80 +/- 0.08, respectively. Thus, for the Y370G enzyme, glycosidic C-O bond cleavage is rate-determining. Analysis of the activity of the Y370G mutant and wild-type enzymes against a substrate [3,4-dihydro-2H-pyrano[3,2-c]pyridinium alpha-d-N-acetylneuraminide (DHP-alphaNeu5Ac)] whose hydrolysis cannot be accelerated by acid catalysis is consistent with these reactions proceeding via S(N)1 and S(N)2 mechanisms, respectively. The overall structure of the Y370G mutant sialidase active site is very similar to the previously reported wild-type structure [Gaskell, A., et al. (1995) Structure 3, 1197-1205], although removal of the tyrosine residue creates two significant changes to the active site. First, the anomeric oxygen atom of the hydrolysis product (beta-N-acetylneuraminic acid) and four water molecules bind in the large cavity created by the Y370G mutation. Second, the side chain of Asn310 moves to make a strong hydrogen bond to one of the bound water molecules.

Use of conformationally restricted pyridinium alpha-D-N-acetylneuraminides to probe specificity in bacterial and viral sialidases.

Investigations into subtle changes in the catalytic activity of sialidases have been performed using enzymes from several different origins, and their results have been compared. This work highlights the potential pitfalls encountered when extending conclusions derived from mechanistic studies on a single enzyme even to those with high-sequence homology. Specifically, a panel of 5 pyridinium N-acetylneuraminides were used as substrates in a study that revealed subtle differences in the catalytic mechanisms used by 4 different sialidase enzymes. The lowest reactivity towards the artificial (pyridinium) substrates was displayed by the Newcastle disease virus hemagglutinin-neuraminidase. Moreover, in reactions involving aryl N-acetylneuraminides, the activity of the Newcastle enzyme was competitively inhibited by the 3,4-dihydro-2H-pyrano[3,2-c]pyridinium compound with a Ki = 58 micromol/L. Alternatively, the 3 bacterial enzymes tested, from Salmonella typhimurium, Clostridium perfringens, and Vibrio cholerae, were catalytically active against all members of the panel of substrates. Based on the observed effect of leaving-group ability, it is proposed that the rate-determining step for kcat (and likely for kcat/Km as well) with each bacterial enzyme is as follows: sialylation, which is concerted with conformational change for V. cholerae; and conformational change for S. typhimurium and C. perfringens.

Mutagenesis of the conserved active-site tyrosine changes a retaining sialidase into an inverting sialidase.

Mutagenesis of the conserved tyrosine (Y370) of the Micromonospora viridifaciens sialidase changes the mechanism of catalysis from retention of anomeric configuration to an unprecedented inverting mechanism in which water efficiently functions as the nucleophile. Three mutants, Y370A, Y370D, and Y370G, were produced recombinantly in Escherichia coli, and all are catalytically active against the activated substrate 4-methylumbelliferyl alpha-D-N-acetylneuraminide. The Y370D mutant was also shown to catalyze the hydrolysis of natural substrate analogues such as 3'-sialyllactose. A comparison of the pH-rate profiles for the wild-type and the Y370D mutant sialidase reveals no major differences, although with respect to the kinetic term k(cat)/K(m), an ionized form of the aspartate-370 enzyme is catalytically compromised. For the wild-type enzyme, the value of the Brønsted parameter beta(lg) on k(cat) is 0.02 +/- 0.03, while for the Y370D mutant sialidase beta(lg) = -0.55 +/- 0.03 for the substrates with bad leaving groups. Thus, for the wild-type enzyme, a nonchemical step(s) is rate-limiting, but for the tyrosine mutant cleavage of the glycosidic C-O bond is rate-determining. The Brønsted slopes derived for the kinetic parameter k(cat)/K(m) display a similar trend (beta(lg) -0.30 +/- 0.04 and -0.74 +/- 0.04 for the wild-type and Y370D, respectively). These results reveal that the tyrosine residue lowers the activation free energy for cleavage of 6'-sialyllactose, a natural substrate analogue, by more than 24.9 kJ mol(-1). Evidence is presented that the mutant sialidases operate by a dissociative mechanism, and the wild-type enzyme operates by a concerted mechanism.