Cloning of the lipooligosaccharide alpha-2,3-sialyltransferase from the bacterial pathogens Neisseria meningitidis and Neisseria gonorrhoeae.

The genes encoding the α-2,3-sialyltransferases involved in lipooligosaccharide biosynthesis from Neisseria meningitidis and Neisseria gonorrhoeae have been cloned and expressed in Escherichia coli. A high sensitivity enzyme assay using a synthetic fluorescent glycosyltransferase acceptor and capillary electrophoresis was used to screen a genomic library of N. meningitidis MC58 L3 in a “divide and conquer” strategy. The gene, denoted lst, was found on a 2.0-kilobase fragment of DNA, and its sequence was determined and then used to design probes to amplify and subsequently clone the corresponding lst genes from N. meningitidis 406Y L3, N. meningitidis M982B L7, and N. gonorrhoeae F62. Functional sialyltransferase was produced from the genes derived from both L3 N. meningitidis strains and the N. gonorrhoeae F62. However, the N. meningitidis M982B L7 gene contained a frameshift mutation that renders it inactive. The expression of the lst gene was easily detected using the enzyme assay, and the protein expression could be detected when an immunodetection tag was added to the COOH-terminal end of the protein. Using the synthetic acceptor N-acetyllactosamine-aminophenyl-(6-(5-(fluorescein-carboxamido)-hexanoic acid amide), the α-2,3 specificity of the enzyme was confirmed by NMR examination of the reaction product. The enzyme could also use synthetic acceptors with lactose or galactose as the saccharide portion. This study is the first example of the cloning, expression, and examination of α-2,3-sialyltransferase activity from a bacterial source.


Mammalian oligosaccharides containing terminal N-acetyl-
neuraminic acid (Neu5Ac) 1 residues are recognized as biologically important carbohydrates for their function as receptors for lectins involved in cellular adhesion, as receptors for toxins, and for certain viruses (1). Some pathogenic bacteria have also been shown to carry sialylated oligosaccharides in their lipooligosaccharides (LOS), which are identical in structure to those found in mammalian glycolipids. Neisseria gonorrhoeae and Neisseria meningitidis LOS contain ␣-2,3-monosialylated lacto-N-neotetraose (2,3), and in Camplyobacter jejuni the structures are variants of a mono-, di-, or tri-sialylated ganglioside (4). The role of these sialylated oligosaccharides in the pathogenesis of N. gonorrhoeae has been clearly demonstrated (2), and although the precise role of similar sialylated LOS in the pathogenesis of N. meningitidis or C. jejuni is not known, it is presumed to be a form of molecular mimicry that aids in the evasion of the host immune response.
There have been extensive studies of the sialyltransferases involved in mammalian glycoconjugate synthesis, where at least eight different enzymes with an ␣-2,3-sialyltransferase activity have been examined either through protein purification or the cloning of the genes (1). In contrast, the bacterial sialyltransferases involved in the synthesis of ␣-2,3-sialylated lacto-N-neotetraose from N. gonorrhoeae or N. meningitidis and the enzyme(s) involved in the sialylation of lipopolysaccharide from C. jejuni have not been purified nor their genes cloned to date. There have been several reports of measurement of the sialyltransferase activity in Neisseria species (3,5), and the enzyme uses the same sugar nucleotide donor, CMP-Neu5Ac, as all the mammalian enzymes. They also appear to be solubilized by Triton X-100, which suggests a membrane association (5,6), and to be low abundance proteins that so far have eluded purification to homogeneity. Attempts have been made to clone these prokaryotic sialyltransferases using nucleic acid probes based on the mammalian "sialyl motif" or using complete mammalian genes (Ref. 3 and references within), but none of these attempts has revealed a sialyltransferase gene. Recently N. gonorrhoeae mutants defective in LOS sialylation have been described (7), but these have not so far enabled the structural gene for the LOS sialyltransferase to be identified. * Part of this work was performed in the laboratory of Dr. E. R. Moxon, which was in part supported by programme grants from the Medical Research Council (UK) and the Wellcome Trust and Action Research. This is National Research Council of Canada Publication Number 39527. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM  Here we describe the first cloning and characterization of a CMP-Neu5Ac:␤-galactoside ␣-2,3-sialyltransferase from the pathogens N. meningitidis and N. gonorrhoeae achieved by the use of a highly sensitive screening procedure based on the expression of enzyme activity. Cloning and Sequencing of the Sialyltransferase from N. meningitidis-The genomic library was prepared using 3-5-kb fragments from a HaeIII partial digest of the chromosomal DNA of N. meningitidis MC58 into ZAPII (Stratagene, La Jolla, CA) as the vector (8). The ZAPII library was plated at low density, and 3600-well isolated plaques were picked in pools of 100. Phage suspensions were made as described previously (9) and used to infect 1.5-ml cultures of E. coli XL1-Blue (in LB medium with 0.2% maltose, 10 mM MgSO 4 , and 2 mM isopropyl-1thio-␤-D-galactopyranoside), which were grown for 4.5 h. Toluene was added to 1%, and the cells were then assayed for sialyltransferase activity as described below. The positive pools were plated, and then plaques were picked in pools of five and analyzed again for activity. Positive pools of five were then used to isolate individual clones expressing sialyltransferase activity. Phagemids carrying the sialyltransferase gene were excised from the positive ZAPII clones using the ExAssist helper phage and the SOLR E. coli strain as described by the supplier Stratagene. The DNA sequence for the 2.0-kb insert of this first plasmid (pNST-01) was determined, and PCR primers based on this sequence were used to amplify the genes from DNA prepared from N. meningitidis 406Y, M982B, and N. gonorrhoeae F62. The primer sequences were 5Ј primer SIALM-5F, (nucleotides 540 -569 in pNST-01 insert sequence, NdeI site shown in bold italics) 43-mer: 5Ј-CTTAGGAGGT-CATATGTTCAATTTGTCGGAATGGAGTTTTAGG-3Ј, and 3Ј primer SIALM-16R, (nucleotides 1688 -1658 of pNST-01 insert sequence, SalI site shown in bold italics) 42-mer: 5Ј-CCTAGGTCGACTCATTAA TTTTTATCGTCAAATGTCAAAATC-3Ј.
Detection of the Sialyltransferase by Western Blotting-The gene product was detected in E. coli by first constructing a plasmid consisting of the lst open reading frame from pNST-01 and the peptide tag for immunodetection with anti-c-myc antibody as described previously (10). This construct was made using the following primers for PCR amplification, 5Ј end primer was the standard M13 "reverse" primer, and 3Ј end primer SIALM-18R (SalI site in italics, and the c-myc tag in bold): 5Ј-CCTAGGTCGACTCATTAGTTCAGGTCTTCTTCGCTGATCAG-TTTTTGTTCATTTTTATCGTCAAATGTCAAAATCGGG-3Ј 78-mer. The PCR product was cloned in the vector pT7-7 (11), and protein expression was then induced with isopropyl-1-thio-␤-D-galactopyranoside. Western blotting was performed as described previously (10).
Measurement of Sialyltransferase Activity-The sialyltransferase activity from N. meningitidis MC58 L3, 406Y L3, and M982B L7 and E. coli carrying pNST plasmids was measured in toluene-treated cells or cell-free extracts prepared as described previously (12). The sialyltransferase acceptors were derived from aminophenylglycosides reacted with 6(5-fluorescein-carboxamido)-hexanoic acid succimidyl ester (FCHASE) and were prepared as described previously (12). Reactions for the enzyme were performed at 37°C in 20 l of 50 mM MES buffer, pH 6.0, 10 mM MnCl 2 , with 0.2 or 1.0 mM labeled acceptor, 0.2 mM CMP-Neu5Ac donor, and various amounts of enzyme, from either crude bacterial extracts or extracts of recombinant E. coli with the cloned gene. The recombinant enzymes were assayed for 10 -120 min, whereas extracts from N. meningitidis were incubated 1-15 h. The reactions were terminated by diluting the reaction 1:100 with 10 mM NaOH. These sam-ples were then diluted appropriately in water prior to analysis by capillary electrophoresis.
Capillary electrophoresis (CE) was performed with a Beckman (Fullerton, CA) P/ACE 5510 equipped with a 3 mW Argon-ion laser-induced fluorescence detector ( excitation ϭ 488 nm; emission ϭ 520 nm). The capillary was bare silica 75 m ϫ 47 cm, with the detector at 40 cm. The capillary was conditioned before each run by washing with 0.2 M NaOH for 2 min, water for 2 min, and 25 mM sodium tetraborate, pH 9.4, for 2 min. Samples were introduced by pressure injection for 2-5 s, and the separation was performed at 15 kV, 75 A. Peak integration was performed with the Beckman System Gold (version 8.1) software.
For rapid detection of enzyme activity, samples from the transferase reaction mixtures were examined by thin layer chromatography on silica-60 TLC plates (Merck). A spot of 0.5-1.0 l from the reaction was air dried, and the plate was developed with ethyl acetate/methanol/ water/acetic acid (7:2:1:0.1). After drying, the acceptor and product spots could be seen by illumination of the plate with a 365 nm UV lamp.
The product R f under these conditions was 0.05.
Preparative Sialyltransferase Reactions-Preparative enzyme reactions were performed as coupled enzyme reactions with the cloned N. meningitidis CMP-Neu5Ac synthetase. 2 The reactions contained 25 mM HEPES, pH 7.5, 0.2 mM dithiothreitol, and 10 mM MgCl 2 , 400 milliunits/ml of CMP-Neu5Ac synthetase, 300 milliunits/ml inorganic pyrophosphatase (Sigma), 1.5 mM CTP, 1.5 mM Neu5Ac, and 50 milliunits of sialyltransferase (based on FCHASE-aminophenyl-LacNAc as the acceptor). The acceptor, FCHASE-aminophenyl-Lac or FCHASE-aminophenyl-LacNAc, was dried down in the tube under vacuum, and the reagents were then added to the tube; the concentration of FCHASEaminophenylglycoside in these reactions was 1 mM. These reactions were performed at 30°C for 3-5 h. After the reaction, the FCHASEaminophenylglycoside was bound to a Sep-Pak C18 reverse phase cartridge (Waters, Milford, MA), desalted by washing with water, and then eluted in 50% acetonitrile.
Determination of the Linkage Specificity of the Sialyltransferase-The product from a preparative sialyltransferase reaction was examined by NMR. Samples for NMR were prepared by the TLC method and were then freeze dried from D 2 O three times prior to collection of the spectra. Prior to lyophilization and exchange with D 2 O, the pH of the sample was adjusted to 7. NMR data collection was performed with a Bruker AMX 600 spectrometer. Spectra were recorded at 300 K in 5-mm tubes at a concentration of 0.5 mg of FCHASE-aminophenylglycoside in 0.6 ml of D 2 O. Chemical shifts in ppm are relative to the methyl resonance of acetone set at 2.225 ppm for 1 H and 31.07 ppm for 13 C. All NMR experiments and spectral analysis were performed as described previously (13).

RESULTS
Detection and Characterization of ␣-2,3-Sialyltransferase Activity from N. meningitidis-The initial part of this work was performed with the N. meningitidis strain 406Y L3, which possesses an LOS identical to that of strain MC58 but has a different capsular type. Both of these strains elaborate the L3 immunotype LOS, which consists of a lacto-N-neotetraose branch with an ␣-2,3-sialic acid on the terminal galactose residue (13). Both of these strains produced easily detectable levels of ␣-2,3-sialyltransferase when using as little as a single colony (10 7 cells) with the CE-based assay. Crude extract from N. meningitidis 406Y L3 was used to prepare material for determination of the linkage of the sialoside being synthesized and the enzyme was verified by NMR of its product to be a ␤-galactoside ␣-2,3-sialyltransferase. By NMR, the complete 1 H assignment of the compounds was performed. It was found that the 1 H chemical shifts (Table I) were similar to those of reported structures containing ␣-NeuAc-(2-3)-Gal (13). Also an NOE across the glycosidic bond H3 ax -sialic acid to H3-Gal confirmed that the ␤-galactoside ␣-NeuAc-(2-3)-Gal linkage was present in the product (Fig. 1).
Variation of the reaction conditions showed that the enzyme had a pH optimum of 6.0 and that the activity was stimulated 2-fold by the addition of either 10 mM MgCl 2 or 3-fold by 10 mM MnCl 2 . However, there are no stringent metal requirements because it was active in the presence of 5 mM EDTA. These same conditions were also optimal for the enzyme from crude extracts of MC58, 406Y, and for the recombinant enzymes from MC58. The natural enzyme was mostly associated with the cell membrane fraction (86% in the cell membrane pellet after centrifugation at 100,000 ϫ g). However, no detergent was required for activity, and in fact many common detergents tested inhibited the enzyme, with the exception of Triton X-100 up to 0.2%. Using this method no activity could be detected in M982B L7 cells.
Cloning and Sequencing of the Sialyltransferase Gene from N. meningitidis MC58 -Using the CE laser-induced fluorescence assay, we observed sialyltransferase activity one time out of five when we infected a 1.5-ml isopropyl-1-thio-␤-D-galactopyranoside-induced E. coli XL1-Blue culture with 1000 pfu from the N. meningitidis MC58 genomic library in ZAPII (Fig.  2). Formation of the product peak in the electropherogram required the addition of CMP-Neu5Ac, and it migrated the same as the sialidase-sensitive product peak formed by the natural enzyme. The peak in the CE electropherogram corresponds to 20 attomoles (2 ϫ 10 Ϫ17 mol) of product. Single clones expressing the sialyltransferase were obtained by a "divide and conquer" strategy sequentially screening pools of 100 pfu from the ZAPII library of MC58, pools of 5 pfu derived from the first positive pool, and finally individual plaques plated at low density. The initial screening yielded two positive pools of 100 pfu out of 36. From one of these pools we screened 60 pools of 5 pfu and obtained three positive pools. From the positive pools of 5 pfu we obtained many individual positive clones and the pBluescript SK Ϫ phagemids excised from them were found to carry a 2.0-kb insert.
The 2.0-kb insert was sequenced on both strands (GenBank accession number U60660) and a BLASTX search was performed in GenBank in order to identify any homology with previously sequenced genes. This analysis revealed two partial open reading frames (nucleotides 1-140 and 1853-2039), located at the opposite ends of the 2.0-kb insert, that were clearly homologous with various bacterial isocitrate dehydrogenases (60 -85% identity) and various bacterial cytochrome cЈ proteins (43-63% identity), respectively. A third open reading frame (nucleotides 573-1685) was designated lst (lipooligosaccharide sialyltransferase) and revealed significant homology to a Haemophilus influenzae gene designated lsg-ORF2 (GenBank accession number M94855). Pair-wise alignment between the translation products of lst and lsg-ORF2 indicated that their amino acid sequences share 29.3% identity and 56.3% similarity (Fig. 3).
The lst gene has two potential start codons, and the second of these is more likely to be used because the amino acid sequence immediately following this start codon appears to be a non- Two electropherograms are superimposed here to illustrate the level of sialyltransferase activity found in a 1.5-ml E. coli culture infected with 1000 pfu from the genomic DNA bank of N. meningitidis MC58 in ZAPII. The thin line is from a run where the reaction contained no CMP-Neu5Ac donor, and the thick line is from a run containing the CMP-Neu5Ac donor. The peak at 6.6 min was shown to comigrate with FCHASE-␣-2,3-sialyl-N-acetyllactosamine. The detector response is expressed in relative fluorescence units (RFU).
Comparison of Sialyltransferase Genes from Different N. meningitidis Strains and N. gonorrhoeae-Isolation of the genes from N. meningitidis 406Y L3 (GenBank accession number U60661), M982B L7 (GenBank accession number U60663), and N. gonorrhoeae F62 (GenBank accession number U60664) was accomplished with PCR primers based on the gene from MC58 L3 (GenBank accession U60660). We found 12 base differences, which results in five amino acid differences between the two genes from the L3 immunotype strains (Fig.  4), 19 differences in the gene from M982B L7 compared with MC58, and 12 differences in the M982B L7 sequence compared with that of 406Y L3. The gene from M982B L7 contains a frameshift mutation at nucleotide 454 and consequently would encode a truncated protein of only 151 amino acids (Fig. 4).
The gene from N. gonorrhoeae F62 shows 63 nucleotide differences compared with the N. meningitidis MC58, 62 nucleotide differences compared with the 406Y L3 gene, and 66 nucleotide differences compared with the M982B L7 gene. These differences in the DNA sequence of the N. gonorrhoeae F62 gene result in 16 and 17 amino acid differences in the protein when compared with the MC58 L3 and 406Y L3, respectively (Fig. 4).
Expression of the Sialyltransferase Gene-We could easily detect enzyme activity in E. coli carrying pNST plasmids, and this expression of the lst gene depended on the vector derived lac promoter because there was no detectable enzyme activity when the gene's orientation was inverted. There was at least 30-fold more enzyme activity from the pNST-01 containing clones compared with N. meningitidis L3 strains. However, the expression of the lst gene was not high enough to permit simple detection of an overexpressed protein by SDS-PAGE analysis. A plasmid was therefore constructed to introduce a c-myc immunodetection peptide tag at the COOH-terminal end of the protein. When this plasmid was used to express the lst gene, we could detect an immunoreactive protein with an M r of 41,000 (Fig. 5), which is slightly shorter than the predicted size of the lst gene product including the c-myc peptide tag (expected M r 44,000).
Acceptor Specificity of the lst Sialyltransferase-The natural acceptor for this enzyme is a terminal N-acetyllactosamine sequence on the lacto-N-neotetraose branch of the LOS from L3 immunotypes. We found that the enzyme would use as acceptors the synthetic saccharides: FCHASE-aminophenyl-Lac-NAc, FCHASE-aminophenyl-Lac, and FCHASE-aminophenyl-Gal. To assess the relative acceptor specificities we assayed for activity using two different substrate concentrations and compared the specific activites of both the native MC58 sialyltransferase and the recombinant form (pNST-01). Comparison of the specific activity reveals a strong preference for the LacNAc containing acceptor (Table II). DISCUSSION The availability of a highly sensitive enzyme assay was instrumental in successful screening of clones expressing the N. meningitidis ␣-2,3-sialyltransferase. The assay uses a glycosyltransferase acceptor, which is easy to synthesize, and does not require specially constructed CE equipment as has been previously described for the ultrasensitive detection of glycosyltransferase reaction products (15). The acceptors used in this study were made from widely available glycosides, and fluorophores and the CE equipment used was commercially available. We were able to reliably detect attomole (10 Ϫ18 mol) quantities of reaction products, which was more than adequate for screening for ␣-2,3-sialyltransferase expression.
The lst gene from MC58 L3 occurs between two genes unrelated to LOS synthesis, isocitrate dehydrogenase and cytochrome c', and is not part of a LOS synthesis operon unlike other N. meningitidis LOS glycosyltransferases (8). This is similar to the situation with the E. coli and N. meningitidis ␣-2,8-polysialyltransferase involved in capsule biosynthesis, although these genes are adjacent to the CMP-Neu5Ac synthetase (16). It is interesting to speculate that the lst gene is found on its own as the result of a transposition event, although we have no evidence for insertion elements or transposon-like sequences flanking the gene. Sequence analysis and data base comparisons showed this gene to be distinct from both the mammalian ␣-2,3-sialyltransferase family, the bacterial ␣-2,8sialyltransferase family, and the bacterial 3-deoxy-␣-D-mannooctulosonic acid transferases, which transfer a related sugar also from a CMP donor. The lst gene product was, however, shown to be similar to the lsg-ORF2 gene product from H. influenzae. Although lsg-ORF2 has been demonstrated to be involved in LOS biosynthesis, it may not encode an ␣-2,3sialyltransferase because it was reported to be involved in the expression of a Gal-GlcNAc LOS epitope (17). For the cloning of the N. gonorrhoeae gene, the F62 strain was used because it has been studied in relation to the role LOS sialylation plays in pathogenesis and because many LOS glycosyltransferases have been shown to be common to both species (8). An examination of the gene derived from N. gonorrhoeae F62 shows only a small number of differences, which is similar to other LOS biosynthesis gene comparisons from N. meningitidis and N. gonorrhoeae (8).
We have observed that the activity from N. meningitidis extracts is associated with the membrane fraction. The protein encoded by the lst gene appears to have an uncleavable signal peptide, and computer aided prediction programs suggest that the sialyltransferase is an integral inner membrane protein (14). However, the original papers describing the sialyltransferase activity from both N. meningitidis and N. gonorrhoeae suggest that the sialyltransferase would be an outer membrane protein on the basis that the enzyme activity is extracted from whole cells by Triton X-100 (5, 6). We have not yet fully purified the recombinant product, but the predicted size of the immunotagged protein is slighty larger than that we observed by SDS-PAGE. The difference between the observed and the expected M r is less than 7% and is within the accuracy of SDS-PAGE. However, the possibility exists that the recombinant protein is truncated at the amino-terminal end, which would result in the loss of the predicted noncleavable leader sequence. Experiments are in progress to determine if the lst protein is expressed intact in E. coli.
That this gene functions in the sialylation of LOS is inferred from an examination of N. meningitidis M982B L7, which appears to be a natural sialyltransferase mutant. The sialyltransferase gene derived from this L7 strain contains a frameshift mutation at nucleotide 454 that renders it inactive in the recombinant plasmid carrying it, which agrees with our observation that sialyltransferase activity cannot be detected in M982B cells. This frameshift is a deletion of a T residue, which is different from the G-tract frameshift mutations observed in the phase variable lgt genes from N. meningitidis and N. gonorrhoeae (8). This strain produces the same lacto-N-neotetraose as the L3 strains do but does not sialylate its LOS. 3 The acceptor specificity for the L3 enzyme with synthetic acceptors shows a strong preference for N-acetyllactosamine over lactose or galactose (Table II). Also the product of the reaction using enzyme from N. meningitidis and FCHASE-LacNAc acceptor was unequivocally determined by NMR to be FCHASE-␣-2,3sialyl-N-acetyllactosamine.
The expression level of the recombinant gene is being optimized, but it should be pointed out that the level of enzyme activity we have produced is approximately 50 units/liter of culture, based on assays with the FCHASE-LacNAc acceptor. Our expression levels are as high as those reported for mammalian sialyltransferases being overexpressed in insect cell cultures (18). We anticipate that optimization of expression of the lst gene in E. coli will yield substantially more enzyme activity than is produced with the mammalian gene constructs. This will be an important improvement for large scale chemienzymatic synthesis of ␣-2,3-sialylated oligosaccharides.
We have shown conclusively that we have cloned the ␣-2,3sialyltransferase gene from the important mucosal pathogens N. meningitidis and N. gonorrhoeae. The availability of this gene will enable defined mutants to be constructed in both species in order to determine the role that LOS sialylation plays in the pathogenesis of these organisms, and the regulation of this important virulence factor can now be studied. The availability of large amounts of an ␣-2,3-sialyltransferase for enzymatic synthesis will help elucidate the role of this important modification of many oligosaccharide structures.

TABLE II
Comparison of acceptor specificities of N. meningitidis ␣-2,3-sialyltransferase Two substrate concentrations were used to assay enzyme activity from both sources. The relative specific activities were calculated from the percent conversion of the substrate to product using the CE assay. The assays were performed so that similar percent conversions are compared for both sources of enzyme.