Purification and properties of cloned Salmonella typhimurium LT2 sialidase with virus-typical kinetic preference for sialyl alpha 2----3 linkages

Diversity of microbial sialic acid metabolism

Sialic acids are structurally unique nine-carbon keto sugars occupying the interface between the host and commensal or pathogenic microorganisms. An important function of host sialic acid is to regulate innate immunity, and microbes have evolved various strategies for subverting this process by decorating their surfaces with sialylated oligosaccharides that mimic those of the host. These subversive strategies include a de novo synthetic pathway and at least two truncated pathways that depend on scavenging host-derived intermediates. A fourth strategy involves modification of sialidases so that instead of transferring sialic acid to water (hydrolysis), a second active site is created for binding alternative acceptors. Sialic acids also are excellent sources of carbon, nitrogen, energy, and precursors of cell wall biosynthesis. The catabolic strategies for exploiting host sialic acids as nutritional sources are as diverse as the biosynthetic mechanisms, including examples of horizontal gene transfer and multiple transport systems. Finally, as compounds coating the surfaces of virtually every vertebrate cell, sialic acids provide information about the host environment that, at least in Escherichia coli, is interpreted by the global regulator encoded by nanR. In addition to regulating the catabolism of sialic acids through the nan operon, NanR controls at least two other operons of unknown function and appears to participate in the regulation of type 1 fimbrial phase variation. Sialic acid is, therefore, a host molecule to be copied (molecular mimicry), eaten (nutrition), and interpreted (cell signaling) by diverse metabolic machinery in all major groups of mammalian pathogens and commensals.

Increasing the hydrophobic interaction between terminal W-motifs enhances the stability of Salmonella typhimurium sialidase. A general strategy for the stabilization of beta-propeller protein fold

Protein engineering of the beta-propeller protein aimed at enhancing the structural stability of the protein was carried out using a monomeric single domain beta-propeller protein, Salmonella typhimurium sialidase, as a model. Ala53 and Ala69 each located at strands B and C of the W1 motif were mutated to Leu and Val, respectively, to increase the hydrophobic interaction between W1 and W6 motifs. The mutants showed enhanced stability towards guanidine hydrochloride and thermal unfolding. Ala53Leu showed higher stability, probably owing to the capability of the mutated Leu to interact extensively with more residues involved in the hydrophobic interactions between the terminal W-motifs. The mutations, which are located far from the active site, have no significant effect on the enzymatic properties. The strategy to enhance the stability proposed here might be applied to the other beta-propeller proteins.

Knockout of mouse beta 1,4-galactosyltransferase-1 gene results in a dramatic shift of outer chain moieties of N-glycans from type 2 to type 1 chains in hepatic membrane and plasma glycoproteins

To understand the contribution of beta 1,4-galactosyltransferase (beta 4Gal-T)-1 to galactosylation in vivo, N-glycans of hepatic membrane glycoproteins and plasma glycoproteins from beta 4Gal-T1 wild-type (beta 4Gal-T1(+/+)) and beta 4Gal-T1 knockout mice were compared. Unexpectedly, glycoproteins from the knockout mice were found to express considerable amounts of sialylated, galactosylated N-glycans. A striking contrast was that galactose residues were largely beta 1,4-linked to GlcNAc residues in the beta 4Gal-T1(+/+) mouse glycans but beta 1,3-linked in the knockout mouse glycans, thus resulting in the shift of the backbone structure from type 2 chain (Gal beta 1-->4GlcNAc) to type 1 chain (Gal beta 1-->3GlcNAc). Detailed analysis of plasma glycoproteins revealed that the expression of sialyl linkage in N-glycans was shifted from the Sia alpha 2-->6Gal to the Sia alpha 2-->3Gal, and oversialylated type 1 chains were, remarkably, found in the knockout mouse glycans. Thus beta 4Gal-T1 deficiency was primarily compensated for by beta1,3-galactosyltransferases, which resulted in different sialyl linkages being formed on the outer chains and altered backbone structures, depending on the acceptor specificities of sialyltransferases. These results suggest that beta 4Gal-T1 in mouse liver plays a central role in the synthesis of type 2 chain and is also involved in the regulation of sialylation of N-glycans. The knockout mice may prove useful in investigation of the mechanism which regulates the tissue-dependent terminal glycosylation.

Adeno-associated virus serotype 4 (AAV4) and AAV5 both require sialic acid binding for hemagglutination and efficient transduction but differ in sialic acid linkage specificity

Adeno-associated virus serotype 4 (AAV4) and AAV5 have different tropisms compared to AAV2 and to each other. We recently reported that alpha 2--3 sialic acid is required for AAV5 binding and transduction. In this study, we characterized AAV4 binding and transduction and found it also binds sialic acid, but the specificity is significantly different from AAV5. AAV4 can hemagglutinate red blood cells from several species, whereas AAV5 hemagglutinates only rhesus monkey red blood cells. Treatment of red blood cells with trypsin inhibited hemagglutination for both AAV4 and AAV5, suggesting that the agglutinin is a protein. Treatment of Cos and red blood cells with neuraminidases also indicated that AAV4 bound alpha 2--3 sialic acid. However, resialylation experiments with neuraminidase-treated red blood cells demonstrated that AAV4 binding required alpha 2--3 O-linked sialic acid, whereas AAV5 required N-linked sialic acid. Similarly, resialylation of sialic acid-deficient CHO cells supported this same conclusion. The difference in linkage specificity for AAV4 and AAV5 was confirmed by binding and transduction experiments with cells incubated with either N-linked or O-linked inhibitors of glycosylation. Furthermore, AAV4 transduction was only blocked with soluble alpha 2-3 sialic acid, whereas AAV5 could be blocked with either alpha 2--3 or alpha 2-6 sialic acid. These results suggest that AAV4 and AAV5 require different sialic acid-containing glycoproteins for binding and transduction of target cells and they further explain the different tropism of AAV4 and AAV5.

Occurrence of gangliosides in the common squid and pacific octopus among protostomia

Acidic lipids from tissues of the common squid Todarodes pacificus and the pacific octopus Octopus vulgaris were characterized. Hepatopancreatic tissues of both animals had complex compositions of resorcinol-positive acidic lipids, many of which became reactive with cholera toxin B subunit and anti-G(M1) antibody after in situ treatment with sialidase on TLC. One of the major acidic lipids in squid tissue was isolated and examined for its structure. This acidic lipid was identified to be the ganglioside G(D1a) based upon the susceptibility to sialidases of different substrate specificity, characterization of reaction products, and electrospray ionization-mass spectrometry of the lipid. Hepatopancreatic tissues of squid and octopus also contained acidic lipids that reacted with A2B5, a monoclonal antibody specific to c-series gangliosides. Cerebral ganglia of both animals expressed resorcinol-positive acidic lipids, though their compositional patterns differed from the hepatopancreatic tissues. N-Acetylneuraminic acid was identified as the main species in lipid-bound sialic acid in both tissues. The contents of lipid-bound sialic acid in cerebral ganglia were significantly lower than those of hepatopancreatic tissues in both animals. The present study presents the first evidence for the occurrence of gangliosides in protostomia.

Probing molecular function of trypanosomal sialidases: single point mutations can change substrate specificity and increase hydrolytic activity

Sialidases are present on the surface of several trypanosomatid protozoan parasites. They are highly specific for sialic acid linked in alpha-(2,3) to a terminal beta-galactose and include the strictly hydrolytic enzymes and trans-sialidases (sialyl-transferases). Based on the structural comparison of the sialidase from Trypanosoma rangeli and the trans-sialidase from T. cruzi (the agent of Chagas' disease in humans), we have explored the role of specific amino acid residues sought to be important for substrate specificity. The substitution of a conserved tryptophanyl residue in the two enzymes, Trp312/313-Ala, changed substrate specificity, rendering the point mutants capable to hydrolyze both alpha-(2,3)- and alpha-(2,6)-linked sialoconjugates. The same mutation abolished sialyl-transferase activity, indicating that transfer (but not hydrolysis) requires a precise orientation of the bound substrate. The exchange substitution of another residue that modulates oligosaccharide binding, Gln284-Pro, was found to significantly increase the hydrolytic activity of sialidase, and residue Tyr119 was confirmed to be part of a second binding site for the acceptor substrate in trans-sialidase. Together with the structural information, these results provide a consistent framework to account for the unique enzymatic properties of trypanosome trans-sialidases.

Analysis of sialo-N-glycans in glycoproteins as 1-phenyl-3-methyl-5-pyrazolone derivatives by capillary electrophoresis

A method for the analysis of the sialo-N-glycans in glycoproteins was established by the electrokinetic chromatography mode of capillary electrophoresis (CE) in sodium dodecyl sulfate (SDS) micelles as 1-phenyl-3-methyl-5-pyrazolone (PMP) derivatives, using sialo-N-glycans in fetuin as a model. Six major and some minor peaks were observed for the N-glycans in fetuin, which were well separated from each other using 50 mM phosphate buffer, pH 6.0, containing SDS to a concentration of 30 mM in an uncoated fused-silica capillary, and these peaks were assigned to sialo-N-glycans having either of the biantennary or beta1-3/beta1-4 linked galactose-containing complex type triantennary N-glycans as the basic structures, by an indirect method based on the assignment of the peaks in high-performance liquid chromatography separated in parallel with CE and peak collation between these two separation methods. The attaching position of the sialic acid residue was determined using the linkage preference of neuraminidase isozymes. The established system is considered to be useful for routine analysis of microheterogeneity of the carbohydrate moiety of this model glycoprotein from the following reasons: (1) the derivatization with PMP proceeds quantitatively under mild conditions without causing release of the sialic acid residue, (2) the derivatives can be sensitively detected by UV absorption, (3) the procedure is simple, rapid and reproducible. Preliminary results of N-glycan analysis for several other glycoproteins under these conditions are also presented.

Chemoenzymatic synthesis of sialyl oligosaccharides with sialidases employing transglycosylation methodology

A series of sialyloligosaccharides was synthesized using the transglycolytic activity of the sialidases from Vibrio cholerae, Clostridium perfringens, Salmonella typhimurium, and Newcastle disease virus. According to their hydrolytic activities the sialidases from V. cholerae and C. perfringens catalyze preferentially the formation of sialyl alpha(2-6)-linkages whereas the sialidases from S.typhimurium and Newcastle disease virus show a distinct preference for alpha(2-3) directed sialylations. Using combined chemical and enzymatic methodologies structures such as T-(Thomsen-Friedenreich) antigen [beta-D-Gal-(1-3)-alpha-D-GalNAc-OThr], Tn-(Thomsen nouveau) antigen (alpha-D-GalNAc-OThr) and beta-D-Gal-(1-4)-alpha-D-2-deoxy-Gal-OMe were sialylated in alpha(2-3)- and alpha(2-6)-positions regioselectively or in high regioisomeric excess and purified by simple isolation procedures. Depending on the enzyme source and acceptor structure yields for transsialylation varied between 10 and 30%.

trans-sialidase of Trypanosoma cruzi: location of galactose-binding site(s)

Trypanosoma cruzi expresses a trans-sialidase on its surface, which catalyzes the transfer of sialic acid from mammalian host glycans to its own surface glycoproteins. It has been proposed that the enzyme consists of three domains prior to a long C-terminal repeating sequence that is not required for enzyme activity. The first of these domains shares significant sequence identity with bacterial sialidases which catalyse the hydrolysis of sialic acid. Here we report the sequence of the N-terminal domains of the TS19y trans-sialidase gene, which was expressed in bacteria with the same specific activity as natural enzyme of T. cruzi. Various deletion mutants of TS19y, without the C-terminal tandem repeat, have been cloned and expressed and their trans-sialidase and sialidase activities measured. These experiments show that all three N-terminal domains are required for full trans-sialidase activity, though only the first is necessary for sialidase activity. Some transferase activity is observed, however, even with the shortest construct comprising the first N-terminal domain. Deletion mutants to probe the role of the N-terminal residues of the first domain suggest that the first 33 residues are also required for trans-sialidase activity, but not for sialidase activity. Molecular modelling of the first N-terminal domain of TS19y based on our structures of bacterial sialidases and site-directed mutations suggests the location of a galactose-binding site within this domain.

The 1.8 A structures of leech intramolecular trans-sialidase complexes: evidence of its enzymatic mechanism

Intramolecular trans-sialidase from leech (Macrobdella decora) is the first member of the sialidase superfamily found to exhibit strict specificity towards the cleavage of terminal Neu5Acalpha2-->3Gal linkage in sialoglycoconjugates. Its release of 2,7-anhydro-Neu5Ac instead of Neu5Ac indicates that it catalyzes an intramolecular trans-sialosyl reaction. Crystal structures of its complexes with an inactive substrate analogue 2-propenyl-Neu5Ac, and with the product 2,7-anhydro-Neu5Ac, have been determined to 1.8 A resolution. The boat conformation of the pyranose observed in the complexes supports the proposed enzymatic mechanism that O7 of an axial 6-glycerol group attacks the positively charged C2 of the intermediate. A generalized mechanism is proposed for the sialidase superfamily.

Identification and characterization of a sialidase released by the salivary gland of the hematophagous insect Triatoma infestans

Sialidases (EC 3.2.1.18) are commonly found in viruses, bacteria, fungi, protozoa, and vertebrates, but not in invertebrates. We have previously reported the presence of a new sialidase activity in the gut of exclusively hematophagous insects of the Triatoma genus, which transmit Chagas' disease (Amino, R., Acosta, A., Morita, O. M., Chioccola, V. L. P., and Schenkman, S. (1995) Glycobiology 5, 625-631). Here we show that this sialidase is present in the salivary gland of Triatoma infestans, and it is released with the saliva during the insect bite. The sialidase was purified to homogeneity (>5000 times) to a specific activity of more than 20 units/mg. It elutes from a gel filtration column with a volume corresponding to the size of 33 kDa, and it migrates as a single 26-kDa band in SDS-polyacrylamide gel electrophoresis, which is unusually smaller when compared with other known sialidases. T. infestans sialidase hydrolyzes preferentially alpha2-->3-linked sialic acids at pH 4-8, with maximal activity between pH 5.5 and 6.5, which is compatible with the optimal pH of secreted sialidases. The sialidase is competitively inhibited by 2-deoxy-2, 3-dehydro-N-acetyl-neuraminic acid (Ki = 0.075 mM) and differently from many sialidases, with exception of Salmonella typhimurium sialidase, it is inhibited competitively by HEPES (Ki = 15 mM). The fact that T. infestans sialidase is released with the saliva and can hydrolyze sialyl-LewisX blood groups, which are the ligands for selectins, suggests that it might have a role in the blood feeding.

The carboxyl-terminal region is a determinant for the intracellular behavior of the chorionic gonadotropin beta subunit: effects on the processing of the Asn-linked oligosaccharides

The placental hormone human CG (hCG) consists of two noncovalently linked alpha- and beta-subunits similar to the other glycoprotein hormones LH, FSH, and TSH. These heterodimers share a common alpha subunit but differ in their structurally distinct beta subunits. The CGbeta subunit is distinguished among the beta subunits by the presence of a C-terminal extension with four serine-linked oligosaccharides (carboxyl terminal peptide or CTP). In previous studies we observed that deleting this sequence decreased assembly of the truncated CGbeta subunit (CGbeta114) with the alpha-subunit and increased the heterogeneity of the secreted forms of the uncombined subunit synthesized in transfected Chinese hamster ovary (CHO) cells. The latter result was attributed to alterations in the processing of the two N-linked oligosaccharides. To examine at what step this heterogeneity occurs, the CGbeta and CGbeta114 genes were transfected into wild-type and mutant CHO cell lines that are defective in the late steps of the N-linked carbohydrate-processing pathway. We show here that removal of the CTP alters the processing of the core mannosyl unit of the subunit to complex forms at both glycosylation sites and that the oligosaccharides contain polylactosamine. Although it has been presumed that there is little intramolecular interaction between the CTP and the proximal domains of the subunit, our data suggest that the CTP sequence participates in the folding of the newly synthesized subunit, which is manifest by the posttranslational changes observed here.

The crystal structure of an intramolecular trans-sialidase with a NeuAc alpha2-->3Gal specificity

BACKGROUND

Intramolecular trans-sialidase from leech (Macrobdella decora) is a unique enzyme which cleaves the terminal neuraminic acid (NeuAc) residue from sialoglycoconjugates, releasing 2, 7-anhydro-neuraminic acid (2,7-anhydro-NeuAc). It is the first enzyme found to exhibit strictly specific cleavage of NeuAc alpha2-->3Gal linkages in sialoglycoconjugates. The release of 2,7-anhydro-NeuAc instead of NeuAc implies a unique mechanism, in which the sialosyl linkage is transferred within the sialoglycoconjugate rather than hydrolyzed. The aims of the structural study were to gain structural insight into the strict specificity and unique mechanism of this unusual enzyme.

RESULTS

. The 2.0 A crystal structure of recombinant leech intramolecular trans-sialidase has been solved by multiple isomorphous replacement. The 1.8 A structure of the enzyme in complex with 2-deoxy-2, 3-didehydro-NeuAc was also solved. The refined model comprising residues 81-769 has a catalytic beta-propeller domain (C), a N-terminal lectin-like domain (II) and an irregular beta-stranded domain (III) inserted into the catalytic domain. The structure reveals several possible carbohydrate-binding features: domain II has a concave face, like that of other sialidases, and there is a suitable surface charge distribution at the domain III-C interface.

CONCLUSIONS

Structural comparisons showed closer evolutionary relationships to bacterial sialidases than to viral neuraminidases. Mainchain and sidechain atoms around Thr593 make the glycerol-binding pocket incapable of accommodating an extended equatorial 6-glycerol group, implying that the 6-glycerol group of the reaction intermediate may occupy an axial position, which is also required by the catalytic mechanism. The steric hindrance introduced by the bulky sidechain of Trp734 above the 2-carboxylate group may explain the lack of water involvement in the cleavage reaction and the substrate specificity.

Two-dimensional mapping of 8-amine-1,3,6-naphthalene trisulfonic acid derivatives of N-linked neutral and sialyloligosaccharides

We describe a simple and sensitive two-dimensional sugar-mapping technique of 8-amine-1,3,6-naphthalene trisulfonic acid derivatives (ANTS derivatives) of neutral and sialyloligosaccharides for structure analysis and characterization of N-linked oligosaccharides using picomoles of samples. The method includes: (1) reductive amination with ANTS of enzymatically released oligosaccharides, (2) simultaneous separation of oligosaccharide derivatives in a fluorophore-assisted carbohydrate electrophoresis and NH2-HPLC column under ion suppression conditions, (3) plotting of the relative migration indexes (X axis) and relative retention times (Y axis), and (4) when necessary, additional exoglycosidase digestion. As illustrated by the glycosylation profiling and structural analysis of alpha 1 anti-trypsin and murine IgG 2a, this methodology fulfills most of the requirements for a complete characterization of neutral and charged oligosaccharides released from N-glycosylated glycoprotein.

Arthrobacter ureafaciens sialidase isoenzymes, L, M1 and M2, cleave fucosyl GM1

Among bacterial, fungal and viral sialidases, the sialidase from Arthrobacter ureafaciens has the unique property of cleaving sialic acids linked to the internal galactose of gangliotetraose. In this study, we examined the ability to cleave the internal sialic acids of GM1 and fucosyl GM1 of sialidases from several bacterial and fungal origins, including Clostridium perfringens and Vibrio cholerae. We found that A. ureafaciens sialidase could liberate the sialic acid of GM1 at the highest rate, and was the only enzyme which could cleave fucosyl GM1 among the sialidases examined. The affinity-purified sialidase derived from the culture medium of A. ureafaciens was comprised of four isoenzymes with different molecular weights and isoelectric points, the isoenzymes that cleaved fucosyl GM1 being L (88 kDa, pl 5.0), M1 (66 kDa, pl 6.2) and M2 (66 kDa, pl 5.5), but not S (52 kDa, pl 6.2) which showed the highest specific activity toward colominic acid among the four isoenzymes.

Detailed oligosaccharide structures of human integrin alpha 5 beta 1 analyzed by a three-dimensional mapping technique

Structures of N-linked oligosaccharides obtained from human integrin alpha 5 beta 1 are described. Integrin alpha 5 beta 1 (4.5 mg) was purified from human placenta and digested using trypsin and chymotrypsin. N-linked oligosaccharides were released from the glycopeptides by digestion with glycoamidase A (from almond). The reducing ends of the oligosaccharides were derivatized with 2-aminopyridine. The pyridylamino-oligosaccharides were separated and these structures were identified by a three-dimensional HPLC mapping technique on three kinds of HPLC columns [Takahashi, N., Nakagawa, H., Fujikawa, K., Kawamura, Y. & Tomiya, N. (1995) Anal. Biochem. 226, 139-146]. Finally, 35 different oligosaccharide structures were identified, 10 of which were neutral, 6 mono-sialyl, 10 di-sialyl, 7 tri-sialyl and 2 tetra-sialyl. The molar ratio of neutral, mono-sialyl, di-sialyl, tri-sialyl and tetra-sialyl oligosaccharides was 20.8%, 24.8%, 27.7%, 18.1% and 8.6%, respectively. High-mannose-type oligosaccharides accounted for only 1.5% of the total. The remaining oligosaccharides were all complex type. The most predominant structure was the diantennary di-alpha-(2,3)-sialyl fucosyl. Major linking of sialic acid was alpha-(2,3)-linkage, and over 50% of all oligosaccharides were fucosylated at the N-acetylglucosamine residue of the reducing end.

Oligosaccharide sequencing based on exo- and endoglycosidase digestion and liquid chromatographic analysis of the products

Exo- and endoglycosidases are used to sequence oligosaccharides and give valuable information on the monosaccharide sequence, together with the anomericity, the stereochemistry, and in some cases, the substitution pattern of the monosaccharides. Both sequential and parallel methods of oligosaccharide sequencing are discussed.

Adherence of Salmonella typhimurium to Caco-2 cells: identification of a glycoconjugate receptor

The mechanism by which Salmonella species adhere to the epithelium of the intestine is not well understood. To identify components on intestinal epithelial cells that may be involved in the initial adherence of Salmonella typhimurium, we correlated patterns of adherence to well-differentiated Caco-2BBe cell monolayers with expression of brush border membrane components and lectin binding sites. This cloned cell line shows heterogeneous expression of sucrase-isomaltase and most lectin receptors. S. typhimurium adhered to a subpopulation of living or formaldehyde-fixed cells with a high multiplicity (up to 150 bacteria per cell). Bacterial binding to selected cells was not correlated with expression of the brush border hydrolases dipeptidyl-peptidase IV and sucrase-isomaltase or with binding of 10 of the 12 lectins tested. However, binding was correlated with the presence of binding sites for peanut agglutinin (PNA) [specific for Gal beta (1-3) GalNAc] and soybean agglutinin (specific for terminal GalNAc). Preincubation of live and fixed Caco-2BBe monolayers with PNA inhibited bacterial binding, while preincubation with soybean agglutinin did not. Electron microscopic analysis demonstrated that the initial adherence of S. typhimurium to Caco-2 cells in vitro involved peripheral components of the glycocalyx on apical microvilli. These results suggest that a Gal beta (1-3)GalNAc epitope recognized by PNA and located in the glycocalyx is involved in the early recognition events between S. typhimurium and Caco-2 cells and that differences in glycosylation patterns among individual epithelial cells may be a determinant in cell-selective adherence of S. typhimurium.

The three domains of a bacterial sialidase: a beta-propeller, an immunoglobulin module and a galactose-binding jelly-roll

BACKGROUND

Sialidases, or neuraminidases, have been implicated in the pathogenesis of many diseases, but are also produced by many non-pathogenic bacteria. Bacterial sialidases are very variable in size, often possessing domains in addition to the catalytic domain. The sialidase from the non-pathogenic soil bacterium Micromonospora viridifaciens is secreted in two forms with molecular weights of 41 kDa or 68 kDa, depending on the nature of the carbohydrate used to induce expression.

RESULTS

We report here the X-ray crystal structures of the 41 kDa and 68 kDa forms of the sialidase from M. viridifaciens at 1.8 A and 2.5 A resolution respectively. In addition, we report a complex of the 41 kDa form with an inhibitor at 2.0 A resolution, and a complex of the 68 kDa form with galactose at 2.5 A. The 41 kDa form shows the canonical sialidase beta-propeller fold. The 68 kDa form possesses two additional domains, one with an immunoglobulin-like fold that serves as a linker to the second, which is homologous to the galactose-binding domain of a fungal galactose oxidase.

CONCLUSIONS

The presence of the additional carbohydrate-binding domain in the 68 kDa form of the bacterial sialidase reported here is a further example of a combination of carbohydrate binding and cleaving domains which we observed in the sialidase from Vibrio cholerae. This dual function may be common, but only to other bacterial and parasitic sialidases, but also to other secreted glycosidases involved in pathogenesis. The bacterium may have acquired both the immunoglobulin module and the galactose-binding module from eukaryotes, as the enzyme shows a remarkable similarity to a fungal galactose oxidase which possesses similar domains performing different functions and assembled in a different order.

Structures of the N-linked oligosaccharides on human plasma vitronectin

The structures of N-linked oligosaccharides present on human plasma vitronectin were elucidated. Oligosaccharides were released from the vitronectin by N-glycosidase F digestion and tagged with 2-aminopyridine; the pyridylamino-oligosaccharides were then fractionated by anion-exchange and reverse-phase HPLC. Ten major pyridylamino-oligosaccharides were isolated. The linkages and locations of sialic acid residues were determined by desialylation with Salmonella sialidase in combination with acid. The asialo forms were then analyzed by two-dimensional sugar mapping, component sugar analysis and 400-MHz 1H-NMR spectroscopy. The major oligosaccharides of human vitronectin were of the diantennary N-acetyllactosamine type, with a lesser amount of the tri- and a small amount of the mono-antennary type, to which 1-3 mol sialic acid residues were linked, mostly through alpha 2-6 linkages, although alpha 2-3 linkages were also present. The possibility that several binding activities of vitronectin can be ascribed to its glycan moiety was discussed, based on the specific features of the N-linked oligosaccharides on human vitronectin revealed here.

Genetic map of Salmonella typhimurium, edition VIII

We present edition VIII of the genetic map of Salmonella typhimurium LT2. We list a total of 1,159 genes, 1,080 of which have been located on the circular chromosome and 29 of which are on pSLT, the 90-kb plasmid usually found in LT2 lines. The remaining 50 genes are not yet mapped. The coordinate system used in this edition is neither minutes of transfer time in conjugation crosses nor units representing "phage lengths" of DNA of the transducing phage P22, as used in earlier editions, but centisomes and kilobases based on physical analysis of the lengths of DNA segments between genes. Some of these lengths have been determined by digestion of DNA by rare-cutting endonucleases and separation of fragments by pulsed-field gel electrophoresis. Other lengths have been determined by analysis of DNA sequences in GenBank. We have constructed StySeq1, which incorporates all Salmonella DNA sequence data known to us. StySeq1 comprises over 548 kb of nonredundant chromosomal genomic sequences, representing 11.4% of the chromosome, which is estimated to be just over 4,800 kb in length. Most of these sequences were assigned locations on the chromosome, in some cases by analogy with mapped Escherichia coli sequences.

Binding of human collectins (SP-A and MBP) to influenza virus

Collectins are a group of soluble proteins each of which has collagenous domains and non-collagenous globular domains, the latter containing the consensus residues found in C-type lectins. Members of the collectin family are the serum proteins mannan-binding protein (MBP), conglutinin, CL-43, and the lung-associated proteins surfactant protein A (SP-A) and surfactant protein D (SP-D). MBP and conglutinin have been shown previously to bind to influenza viruses and to inhibit the infectivity and haemagglutinating activity of influenza viruses. We report here that the lung protein SP-A, like MBP, can bind to influenza virus (strain A/X31) through its lectin domain and inhibit the virus-mediated agglutination of red cells. The binding of SP-A or MBP to influenza virus was saturable, concentration-dependent, and required the presence of Ca2+ ions. Ligand-blot analysis, using MBP as ligand, of the virus lysate indicated that MBP binds to a 68 kDa viral species. The 68 kDa species was isolated to homogeneity and was shown to be the viral neuraminidase. The purified 68 kDa species inhibited the binding of both MBP and SP-A to influenza virus.

Expression cloning of a CMP-NeuAc:NeuAc alpha 2-3Gal beta 1-4Glc beta 1-1'Cer alpha 2,8-sialyltransferase (GD3 synthase) from human melanoma cells

Using an expression cloning approach, we have isolated a cDNA encoding GD3 synthase (CMP-NeuAc:NeuAc alpha 2-3Gal beta 1-4Glc beta 1-1'Cer alpha 2,8-sialyltransferase, EC 2.4.99.8), which is a key regulatory enzyme determining the prominence of the ganglioside biosynthesis pathway. The cloned cDNA encodes a 341-amino acid protein containing a single transmembrane domain at its N-terminal region, suggesting that the protein has a type II transmembrane topology. The sequence of alpha 2,8-sialyltransferase showed a high level of similarity with other sialyltransferases at two conserved regions typical in the sialyltransferase family. Transfected cells containing the cloned cDNA expressed GD3 ganglioside on the cell surface, which was detectable with specific anti-GD3 antibody by immunofluorescence and immunostaining after separation of isolated glycolipids on thin-layer chromatography. The cDNA hybridized to a single mRNA species of 2.4 kb in melanoma cells. This sialyltransferase is distinctive in catalyzing the formation of the alpha 2-8 linkage of sialic acids.

Microbial sialidases: does bigger always mean better?

Sialidases are a superfamily of N-acylneuraminate-releasing (sialic-acid-releasing) exoglycosidases found mainly in higher eukaryotes and in some, mostly pathogenic, viruses, bacteria and protozoans. The functions of sialidases are poorly understood and, until recently, their biochemical and evolutionary relationships were unclear. A comparative approach has demonstrated the remarkable similarities and differences between nonviral sialidases, and is providing clues about their functions.

In vitro synthesis of disialoganglioside (GD1 alpha) from asialo-GM1 using sialyltransferases in rat liver Golgi vesicles

Two gangliosides were efficiently synthesized from asialo-GM1 (Gal beta 1-3GalNAc beta 1-4Gal beta 1-4Glc beta 1-1 Cer) and cytidine 5'-phosphate-N-acetylneuraminic acid (CMP-NeuAc) by using sialyltransferases in rat liver Golgi vesicles in vitro. These gangliosides were rapidly purified by a combination of anion exchange and reverse-phase column chromatographies. The ganglioside structures were determined by TLC analysis, treatment with a sialidase from Salmonella typhimurium LT2, which specifically hydrolyzes alpha 2-3 N-acetylneuraminic acid (NeuAc alpha 2-3) linkages, TLC immunostaining, and 1H-NMR spectroscopy. One of the gangliosides was identified as GD1 alpha [Neu-Ac alpha 2-3Gal beta 1-3(NeuAc alpha 2-6)GalNAc beta 1-4Gal beta 1-4Glc beta 1-1 Cer]. The other ganglioside was determined to be GM1b (NeuAc alpha 2-3Gal beta 1-3GalNAc beta 1-4Gal beta 1-4Glc beta 1-1 Cer), which has been reported in a previous study [Pohlentz, G., Klein, D., Schmitz, D., Schwarzmann, G., Peter-Katalinic, J. & Sandhoff, K. (1988) Biol. Chem. Hoppe-Seyler 369, 55-63]. Finally, GM1b and GD1 alpha were obtained from asialo-GM1 as a starting material in 8.1% and 1.2% overall yields, respectively. This study also suggests that the novel synthetic pathway asialo-GM1-->GM1b-->GD1 alpha may exist in rat liver.

Structures of the N-linked oligosaccharides on porcine plasma vitronectin

The structures of N-linked oligosaccharides, especially the distribution of sialic acid species, present on porcine plasma vitronectin were elucidated. Oligosaccharides were released from the vitronectin by N-glycosidase F digestion and tagged with 2-aminopyridine, and the pyridylamino-oligosaccharides were fractionated by anion-exchange and reverse-phase HPLC. Nine major pyridyl-amino-oligosaccharides were isolated. The linkages and locations of sialic acids were determined by a novel approach involving desialylation with Salmonella sialidase in combination with acid desialylation. After desialylation, the asialo-forms were analyzed by two-dimensional sugar mapping, component sugar analysis and 400-MHz 1H-NMR spectroscopy. The major oligosaccharides of porcine vitronectin were of the fucosylated biantennary type, with a small amount of the triantennary N-acetyllactosamine type, to which 1-3 mol sialic acids was linked. Sialic acids were linked predominantly through alpha 2-6 linkages, although alpha 2-3 linkages were also present, and fucose was linked to the innermost N-acetylglucosamine through an alpha 1-6 linkage. It was found that every pyridylamino-oligosaccharide population contained N-glycolylneuraminic acid and N-acetylneuraminic acid in a molar ratio of 1:2-9, and that N-glycolylneuraminic acids were located predominantly on the Man alpha 1-6 arm.

The cell surface receptor is a major determinant restricting the host range of the B-lymphotropic papovavirus

The B-lymphotropic papovavirus (LPV) productively infects only a subset of human B-lymphoma-derived cell lines while transfection of the viral genome yields infectious viral particles in a much wider variety of human hematopoietic cell lines. We have analyzed the contribution of a putative LPV receptor on the cell surface of B-cell lines in restricting the virus host range. In order to establish a quantitative virus binding assay for LPV, infectious virus particles were highly purified by metrizamide equilibrium density centrifugation and used as immunogens to raise seven mouse monoclonal antibodies specific for LPV VP1. Virus particle binding was quantitated in an indirect, nonradioactive assay with an LPV VP1-specific enzyme-linked immunosorbent assay. Binding of LPV particles to permissive human B-lymphoma cell line BJA-B occurred within minutes. Kinetics and capacity of binding were similar at 4 and 37 degrees C. A BJA-B cell was estimated to bind approximately 600 virus particles at conditions under which 50% of the administered virus was bound. The sialidase and trypsin sensitivities of the cellular virus binding moiety show that sialylated and proteinaceous components are necessary components of the LPV receptor on BJA-B cells. Despite a high binding capacity of BJA-B cells for simian virus 40, LPV binding was not significantly affected by a 20-fold excess of simian virus 40 particles, indicating that these related polyomaviruses do not bind to the same receptor on BJA-B cells. Reduction of LPV binding to sialidase-pretreated BJA-B cells was accompanied by a similar reduction of infection, indicating that virus binding may be a limiting factor in the LPV replicative cycle. The two highly LPV-permissive human B-lymphoma cell lines BJA-B and Namalwa displayed high virus binding whereas low and nonpermissive hematopoietic cell lines showed reduced or undetectable virus binding. We conclude that the inability of LPV particles to productively infect the nonpermissive human hematopoietic cell lines analyzed is probably due to the absence or insufficient expression of a functional cell surface receptor.

Structural characterization of the N-linked carbohydrate chains from mouse zona pellucida glycoproteins ZP2 and ZP3

N-linked oligosaccharides of mouse zona pellucida glycoproteins ZP2 and ZP3 were prepared as pyridylaminated derivatives. Anion-exchange HPLC revealed that 95% or more of them are acidic. About 80% of these acidic chains were neutralized by digestion with Arthrobacter ureafaciens sialidase and a small amount of sulfate was found in the residual acidic chains. The neutralized fractions (SN fractions) from ZP2 and ZP3 were similarly fractionated by size-fractionation HPLC into many peaks having different numbers of LacNAc in the non-reducing regions. The SN fractions from ZP2 were cleaved by endo-beta-galactosidase into the core and non-reducing regions. The tri- and tetra-antennary complex-type chains with a Fuc residue were predominant in the core region. On the other hand, four kinds of fragment containing the LacNAc sequence were obtained from the non-reducing region. Sialic acids were shown to be linked to the core fragments, as well as the non-reducing fragments. The N-linked oligosaccharides of ZP3 were suggested to have essentially the same structures as those of ZP2.

Crystal structure of a bacterial sialidase (from Salmonella typhimurium LT2) shows the same fold as an influenza virus neuraminidase

Sialidases (EC 3.2.1.18 or neuraminidases) remove sialic acid from sialoglycoconjugates, are widely distributed in nature, and have been implicated in the pathogenesis of many diseases. The three-dimensional structure of influenza virus sialidase is known, and we now report the three-dimensional structure of a bacterial sialidase, from Salmonella typhimurium LT2, at 2.0-A resolution and the structure of its complex with the inhibitor 2-deoxy-2,3-dehydro-N-acetylneuraminic acid at 2.2-A resolution. The viral enzyme is a tetramer; the bacterial enzyme, a monomer. Although the monomers are of similar size (approximately 380 residues), the sequence similarity is low (approximately 15%). The viral enzyme contains at least eight disulfide bridges, conserved in all strains, and binds Ca2+, which enhances activity; the bacterial enzyme contains one disulfide and does not bind Ca2+. Comparison of the two structures shows a remarkable similarity both in the general fold and in the spatial arrangement of the catalytic residues. However, an rms fit of 3.1 A between 264 C alpha atoms of the S. typhimurium enzyme and those from an influenza A virus reflects some major differences in the fold. In common with the viral enzyme, the bacterial enzyme active site consists of an arginine triad, a hydrophobic pocket, and a key tyrosine and glutamic acid, but differences in the interactions with the O4 and glycerol groups of the inhibitor reflect differing kinetics and substrate preferences of the two enzymes. The repeating "Asp-box" motifs observed among the nonviral sialidase sequences occur at topologically equivalent positions on the outside of the structure. Implications of the structure for the catalytic mechanism, evolution, and secretion of the enzyme are discussed.

The sialidase superfamily and its spread by horizontal gene transfer

Sialidases (neuraminidases, EC 3.2.1.18) belong to a class of glycosyl hydrolases that release terminal N-acylneuraminate (sialic acid) residues from glycoproteins, glycolipids, and polysaccharides. These enzymes are common in animals of the deuterostomate lineage (Echinodermata through Mammalia) and also in diverse microorganisms that mostly exist as animal commensals or pathogens. Sialidases, and their sialyl substrates, appear to be absent from plants and most other metazoans. Even among bacteria, sialidase is found irregularly so that related species or even strains of one species differ in this property. This unusual phylogenetic distribution makes sialidases interesting for evolutionary studies. The biochemical diversity among bacterial sialidases does not indicate close relationships. However, at the molecular level, homologies are detectable, supporting the hypothesis of a common sialidase origin and thus of a sialidase superfamily. Some findings indicate that sialidase genes were recently transferred via phages among bacteria. The proposal of a sialidase origin in higher animals is suggested by the presence of apparently homologous enzymes in this kingdom, supporting the idea that some microbes may have acquired the genetic information during association with their animal hosts.

Complete nucleotide sequence of the bacteriophage K1F tail gene encoding endo-N-acylneuraminidase (endo-N) and comparison to an endo-N homolog in bacteriophage PK1E

Endo-N-acylneuraminidase (endo-N) is a phage-encoded depolymerase that degrades the alpha (2-8)-linked polysialic acid chains of K1 serotypes of Escherichia coli and vertebrate neural cell adhesion molecules. We have determined the DNA sequence of the bacteriophage K1F tail protein structural gene, which codes for a polypeptide of 920 residues. Purification of the tail protein yields a 102-kDa species upon denaturing gel electrophoresis and detection by Western immunoblot analysis. An identical polypeptide was detected by Western blot analysis of K1F virions. Peptide sequencing confirmed that the open reading frame determined by nucleotide sequencing encodes endo-N. Immunoelectron microscopy with neutralizing antibodies raised against the depolymerase confirmed that endo-N is a component of the K1F tail apparatus. Antibodies in the serum cross-reacted with endo-N from another K1-specific phage, PK1E, demonstrating the presence of shared epitopes. Homology between K1F and PK1E endo-N was confirmed by Southern, Northern (RNA), and Western blot analyses. The endo-N amino-terminal domain is homologous to the amino termini of phage T7 and T3 tail proteins, indicating by analogy that this domain functions in attachment of endo-N to the K1F virion's head. A central domain of 495 residues has weak similarity to sea urchin aryl sulfatase, suggesting that this region may contain the endo-N catalytic site. Failure to detect homology between the PK1E homolog and the carboxy-terminal domain of K1F endo-N is consistent with the central domain's involvement in binding and catalysis of polysialic acid. These results provide the initial molecular and genetic description of polysialic acid depolymerase, which has so far been detected only in K1-specific phage.

Isolation and properties of the natural and the recombinant sialidase from Clostridium septicum NC 0054714

The natural sialidase of Clostridium septicum was purified and characterized in parallel with the recombinant enzyme expressed by Escherichia coli. The two enzymes exhibit almost identical properties. The maximum hydrolytic activity was measured at 37 degrees C in 60 mM sodium acetate buffer, pH 5.3. Glycoproteins like fetuin and saponified bovine submandibular gland mucin, most of them having alpha(2-6) linked sialic acids, are preferred substrates, while sialic acids from gangliosides, sialyllactoses, or the alpha(2-8) linked sialic acid polymer (colominic acid) are hydrolysed at lower rates. alpha(2-3) Linkages are more rapidly hydrolysed than alpha(2-6) bonds of sialyllactoses. The cleavage rate is markedly reduced by O-acetylation of the sialic acid moiety. These properties are similar to those of other secreted clostridial sialidases. The enzyme exists in mono-, di- and trimeric forms, the monomer exhibiting a molecular mass of 125 kDa, which is close to the protein mass of 111 kDa deduced from the nucleotide sequence of the cloned gene.

An immunoassay for the rapid and specific detection of three sialidase-producing clostridia causing gas gangrene

A rapid and methodologically unusual diagnostic test was developed for the specific detection of Clostridium perfringens, C. septicum and C. sordellii, which cause clostridial myonecrosis. Sialidases (EC 3.2.1.18) secreted by these bacterial species were bound to polyclonal antibodies raised against the respective enzyme and immobilized onto microtiter plates. The activity of bound sialidase was determined with the fluorogenic substrate 4-methylumbelliferyl-alpha-D-N-acetylneuraminic acid. The assay permits the detection of a minimum sialidase activity of about 0.1-1 mU/ml of sample solution within 2 h. The sensitivity of the test was reduced by about three-fold for sialidase activities in samples containing 50% serum. Only a few, low cross-reactivities, which never exceeded 10% of the homologous reaction, were observed with 12 sialidases from other bacterial sources. Clinical isolates of the three clostridial species were analysed by this assay and gave positive signals in the homologous test. The assay for the detection of C. perfringens was applied to nine samples from patients suspected to be suffering from clostridial myonecrosis. There was a high correlation between the results of the immunoassay and the bacteriological analysis of infection.

Quantitative mapping of the N-linked sialyloligosaccharides of recombinant erythropoietin: combination of direct high-performance anion-exchange chromatography and 2-aminopyridine derivatization

A rapid quantitative analysis of the sialylated N-linked oligosaccharides of recombinant erythropoietin (EPO) expressed in Chinese hamster ovary (CHO) cells has been developed. The procedure utilizes a glycoamidase (glycopeptidase F) to release all of the N-linked oligosaccharides from the native glycoprotein, followed by direct chromatographic analysis using high-performance anion-exchange chromatography (HPAEC) with pulsed amperometric detection. The eight sialyloligosaccharides isolated from HPAEC were characterized by derivatizing with 2-aminopyridine followed by two-dimensional HPLC mapping of the pyridylaminated asialooligosaccharides (Tomiya et al., 1988, Anal. Biochem. 171, 73-90). Seven kinds of complex-type asialooligosaccharides were identified ranging from a biantennary structure to N-acetyllactosamine-extended tetraantennary structure. Approximately 3% of the terminal galactose residues of the oligosaccharides released from EPO were not sialylated whereas 97% contained an alpha(2-->3)-linked sialic acid. Quantitative oligosaccharide mapping of four different lots of EPO from CHO cells was performed to quantify the molar balance and distribution of the N-linked oligosaccharides. The sialyloligosaccharides were distributed with approximately 5% disialylated (single type), 20% trisialylated (six types), and 75% tetrasialylated (four types) oligosaccharides with an average molar recovery of 85% starting from 750 pmol of EPO.

Purification, crystallization and preliminary crystallographic study of neuraminidase from Vibrio cholerae and Salmonella typhimurium LT2

The nanH genes of Vibrio cholerae and Salmonella typhimurium LT2 coding neuraminidase were cloned separately in Escherichia coli, and the expression products purified. Single crystals of the V. cholerae neuraminidase were obtained using the hanging drop vapour diffusion method with polyethylene glycol as precipitant at pH 7.2. The crystals belong to the orthorhombic space group P2(1)2(1)2(1), with unit cell dimensions a = 71.9 A, b = 79.0 A, c = 165.7 A, and with one molecule in the asymmetric unit. Diffraction extends to at least 2.5 A. Single crystals of the S. typhimurium neuraminidase were obtained by hanging drop with potassium phosphate as precipitant at pH 7.2. The crystals also belong to the orthorhombic space group P2(1)2(1)2(1), with unit cell dimensions a = 47.4 A, b = 82.8 A, c = 92.4 A, and with one molecule in the asymmetric unit. Diffraction extends to at least 1.8 A.

Photolabelling of Salmonella typhimurium LT2 sialidase. Identification of a peptide with a predicted structural similarity to the active sites of influenza-virus sialidases

The sialidase from Salmonella typhimurium LT2 was characterized by using photoaffinity-labelling techniques. The well-known sialidase inhibitor 5-acetamido-2,6-anhydro-3,5-dideoxy-D-glycero-D-galacto-non- 2-enonic acid (Neu5Ac2en) was modified to contain an amino group at C-9, which permitted the incorporation of 4-azidosalicylic acid in amide linkage at this position. Labelling of the purified protein with the radioactive (125I) photoprobe was determined to be highly specific for a region within the active-site cavity. This conclusion was based on the observation that the competitive inhibitor Neu5Ac2en in the photolysis mixture prevented labelling of the protein. In contrast, compounds with structural and chemical features similar to the probe and Neu5Ac2en, but which were not competitive enzyme inhibitors, did not affect the photolabelling of the protein. The peptide interacting with the probe was identified by CNBr treatment of the labelled protein, followed by N-terminal sequence analysis. Inspection of the primary structure of the protein, predicted from the cloned structural gene for the sialidase [Hoyer, Hamilton, Steenbergen & Vimr (1992) Mol. Microbiol. 6, 873-884] revealed that the label was incorporated into a 9.6 kDa fragment situated within the terminal third of the molecule near the C-terminal end. Secondary-structural predictions using the Garnier-Robson algorithm [Garnier, Osguthorpe & Robson (1978) J. Mol. Biol. 120, 97-120] of the labelled peptide revealed a structural similarity to the active site of influenza-A- and Sendai-HN-virus sialidases with a repetitive series of alternating beta-sheets connected with loops.

Cloning, sequencing and distribution of the Salmonella typhimurium LT2 sialidase gene, nanH, provides evidence for interspecies gene transfer

The Salmonella typhimurium LT2 sialidase (neuraminidase, EC 3.2.1.18) structural gene, nanH, has been cloned and sialidase overproduced from multicopy plasmids in Escherichia coli. Sialidase expression was regulated positively by cAMP. In contrast, certain Tn1000 insertions located upstream of nanH coding sequences reduced sialidase activity. A nanH chromosomal insertion mutation constructed by marker exchange demonstrated a single sialidase gene copy in S. typhimurium LT2. The complete nucleotide sequence of nanH, encoding a 41,300 dalton polypeptide, was determined and the derived primary structure was similar to sialidases from Clostridium perfringens, Clostridium sordellii, Bacteroides fragilis, and Trypanosoma cruzi. Comparative sequence analysis, including codon usage and secondary structure predictions, indicated that the S. typhimurium and clostridial sialidases are homologous, strongly suggestive of an interspecies gene transfer event. At least two primary sequence motifs of the bacterial enzymes were detected in influenza A virus sialidases. The predicted secondary structure of the bacterial enzymes was strikingly similar to viral sialidase. From the population distribution of nanH detected within a collection of salmonellae, it was apparent that S. typhimurium obtained its nanH copy most recently from Salmonella arizonae. S. typhimurium LT2 is thus a genetic mosaic that differs from other strains of even the same serotype by nanH plus potentially additional characters linked to nanH. These results have relevance to the evolution and function of sialidases in pathogenic microbes, and to the origin of the sialic acids.

Functional analysis of the sialyltransferase complexes in Escherichia coli K1 and K92

The polysialyltransferase (polyST) structural gene, neuS, for poly alpha 2,8sialic acid (PSA) capsule synthesis in Escherichia coli K1 was previously mapped near the kps region 1 and 2 junction (S. M. Steenbergen and E. R. Vimr, Mol. Microbiol. 4:603-611, 1990). Present Southern and colony blot hybridization results confirmed that neuS was a region 2 locus and indicated apparent homology with neuS from E. coli K92, bacteria that synthesize a sialyl alpha 2,8-2,9-linked polymer. A K1- mutant with an insertion mutation in neuS was complemented in trans by K92 neuS, providing direct evidence that neuS encoded the PSA polymerase. A 2.9-kb E. coli K1 kps subclone was sequenced to better characterize polyST. In addition to neuS, the results identified a new open reading frame, designated neuE, the linker sequence between regions 1 and 2, and the last gene of region 1, kpsS. The kpsS translational reading frame was confirmed by sequencing across the junction of a kpsS'-lacZ+ fusion. PolyST was identified by maxicell analysis of nested deletions and coupled in vitro transcription-translation assays. PolyST's derived primary structure predicted a 47,500-Da basic polypeptide without extensive similarity to other known proteins. PolyST activity was increased 31-fold and was membrane localized when neuS was cloned into an inducible expression vector, suggesting, together with the polyST primary structure, that polyST is a peripheral inner membrane glycosyltransferase. However, polyST could not initiate de novo PSA synthesis, indicating a functional requirement for other kps gene products. The existence of a sialyltransferase distinct from polyST was suggested by identification of a potential polyprenyl-binding motif in a C-terminal membrane-spanning domain of the predicted neuE gene product. Direct evidence for a quantitatively minor sialyltransferase activity, which could function to initiate PSA synthesis, was obtained by phenotypic analysis of mutants with multiple defects in sialic acid synthesis, degradation, and polymerization. The results provide an initial molecular description of K1 and K92 sialyltransferase complexes and suggest a possible common function for accessory kps gene products.