Disrupted Synthesis of a Di-N-acetylated Sugar Perturbs Mature Glycoform Structure and Microheterogeneity in the O-Linked Protein Glycosylation System of Neisseria elongata subsp. glycolytica

ABSTRACT

The genus Neisseria includes three major species of importance to human health and disease (Neisseria gonorrhoeae, Neisseria meningitidis, and Neisseria lactamica) that express broad-spectrum O-linked protein glycosylation (Pgl) systems. The potential for related Pgl systems in other species in the genus, however, remains to be determined. Using a strain of Neisseria elongata subsp. glycolytica, a unique tetrasaccharide glycoform consisting of di-N-acetylbacillosamine and glucose as the first two sugars followed by a rare sugar whose mass spectrometric fragmentation profile was most consistent with di-N-acetyl hexuronic acid and a N-acetylhexosamine at the nonreducing end has been identified. Based on established mechanisms for UDP-di-N-acetyl hexuronic acid biosynthesis found in other microbes, we searched for genes encoding related pathway components in the N. elongata subsp. glycolytica genome. Here, we detail the identification of such genes and the ensuing glycosylation phenotypes engendered by their inactivation. While the findings extend the conservative nature of microbial UDP-di-N-acetyl hexuronic acid biosynthesis, mutant glycosylation phenotypes reveal unique, relaxed specificities of the glycosyltransferases and oligosaccharyltransferases to incorporate pathway intermediate UDP-sugars into mature glycoforms.

IMPORTANCE Broad-spectrum protein glycosylation (Pgl) systems are well recognized in bacteria and archaea. Knowledge of how these systems relate structurally, biochemically, and evolutionarily to one another and to others associated with microbial surface glycoconjugate expression is still incomplete. Here, we detail reverse genetic efforts toward characterization of protein glycosylation mutants of N. elongata subsp. glycolytica that define the biosynthesis of a conserved but relatively rare UDP-sugar precursor. The results show both a significant degree of intra- and transkingdom conservation in the utilization of UDP-di-N-acetyl-glucuronic acid and singular properties related to the relaxed specificities of the N. elongata subsp. glycolytica system

INTRODUCTION

Glycoconjugates varying in function and structure are common on surfaces of bacterial cells. Given their prevalence and significance, it is interesting to understand the forces driving glycan diversification. Defining the selective pressures is difficult because carbohydrates are secondary gene products and the biosynthetic pathways and the glycans themselves are not always completely characterized. Systems involving lipopolysaccharides, capsules, and cell wall-associated moieties show robust inter- or intrastrain glycan variation, suggesting that such diversification is adaptive. However, strict correlations between glycosylation genotypes and phenotypes have been difficult to establish. In the same vein, reconciling the glycosylation gene repertoire with particular bacterial populations is challenging. Thus, there remain significant knowledge gaps relating to both the short- and the long-term evolutionary trajectories in these systems.

Bacterial protein glycosylation systems in which multiple, diverse proteins are targeted for modification are continually being identified. In particular, studies have strived to elucidate how these systems are evolving at the species and genus levels. In the case of N-linked protein glycosylation exhibited by species within the genus Campylobacter, there is conservation of a basic oligosaccharide glycoform within species of the thermotolerant taxa (1). In contrast, an unexpected diversity of N-glycans varying in length and structure was seen in nonthermotolerant Campylobacter species (1). In Acinetobacter, multiple distinct glycoforms associated with general O-linked protein glycosylation have been found (2), and a study of O-linked protein glycosylation in Bacteroidetes revealed that while core glycan structures are conserved across the phylum, terminal glycans are clearly divergent (3).

O-linked protein glycosylation (Pgl) systems expressed by the important human bacterial pathogens Neisseria gonorrhoeae and Neisseria meningitidis have been well characterized at both the genetic and biochemical levels (Fig. 1). From these studies, a consensus model for neisserial pgl-dependent protein glycosylation has been identified (4–8). A NAD⁺-dependent dehydratase (PglC) and aminotransferase (PglD) convert UDP-GlcNAc to UDP-2-acetamido-4-amino-2,4,6-trideoxy-α-d-glucose (8). The bifunctional enzyme (PglB) then catalyzes the amino acetylation of the above-mentioned precursor to form UDP-di-N-acetylbacillosamine (diNAcBac) and the subsequent transfer of the phosphosugar to the lipid carrier undecaprenyl phosphate (Und-P) (8). PglB2, encoded by pglB2 alleles found in some N. meningitidis strains, contains an altered C-terminal domain involved in the transfer of a glycerol moiety (instead of the acetyl group) to produce 4-glyceramido-2-acetamido-2,4,6-trideoxy-αd-hexose (GATDH) (9). Subsequent elaboration of these undecaprenyl diphosphate (Und-PP) monosaccharide sugars occurs via two parallel pathways using distinct glycosyltransferases. One involves PglH or its allelic variant-encoded PglH2, which attaches a Glc or GlcNAc, respectively, to the Und-PP-monosaccharides. The second pathway utilizes the PglA and PglE glycosyltransferases to add successive Gal units to produce a trisaccharide (4, 10). The PglH/PglH2-generated Und-PP-disaccharides are recalcitrant to further extension by PglE (5, 6). However, strains simultaneously expressing both PglA and PglH can generate both glycoforms, and all di- and trisaccharide forms can be further modified via O-acetylation mediated by PglI (4, 5, 11).

FIG 1.

Comparison of pgl gene content and synteny between N. gonorrhoeae strain FA1090 and N. elongata subsp. glycolytica strain ATCC 29315. The core pgl gene locus of N. elongata subsp. glycolytica contains pglG, pglH, a split, two-ORF version of pglB, pglC, pglJ, and pglK. pglD, pglO, and the linked pglN and pglM alleles are located separately outside the core locus. Gonococcal genomes lack orthologues of pglJ, -K, -N, or -M but do have pglA and pglE (which are absent in N. elongata subsp. glycolytica strain ATCC 29315).

O-linked protein glycosylation is not a classical virulence determinant as it is not differentially distributed among disease-associated and carriage strains of N. meningitidis and is found in nonpathogenic species such as Neisseria lactamica and Neisseria elongata subsp. glycolytica (7). Despite this, the glycoform structure appears to be under positive, diversifying selection in both N. gonorrhoeae and N. meningitidis, and major shifts in pgl genotypes have been documented within otherwise stable clonal complexes due to horizontal gene transfer (12, 13).

We sought to gain a broader understanding of pgl biology by examining a strain of N. elongata subsp. glycolytica and identified an unrecognized tetrasaccharide glycoform consisting of di-N-acetylbacillosamine, glucose, di-N-acetylhexuronic acid, and N-acetylhexosamine [diNAcBac-Glc- HexNAc(3NAc)A-HexNAc] (14). In that background, the pglBCD-pglH pathway generates a disaccharide, diNAcBac-Glc, and is extended by the product of pglG, a gene found in the pgl loci of most pathogenic Neisseria spp., to form diNAcBac-Glc-HexNAc(3NAc)A (6, 7) (Fig. 1). The glycosyltransferase responsible for the addition of the terminal HexNAc in the N. elongata subsp. glycolytica glycoform has yet to be identified.

Glycan serotyping utilizing previously established antibodies and mass spectrometric (MS) data identified the glycan components of the tetrasaccharide with the exception of that defined as diNAcHexA, for which only MS-based fragmentation data were available (14). Conserved pathways for HexNAc(3NAc)A biosynthesis have been defined based on data collected from mutagenesis and biochemical experiments in Pseudomonas aeruginosa and Bordetella pertussis, where it is found in the lipopolysaccharides (LPS) of some strains (15), and in archaeal Methanococcus species, where it is a component of the N-linked protein glycosylation glycan (16). Studies in P. aeruginosa PAO1 have shown that GlcNAc(3NAc)A is synthesized by four enzymes in the Wbp pathway, sequentially using WbpA, WbpB, WbpE, and WbpD, starting from the precursor UDP-GlcNAc (Fig. 2) (15). WbpA oxidizes the C-6 hydroxyl group to form UDP-GlcNAcA, followed by the action of the dehydrogenase WbpB and aminotransferase WbpE to yield UDP-GlcNAc(3NH₂)A. WbpB and WbpE act as a dehydrogenase/aminotransferase pair converting UDP-GlcNAcA to UDP-GlcNAc(3NH₂)A in a coupled reaction via a NAD(+) recycling pathway that uses α-ketoglutarate as an oxidant. Next, the acetyltransferase WbpD catalyzes acetylation of UDP-GlcNAc(3NH₂)A to yield UDP-GlcNAc(3NAc)A (Fig. 2). To confirm the identity of the third sugar moiety of the N. elongata subsp. glycolytica oligosaccharide, we identified genes orthologous to those required for HexNAc(3NAc)A synthesis in these other microbes. Here, we report the isolation of mutants with defined mutations in the N. elongata subsp. glycolytica genes and the ensuing effects on pgl glycoform expression.

FIG 2.

Proposed biosynthetic pathway of UDP-GlcNAc(3NAc)A in N. elongata subsp. glycolytica. The N. elongata subsp. glycolytica strain ATCC 29315 genome contains four ORFs whose corresponding products could function in UDP-GlcNAc(3NAc)A biosynthesis. The experimentally defined P. aeruginosa WbpABED pathway components are shown in gray (15). Identity percentages are shown in parentheses. N. elongata subsp. glycolytica Pgl homologue designations: Nelon10580, PglJ; Nelon10585, PglK; Nelon05890, PglM; and Nelon05885, PglN.

RESULTS

Identification of genes involved in UDP-HexNAc(3NAc)A biosynthesis.

Using the P. aeruginosa PAO1 Wbp proteins in queries of the annotated N. elongata subsp. glycolytica genome sequence, significantly related open reading frames (ORFs) were found for the first four enzymes in the pathway (Fig. 2). NELON_10580 shared 76% sequence identity to WbpA, while NELON_10585 shared 20% sequence identity to WbpB. The two associated genes were tandemly arrayed just 3′ of pglC (encoding an aminotransferase essential for the synthesis of UDP-N,N'-diacetylbacillosamine). NELON_05890 and NELON_05885 shared 75% and 76% similarities to WbpE and WbpD, respectively, and their genes were also oriented tandemly at a locus far removed from the so-called core pgl locus (Fig. 1). Using UniProtKB, all four N. elongata subsp. glycolytica ORFs were annotated, and their products were predicted to have enzymatic activity consistent with the Wbp enzymes. We were unsuccessful in identifying an N. elongata subsp. glycolytica orthologue for the WbpI epimerase. Amino acid alignments of the predicted products from N. elongata subsp. glycolytica with their counterparts in both P. aeruginosa PAO1 and B. pertussis are shown in Fig. S1 in the supplemental material.

Cross-genus complementation restores OSA biosynthesis in P. aeruginosa.

To verify that the four identified N. elongata subsp. glycolytica gene-encoded enzymes can support the synthesis of UDP-GlcNAc(3NAc)A, we attempted to complement a panel of defined P. aeruginosa PAO1 mutants bearing corresponding null mutations in the O-antigen pathway. We introduced plasmids expressing each of the N. elongata subsp. glycolytica ORFs into the corresponding P. aeruginosa PAO1 wbpA, wbpB, wbpE, and wbpD mutant strains and investigated the status of O-specific antigen (OSA) expression. In P. aeruginosa, smooth LPS containing OSA runs on an SDS-polyacrylamide gel as a ladder that represents individual LPS species with differing numbers of O-antigen repeats (15). The LPS forms can be specifically detected by their reactivity in immunoblotting utilizing monoclonal antibody (MAb) MF15-4. Previous work has shown that restitution of OSA O-antigen expression requires plasmid-based transcomplementation with the individual corresponding gene (15). We prepared LPS from the wbp mutants and strains complemented with the putative N. elongata subsp. glycolytica orthologues and carried out immunoblotting analyses. LPS of PAO1 mutants individually lacking wbpA, wbpB, wbpE, or wbpD failed to react with MAb MF15-4, whereas LPS purified from strains complemented with the putatively orthologous N. elongata subsp. glycolytica pgl genes showed reactivity with the MAb (Fig. 3). These results demonstrate that the N. elongata subsp. glycolytica genes are orthologous to their P. aeruginosa PAO1 counterparts, and we have designated them, according to pgl nomenclature, pglJ (NELON_10580), pglK (NELON_10585), pglM (NELON_05890), and pglN (NELON_05885).

FIG 3.

Cross-complementation of P. aeruginosa PAO1 wbp null mutants by the respective N. elongata subsp. glycolytica pgl genes. Immunoblotting of LPS preparations was performed using MAb MF15-4, which is specific against O-specific antigen (OSA) of P. aeruginosa.

Effects of pgl null mutations on N. elongata subsp. glycolytica protein glycosylation.

We next examined the effects of loss-of function-mutations in the newly identified N. elongata subsp. glycolytica pgl genes on glycoform expression and protein glycosylation. Accordingly, we constructed a panel of N. elongata subsp. glycolytica mutants in which each of the four genes was individually deleted using a previously described strategy (14). These mutations were created in an N. elongata subsp. glycolytica background expressing a nirK allele in which its ORF was tagged with a C-terminal hexahistidine (6×His) extension. NirK was then affinity purified from all four mutant backgrounds, subjected to proteolytic digestion, and examined by liquid chromatography-tandem mass spectrometry (LC-MS²) analysis. As previously described (14), the MS² spectrum of the wild-type N. elongata subsp. glycolytica tetrasaccharide displays a series of glycan-specific oxonium reporter ions: diNAcBac at m/z 211.110, diNAcBac-Hex at m/z 391.170, diNAcBac-Hex-diNAcHexA at m/z 649.254, and HexNAc at m/z 204.086.

The LC-MS² extracted-ion chromatograms (XIC) of chymotrypsin-generated NirK-His from a pglJ background revealed overlapping peaks representing diNAcBac and diNAcBac-Hex glycan reporter ions (Fig. 4A). This indicated that the glycopeptides detected were modified solely with a diNAcBac-Hex disaccharide and that the pglJ mutant phenocopied a pglG null mutant (14). Confirmation that the peaks representing the glycan reporter ions were derived from glycopeptides can be seen in a representative MS² spectrum selected from the XIC because of the overlapping reporter ion peaks. In the chromatogram of the chymotrypsin-generated glycopeptide 371KGTGAAAGAASGASGASAPAAPASSASGSSNPY403 (Fig. 4B), two glycan reporter ions representing diNAcBac (at m/z 211.1) and diNAcBac-Hex (at m/z 391.2) (i.e., a disaccharide) were detected in the low-mass area.

FIG 4.

Effect of pgl gene mutations on glycan structure. Shown are LC-MS² chromatograms of proteolytically derived peptides from affinity-purified NirK from pgl mutants. Total ion chromatogram (TIC) intensity values represent the amount of peptides entering the mass spectrometer. The selected ion chromatograms (XIC) are of the four glycan reporter ions characteristic for a tetrasaccharide, as well as the predicted glycan reporter ions originating from oligosaccharides of the pglJKMN pathway: diNAcBac at m/z 211.108 (1), diNAcBac-Hex at m/z 391.170 (2), diNAcBac-Hex-HexNAcA at m/z 608.229 (3), diNAcBac-Hex-(Keto)HexNAcA at m/z 607.222 (4), diNAcBac-Hex-(amino)HexNAcA at m/z 607.245 (5), diNAcBac-Hex-diNAcHexA at m/z 649.254 (6), and HexNAc at m/z 204.086 (7). All glycan reporter ions were searched with a 10-ppm window. The MS² spectrum demonstrates the presence of glycan reporter ions (marked in boldface type and numbered as described above). All XIC were normalized according to their abundance, either to the most abundant glycan reporter ion or, for the less abundant reporter ion, to the abundance of the overall least abundant reporter ion, the diNAcBac-Hex-(amino)HexNAcA ion at m/z 607.245. (A) LC-MS² chromatogram of trypsin-derived peptides from affinity-purified NirK-His from a pglJ background (RV727). The total ion chromatogram intensity value was set at 4E8. All selected ion chromatogram values were normalized to the most abundant glycan reporter ion, diNacBac-Hex at 2.0E5. (B) MS² spectrum of the singly modified chymotrypsin-generated peptide ³⁷¹KGTGAAAGAASGASGASAPAAPASSASGSSNPY⁴⁰³ from NirK-His from panel A carrying a diNAcBac-Hex disaccharide. (C) LC-MS² chromatogram of chymotrypsin-derived peptides from affinity-purified NirK-His from a pglK background (RV731). The total ion chromatogram intensity value was set at 4E8. Abundant XIC values were normalized to the most abundant glycan reporter ion, diNAcBac-Hex-HexNAcA at 2.0E5. (D) MS² spectrum of the singly modified chymotrypsin-generated peptide ³⁷¹KGTGAAAGAASGASGASAPAAPASSASGSSNPY⁴⁰³ from NirK-His from panel C carrying a diNAcBAc-Hex-HexNAcA-HexNAc tetrasaccharide. (E) LC-MS² chromatogram of chymotrypsin-derived peptides from affinity-purified NirK-His from a pglM background (RV735). The XIC intensity value was set at 4E8. Abundant XIC values were normalized to the most abundant glycan reporter ion, diNAcBac-Hex-HexNAcA at 2.0E5. (F) MS² spectrum of the chymotrypsin-generated peptide ³⁷¹KGTGAAAGAASGASGASAPAAPASSASGSSNPY⁴⁰³ from NirK-His from panel E carrying a diNAcBAc-Hex-HexNAcA-HexNAc tetrasaccharide. (G) LC-MS² chromatogram of chymotrypsin-derived peptides from affinity-purified NirK-His from a pglN background (RV739). The total ion chromatogram intensity value was set at 4E8. The XIC intensity value was set at 4E8. Abundant XIC values were normalized to the most abundant glycan reporter ion, diNAcBac-Hex-HexNAcA at 2.0E5. (H) MS² spectrum of the chymotrypsin-generated peptide ³⁷¹KGTGAAAGAASGASGASAPAAPASSASGSSNPY⁴⁰³ from NirK-His from panel G carrying a diNAcBAc-Hex-(amino)HexNAcA-HexNAc tetrasaccharide.

Based on the data from prior studies and the complementation data presented here, we anticipated that subsequent disruption of the GlcNAc(3NAc)A biosynthetic pathway in N. elongata subsp. glycolytica mutants might lead to incorporation of oligosaccharide moieties with distinct, signature reporter ions for each blocked step, i.e., diNAcBac-Hex-HexNAcA (m/z 608.229), diNAcBac-Hex-HexNAc(3keto)A (m/z 607.245), and diNAcBac-Hex-HexNAc(3NH₂)A (m/z 607.222), each together with a terminal HexNAc (m/z 204.1). The LC-MS² chromatogram and representative spectrum of the corresponding glycopeptide from the pglK background displayed reporter ions representing a tetrasaccharide with HexNAcA (at m/z 218.1) rather than the diNAcHexA at the third position, at m/z 608.2 (together with diNAcBac [m/z 211.1], diNAcBac-Hex [m/z 391.2], and a HexNAc [m/z 204.1]) (Fig. 4C and D). In addition, this sample yielded evidence for microheterogeneity as seen by the simultaneous presence of a population of peptides bearing just the disaccharide diNAcBac-Hex (m/z 391.2) across the XIC of the specific glycan reporter ions. Interestingly, the same tetrasaccharide reporter ions and signs of disaccharide-associated microheterogeneity seen in the pglK background were detected in both the chromatogram and spectrum of the NirK-His glycopeptide from the pglM background (Fig. 4E and F). Identical findings for the latter two mutants could be attributed to the likelihood that N. elongata subsp. glycolytica PglK and PglM act as a dehydrogenase/aminotransferase pair that converts UDP-GlcNAcA to UDP-GlcNAc(3NH₂)A in an obligatory, coupled reaction involving NAD(+) recycling, as previously detailed for WbpB and WbpE in P. aeruginosa (17).

For the glycopeptide from the pglN background, reporter ions consistent with the presence of the two distinct glycoforms seen in the pglK and pglM samples were evident. In addition, signals indicative of HexNAc(NH₂)A-containing tetrasaccharide were seen in the chromatogram (Fig. 4G and H). In the MS² spectrum of the triply charged precursor ion at m/z 1172.907 (Fig. 4H), dominant reporter ions at m/z 607.3 and at m/z 810.4 corresponding to diNAcBac-Hex-HexNAc(NH₂)A and diNAcBac-Hex-HexNAc(NH₂)A-HexNAc, respectively, were detected. However, it is likely that the MS spectrum (Fig. 4H) contains the fragmentation pattern of both a diNAcBac-Hex-HexNAc(NH₂)A-HexNAc and the diNAcBac-Hex-HexNAcA-HexNAc seen in both the pglM and pglK backgrounds. This is based on the finding that the ¹³C isotope of each of m/z 607.3 and m/z 810.4 (i.e., m/z 608.3 and m/z 811.4, respectively) is more abundant than predicted. These observations suggest the presence of diNAcBac-Hex-HexNAcA and diNAcBac-Hex-HexNAcA-HexNAc glycoforms along with the presence of the reporter ion at m/z 218.1 corresponding to HexNAcA.

Together, these results show that PglG has relaxed specificity as evidenced by its ability to incorporate altered HexNAcA-derived sugars resulting from disruptions in the pglJKMN biosynthetic pathway. Thus, pglG and pglJ seem to fulfil the minimal requirements for further elaboration of the disaccharide form. Moreover, the unidentified glycosyltransferase responsible for tetrasaccharide formation must also have relaxed specificity to recognize Und-PP-trisaccharide substrates terminating with either HexNAcA, HexNAc(NH₂)A, or HexNAc(NAc)A.

To corroborate the MS findings, we examined the status of N. elongata subsp. glycolytica NirK in a panel of pgl-relevant backgrounds by immunoblotting using an antibody specific for a tetra-His epitope and a rabbit polyclonal serum specific for diNAcBac-Glc disaccharide epitopes (pDAb2). As previously reported (14), detection using the poly-His-recognizing antibody revealed a clear increase in the relative mobility of tagged-NirK in a pglC null background (totally defective in glycosylation) versus that seen for the wild-type background. In neither case did NirK-His react with the pDAb2 antibody. In the pglH and pglG backgrounds, NirK showed slightly decreased mobility relative to that seen in the pglC background (the mono- and disaccharide glycoforms are difficult to resolve on an SDS-polyacrylamide gel) but increased mobility compared to that in the wild-type background. However, as expected, only in the pglG background was NirK reactive with antiserum pDAB2, which is specific for the diNAcBac-Glc disaccharide (Fig. 5).

FIG 5.

Immunoblotting analyses of N. elongata subsp. glycolytica pgl mutants reveals associated, altered glycoprotein mobility. (Top) Immunoblot analysis of whole-cell lysates from wild-type (wt) (KS992), pglC (KS994), pglH (KS1052), pglG (KS1034), pglJ (RV727), pglK (RV731), pglM (RV735), and pglN (RV739) strains using a tetra-His epitope-recognizing antibody (detecting the NirK-His glycoprotein). (Bottom) Immunoblot analysis of the same samples as those described above with the diNAcBac-Glc epitope-recognizing polyclonal antibody pDAb2.

NirK from the pglJ background phenocopied the reactivity pattern of the pglG background, as no discernible differences between the two backgrounds toward either the anti-His or the pDAb2 antibodies were detected (Fig. 5). In contrast, NirK from pglK, pglM, and pglN all showed similar relative mobilities intermediate to those of the wild-type and pglG backgrounds using SDS-PAGE. The “ladder-like banding” observed here was also detected by both the anti-His and pDAb2 antibodies (Fig. 5). In the case of NirK in a pglG background, earlier MS work revealed microheterogeneity as evidenced by the detection of both mono- and disaccharide glycans (14), and similar findings are reported here for the pglK, pglM, and pglN backgrounds (Fig. 4). In this context, it is also important to recall that N. elongata subsp. glycolytica NirK has at least five glycan attachment sites (14). Hence, the observed banding pattern exhibited in these backgrounds could be indicative of a variable number of sites occupied by glycans per protein as well as proteins carrying differing glycoforms or even individual proteins carrying mixtures of glycoforms.

Other glycoproteins detected by the diNAcBac-Glc disaccharide antibodies in the pglK, pglM, and pglN backgrounds also show signs of altered migration versus those in the other backgrounds (for example, the CcoP glycoprotein band that migrates just above the NirK band). Therefore, the observed electrophoretic migration anomalies are not unique to NirK.

The immunoblotting data from the pglJ, pglK, pglM, and pglN backgrounds were consistent with the expression of NirK bearing differing glycoforms or mixtures of glycoforms per protein. Based on these data alone, however, one cannot formally rule out the potential influence of differences in site occupancy (microheterogeneity) potentially occurring in these backgrounds on SDS-PAGE migration.

DISCUSSION

In this study, we identified four genes involved in the synthesis of the rare sugar 2,3-diacetamido-2,3-dideoxy-d-glucuronic acid, which is a component of the tetrasaccharide expressed by the broad-spectrum O-linked protein glycosylation system of N. elongata subsp. glycolytica. These findings reveal an extension of a highly conserved pathway for the production of HexNAc(3NAc)A sugar and its derivatives associated with LPS structures in some strains of P. aeruginosa (17) and B. pertussis (15) and protein N-linked glycans in the archaeal species Methanococcus voltae (16) and Methanococcus maripaludis (18). Although we lack nuclear magnetic resonance (NMR) data to define glycan stereochemistry, our results, together with the apparent absence of any ORF within the N. elongata subsp. glycolytica genome sharing significant relatedness to established UDP-GlcNAc(3NAc)A epimerases [converting the glycan to a UDP-ManNAc(3NAc)A], suggest that the likely end product of the N. elongata subsp. glycolytica hexuronic acid pathway is a glucuronic acid-based sugar. HexNAc(3NAc)A- and diNAcBac-containing glycoforms have also been identified in the general O-linked protein glycosylation systems of some Acinetobacter species (2). Although Neisseria falls within the betaproteobacteria and Acinetobacter within the gammaproteobacteria, it seems likely that there may be some common ancestry behind their shared glycans and pgl gene repertoires.

The pathway for GlcNAc(3NAc)A sugar synthesis based on biochemistry experiments is well defined in other bacterial biosynthesis systems (15, 17, 18), all of which entail stepwise assembly of a glycan onto polyprenyl-phosphates. In contrast, the in vivo steps at which defects in particular genes and their products associated with GlcNAc(3NAc)A synthesis might disrupt downstream oligosaccharide synthesis are unclear. Formally, mutations causing defects in synthesis of intermediates through to UDP-GlcNAc(3NAc)A from UDP-GlcNAcA could result in failure of relevant glycosyltransferases to recognize these precursors, failure of subsequently acting glycosyltransferases to utilize lipid-linked oligosaccharides bearing the aberrant sugar, or failure of the targeting oligosaccharyltransferase to utilize an altered lipid-linked precursor. In both P. aeruginosa and B. pertussis, disruption of the genes associated with the UDP-diNAcManA pathway cause lack of insertion of any glycan into OSA LPS and LPS, respectively (15). In the M. maripaludis N-linked system, glycoproteins from mutants individually lacking any of the four enzymes required for the synthesis of UDP-GlcNAc(3NAc)A from UDP-GlcNAc carry only a truncated monosaccharide (18). The corresponding N. elongata subsp. glycolytica pgl system mutants showed clear evidence of usage of the noncanonical UDP-sugar. While the pglJ dehydrogenase null mutant-derived glycoprotein bore only the diNAcBaC-Glc disaccharide, the glycoproteins from mutants lacking PglK, PglM, or PglN carried mixtures of di-, tri-, and tetrasaccharide glycoforms. In addition, the oxonium ion fragmentations patterns for the tri- and tetrasaccharide glycoforms in LC-MS² analyses were consistent with the presence of the glucuronic acid-derived intermediate predicted from a disrupted biosynthetic pathway. It follows then that the PglG glycosyltransferase has relaxed specificity for the UDP-sugar donor that minimally requires UDP-glucuronic acid. Moreover, the unidentified glycosyltransferase responsible for the addition of the fourth sugar should also have relaxed specificity for the Und-PP-linked substrate, minimally requiring a terminally situated glucuronic acid.

In conclusion, we have identified a conserved enzymatic pathway for the biosynthesis of di-N-acetylated glucuronic acid in a deeply branching species in the genus Neisseria. A current working model of the O-linked glycosylation pathway as expressed in N. elongata subsp. glycolytica is presented in Fig. 6. The findings presented here show that both PglG and PglJ are required for the extension of the growing oligosaccharide chain before the full-length tetrasaccharide glycan can be transferred to its target glycoproteins. We find it interesting that this and related moieties are so widely but relatively sparsely distributed as an oligosaccharide constituent across bacterial O-linked protein glycosylation, LPS biosynthetic, and capsular polysaccharide systems as well as in N-linked protein glycosylation in archaeal species. However, we currently have little knowledge of the structure-function relationships accounting for the broad but restricted distribution of this modified sugar form. Furthermore, hexuronic acid-related sugars are uncommon as components of oligosaccharides in neisserial species, having been identified only in the meningococcal capsular polysaccharides of serogroup I and K variants as N-acetylmannosaminuronic acid units (19). That said, there is relatively little information on the distribution of O-linked protein glycosylation and potential associated glycoforms in commensal neisserial species. The findings and methods developed here should help address these concerns.

FIG 6.

Current working model of the broad-spectrum O-linked glycosylation pathway expressed in N. elongata subsp. glycolytica. The findings discussed in this paper show that PglJ, PglK, PglM, and PglN are all required for the biosynthesis of UDP-GlcNAc(3NAc)A. The PglG glycosyltransferase uses this as a donor to extend the disaccharide chain. A yet-to-be-identified glycosyltransferase (Pgl?) transfers the final HexNAc onto the terminal position to generate the tetrasaccharide, which then gets flipped by PglF into the periplasm. The PglO oligosaccharyltransferase transfers the glycan onto glycoproteins.

MATERIALS AND METHODS

Bacterial strains and culture conditions.

The strains used are described in Table S1 in the supplemental material and were derived from N. elongata subsp. glycolytica strain ATCC 29315. N. elongata subsp. glycolytica strains were cultured on conventional GC plates (Difco) as previously described (14). Mutants were obtained by transforming N. elongata subsp. glycolytica with crude cell lysates, PCR-amplified DNA, or purified plasmid DNA. Transformation of N. elongata subsp. glycolytica was performed following a protocol described previously for N. gonorrhoeae (20), and mutants were selected on plates containing appropriate antibiotics as follows: kanamycin (50 μg/ml), tetracycline (20 μg/ml), and chloramphenicol (30 μg/ml). The antibiotics used for selection in Escherichia coli were kanamycin (50 µg/ml) and tetracycline (15 µg/ml), and those for P. aeruginosa strains were gentamicin (200 µg/ml) and tetracycline (100 µg/ml). Plasmid transformations in P. aeruginosa were done using an adapted electroporation protocol detailed previously (21), while for E. coli plasmids were chemically transformed into TOP10 cells (Invitrogen).

Plasmid and strain construction.

Information about N. elongata subsp. glycolytica-related plasmids and strains used in this study is in the supplemental material.

Construction of cross-complementation plasmids.

For complementation of the defined wbpA, wbpB, wbpD, and wbpE P. aeruginosa deletion mutants (15), N. elongata subsp. glycolytica genes were cloned into the pUCP26 shuttle vector and used for P. aeruginosa complementation analyses. An EcoRI restriction site and the optimized Shine-Dalgarno sequence AGGAGGACAAGCT were included in the forward primer (which contained the start codon of the respective gene). Reverse primer pairs included a short sequence downstream of the stop codon and a HindIII restriction site. Primer sequences are listed in Table S2 in the supplemental material. PCR products were digested accordingly and ligated into an EcoRI/HindIII-digested pUCP26 vector. The resulting plasmids were chemically transformed into TOP10 E. coli cells (Invitrogen) and isolated using Qiagen’s small-scale miniprep kit, and DNA was sequenced at GATC Sanger sequencing (GATC Biotech, Germany).

Cross-complementation studies and analysis of LPS.

Plasmids containing N. elongata subsp. glycolytica genes were transformed into the corresponding P. aeruginosa knockout strains. The unmodified pUCP26 vector was used as a negative control. LPS samples from the resulting strains were then used in complementation analyses by immunoblotting. A small-scale LPS preparation of P. aeruginosa strains was prepared as described previously (22). In brief, cells were harvested from overnight liquid cultures, and equal numbers of cells (as measured by optical density [OD]) were resuspended in lysis buffer (22). Samples were then boiled for 30 min, treated with proteinase K, and incubated at 55°C overnight. Equal volumes of LPS samples were run on a 12% Criterion Bis-Tris protein gel (Bio-Rad) and immunoblotting was conducted with the primary monoclonal antibody MAb MF 15-4 (specific for B-band O-antigen) (23).

Affinity purification of glycoproteins and in-gel digestion.

Purification of His-tagged proteins and in-gel digestion of Coomassie blue-stained gel slices containing purified proteins with chymotrypsin (Sigma) were performed as previously described (24, 25).

Reverse-phase LC-MS² analysis of tryptic peptides.

Reverse-phase nanoflow LC-MS and MS² analysis (nano-LC-MS²) of chymotrypsin-derived peptides was performed as previously described (14).

Immunoblotting of whole-cell lysates.

Whole-cell lysates were prepared from overnight cultures grown for 16 to 18 h. Cells were scraped from plates, resuspended in water, and diluted to a final OD of 1, when an equal volume of sample loading buffer was added. Lysates were heated at 80°C for 15 min and saved at −20°C. Samples were run on a 12% Criterion Bis-Tris protein gel (Bio-Rad), and immunoblotting was done with primary antibodies pDAB2 (diNAcBac-Glc epitope-recognizing polyclonal antibody) (5) or tetra-His-recognizing antibodies (Qiagen) as previously described (14).

Genome databases.

Genome analysis was done by using deposited sequences found in the Bacterial Isolate Genome Sequence Database (BIGSdb) (26) and from the National Center for Biotechnology Information (NCBI) database servers. N. elongata subsp. glycolytica accession numbers were found using the Universal Protein Resource (UniProt) (27).