Amino acid 310 determines the donor substrate specificity of serogroup W-135 and Y capsule polymerases of Neisseria meningitidis

Authors

  • Heike Claus,

    Corresponding author
    1. Institut für Hygiene und Mikrobiologie, Universität Würzburg, Josef-Schneider-Straße 2, 97080 Würzburg, Germany.
    Search for more papers by this author
    • These authors contribute equally to this work.

  • Katharina Stummeyer,

    1. Abteilung Zelluläre Chemie, Medizinische Hochschule Hannover, Carl-Neuberg-Straße 1, 30625 Hannover, Germany.
    Search for more papers by this author
    • These authors contribute equally to this work.

  • Julia Batzilla,

    1. Institut für Hygiene und Mikrobiologie, Universität Würzburg, Josef-Schneider-Straße 2, 97080 Würzburg, Germany.
    Search for more papers by this author
    • Present address: Max-von-Pettenkofer-Institut, Ludwig-Maximilians-Universität München, Pettenkoferstraße 9a, 80336 München, Germany.

  • Martina Mühlenhoff,

    1. Abteilung Zelluläre Chemie, Medizinische Hochschule Hannover, Carl-Neuberg-Straße 1, 30625 Hannover, Germany.
    Search for more papers by this author
  • Ulrich Vogel

    1. Institut für Hygiene und Mikrobiologie, Universität Würzburg, Josef-Schneider-Straße 2, 97080 Würzburg, Germany.
    Search for more papers by this author

*E-mail hclaus@hygiene.uni-wuerzburg.de; Tel. (+49) 931 201 46036; Fax (+49) 931 201 46445.

Summary

The capsular polysaccharides of serogroup W-135 and Y meningococci are sialic acid-containing heteropolymers, with either galactose or glucose as the second sugar residue. As shown previously, sequences of the predicted enzymes that catalyse capsule polymerization, i.e. SiaDW-135 and SiaDY, differ in only a few amino acids. By in vitro assays with purified recombinant proteins, SiaDW-135 and SiaDY were now confirmed to be the capsule polymerases harbouring both hexosyltransferase and sialyltransferase activity. In order to identify amino acids crucial for substrate specificity of the capsule polymerases, polymorphic sites were narrowed down by DNA sequence comparisons and subsequent site-directed mutagenesis. Serogroup-specific amino acids were restricted to the N-terminal part of the proteins. Exclusively amino acid 310, located within the nucleotide recognition domain of the enzymes' predicted hexosyltransferase moiety, accounted for substrate specificity as shown by immunochemistry and in vitro activity assay. Pro-310 determined galactosyltransferase activity that resulted in a serogroup W-135 capsule and Gly-310 determined glucosyltransferase activity that resulted in a serogroup Y capsule. In silico analysis revealed a similar amino acid-based association in other members of the same glycosyltransferase family irrespective of the bacterial species.

Introduction

Neisseria meningitidis is a causative agent of bacterial meningitis and septicaemia worldwide. The species comprises at least 12 different serogroups that are based on antigenic variation of the capsular polysaccharides (CPSs) (Frosch and Vogel, 2006). Five serogroups, A, B, C, W-135 and Y, account for most cases of disease (Rosenstein et al., 2001). The emergence of Hajj-related serogroup W-135 strains in Africa has provided more focus to this serogroup (Taha et al., 2000; Aguilera et al., 2002; Decosas and Koama, 2002; Lingappa et al., 2003; Boisier et al., 2005; von Gottberg et al., 2008). Furthermore, serogroup Y has recently developed a high prevalence in the USA, where it has increased from 2% during 1989–1991 to 37% during 1997–2002 (Rosenstein et al., 1999; Pollard and Scheifele, 2001; Bilukha and Rosenstein, 2005). Of note, the American Advisory Committee on Immunization Practices launched the recommendation of a meningococcal tetravalent polysaccharide conjugate vaccine including the serogroup Y component in 2005 (Bilukha and Rosenstein, 2005).

The CPSs of serogroup B, C, W-135 and Y meningococci contain N-acetyl neuraminic acid (Neu5Ac, sialic acid) as a structural unit. In serogroups B and C, the CPS is a Neu5Ac homopolymer with unique linkages, i.e. α-2→8 in serogroup B and α-2→9 in serogroup C (Bhattacharjee et al., 1975). Serogroup W-135 and Y CPSs are both heteropolymers. The serogroup W-135 CPS is composed of a galactose (Gal) Neu5Ac disaccharide repeating unit [→6)-α-d-Galp-(1→4)-α-Neu5Ac-(2→]n whereas the serogroup Y CPS is composed of a glucose (Glc) Neu5Ac disaccharide repeating unit [→6)-α-d-Glcp-(1→4)-α-Neu5Ac-(2→]n (Bhattacharjee et al., 1976). In all four serogroups the genes siaA–C encode the enzymes required for activated sialic acid synthesis whereas the serogroup-specific siaD, i.e. siaDB, siaDC, siaDW-135 and siaDY, encodes the transferase that catalyses polymerization of the CPS (Edwards et al., 1994; Claus et al., 1997; Swartley et al., 1997; Frosch and Vogel, 2006).

In 1997, we published the sequences of the genes that encode the serogroup W-135 and Y capsule polymerases, i.e. siaDW-135 and siaDY respectively (Claus et al., 1997). With more than 3 kb, the size of both genes was remarkable, since, in comparison, the polymerase genes of serogroups B and C comprise only approximately 1.5 kb. Based on the observed differences in gene length it is tempting to speculate that siaDW-135 and siaDY encode bi-functional enzymes that evolved by fusion of a hexosyl- and a sialyltransferase gene. However, the enzymes share no sequence similarity with known sialyltransferases and so far, no functional data have been reported. siaDW-135 and siaDY harbour only a few non-synonymous polymorphisms, highlighting their close relationship. This sequence information is used for molecular serogroup typing in medical microbiology which has become increasingly important because in the UK up to 47% of the serogroup assignments are obtained by culture-independent techniques (Borrow et al., 1998; Gray et al., 2006).

Since CPSs of serogroups W-135 and Y differ only in the presence of Gal and Glc, respectively, the molecular basis of substrate specificity of SiaDW-135 and SiaDY might resides in the hexosyltransferase domain of the capsule polymerases. The N-terminal parts of SiaDW-135 and SiaDY share sequence similarity with the cluster of orthologous group (COG) 0438-like glycosyltransferases (GTs). This COG comprises predicted GTs that are related to the α1,3-glucosyltransferase WaaG of Escherichia coli, an enzyme that is involved in the biosynthesis of the lipopolysaccharide (LPS) core structure (http://www.ncbi.nlm.nih.gov/COG/) (Yethon et al., 2000). GTs (EC 2.4.x.y), or more specifically, hexosyltransferases (EC 2.4.1), both in eukaryotes and in prokaryotes are involved in the synthesis of glycoconjugates, i.e. polysaccharides, glycoproteins and glycolipids. All bacterial GTs participate in the formation of exopolysaccharides, LPSs or CPSs. They catalyse the formation of glycosidic bonds by transferring activated sugar residues (i.e. UDP/TDP-hexoses) on acceptor molecules. GTs can be divided into subgroups of retaining and inverting enzymes with either retention of the anomeric configuration of the α-linked nucleotide sugar or inversion of the configuration. Due to the large number of different donor sugars and acceptor molecules GTs exhibit a high specificity and consequently enormous sequence variability. Nonetheless, Campbell et al. (1997) have developed a classification scheme based on the relative substrate/product stereochemistry and on sequence homologies which is known as the CAZy scheme (http://www.cazy.org/). To date, 91 GT families have been assigned. GTs within each family are presumed to be structurally and mechanistically similar. SiaDW-135 and SiaDY, the enzymes that catalyse the CPS polymerization of serogroup W-135 and Y meningococci, respectively, have been assigned to the GT4 family of the CAZy scheme which comprises retaining GTs.

In the present study, full-length DNA sequences of siaDW-135 and siaDY of representative meningococcal strains were obtained and analysed. Subsequent site-directed mutagenesis revealed that the amino acid at position 310 of SiaDW-135 and SiaDY determined the donor substrate specificity, for which immunochemical and functional evidence is provided.

Results

Sequence analysis and allelic variations of SiaDW-135 and SiaDY

SiaDW-135 and SiaDY catalyse the biosynthesis of the heteropolymeric CPSs of the meningococcal serogroups W-135 and Y, which are composed of alternating Gal/sialic acid and Glc/sialic acid moieties respectively. Both enzymes are large proteins (1037 aa) which share no significant homologies with the much smaller capsule polymerases of serogroups B (SiaDB, 495 aa) and C (SiaDC, 492 aa), which catalyse the biosynthesis of sialic acid homopolymers. The N-terminal part of SiaDW-135 and SiaDY is related to hexosyltransferases of the GT4 family, whereas the C-terminal part shares no significant sequence similarity to sialyltransferases or any other protein with known function.

To obtain a more detailed picture of existing allelic variations, DNA sequences of siaDW-135 and siaDY were obtained from 34 meningococcal isolates that had been previously characterized by multilocus sequence typing (MLST) (Claus et al., 2004). MLST is a tool to genetically type bacteria and fungi by DNA sequencing of the genes that encode neutral housekeeping enzymes. Related sequence types (STs) are grouped into clonal complexes (cc) (Maiden, 2006). The 34 strains analysed (18 serogroup W-135 and 16 serogroup Y strains) were assigned to 17 different STs with 15 of them belonging to eight cc: ST-11 cc (n = 6), ST-22 cc (n = 8), ST-23 cc (n = 7), ST-92 cc (n = 2), ST-167 cc (n = 2), ST-174 cc (n = 4), ST-334 cc (n = 1) and ST-41/44 cc (n = 1). The DNA sequences that encoded the capsule polymerases clearly divided into two distinct clusters representing siaDW-135 and siaDY. Only two alleles diverged from the siaDY cluster because they harboured four amino acids at positions 50, 80, 111 and 132 that were otherwise found exclusively in serogroup W-135 strains (Fig. 1A). However, there was limited allelic diversity in the genes that encoded the capsule polymerases with only 56 polymorphic sites (1.8%) among 3114 bp in contrast to 292 polymorphic sites (8.9%) spread over 3282 bp of the concatenated housekeeping gene fragments sequenced for MLST. Interestingly, the polymorphic sites were non-randomly distributed and predominantly found at the 5′ end of the gene whereas the 3′ end of the gene was extremely conserved.

Figure 1.

Amino acid polymorphisms of SiaDW-135 and SiaDY.
A. Representation of all amino acid polymorphisms identified in 18 serogroup W-135 and 16 serogroup Y meningococcal strains respectively. The positions of the polymorphic sites are shown above the amino acids in vertical format. The number of strains that harboured the different alleles is indicated. Amino acids that diverged from the consensus are shown in bold letters. Consistent amino acid differences between SiaDW-135 and SiaDY are underlined in the consensus sequences and the critical amino acid at position 310 identified in this study is shaded in grey.
B. Selection of target sites for site-directed mutagenesis.
C. A single spontaneous nucleic acid point mutation in siaDY resulted in an amino acid exchange (G310S) and subsequent expression of a serogroup with mixed serogroup W-135 and Y capsule composition (Tsang et al., 2008). The nucleic acid sequences of the EX7E motif of the known siaDW-135/Y variants were aligned with polymorphisms in comparison with siaDY shown in bold letters. Position 310 is boxed.

In total, 24 polymorphic sites were identified among the 1037 amino acids that characterized six SiaDW-135 and seven SiaDY alleles (Fig. 1A). Of these, 16 were found in the N-terminal and eight in the C-terminal part of the proteins. Only nine sites were consistently different between SiaDW-135 and SiaDY, all clustered between positions 305 and 343 within the putative N-terminal hexosyltransferase domain.

Specificity of SiaDW-135 and SiaDY for either galactose or glucose depends on a single amino acid

Although SiaDW-135 and SiaDy are almost identical proteins, they catalyse the synthesis of different CPSs. While serogroup W-135 CPS is composed of Gal and sialic acid, serogroup Y CPS is composed of Glc and sialic acid. The identification of nine consistently different sites between SiaDW-135 and SiaDY, all clustering between positions 305 and 343 (Fig. 1A), strongly suggested the implication of this region of the polymerase in Gal or Glc specificity. To further analyse the functional importance of the polymorphic region, five of the nine polymorphic sites were targeted by site-directed mutagenesis. Positions 305, 310, 327, 335 and 343 of SiaDY were individually mutated to the respective amino acid found in SiaDW-135 (Fig. 1B). Mutations were introduced into the siaDY gene located on a plasmid together with the gene's flanking sequences at the 5′ and 3′ ends, and a kanamycin-resistance cassette for antibiotic selection. This enabled the generation of mutant meningococci in which the siaDY gene was replaced by a mutated gene through homologous recombination. Capsule expression of the mutant strains was first monitored by slide agglutination using serogroup-specific monoclonal antibodies. As shown in Fig. 1B, mutation of the amino acids at positions 305, 327, 335 and 343 had no effect on the capsular phenotype. However, a single-point mutation at position 310 (G310P) resulted in a switch of the phenotype from serogroup Y to W-135. For a more detailed analysis of the phenotype switch, a quantitative whole-cell ELISA was performed. This assay also included the corresponding SiaDW-135 mutant, which carried the amino acid exchange P310G. As depicted in Fig. 2, glycine or proline substitutions at position 310 resulted in a switch of the capsular phenotype for both enzymes. G310P mutation of the serogroup Y strain resulted in a complete loss of reactivity with the serogroup Y-specific antibody, whereas the mutant reacted well with the serogroup W-135-specific antibody. Accordingly, the serogroup W-135 strain gained reactivity with the serogroup Y-specific antibody by a P310G mutation. In summary, the immunochemistry studies demonstrated that substitution of the amino acid located at position 310 of SiaDW-135 and SiaDy is sufficient to switch the capsular serogroup between W-135 and Y.

Figure 2.

Capsule expression in wild-type and mutant strains. Capsule expression was quantified by ELISA using monoclonal antibodies towards the capsular polysaccharides, i.e. mab1509 reacts with the serogroup W-135 CPS and mab1938 with the serogroup Y CPS respectively. Equal bacterial loading was controlled with antibodies directed against the major outer membrane protein PorA, i.e. mab P1.10 for the serogroup W-135 wild-type strain and its mutants and P1.2 for the serogroup Y wild-type strain and its mutants. Each value represents the average of four independent experiments with the standard deviation.

To confirm the immunochemistry results, we performed in vitro activity assays with purified recombinant SiaDW-135 and SiaDY enzymes. Wild-type and mutant proteins that carried an N-terminal StrepII- and a C-terminal His6-tag were expressed in E. coli BL21 (DE3) and isolated from crude lysates by immobilized metal ion affinity chromatography. The single-step purification yielded partially degraded but relatively pure protein as shown by SDS-PAGE and Western blot analysis (Fig. 3A). Full-length SiaDW-135 and SiaDy proteins migrated with the expected molecular mass of 120 kDa and were detected by anti-StrepII and anti-His6 immunostaining. In contrast, all additional protein bands migrated faster and were exclusively detected by the anti-His6 antibody, indicating that these bands corresponded to N-terminal degradation products. Initial trials to utilize the N-terminal StrepII-tag for purification failed, suggesting that the N-terminal tag is not accessible in the native protein.

Figure 3.

Purification and activity of recombinant wild-type and mutant SiaDW-135 and SiaDY.
A. Recombinant SiaDW-135 and SiaDY carrying an N-terminal StrepII and a C-terminal His6-tag were expressed in E. coli BL21 (DE3) and purified by immobilized metal ion affinity chromatography. Lysates (L) and purified proteins (P) were analysed by Coomassie-stained SDS-PAGE (top). Purified proteins were additionally analysed by Western blot directed against the N-terminal StrepII-tag (bottom left) and the C-terminal His6-tag (bottom right). Protein bands corresponding to full-length SiaDW-135 and SiaDY are indicated by an arrow.
B. Enzymatic activities of the purified enzymes were determined using a radiochemical assay in the presence of CMP-[14C]Neu5Ac. Reaction mixtures additionally contained UDP-hexose donor substrates (UDP-Gal or UDP-Glc) and acceptor substrates (purified CPS of either serogroup W-135 or serogroup Y) as indicated. Samples were separated by descending paper chromatography, analysed by scintillation counting (6800 dpm maximum signal, 35 dpm background) and specific activities for Neu5Ac-transfer were calculated. Each value represents the average of three independent determinations with the standard deviation indicated.

Enzymatic activities of purified wild-type and mutant enzymes were measured in a radiocarbon incorporation assay using CMP-[14C]Neu5Ac as a donor substrate (Weisgerber and Troy, 1990; Freiberger et al., 2007). As shown in Fig. 3B, enzymatic activity was detected for all tested enzyme variants when CMP-[14C]Neu5Ac was used in combination with either UDP-Gal or UDP-Glc and the respective CPS as an acceptor substrate. No significant incorporation was found in the absence of the CPS acceptor or when CMP-[14C]Neu5Ac was provided as the exclusive donor substrate. Similar results were obtained in parallel experiments, in which unlabelled CMP-Neu5Ac was used in combination with either UDP-[3H]-Gal or UDP-[3H]-Glc (data not shown). Together, these findings provided strong evidence that SiaDW-135 and SiaDY are bi-functional enzymes, which combine the catalytic capacity for hexosyl- and sialyltransfer. Both wild-type enzymes displayed a strong preference for incorporating the serogroup-specific hexose into the growing polysaccharide chain, which clearly demonstrated galactosyltransferase (Gal-T) activity for SiaDW-135 and glucosyltransferase (Glc-T) activity for SiaDY. Mutant forms of SiaDW-135 and SiaDY showed reduced activity, but in line with the serogroup switch observed in vivo, donor substrate preferences were altered by amino acid exchanges at position 310. SiaDY-G310P displayed a change of donor sugar preference from UDP-Glc to UDP-Gal whereas a change from UDP-Gal to UDP-Glc was detectable for the SiaDW-135-P310G mutant.

In summary, we showed that substrate specificity of the hexosyltransferase activity was mediated by a single amino acid located in the N-terminal part of SiaDW-135 and SiaDY. Thus, the sialyltransferase activity of these bi-functional enzymes might be located in the C-terminal portion. However, initial trials to dissociate the two domains failed since separate expression resulted exclusively in enzymatically inactive proteins (data not shown).

Identification of a previously unknown polymorphism with relevance for the donor sugar specificity of galactosyl- and glucosyltransferases

The amino acid sequences of SiaDW-135 and SiaDY were aligned to GT4 family Gal-Ts and Glc-Ts retrieved from the CAZy database (http://www.cazy.org/), for which functional data concerning their substrate specificities were available. Orthologous proteins of different species were included only when their amino acid sequences differed by more than 50%. In the obtained multiple sequence alignment, amino acid 310, which was found to determine the donor substrate specificity of SiaDW-135 and SiaDY towards Gal or Glc, respectively, resided between two conserved glutamic acid residues at position four of an EX7E motif. This motif is a central part of the nucleotide recognition domain NRD1α of retaining GTs. Position four of the EX7E motif was occupied by a glycine in 10 of 11 Glc-Ts, whereas a proline was found at this position in seven of eight Gal-Ts (Table 1). The association of glycine with Glc-T and proline with Gal-T activity in a variety of enzymes from several organisms supports the observations made herein by site-directed mutagenesis. Moreover, this finding suggested that the molecular basis for donor substrate specificity observed for SiaDY and SiaDW-135 applies to other Glc-Ts and Gal-Ts. Thus, an EX2GX4E motif might in general indicate Glc-T activity, whereas an EX2PX4E motif points towards Gal-T activity. As sole exception within the Gal-Ts, the α-1,6-galactosyltransferase EpsG of Lactobacillus helveticus NCC2745 (Accession No. CAC07464) harboured a serine at the relevant position (Table 1).

Table 1.  Retaining glycosyltransferases of the GT4 family with biochemically or immunologically proven function that exhibited a correlation between their donor sugar specificity and the amino acid within the EX7E motif that corresponded to position 310 of SiaDW-135 and SiaDyThumbnail image of

Interestingly, the serogroup W-135 meningococcal strain DE9555 submitted to the German national reference laboratory for meningococci in 2003 revealed a SiaDY consensus sequence that also carried a serine at position 310 (Fig. 1C). This finding raised the hypothesis that a SiaDY can be functionally converted from a Glc-T to a Gal-T not only by proline, but also by serine at position 310. To prove this point, the EcoRI fragment of the siaD gene of strain DE9555, which harboured all relevant polymorphic sites (Fig. S1), was integrated into a serogroup Y strain by homologous recombination. Subsequent analysis of capsule expression in the quantitative whole-cell ELISA indeed revealed a switch from serogroup Y to W-135 (Fig. 2). Purified recombinant SiaDY-G310S showed a drastically decreased activity compared with the wild-type enzymes; however, UDP-Gal was still slightly favoured over UDP-Glc (Fig. 3B).

Discussion

In the present study, we demonstrated that the two closely related meningococcal capsule polymerases SiaDW-135 and SiaDY are bi-functional enzymes. In an in vitro incorporation assay performed with purified recombinant SiaDW-135 and SiaDY, both enzymes showed hexosyl- and sialyltransferase activity, providing strong evidence that also in vivo polymerization of the W-135 and Y CPS depends exclusively on these enzymes. Moreover, we elucidated that the molecular basis for donor substrate specificity towards UDP-Gal or UDP-Glc relied on a single amino acid exchange at position 310. While a proline at this position was tightly associated with incorporation of Gal, a glycine altered the donor specificity towards UDP-Glc, resulting in the synthesis of Glc-containing heterosialopolymers.

Position 310 is located within the N-terminal domain of SiaDW-135 and SiaDY. Based on sequence similarities between this domain and other hexosyltransferases, both polymerases have been assigned to the GT4 family which comprises retaining GTs. The growing number of GT crystal structures allowed a further classification into structural superfamilies with members of the GT4 family falling into the GT-B superfamily. The GT-B fold is characterized by two similar Rossmann domains with the N-terminal domain providing the acceptor binding site and the C-terminal domain providing the donor binding site (Bourne and Henrissat, 2001; Coutinho et al., 2003). Based on sequence comparisons, a conserved domain has been identified in UDP/TDP-GTs which is responsible for the recognition of the donor sugar-nucleotide and is termed ‘nucleotide recognition domain’ (NRD) (Kapitonov and Yu, 1999). NRD1α in retaining GTs contains an EX7E motif whereas NRD1β in inverting GTs contains an R/HX7E motif. Interestingly, position 310 of SiaDW-135 and SiaDY is located at the fourth position within the EX7E motif. So far, the impact of proline or glycine at this position on substrate specificity of GT4 family members has not been addressed. However, our query of the CAZy database for Gal-Ts and Glc-Ts of the GT4 family suggests that specificity towards UDP-Gal and UDP-Glc is commonly associated with the presence of an EX2PX4E and an EX2GX4E motif respectively. Recently, the crystal structure of the GT4 family member WaaG has been determined in complex with the donor substrate analogue UDP-2-deoxy-2-fluoro glucose (UDP-2FGlc) (Martinez-Fleites et al., 2006). WaaG of E. coli is an α-1,3 glucosyltransferase that catalyses a key step in LPS synthesis and harbours an EX2GX4E motif. Analysis of the crystal structure revealed that G284, which corresponds to G310 in SiaDY, is involved in recognition of the hydroxyl group at the C4 carbon atom of Glc (Fig. S2). Since Gal and Glc exclusively differ in the configuration of this hydroxyl group, the structural data imply a key role of the glycine residue in determining substrate specificity. This strongly supports the findings of our study and provides fundamental evidence that the amino acid in position four of the EX7E motif confers substrate specificity towards UDP-Gal or UDP-Glc not only in SiaDW-135 and SiaDY, but also in other Gal-Ts and Glc-Ts of the GT4 family. In the crystal structure of WaaG, the first four amino acids of the EX7E motif are located on an active-site loop that is followed by an α-helical structure. Strikingly, G284 is positioned directly at the intersection between the two secondary structure elements. Glycine residues are known to confer conformational flexibility due to the lack of a side-chain. Substitution of glycine by proline at this particular position might favour binding of Gal by introducing steric hindrance or structural changes of the protein backbone that interfere with binding of the corresponding C4 epimer.

Consistent with our observation that proline and glycine at position 310 of SiaDW-135 and SiaDY, respectively, determined donor substrate specificity, we were able to convert SiaDW-135 into a glucosyl-/sialyltransferase by P310G exchange and SiaDY into a galactosyl-/sialyltransferase by G310P substitution. However, although the respective mutant enzymes showed the expected switch in donor substrate specificity in vivo and in vitro, the introduced amino acid exchange was accompanied by reduced enzymatic activity. This demonstrates that proline and glycine at position 310 are indeed crucial for defining specificity towards Gal and Glu, respectively, but that other amino acid residues contribute to efficient binding and/or transfer of the donor substrate. Thus, the introduction of a glycine into the amino acid context of SiaDW-135 is not equivalent with a naturally occurring SiaDY which in addition contains several other polymorphisms as highlighted in this study.

Noteworthy, we identified a meningococcal strain that harboured a SiaDY with a serine at position 310. This strain expressed a serogroup W-135 CPS, which indicated that a serine at position 310 conferred Gal-T activity. Artificial introduction of a G310S substitution into SiaDY of a serogroup Y strain confirmed this observation and resulted in a capsule switch from serogroup Y to W-135 in vivo. Among Gal-Ts and Glc-Ts of the GT4 family, a serine within the EX7E motif is a rare exception, suggesting that this residue is unfavourable for enzymatic activity. In line with this presumption, the variant SiaDY-G310S showed drastically reduced activity in vitro. Moreover, in contrast to the serogroup switch observed in vivo, UDP-Gal was only marginally preferred over UDP-Glc in vitro. This might be due to different reaction conditions in vivo and in vitro, e.g. different UDP-Glc and UDP-Gal concentrations, or result from interactions with other components of the capsule biosynthesis machinery that stabilize SiaDY-G310S in vivo.

During the preparation of this article Tsang et al. (2008) also reported on a meningococcal strain that reacted with an anti-serogroup W-135 antibody but harboured a SiaDY with three polymorphisms in comparison with SiaDW-135. Mutational analysis to identify the serine at position 310 as a critical amino acid for CPS composition was not reported. On the other hand, the authors analysed the monosaccharide composition of the CPS of this strain and identified sialic acid together with both Gal and Glc in a ratio of approximately 80%:20%. Furthermore, a polyclonal antiserum against the respective CPS was reactive with both serogroup W-135 and Y isolates. This finding is in agreement with our in vitro data obtained for SiaDY-G310S and indicates that a serine at position 310 does not allow strict discrimination between UDP-Gal and UDP-Glc. The strains analysed by Tsang et al. belong to ST-3923 and ST-23, whereas our isolate belongs to ST-3015, which indicates that they are not related to each other (Tsang et al., 2008). Thus, no expansion of meningococcal strains with this unusual CPS occurred. As stated by Tsang et al. meningococcal isolates with this rare CPS composition are supposed to be susceptible to immunity induced by the available polyvalent vaccines.

In the present study, DNA sequencing of siaDW-135 and siaDY of numerous meningococcal isolates was conducted. Only 1.8% (n = 56) of all sites of the capsule polymerase DNA sequences were polymorphic. Remarkably, polymorphic sites were predominantly found in the N-terminal hexosyltransferase domain, whereas the C-terminal part of the enzyme was highly conserved. One possible explanation for the limited genetic diversity observed in the 3′ part of the gene is that recently a gene adjacent to it has spread in the meningococcal population. Because of hitchhiking of the 3′ part, its genetic diversity has been reduced. This so-called selective sweep and its consequences for genetic diversity at defined sites of the chromosome has been proposed previously (Smith and Haigh, 1974). A candidate gene adjacent to both siaDW-135 and siaDY is the oatWY gene that encodes the capsular O-acetyltransferase of both CPSs (Claus et al., 2004). Of each ST represented in the strain collection reported here, one strain was selected for analysis of the genetic diversity of oatWY, and no polymorphism at all was detected (data not shown). Therefore, one may hypothesize that oatWY has recently swept through the meningococcal population, and has given rise to some kind of selective benefit, and thereby has reduced the genetic variability of adjacent sequences that are horizontally transferred along with oatWY. This hypothesis explains the uneven distribution of polymorphic sites observed in siaDW-135 and siaDY.

For medical microbiologists and molecular epidemiologists, culture-independent serogrouping by molecular tools has become common practice because meningococci frequently cannot be cultivated from blood or cerebrospinal fluid collected after administration of antibiotics. Serogroup determination is pivotal for laboratory surveillance of disease and national allocation of resources for vaccine development and distribution. With the data reported here, molecular serogroup assignment of W-135 and Y meningococci should be adapted and focused on the analysis of position 310 of SiaDW135/Y. In this context, it is important to carefully examine exceptions from the rule, such as the Ser-310 variant described here and most recently by Tsang et al. (2008). These isolates would have been identified as serogroup Y by applying the molecular genogrouping as described by Borrow et al. (1998). Such variants, albeit difficult to address in diagnostic procedures, are of basic biological interest since they increase our insight into structure–function relationships of these enzymes.

Experimental procedures

Bacterial strains and growth conditions

The serogroup W-135 and Y meningococcal wild-type strains 171 and 172 used in this study have been described previously (Claus et al., 2004). The unencapsulated strains were generated with plasmid pHC4siaD:TnMax5 as described previously (Ram et al., 2003). Meningococcal strains were grown at 37°C and 5% CO2 on GC agar supplemented with PolyViteX (bioMerieux), and, when appropriate, with kanamycin (100 μg ml−1).

Escherichia coli strain DH5α was used for propagating cloned DNA. E. coli strains were grown at 37°C on LB agar supplemented with ampicillin (100 μg ml−1) or kanamycin (30 μg ml−1) for plasmid selection.

DNA sequence analysis of the siaD genes

The siaD genes of 18 serogroup W-135 and 16 serogroup Y strains (Claus et al., 2004) were amplified by PCR with primer pairs UE12/HC65 and HC49/HC62 respectively (Table S1). The amplification with UE12/HC65 was performed under the following conditions: initial denaturation at 94°C for 10 min, 36 cycles of denaturation at 94°C for 1 min, annealing at 54°C for 1 min, and polymerization at 72°C for 3 min, final polymerization for 10 min. The amplification with HC49/HC62 was carried out with an annealing temperature of 50°C and a polymerization time of 1 min. Subsequently, the PCR products were sequenced by primer walking with oligonucleotides listed in Table S1 using dye terminator technology (BigDye® Terminator v1.1 Cycle Sequencing Kit, Applied Biosystems) and an ABI-Prism 377 autosequencer. DNA sequences were analysed with DNAstar (Lasergene) and submitted to the EMBL Nucleotide Sequence Database under the Accession Nos AM982801AM982812. DNA and protein sequences were aligned using clustal w. The neighbour-joining tree of the siaD gene sequences was generated using the Mega 3.1 software package available at http://www.megasoftware.net/ (Kumar et al., 2004).

Plasmid construction

Plasmids that harboured the mutated meningococcal DNA flanked by wild-type DNA and an antibiotic-resistance marker were used for transformation. The plasmid for allelic exchange of siaD between serogroup W-135 and Y meningococci was constructed as follows. A DNA fragment that comprised the 3′ end of siaC, the complete siaDY, the complete oatWY and parts of the IS1016-like transposase was excised from pHC5 (Claus et al., 1997) with the restriction enzymes SacI and NdeI (New England Biolabs) (Fig. S1). The 5′ overhang of the NdeI site was filled in to form blunt ends with T4 DNA polymerase. Subsequently, the DNA fragment was cloned into pBluescript II SK (+) (pBS) (Stratagene) restricted with SacI and EcoRV which resulted in pJP1. For selection in meningococci, the HincII-restricted kanamycin-resistance cassette of pUC4K (GE Healthcare) was inserted into the SpeI site filled in with T4 DNA polymerase of the oatWY gene (Claus et al., 2004) downstream of siaDY which resulted in pJP2 (Table S2). oatWY encodes a capsule O-acetyltransferase and is of no relevance for studying capsule expression. To generate a similar plasmid that harboured the hexosyltransferase domain of siaDW-135 instead of that of siaDY, the EcoRI fragment of pJP2 which comprised the 5′ region of siaDY (Fig. S1) was replaced by the EcoRI fragment of pHC4 (Claus et al., 1997) which comprised the 5′ region of siaDW-135 which resulted in pGH9 (Table S2). Following transformation in meningococci, the donor DNA and the resistance gene integrated into the chromosome via double-cross-over homologous recombination. Successful integration of the mutated DNA fragment was confirmed by DNA sequencing.

Site-directed mutagenesis of siaDW-135 and siaDY

Site-directed mutagenesis was performed by PCR using the QuikChangeTM site-directed mutagenesis kit (Stratagene). For mutagenesis of positions 305, 310, 327, 335 and 343 of SiaDY, the EcoRI fragment of pJP2 was cloned into pBS restricted with EcoRI, mutated and finally reintegrated into the pJP2ΔEcoRI remnants of pJP2 (Fig. S1). P310G mutagenesis of SiaDW-135 was performed with the siaDW-135EcoRI fragment of pHC4 cloned into pBS restricted with EcoRI. Subsequently, the mutated EcoRI fragment was integrated into the pJP2ΔEcoRI remnants of pJP2 which resulted in pGH8 (Table S2). Correct mutagenesis was confirmed by DNA sequencing of the entire EcoRI fragment. Finally, all plasmids were used for transformation of the respective meningococcal strains. Homologous recombination of the meningococcal DNA was verified by PCR analysis and subsequent DNA sequencing.

Phenotypic analysis of meningococcal mutants by ELISA

The meningococcal wild-type and mutant strains were analysed for capsule expression by ELISA, with mab1509 specific to serogroup W-135 and mab1938 specific to serogroup Y, as described previously (Vogel et al., 1998). For quantification of the capsule expression, the amount of bacteria was estimated by using antibodies against the meningococcal outer membrane protein PorA, i.e. P1.10 for the serogroup W-135 strain and P1.2 for the serogroup Y strain (National Institute for Biological Standards and Control, UK) (Poolman et al., 1995).

Cloning of SiaDW-135 and SiaDY expression plasmids

siaDW-135 and siaDY were amplified by PCR from plasmids pHC4 and pHC5 (Claus et al., 1997), respectively, using the oligonucleotides KS272 and KS273. The PCR products were ligated between the BamHI and XhoI sites of the expression vector pET22b-Strep derived from pET-22b (Novagen) (Schwarzer et al., 2007). The resulting constructs (pET22b-Strep-NmW135 and pET22b-Strep-NmY) carried an N-terminal Strep-tag II followed by a thrombin cleavage site and a C-terminal His6-tag. Expression plasmids for mutant enzymes were generated by ligating the 1.2 kb EcoRI fragments of plasmids pGH8, pJP2-G310P and pCM5 into the respective sites of pET22b-Strep-NmW135 or pET22b-Strep-NmY to yield pET22b-Strep-NmW135(P310G), pET22b-Strep-NmY(G310P) and pET22b-Strep-NmY(G310S) respectively (Table S2). The sequence identity of all constructs was confirmed by sequencing.

Expression and purification of recombinant CPS polymerases

Freshly transformed E. coli BL21 (DE3) was grown at 15°C and 225 r.p.m. in autoinducing ZYM-5052 medium that contained 100 μg ml−1 carbenicillin (Studier, 2005). Cells were harvested after 78 h (6000 g, 15 min, 4°C), washed once with PBS and stored at −20°C. Bacterial pellets from 250 ml of cultures were re-suspended in binding buffer (50 mM Tris pH 8.0, 300 mM NaCl) supplemented with protease inhibitors (40 mg ml−1 Bestatin, 1 μg ml−1 Pepstatin and 1 mM PMSF) to give a final volume of 15 ml. Cells were disrupted by sonication and samples were centrifuged (16 000 g; 30 min, 4°C). Lysates were filtered (Sartorius Minisart 0.8 μm) and recombinant proteins were bound to 1 ml HisTrap affinity columns (GE Healthcare). After washing with 10 column volumes of washing buffer (50 mM Tris, pH 8.0; 300 mM NaCl, 50 mM imidazole) bound proteins were eluted (50 mM Tris pH 8.0, 300 mM NaCl, 150 mM imidazole). Fractions containing the recombinant proteins were pooled and passed through a desalting column (HiPrep 26/10, GE Healthcare) equilibrated in buffer E (50 mM Tris pH 7.5, 300 mM NaCl, 2 mM DTT). Finally, proteins were concentrated to 2 mg ml−1 using Amicon Ultra centrifugal devices (Millipore), flash-frozen in liquid nitrogen and stored at −80°C.

SDS-PAGE and immunoblotting

SDS-PAGE was performed under reducing conditions using 2.5% (v/v) β-mercaptoethanol and 1.5% (w/v) SDS. For Western blot analysis samples and standard proteins were blotted onto PVDF membranes (Millipore). Proteins containing an N-terminal StrepII-tag were detected by Strep-Tactin alkaline phosphatase conjugate (Strep-Tactin® AP conjugate; IBA) according to the manufacturer's guidelines. His-tagged proteins were detected with 1 μg ml−1 penta-His antibody (Qiagen) followed by 50 ng ml−1 goat anti-mouse IR680 antibody (LI-COR) and imaged according to the recommendations of the Odyssey infrared imaging system (LI-COR).

In vitro activity assay

SiaDW-135 and SiaDY activities were analysed using a radiochemical assay based on a polysialyltransferase test described previously (Weisgerber and Troy, 1990; Freiberger et al., 2007). Purified recombinant proteins (30 μg) were assayed in reaction buffer (20 mM Tris/HCl pH 8.0, 10 mM MgCl2, 1 mM DTT) in the presence of 1 mM radiocarbon labelled CMP-[14C]Neu5Ac (0.13 mCi mmol−1, GE Healthcare), 2 mM of either UDP-Gal or UDP-Glc (Sigma) and 0.5 μg of CPS of either serogroup W-135 or serogroup Y [kindly provided by Pasteur Mérieux Connaught (now Sanofi Pasteur)] in a total volume of 37.5 μl. Samples were incubated at room temperature and enzymatic activity was determined at appropriate time intervals (0, 2, 4, 6 and 10 min) by mixing 5 μl of aliquots of the reaction solution with 5 μl of chilled ethanol (96%). Samples were spotted on Whatman 3MM CHR paper and the chromatographically immobile 14C-labelled reaction products were quantified by scintillation counting following descending paper chromatography in 96% ethanol/1 M ammonium acetate, pH 7.5 (7:3, v/v).

Acknowledgements

We thank Gabriele Heinze, Andrea Bethe and Christine Meinhardt for expert technical assistance. This work was supported by grants of the Deutsche Forschungsgemeinschaft to U.V. (VO 718-3-1 and VO 718-4-4) and to M.M. (MU 1774/2–1). Rita Gerardy-Schahn and Matthias Frosch are thanked for helpful discussions and continuous support. The Carbohydrate Active Enzymes database (http://www.cazy.org/) (Coutinho and Henrissat, 1999) was used to analyse the GT family of SiaDW-135/Y. This publication made use of the Neisseria Multi Locus Sequence Typing website (http://pubmlst.org/neisseria/) developed by Keith Jolley and Man-Suen Chan and sited at the University of Oxford (Jolley et al., 2004). Development of this site has been funded by the Wellcome Trust and European Union.

Ancillary