Characterization of the Structurally Diverse N-Linked Glycans of Campylobacter Species

Abstract

The Gram-negative bacterium Campylobacter jejuni encodes an extensively characterized N-linked protein glycosylation system that modifies many surface proteins with a heptasaccharide glycan. In C. jejuni, the genes that encode the enzymes required for glycan biosynthesis and transfer to protein are located at a single pgl gene locus. Similar loci are also present in the genome sequences of all other Campylobacter species, although variations in gene content and organization are evident. In this study, we have demonstrated that only Campylobacter species closely related to C. jejuni produce glycoproteins that interact with both a C. jejuni N-linked-glycan-specific antiserum and a lectin known to bind to the C. jejuni N-linked glycan. In order to further investigate the structure of Campylobacter N-linked glycans, we employed an in vitro peptide glycosylation assay combined with mass spectrometry to demonstrate that Campylobacter species produce a range of structurally distinct N-linked glycans with variations in the number of sugar residues (penta-, hexa-, and heptasaccharides), the presence of branching sugars, and monosaccharide content. These data considerably expand our knowledge of bacterial N-linked glycan structure and provide a framework for investigating the role of glycosyltransferases and sugar biosynthesis enzymes in glycoprotein biosynthesis with practical implications for synthetic biology and glycoengineering.

INTRODUCTION

The asparagine-linked glycosylation of proteins is common in eukaryotes and is involved in protein folding, quality control, secretion, and sorting (10). Glycosylation involves the transfer of an oligosaccharide from a lipid carrier to asparagine residues within the consensus sequence N-X-S/T (X is any residue except proline) and takes place in the endoplasmic reticulum (ER) lumen as nascent polypeptides are secreted across the ER membrane (26). Analogous protein N-glycosylation reactions also occur in a restricted set of species from the archaeal and bacterial domains (3, 37). In eukaryotes, protein N-glycosylation usually involves a multiprotein complex (34), but only a single subunit, the STT3 protein of the yeast complex, has bacterial/archaeal orthologues, namely, PglB and AglB, respectively. The STT3, PglB, and AglB proteins all have multiple transmembrane domains within the amino terminus and a soluble carboxy terminus (3, 31, 32, 34, 37). The STT3 protein is the catalytic subunit of the yeast oligosaccharyltransferase (OTase) responsible for transferring oligosaccharide from the lipid carrier onto the target protein, but PglB and AglB act as single-subunit OTases (1, 18, 50).

The prototypical bacterial N-linked OTase is the Campylobacter jejuni PglB protein. The C. jejuni pglB gene is located within a cluster of genes termed the protein glycosylation (pgl) locus (45). This locus encodes all of the necessary enzymes for N-linked protein glycosylation with a heptasaccharide consisting of GalNAc–α1,4-GalNAc–α1,4(Glcβ1,3)-GalNAc–α1,4-GalNAc–α1,4-GalNAc–α1,3-Bac (Bac is bacillosamine or 2,4,-diacetamido-2,4,6-trideoxyglucose). Functional analysis of the genes from the pgl locus has revealed that the heptasaccharide is synthesized by five glycosyltransferases encoded by the pglA, pglC, pglH, pglI, and pglJ genes (6, 7, 30, 38). The lipid-linked heptasaccharide is transferred across the bacterial inner membrane and into the periplasm by a “flippase,” PglK (2), where it is covalently linked to asparagine residues within an extended D/E-X-N-X-S/T sequon by the PglB protein (4, 22, 23).

More than 50 C. jejuni N-linked glycoproteins with diverse functions have been identified (29, 35, 42, 51). Mutants deficient in N-linked glycosylation are less competent (28), have a reduced capacity to invade in vitro-cultured epithelial cells, and are less able to colonize chicks and mice (11, 16, 17, 43). It has also been proposed that PglB-mediated release of free oligosaccharide in the periplasm plays a role in osmotic protection (36). Finally N-linked-heptasaccharide-specific recognition by macrophages is thought to have an immunosuppressive role (46). From these findings, it is clear that the role of N-linked protein glycosylation in campylobacters and related species is likely to be both significant and complex.

The C. jejuni pgl locus is fully functional when transferred to Escherichia coli, allowing glycosylation with heptasaccharide of both C. jejuni and non-C. jejuni proteins that contain the N-linked glycosylation sequon (50). When structurally variant lipid-linked glycans are made available, the C. jejuni PglB OTase is able to transfer many, but not all, of these onto proteins (5, 49). Such flexibility makes this an attractive approach for the development of an in vivo glycoconjugate synthesis system (14). The key structural requirement for PglB-mediated glycan transfer is the presence of an acetamido group on the C-2 carbon of the reducing-end sugar (5, 49).

Recently released bacterial genome sequence data indicate that many members of the Deltaproteobacteria-Epsilonproteobacteria subdivision of the Proteobacteria contain pglB orthologues (13, 15, 25, 37, 44). Prominent genera within the Epsilonproteobacteria subdivision include Arcobacter, Campylobacter, and Helicobacter. Species of the Arcobacter genus lack pglB orthologues, and within the Helicobacter genus, only three closely related species (H. pullorum, H. canadensis, and H. winghamensis) contain pglB orthologues; we have characterized the H. pullorum-associated N-linked glycan (15). In contrast, all Campylobacter species with available genome sequence data contain one or, in a small number of cases, two pglB orthologues. Aside from C. jejuni and, to a much lesser extent, Campylobacter lari (41), other Campylobacter-encoded N-linked protein glycosylation systems are as yet completely uncharacterized.

In this study, we set out to provide a broader perspective on N-linked protein glycosylation within the Campylobacter genus as a whole. We employed an in vitro peptide N-glycosylation assay combined with mass spectrometry (MS) to rapidly characterize N-linked glycan structures from nine Campylobacter species. This considerably expands our knowledge of bacterial N-linked protein glycosylation systems, providing a further set of glycoactive enzymes for potential biotechnological exploitation.

MATERIALS AND METHODS

Bacterial strains and plasmids.

All E. coli strains were grown in Luria-Bertani (LB) broth or on LB agar plates including antibiotic supplements as appropriate, and all Campylobacter species were grown on Columbia agar (Oxoid Ltd.) containing 5% defibrinated horse blood (TCS Biosciences). C. jejuni NCTC 11168H, Campylobacter coli RM2228, C. lari RM2100, and Campylobacter upsaliensis NCTC11541 were all grown at 42°C in a VA500 workstation (Don Whitley Ltd.) maintaining a microaerobic atmosphere of 85% N₂, 10% CO₂, and 5% O₂. Campylobacter helveticus NCTC 12470, Campylobacter hyointestinalis subsp. hyointestinalis NCTC 11608, Campylobacter fetus subsp. fetus NCTC 10842, and Campylobacter lanienae NCTC 13004 were grown at 42°C in gas jars under microaerobic conditions generated using the Campygen system (Oxoid Ltd.) supplemented with ∼10% H₂. Campylobacter concisus NCTC 11485 was grown in the same manner but at 37°C, and Campylobacter sputorum biovar sputorum NCTC 11528 was grown anaerobically in a gas jar with H₂ at 42°C.

Insertional inactivation of C. fetus pglB.

A 2.7-kb PCR product was amplified from C. fetus genomic DNA using primers CfetpglB-F (5′-CTTTTGATATTAGGCGAGGG-3′) and CfetpglB-R (5′-TACACCTTAGCGCTTGGATC-3′). The resultant PCR product, from ∼420 bp upstream of the predicted start codon of the pglB gene to 12 bp upstream of the predicted stop codon, was ligated into pGEMTEasy (Stratagene) and digested with XbaI, removing a 1.2-kb internal region of the pglB gene. A kanamycin resistance cassette (aphA) lacking a transcriptional terminator was ligated into the XbaI-digested vector in the same transcriptional orientation as the pglB gene. The resultant plasmid was electroporated into electrocompetent C. fetus cells as described previously (47). Kanamycin-resistant transformants were isolated, and integration by a double-crossover event was verified by PCR. Approximately 1 in 20 recombination events was a double crossover, as previously reported (9).

Glycan detection with hR6 antiserum and soybean agglutinin (SBA) lectin.

Campylobacters were grown as described above, and cells were harvested from agar plates, standardized by optical density at 600 nm, and heated to 95°C for 5 min in SDS-PAGE loading buffer. These whole-cell lysates were subjected to 12% SDS-PAGE, transferred to polyvinylidene difluoride membrane, and probed either with hR6 antiserum (S. Amber and M. Aebi, unpublished data) followed by an IRDye 800CW goat anti-rabbit IgG secondary antibody (Li-Cor) or with biotin-conjugated lectin SBA (Vector Labs) followed by IRDye 800CW streptavidin (Li-Cor). Blots were imaged using a Li-Cor Odyssey Infrared Imaging System.

In vitro OTase assays.

In vitro OTase assays were carried out as previously described (15, 21). In brief, a standard in vitro OTase assay mixture contained 26 μl of a Triton X-100-solubilized membrane preparation, 3 μl of 30 μM peptide (6-carboxyfluorescein–ADQNATA–NH₂), and 1 μl of 150 mM MnCl₂. When the glycopeptide yield was low, reaction mixtures were supplemented with 3 μl of a Triton X-100-solubilized membrane preparation from E. coli SCM7 (2) expressing the C. jejuni pglB gene from plasmid pMAF10 (5). Assay mixtures were incubated overnight at 30°C with agitation, and reaction products were analyzed by Tris-Tricine SDS-PAGE with a 16.5% resolving gel containing 6 M urea and a 10% spacer gel (39). Gels were visualized using a Typhoon Trio Plus Imager (GE Healthcare) with excitation at 532 nm (green laser) and a 526-nm short-pass emission filter.

Purification and matrix-assisted laser desorption ionization–time of flight (MALDI-TOF) MS analysis of biotinylated peptides.

Biotinylated peptides (fluorescein isothiocyanate [FITC]-ADQNATAK-biotin) were purified from OTase assay mixtures using streptavidin-coated magnetic beads (Pierce) as previously described (15) but with the following modifications. When efficiency of transfer was low, some assays were carried out in larger quantities and/or with an increased amount of membrane preparation relative to that of peptide (up to 10-fold). In such cases, the quantity of magnetic beads was increased in accordance with the volume of the membrane preparation used. Samples were eluted from streptavidin-coated beads by incubation in 10 μl 70% acetonitrile–5% formic acid–125 μM biotin for 30 min at room temperature and desalted on a C₁₈ ZipTip (Millipore). MALDI-TOF MS and MALDI-LIFT-TOF/TOF MS spectra were acquired by laser-induced dissociation (LID) using a Bruker Ultraflex II mass spectrometer in the positive-ion reflection mode with 50 mM α-cyano-4-hydroxycinnamic acid (in 80% acetonitrile) as the matrix. Data were analyzed with FlexAnalysis 3.0 software (Bruker Daltonics).

RESULTS

Diversity of Campylobacter pgl gene loci.

Available genome sequence data indicate that all Campylobacter species encode a pglB orthologue located within a cluster of genes many of which are orthologues of genes from the well-characterized C. jejuni pgl gene locus (25, 37) (Fig. 1). This gene cluster in C. coli, the species most closely related to C. jejuni, has an essentially identical overall arrangement of genes, and individual gene products show significant levels of amino acid sequence identity (82 to 97%) to the C. jejuni pgl locus-encoded gene products (data not shown). Similarly, the pglB gene-containing locus from C. lari, also closely related to C. jejuni, is very similar in overall organization but with lower levels of orthologue sequence identity (56 to 83%; data not shown). The C. lari locus also lacks an orthologue of the C. jejuni pglI gene encoding a glucosyltransferase involved in heptasaccharide biosynthesis (30, 38, 41). More significant differences in pgl locus gene content and arrangement are evident in Campylobacter species more distantly related to C. jejuni (Fig. 1; data not shown). For example, in C. upsaliensis and C. fetus, although the overall arrangement of the putative pgl locus is still similar to that of the C. jejuni locus, there are insertions of 21 and 24 open reading frames (ORFs), respectively, between the putative pglD and pglE genes (Fig. 1). These insertions are apparently unrelated to each other and encode putative proteins with no similarity to enzymes involved in sugar biosynthesis or transport and are therefore likely not involved in N-linked protein glycosylation (data not shown). Comparison of the C. fetus and C. jejuni loci reveals further divergence, with additional ORFs encoding putative glycosyltransferases and sugar biosynthesis enzymes upstream of pglB (Fig. 1). Loci from C. gracilis, C. hominis, C. showae, and C. curvus also contain similarly located additional ORFs encoding putative glycosyltransferases and enzymes involved in sugar biosynthesis. Several Campylobacter species (C. concisus, C. curvus, and C. gracilis) possess two pglB orthologues, and these can be located within the pgl locus (C. concisus, C. curvus) or one of the two may be located outside the pgl locus (C. gracilis).

Fig 1.

Inferred phylogeny (A) and putative pgl loci (B) of Campylobacter species. (A) A phylogenetic tree was generated using approximately 1,300 bases of 16S rRNA gene sequence from Campylobacter species and E. coli K-12 strain W3110 as an outlier. Sequences were aligned using ClustalW2 (27) and a neighbor-joining phylogenetic tree produced using MEGA5 software (24). Species in bold type were analyzed further in this study, and the genome sequences of those underlined are available. (B) The C. jejuni NCTC 11168 PglB amino acid sequence was used as the query in a BLAST search against the nonredundant protein database to identify further Campylobacter pglB genes. For species whose genomes have been sequenced, predicted amino acid sequences from ORFs surrounding the pglB genes were used as queries in BLAST searches to identify potential functions. Where BLAST searches resulted in hits with C. jejuni Pgl proteins, a Needleman-Wunsch pairwise global alignment (NCBI) of the query and the corresponding C. jejuni NCTC 11168 orthologue was performed. Where the amino acid identity was >30% (gaps, <10%), ORFs were labeled according to the C. jejuni pgl nomenclature (data not shown). Those with predicted functions involved in polysaccharide biosynthesis are identified as follows: putative N-linked OTases or PglB proteins, black shading; sugar biosynthesis enzymes, no shading; glycosyltransferases, light shading; transporters, dark shading; unrelated genes or genes of unknown function, grey outline with no shading. Excluded are Campylobacter rectus, which has sequencing errors in the pglB gene (37), and C. fetus subsp. venerealis, which has the same gene arrangement in this locus as C. fetus subsp. fetus.

Reactivity of diverse Campylobacter species with C. jejuni glycan interacting reagents.

The C. jejuni N-linked heptasaccharide can be detected on glycoproteins via interaction with lectins such as SBA (29) which bind to terminal GalNAc residues or a polyclonal antiserum (hR6) specific for the terminal two or three GalNAc residues of the C. jejuni N-linked glycan (Amber and Aebi, unpublished). When Western blot assays of whole-cell lysates of Campylobacter species were probed with hR6 antiserum, those from C. jejuni, C. coli, C. lari, C. helveticus, and C. upsaliensis were strongly immunoreactive (Fig. 2A). The dependence of hR6 reactivity on the presence of N-linked glycan was confirmed by the complete lack of signal (Fig. 2A) in a previously characterized C. jejuni pglB knockout mutant (30, 51). Interestingly, whole-cell extracts from C. hyointestinalis, C. fetus, C. lanienae, C. sputorum, and C. concisus were unreactive with the hR6 antiserum, suggesting structural differences in any N-linked glycans produced by these species. The lectin SBA produced essentially the same pattern and specificity of binding, although for some species there were additional bands that we attribute to nonspecific interactions (Fig. 2B). These data suggest that the N-linked glycans from the phylogenetically closely related species C. jejuni, C. coli, C. lari, C. helveticus, and C. upsaliensis all have structural features in common. In contrast, the species more distantly related to this group either do not produce N-linked glycoproteins or produce structurally diverse N-linked glycans that do not interact with either antiserum hR6 or lectin SBA. The latter explanation is strongly favored, at least for C. fetus and C. concisus, given the presence of putative N-linked glycosylation loci in their respective genome sequences (Fig. 1). Furthermore, these loci differ significantly in gene content from the C. jejuni prototype (Fig. 1), indicating potential for the biosynthesis of structurally distinct N-linked glycans that may well not interact with reagents with some degree of specificity for the C. jejuni N-linked glycan.

Fig 2.

Reactivity of Campylobacter proteins with C. jejuni N-linked-glycan-specific reagents. Western blot assays of whole-cell lysates from C. jejuni (C. jej.) and the corresponding isogenic pglB knockout mutant (C. jej. pglB::kan) along with C. coli (C. col.), C. lari (C. lar.), C. hyointestinalis (C. hyo.), C. helveticus (C. hel.), C. upsaliensis (C. ups.), C. fetus (C. fet.), C. lanienae (C. lan.), C. sputorum (C. spu.), and C. concisus (C. con.) were probed with C. jejuni N-linked-glycan-specific hR6 antiserum (A) or SBA lectin (B).

In vitro N-linked glycosylation in Campylobacter species.

We recently employed a modification of an in vitro OTase assay (21) to investigate N-linked glycosylation processes in H. pullorum (15), and we have now used a similar approach to characterize N-linked protein glycosylation in nine Campylobacter species. Detergent-solubilized membrane preparations from individual Campylobacter species were incubated with a fluorescently labeled peptide, ADQNATA, incorporating a C. jejuni-optimized N-glycosylation acceptor sequon (4). Products were separated by Tris-Tricine SDS-PAGE in which glycopeptides displayed reduced electrophoretic mobility compared to that of the unmodified peptide (Fig. 3). Each of the Campylobacter species produced at least one band with reduced electrophoretic mobility, although some are relatively faint, indicating varying efficiencies of individual reactions (Fig. 3A). In order to improve the efficiency of glycan transfer, reaction mixtures were supplemented with a membrane preparation from E. coli SCM7 expressing the C. jejuni pglB gene (see Materials and Methods for details). When resultant products were analyzed by gel electrophoresis, the overall banding pattern was essentially the same as that produced by reaction mixtures containing only Campylobacter-derived membrane preparations (compare Fig. 3A and B), but for a number of species (C. hyointestinalis, C. helveticus, C. upsaliensis, C. lanienae, and C. concisus), the fluorescence intensity of the derived glycopeptide bands was significantly increased, indicating enhanced transfer of glycan, presumably due to C. jejuni PglB activity.

Fig 3.

OTase activities of Campylobacter species. Campylobacter membrane preparations were solubilized in detergent and assayed for N-linked OTase activity using a fluorescently labeled peptide (ADQNATA). Reaction products were separated by Tris-Tricine SDS-PAGE. The unmodified peptide (arrow) runs toward the bottom of the gel, while derived glycosylated peptides display reduced mobility. For reference, the lowest-mobility product produced by C. jejuni membranes corresponds to the peptide modified with a heptasaccharide (15). The OTase activity displayed by membrane preparations from a range of Campylobacter species (A) was, in a number of cases, significantly increased when membranes prepared from E. coli expressing C. jejuni pglB were added to reaction mixtures (B). The species tested were C. jejuni (C. jej), C. coli (C. col.), C. lari (C. lar.), C. hyointestinalis (C. hyo.), C. helveticus (C. hel.), C. upsaliensis (C. ups.), C. fetus (C. fet.), C. lanienae (C. lan.), C. sputorum (C. spu.), and C. concisus (C. con.).

The differing mobilities of the presumed glycopeptide products (Fig. 3) indicated potential structural differences in N-linked glycans, and these were further structurally characterized (see below). In parallel experiments emploing synthetic peptides with either an Asn-to-Gln substitution or an Asp-to-Ala substitution at the −2 position of the N-glycosylation sequon, bands with reduced electrophoretic mobility were not generated by any of the Campylobacter species (unpublished data). The requirement for an Asn residue is consistent with an N-linked modification (structurally confirmed [see below]), and the requirement for an Asp residue suggests that all Campylobacter PglB proteins require a negatively charged residue at the −2 position, as demonstrated for C. jejuni PglB (23, 35).

Structural diversity in Campylobacter N-linked glycans.

Using a biotin-tagged sequon-containing peptide (FITC-ADQNATAK-biotin) that can be readily purified from OTase reaction mixtures prior to analysis by MS (15), we attempted to structurally characterize N-linked glycans of glycopeptides produced by diverse Campylobacter species. For a number of species (C. upsaliensis, C. helveticus, C. lanienae, and C. fetus), the low yield of glycopeptide produced was insufficient to generate reliable MS data and two distinct approaches were employed to increase glycopeptide production. The first involved combining membrane preparations from an E. coli strain expressing the C. jejuni pglB gene with Campylobacter membrane preparations as described above, and this modification significantly increased the glycopeptide yield for the majority of the species. For C. fetus, an alternative approach involving the construction of a C. fetus pglB insertional knockout mutant was required (see Materials and Methods). When membrane preparations from the C. fetus pglB::aphA insertional knockout mutant and the corresponding wild-type strain were combined, significantly increased glycopeptide product was observed (data not shown), presumably due to the accumulation of lipid-linked oligosaccharide (LLO) on membranes derived from the pglB mutant strain (15). By these approaches, we were able to produce sufficient glycopeptides from the set of nine Campylobacter species for subsequent MS characterization.

During MALDI-TOF MS analysis, the reporter peptide produces a number of predominant peaks, including m/z 1,589 (theoretical mass, 1,588 Da), 1,556, and 1,199, the latter of which is due to the total loss of FITC (15). Purified glycopeptides produced by Campylobacter species were characterized by MALDI-LIFT-TOF/TOF MS, and representative spectra are presented (Fig. 4). For each species, a single glycopeptide was detected with glycan fragmentation patterns containing a peak with an m/z value of either 1,199 or 1,556 at the lower-mass end of the spectrum. Peaks with m/z values of 1,182 and 1,539 (17 Da less than 1,199 and 1,556, respectively) were also observed, characteristic of the fragmentation of a side-chain amide bond of an N-linked glycan (Fig. 4). Therefore, most glycopeptide parent ions consisted of intact peptide plus glycan, and under tandem MS analysis, the FITC moiety fragmented (spectra containing an m/z 1,199 peak), while some did not lose the whole moiety (spectra containing an m/z 1,556 peak). In the case of C. lari, the FITC was not present in the parent ion.

Fig 4.

MS analysis of Campylobacter in vitro-generated N-linked glycans. In vitro glycosylation of a biotin-labeled fluorescent peptide (FITC-ADQNATAK-biotin) was analyzed by MALDI-TOF MS. Panels A through to I are MALDI-LIFT-TOF/TOF MS-generated spectra derived from glycopeptides produced by the indicated Campylobacter species. Below the spectra are the corresponding inferred N-linked glycan structures. Fragments corresponding to peptides with no glycan (arrows) generated peaks with an m/z value of either 1,556 or 1,199, depending on the presence or absence of the labile FITC moiety. The m/z values of peaks corresponding to fragment ions resulting from the sequential loss of sugar residues are indicated in the spectra. a.u., arbitrary units.

The MALDI-LIFT-TOF/TOF MS fragmentation patterns produced were consistent with sequential loss of sugar residues due to cleavage at glycosidic bonds, as indicated in Fig. 4. All Campylobacter N-linked oligosaccharides were composed of residues with masses of 228 Da (diacetamidotrideoxyhexose [DATDH]), 203 Da (N-acetylhexosamine [HexNAc]), and 162 Da (hexose [Hex]), apart from a single residue of 217 Da present in the C. concisus glycan. Comparison of the spectra shows that a variety of N-linked glycans are generated among Campylobacter species (Fig. 4). Key conserved features include the presence of the 228-Da residue as the reducing-end sugar of a HexNAc-rich oligosaccharide. More variable features include the number of sugar residues present in the N-linked oligosaccharide, with C. coli, C. upsaliensis, C. helveticus, C. sputorum, and C. concisus generating N-linked heptasaccharides; C. lari, C. hyointestinalis, and C. fetus generating hexasaccharides; and C. lanienae generating a pentasaccharide. Also, seven species (C. coli, C. upsaliensis, C. helveticus, C. hyointestinalis, C. fetus, C. sputorum, and C. concisus) produced oligosaccharides with a single branched sugar, a hexose residue in all species except for C. sputorum, with a branching HexNAc residue. Linear N-linked oligosaccharides were produced by C. lari and C. lanienae.

Overall, we have demonstrated experimentally that N-linked glycosylation is a conserved feature of Campylobacter species. It is also evident from MS data that the differing reactivities of whole-cell lysates (Fig. 2) and mobilities of in vitro-generated glycopeptides (Fig. 3) are a reflection of structural diversity in Campylobacter N-linked glycans, particularly at the nonreducing end (Fig. 4).

DISCUSSION

The extensively characterized C. jejuni N-linked protein glycosylation system is now established as both a model for understanding protein glycosylation and a biotechnological tool for in vivo generation of glycoconjugates (5, 12, 14, 50). However, genome sequence data indicate that not only C. jejuni but all Campylobacter species contain a gene cluster encoding a putative N-linked protein glycosylation system. In this study, we have demonstrated the functionality of these Campylobacter N-linked protein glycosylation systems and structurally characterized the associated N-linked glycans, thereby highlighting their conserved and more variable features.

In order to investigate N-linked glycosylation processes in multiple Campylobacter species, we employed a simple in vitro peptide glycosylation assay requiring only a source of bacterial membranes containing both an LLO and the integral membrane OTase protein along with a target peptide with the appropriate glycosylation sequon (21). An important assumption with this assay is that only LLOs that are involved in N-linked protein glycosylation will be used as donors by the OTase rather than other LLOs involved in glycolipid biosynthesis, for example. Though we have not formally excluded this possibility, the data obtained strongly suggest that this is a valid assumption. With this approach, we have characterized in vitro-generated, N-linked glycan from nine Campylobacter species. The observed varying levels of peptide glycosylation activity with membranes from different Campylobacter species (Fig. 3A) may have a number of explanations, but given that for most species this activity could be significantly increased by the addition of membranes derived from an E. coli strain expressing C. jejuni pglB (Fig. 3B), we suggest that either a lack of sufficient native PglB or the C. jejuni-optimized peptide employed (4) is not optimal for all Campylobacter PglB proteins. Using this assay, it was also clear that an acidic residue at the −2 position of the N-glycosylation sequon required for glycosylation in C. jejuni (23) is also necessary for glycosylation in all other Campylobacter species (data not shown), although this was not absolutely required by C. lari in vivo (41). The principal limitation of the peptide glycosylation assay as employed herein is that the analytical scale of the in vitro reactions limits further structural investigation of N-linked glycan by nuclear magnetic resonance approaches; however, in all cases, sufficient glycopeptide was generated for structural analysis by MS.

MS-based analyses indicated that a conserved feature of all of the Campylobacter N-linked glycan structures elucidated was the presence of a 228-Da residue, assumed to be DATDH, as the reducing-end sugar. In C. jejuni, the 228-Da residue is the DATDH sugar bacillosamine, synthesized from GlcNAc by-products of the pglDEF genes and transferred onto a lipid carrier by the glycosyltransferase PglC (38, 40). Orthologues of the C. jejuni pglCDEF genes are present in all Campylobacter genome sequences (Fig. 1), suggesting that the 228-Da DATDH residue in these species is also bacillosamine, although this requires verification. At the nonreducing end of the in vitro-generated, N-linked glycans, MS data demonstrated structural variation that correlated to some degree with inferred evolutionary relationships of the corresponding bacterial species (Fig. 1A and 5). Thus, N-linked glycans from relatively closely related Campylobacter species (C. jejuni, C. coli, C. upsaliensis, and C. helveticus) were indistinguishable by MS while those from species more distantly related to this group (C. fetus, C. hyointestinalis, C. lanienae, C. sputorum, and C. concisus) displayed differing arrangements of Hex and HexNAc sugars at the nonreducing terminus, with the C. concisus N-linked glycan containing a residue of 217 Da that is not found in other Campylobacter species (Fig. 5). The N-linked glycans from species of the genera Helicobacter and Wolinella (Fig. 5) display further structural diversity, including the presence of unusual sugar residues of 216, 217, and 232 Da (15). A sugar residue of 216 Da was detected in the O-antigens of Francisella tularensis (48), Pseudomonas aeruginosa O6 (20), and E. coli O121 (19) and is known to be an N-acetylhexuronamide (HexNAcAN). In P. aeruginosa, HexNAcAN is synthesized from nucleotide-activated HexNAc through conversion by WbpO to an N-acetylhexuronic acid (HexNAcA) of 217 Da, followed by conversion to the 216-Da HexNAcAN residue by the amidotransferase WbpS (19, 33, 52). While we have no further structural data on the 216- and 217-Da sugars from H. pullorum and W. succinogenes N-linked glycans, both species possess orthologues of the P. aeruginosa wbpO and wbpS genes located close to orthologues of C. jejuni pgl genes. Furthermore, the presence of a wbpO, but not a wbpS, orthologue in C. concisus (Fig. 1B) is consistent with a 217-Da (presumably HexNAcA) but not a 216-Da residue in the N-linked glycan.

Fig 5.

Schematic overview of epsilonproteobacterial N-linked glycan structures. The C. jejuni N-linked glycan structure is from reference 51, the C. lari structure is from reference 41 and this study, the H. pullorum structure is from reference 15, and the H. winghamensis and W. succinogenes structures are unpublished data from our laboratory. All other structures are from this study. The abbreviated species names C. jej., C. col., C. ups., C. hel., C. fet., C. hyo., C. lar., C. lan., C. spu., C. con., H. pul., H. win., and W. suc. correspond to C. jejuni, C. coli, C. upsaliensis, C. helveticus, C. fetus, C. hyointestinalis, C. lari, C. lanienae, C. sputorum, C. concisus, H. pullorum, H. winghamensis, and W. succinogenes, respectively. Unusual sugar residues are indicated by their respective masses in daltons.

Structural data presented herein indicate that Campylobacter N-linked glycan structures vary at the species level and above. There is also little or no evidence from the now considerable number of C. jejuni genome sequences of genetic variation at the pgl locus, suggesting intraspecies conservation of the N-linked heptasaccharide. This is in contrast to the intraspecies and even intrastrain structural variation of glycan components of capsular polysaccharide, lipooligosaccharide (LOS), and the flagellin glycoprotein in C. jejuni (8). This intraspecies structural variation is due to differences in gene content among strains of C. jejuni within the specific loci responsible for the biosynthesis of LOS, capsule, and the glycan component of flagellin glycoprotein. These loci also contain many homopolymeric-tract-containing, phase-variable genes encoding products involved in the biosynthesis of these structures, so generating intrastrain structural variation. There is no evidence of phase-variable genes within Campylobacter pgl loci. This relatively limited structural diversity of Campylobacter N-linked glycans indicates that they are not under the same degree of selective pressure as capsule, LOS, and flagellin glycan. This relative structural conservation also makes N-linked glycans potential subunit vaccine components and targets for antibody-based detection methods. For these approaches, the multiplicity of N-linked glycan-containing proteins and their location at the bacterial surface are advantageous. Despite this relative conservation of Campylobacter N-linked glycosylation systems, there appears to be limited conservation among target glycoproteins. Thus, BLAST searches with each of the 55 known C. jejuni NCTC 11168 glycoprotein amino acid sequences against predicted proteomes from available genome sequences of C. coli, C. lari, C. upsaliensis, C. fetus, and C. concisus identified 39 homologues present in all of the species. However, only 15 of these proteins also have an extended N-glycosylation sequon of D/E-X-N-X-S/T within the predicted amino acid sequences from all six Campylobacter species. This suggests that there is only partial overlap in the glycoproteomes of Campylobacter species.

The major biotechnological application of the C. jejuni N-linked protein glycosylation technology is in developing systems for in vivo glycoconjugate synthesis (12, 14, 31). The C. jejuni OTase PglB is central to this approach, and our data demonstrate that there are other PglB proteins from non-C. jejuni campylobacters that may be similarly useful. This in vivo glycoconjugate synthesis technology is limited primarily by the restricted capacity of the C. jejuni PglB OTase to transfer only glycans with specific sugars at the reducing end (5, 49). Our data on the structure of Campylobacter N-linked glycans, specifically, the universal presence of a 228-Da sugar residue at the reducing end, suggest that all Campylobacter PglB proteins may have similar limited substrate specificity. Thus, characterization of N-linked OTases present in species of Deltaproteobacteria and Epsilonproteobacteria but more distantly related to the Campylobacter genus may be more productive in the search for alternate substrate specificities of biotechnological potential. However, it may well be worthwhile to evaluate individual Campylobacter N-linked OTases for increased activity or yield in specific biotechnological applications.