Characterization of N-Linked Protein Glycosylation in Helicobacter pullorum

Abstract

The first bacterial N-linked glycosylation system was discovered in Campylobacter jejuni, and the key enzyme involved in the coupling of glycan to asparagine residues within the acceptor sequon of the glycoprotein is the oligosaccharyltransferase PglB. Emerging genome sequence data have revealed that pglB orthologues are present in a subset of species from the Deltaproteobacteria and Epsilonproteobacteria, including three Helicobacter species: H. pullorum, H. canadensis, and H. winghamensis. In contrast to C. jejuni, in which a single pglB gene is located within a larger gene cluster encoding the enzymes required for the biosynthesis of the N-linked glycan, these Helicobacter species contain two unrelated pglB genes (pglB1 and pglB2), neither of which is located within a larger locus involved in protein glycosylation. In complementation experiments, the H. pullorum PglB1 protein, but not PglB2, was able to transfer C. jejuni N-linked glycan onto an acceptor protein in Escherichia coli. Analysis of the characterized C. jejuni N-glycosylation system with an in vitro oligosaccharyltransferase assay followed by matrix-assisted laser desorption ionization (MALDI) mass spectrometry demonstrated the utility of this approach, and when applied to H. pullorum, PglB1-dependent N glycosylation with a linear pentasaccharide was observed. This reaction required an acidic residue at the −2 position of the N-glycosylation sequon, as for C. jejuni. Attempted insertional knockout mutagenesis of the H. pullorum pglB2 gene was unsuccessful, suggesting that it is essential. These first data on N-linked glycosylation in a second bacterial species demonstrate the similarities to, and fundamental differences from, the well-studied C. jejuni system.

Glycosylation is one of the most common protein modifications, and eukaryotes glycosylate many of their secreted proteins with asparagine or N-linked glycans. This process is thought to have diverse roles in protein folding, quality control, protein secretion, and sorting (13). Eukaryotic glycosylation takes place at the luminal side of the endoplasmic reticulum (ER) membrane, where a preassembled oligosaccharide is transferred from a lipid carrier to asparagine residues within an N-X-S/T consensus sequence, where X can be any amino acid except proline (19). The coupling of glycan to the protein takes place cotranslationally as nascent polypeptide chains cross the ER membrane via a translocon apparatus (5). This reaction involves a protein complex of at least eight subunits (49), with the STT3 protein (50, 52) apparently acting as the central enzyme in the process of N-linked protein glycosylation (29, 48). The STT3 protein consists of an amino terminus with multiple membrane-spanning domains and a carboxy-terminal region containing the highly conserved WWDYG amino acid sequence motif (15).

The first prokaryotic glycoproteins were described for archaeal species over 30 years ago (26), and for some time it was thought that protein glycosylation was a eukaryotic and archaeal, but not a bacterial, trait. However, there are now many examples of protein glycosylation in species from the domain Bacteria. For example, general O-linked protein glycosylation systems in which functionally diverse sets of proteins are glycosylated via a single pathway have recently been identified in Neisseria and Bacteroides spp. (8, 21, 44). The most-well-characterized bacterial species with respect to protein glycosylation is the enteropathogen Campylobacter jejuni, which encodes an O-linked system that glycosylates the flagellin protein of the flagellar filament along with the first described bacterial N-linked glycosylation system (39).

The C. jejuni N-linked glycosylation pathway is encoded by genes from a single protein glycosylation, or pgl, locus (38). The glycosylation reaction is thought to occur at the periplasmic face of the bacterial inner membrane mediated by the product of the STT3 orthologue pglB (46). The C. jejuni heptasaccharide glycan is assembled on a lipid carrier in the cytoplasm through the action of glycosyltransferases encoded by the pglA, pglC, pglH, pglJ, and pglI genes (11, 12, 24, 31). This lipid-linked oligosaccharide (LLO) is then “flipped” into the periplasm by the pglK gene product, or “flippase” (1), and transferred by PglB onto an asparagine residue within an extended D/E-X-N-X-S/T sequon (19). Many C. jejuni periplasmic and surface proteins of diverse function are N glycosylated (51), yet the function of glycosylation remains elusive. Unlike in eukaryotes, this process occurs posttranslationally, and the surface location of the sequon in folded proteins appears to be required for glycosylation (20).

The C. jejuni pgl gene locus can be transferred into Escherichia coli, and the corresponding gene products will function to transfer the heptasaccharide onto asparagine residues of coexpressed C. jejuni glycoproteins as well as non-C. jejuni proteins containing the appropriately located acceptor sequon (19, 46). When alternative lipid-linked glycans are present, such as those involved in lipopolysaccharide biosynthesis, glycans with diverse structure can also be transferred onto proteins (7). Although there are limitations, particularly with regard to the apparent structural requirement for an acetamido group on the C-2 carbon of the reducing end sugar (7, 47), this is still a significant advance toward tractable in vivo systems for glycoconjugate synthesis. The identification and characterization of further bacterial PglB proteins with potentially diverse properties would considerably expand the utility of such systems. Data from genome sequencing indicate that pglB orthologues are found in species closely related to C. jejuni, such as Campylobacter coli, Campylobacter lari, and Campylobacter upsaliensis (40), as well as in the more distantly related species Wolinella succinogenes (2). These species are members of the phylogenetic grouping known as the epsilon subdivision of the Proteobacteria, or Epsilonproteobacteria, consisting of the well-established genera Campylobacter, Helicobacter, Arcobacter, and Wolinella, which are often associated with human and animal hosts, as well as a number of newly recognized groupings of environmental bacteria often found in sulfidic environments (3). However, not all species of Epsilonproteobacteria contain pglB orthologues, and until recently, all characterized Helicobacter species lacked pglB genes.

Given the considerable interest in exploiting bacterial protein glycosylation, especially the C. jejuni N-linked glycosylation system, for generating glycoconjugates of biotechnological and therapeutic potential, the functional characterization of newly discovered pglB orthologues is a priority. In this report we describe the application of an in vitro oligosaccharyltransferase assay to investigate N-linked glycosylation initially in C. jejuni, where the utility of this approach was demonstrated, and then in Helicobacter pullorum, demonstrating that one of the two H. pullorum PglB enzymes is responsible for N-linked protein glycosylation with a pentasaccharide glycan.

MATERIALS AND METHODS

Bacterial strains and plasmids.

Escherichia coli strains were grown in Luria-Bertani (LB) broth or on LB agar plates. Campylobacter jejuni NCTC 11168 and Helicobacter pullorum NCTC 12824 strains were grown on Columbia agar containing 5% defibrinated horse blood (TCS Biosciences) at 42°C in a modified atmosphere consisting of 85% N₂, 10% CO₂, and 5% O₂ generated with a VA500 workstation (Don Whitley Ltd.).

Attempted construction of H. pullorum pglB mutants.

An approximately 750-bp internal fragment of pglB1 was PCR amplified with primers pglB1F (5′-GCT TTT TTG GGC TAG TTT TGC-3′) and pglB1R (5′-CAC AAA TGC GGG AAT ACG CT-3′). Similarly a 1,992-bp internal fragment of pglB2 was amplified with primers pglB2F (5′-ATT CAA GAG TTG TTG AGC GC-3′) and pglB2R (5′-TAC AAT CTC TCT CCC TTC CC-3′). Both amplicons were cloned into plasmid pGEM-T Easy, and a kanamycin resistance cassette (aphA) lacking a transcriptional terminator (42) was ligated into HindIII sites within the central region of the cloned pglB fragments. The resultant plasmids were electroporated into electrocompetent H. pullorum cells as described previously (43). Genomic DNA was isolated from kanamycin-resistant transformants by using the ArchivePure DNA Cell/Tissue kit (Flowgen), and integration by a double-crossover event was checked by PCR.

Construction of an H. pullorum pglDEF mutant.

The pglD, pglE, and pglF gene homologues were disrupted in a single step. A 5′ region extending upstream of pglD (pglDEF-U) was PCR amplified with primers pglDEF-UF (5′-CTA TCA GAA ATA GGA TAA G-3′) and pglDEF-UR (5′-AGG AAG CTT GAT GCT TGG CAC CTC AAA AC-3′), incorporating a 3′ HindIII site (underlined). Similarly, a 3′ region extending downstream of pglF (pglDEF-D) was PCR amplified by using primers pglDEF-DF (5′-CTG AAG CTT GCA GGC TTG TTT TGC-3′) and pglDEF-DR (5′-TGC GCT TAA AAT ATA AGG C-3′) to include a 5′ HindIII site (underlined). The two PCR products were mixed in equimolar concentrations, digested with HindIII, cleaned, ligated, and used as a template DNA for PCR using primers pglDEF-UF and pglDEF-DR. The resultant PCR product was cloned into pGEM-T Easy, and a kanamycin resistance cassette was inserted at the HindIII site. The plasmid was used to transform H. pullorum cells, and correct integration was confirmed as described above for the pglB genes. The resulting strain possesses only the 5′ 321 bp of the pglD gene, no pglE gene, and only the 3′ 359 bp of the pglF gene.

Preparation of solubilized membranes.

Solubilized lipid-linked oligosaccharide-containing membrane fractions were prepared as described previously (18). Campylobacter jejuni and H. pullorum were grown on blood agar plates and harvested after 2 and 4 days, respectively. To produce E. coli membranes containing C. jejuni PglB, strain SØ874 containing plasmid pMAF10 (7) was grown at 37°C, with shaking, in LB agar to an optical density at 600 nm (OD₆₀₀) of 0.5, whereupon cultures were induced with 0.2% arabinose for 16 h, before harvesting by centrifugation. Cells were resuspended in a solution containing 50 mM Tris-HCl (pH 7.0) and 25 mM NaCl and disrupted by using a French press (Thermo Scientific), and unbroken cells were removed by two rounds of centrifugation at 40,000 × g for 20 min at 4°C. The supernatant was ultracentrifuged at 100,000 × g for 1 h at 4°C, and the resulting membrane pellet was resuspended in a solution containing 50 mM Tris-HCl (pH 7.0), 25 mM NaCl, and 1% Triton X-100 (Sigma) at 2 ml per gram of starting wet cell pellet. Following a 1-h incubation at room temperature, detergent-solubilized samples were ultracentrifuged at 100,000 × g for 1 h at 4°C, and the supernatants were employed in the assay described below.

Oligosaccharyltransferase assay.

Assays were performed essentially as described previously (18). Typically, a solution containing 3 μl of 30 μM fluorescently labeled peptide, 1 μl of 150 mM MnCl₂, and 26 μl of solubilized membrane preparation was incubated at 30°C overnight with agitation. Reactions were stopped by the addition of 10 μl 4× SDS-PAGE sample buffer to the mixture, the mixture was heated to 95°C for 5 min, and 20-μl samples were analyzed by Tricine-SDS-PAGE (34) in minigels (8 cm by 8 cm) consisting of a 16.5% resolving gel containing 6 M urea, a 10% spacer gel, and a 4% stacking gel. Typical run conditions were 30 V for 30 min followed by 200 V for 60 min. Fluorescent products were visualized by using a Typhoon Trio Plus imager (GE Healthcare) with excitation at 532 nm (green laser) and a 526-nm SP emission filter.

Purification and MALDI-MS analysis of biotinylated peptides.

Biotinylated peptides were purified from oligosaccharyltransferase assay mixtures by using streptavidin-coated magnetic beads (Pierce) as described previously (10). Following three washes in 50 mM Tris-HCl (pH 7.4) containing 0.1% bovine serum albumin (BSA), beads were resuspended in the assay mixture described above and incubated for 30 min at room temperature. Beads were then washed twice in buffer A (50 mM Tris-HCl [pH 7.4], 1 mM dithiothreitol [DTT], 0.1% BSA), twice in buffer A containing 0.1% SDS, twice in buffer A containing 1 M NaCl, and, finally, three times in 1 mM DTT. Beads were resuspended in 25 μl H₂O containing a 2-fold excess of free biotin and incubated at room temperature for 15 min. Following three washes in milliQ-H₂O, peptides were eluted from the beads by either boiling in 1× SDS-PAGE loading buffer for 5 min or incubation in a 50 mM α-cyano-4-hydroxycinnamic acid (in acetonitrile-H₂O at a dilution of 4:1, vol/vol) matrix-assisted laser desorption ionization (MALDI)-mass spectrometry (MS) matrix. Peptides eluted in the MALDI-MS matrix were separated from magnetic beads by using a magnet and spotted directly onto a metal MALDI target plate. MALDI-tandem time of flight (TOF/TOF)-MS and tandem MS (MS/MS) spectra were acquired by using a Bruker Ultraflex II mass spectrometer in the positive-ion reflection mode. Spectra were viewed and analyzed by using FlexAnalysis 3.0 software (Bruker Daltonics).

Cloning and expression of Cj0114.

The Cj0114 coding sequence was PCR amplified from C. jejuni NCTC 11168 genomic DNA with primers CJ0114-F (5′-ATG AAA AAA ATA TTC ACA GTA GCT C-3′) and CJ0114-R (5′-TTA GTG ATG GTG ATG GTG ATG TTT TCT ATT AGG TGA AGC TTT TG-3′ [the underlined sequence encodes a 6×His tag]). The amplicon was cloned into pETBlue-1 (Novagen), inserts in the correct orientation for expression from the T7 promoter were identified by PCR and sequence analysis, and the resultant plasmid (pET0114) was transformed into E. coli BL21-AI cells (Invitrogen). Following growth in LB medium to an OD₆₀₀ of 0.5, protein expression was induced with 1 mM isopropyl-β-d-thiogalactopyranoside (IPTG) for 4 h, cells were harvested, and His-tagged Cj0114 was detected with penta-His antibody (Qiagen) on Western blots of whole-cell lysates resolved on 10% NuPAGE Novex Bis-Tris gels (Invitrogen).

Cloning and expression of Helicobacter pullorum pglB genes.

The pglB1 coding sequence was amplified from H. pullorum genomic DNA with primers HppglB1EcoRI (5′-AAT GAA TTC TTG TTG ATA GCG GTA TTT GT-3′ [boldface type indicates an EcoRI site]) and HppglB1NcoI-HA (5′-AAT CCA TGG TTA AGC GTA ATC TGG AAC ATC GTA TGG GTA TTG TTT GAG TTT ATA AAT C-3′ [the underlined sequence encodes a hemagglutinin {HA} tag, and boldface type indicates an NcoI site]). The pglB2 coding sequence was amplified from H. pullorum genomic DNA with primers HppglB2EcoRI (5′-AAT GAA TTC TTG TTG AGC GCA AAA AGT ATG-3′ [boldface type indicates an EcoRI site]) and HppglB2NcoI-HA (5′-AAT CCA TGG TTA AGC GTA ATC TGG AAC ATC GTA TGG GTA TTT CCT CAC TTG ATA TAC CAC-3′ [the underlined sequence encodes an HA tag, and boldface type indicates an NcoI site]). Both the pglB1 and pglB2 amplicons were digested with EcoRI/NcoI and cloned into EcoRI/NcoI-digested pMLBAD (22) to generate pMLHP1 and pMLHP2, respectively. The expression of HA-tagged PglB proteins was confirmed by Western blot analysis of whole-cell lysates with an anti-HA-horseradish peroxidase (HRP)-coupled antibody (Roche).

Expression and purification of Cj0114.

Electrocompetent E. coli SØ874 (28) cells were transformed with pET0114 and the pglB-containing plasmids described above. Following growth in LB medium with 0.1% glucose to an OD₆₀₀ of 0.5, protein expression was induced with 1 mM IPTG and 0.2% arabinose for 16 h. Cells were harvested and lysed with BugBuster (Novagen) in a solution containing 50 mM NaH₂PO₄, 300 mM NaCl, and 10 mM imidazole (pH 8.0) supplemented with 0.1% Tween, 1 mg/ml lysozyme, and 1 μl/ml Benzonase nuclease (Novagen). Following incubation for 30 min at room temperature with shaking, cell debris was removed by centrifugation at 13,000 rpm. Fifty microliters of Ni-nitrilotriacetic acid (NTA) agarose (Qiagen) was added to the supernatant, and the mixture was incubated for 1 h at 4°C with shaking. The suspension was centrifuged at 5,000 rpm for 2 min through a spin filter column (Fisher), and the column was washed five times with 500 μl of a solution containing 50 mM NaH₂PO₄, 300 mM NaCl, 20 mM imidazole (pH 8.0), and 0.1% Tween, and bound protein was eluted with 50 μl of a solution containing 50 mM NaH₂PO₄, 300 mM NaCl, 250 mM imidazole (pH 8.0), and 0.1% Tween.

Western blot detection of Cj0114.

Purified Cj0114-His was heated to 100°C with 1× SDS-PAGE loading buffer, separated by SDS-PAGE, transferred onto nitrocellulose, and detected with penta-His antibody (Qiagen) according to the manufacturer's instructions.

RESULTS

N-linked oligosaccharyltransferases in the Epsilonproteobacteria.

The key enzyme in the N-linked glycosylation reaction is the oligosaccharyltransferase encoded in C. jejuni by the pglB gene (11, 46). A BLAST search of the nonredundant protein sequence database with the C. jejuni PglB amino acid sequence retrieved over 30 bacterial pglB orthologues. All predicted PglB sequences contained multiple membrane-spanning domains at the N terminus and a C-terminal domain containing the conserved sequence motif WWDYG, which is thought to be involved in catalysis (46). Notably, all bacterial pglB orthologues were from species belonging to the epsilon or delta subdivision of the Proteobacteria. All Campylobacter species for which genome sequence data are available, along with the related species Wolinella succinogenes (2), contain pglB orthologues, with three phylogenetically related Campylobacter species (Campylobacter concisus, Campylobacter curvus, and Campylobacter gracilis) containing two pglB orthologues. A number of environmental species from the Deltaproteobacteria and Epsilonproteobacteria, including sulfate-reducing bacteria and species isolated from deep-sea thermal vents, also contain pglB orthologues (4, 27, 33, 36).

Among Helicobacter species, genomes of the well-studied H. pylori (41) and less-well-characterized Helicobacter species, such as H. acinonychis, H. mustelae, H. hepaticus (37), H. cinaedi, and H. bilis, do not contain pglB orthologues. However, using PCR with degenerate primers based on C. jejuni and W. succinogenes pglB gene sequences, we identified pglB-like genes in H. canadensis and H. pullorum (J. L. Fothergill, A. J. Lawson, and D. Linton, unpublished data), and in order to further characterize these genes, we sequenced the genomes of both species (25; our unpublished data). More recently, draft genome sequences of a second strain of H. pullorum and of H. canadensis along with the closely related species H. winghamensis have been released (http://www.broadinstitute.org/annotation/genome/Helicobacter_group/MultiHome.html). All three of these phylogenetically related Helicobacter species possess two pglB orthologues (pglB1 and pglB2) located in different regions of the bacterial chromosome. The levels of sequence identity between predicted Helicobacter PglB1 and PglB2 amino acid sequences are relatively low, with the H. pullorum PglB1 and PglB2 sequences having only 23% sequence identity (Fig. 1).

FIG. 1.

Schematic representation of the C. jejuni pgl gene N-glycosylation locus and orthologues present in H. pullorum. The percent values below open reading frames are the levels of identity between amino acid sequences encoded by H. pullorum and C. jejuni pgl gene orthologues and between H. pullorum PglB1 and PglB2. Arrows representing genes are not to scale.

In the C. jejuni N-linked glycosylation system, all genes necessary for the glycosylation reaction are located in a single locus (46) (Fig. 1), and similar arrangements have been found for other Campylobacter species and W. succinogenes (2). However, in H. pullorum, H. canadensis, and H. winghamensis, neither of the pglB genes is located near orthologues of the C. jejuni pgl genes (Fig. 1). Indeed, Helicobacter pglB genes are not physically associated with any genes encoding proteins potentially involved in glycosylation, such as glycosyltransferases or enzymes involved in sugar biosynthesis. Thus, there is no single “protein glycosylation locus” in these Helicobacter species, as found for C. jejuni (38, 46) and seemingly other Campylobacter species. Orthologues of other C. jejuni pgl genes are present in the pglB-containing Helicobacter species but at three further distinct loci: one locus containing pglA and pglC orthologues; a second containing pglD, pglE, pglF, and pglH orthologues; and a third containing pglI and pglJ orthologues in association with an additional predicted glycosyltransferase (termed pglL) and an orthologue of the C. jejuni pglK gene encoding a predicted ABC transporter (Fig. 1). The multiplicity of Helicobacter pglB genes and their novel arrangement led us to investigate their function in more detail.

The H. pullorum pglB1, but not the pglB2, gene is able to partially complement the C. jejuni pglB gene.

To investigate H. pullorum PglB function, we exploited the capacity of the C. jejuni N-linked glycosylation locus to function in E. coli (46). The target C. jejuni glycoprotein (Cj0114) contains four potential N-glycosylation sites (Fig. 2 A). The expression of the Cj0114 gene in E. coli along with a functional C. jejuni N-linked protein glycosylation locus produced a tetraglycosylated protein (Fig. 2B). In order to test for the complementation of C. jejuni PglB function, a version of the C. jejuni N-linked glycosylation locus with a kanamycin cassette inserted into the pglB gene (24) was coexpressed in E. coli along with the Cj0114 gene, and this abolished N glycosylation (Fig. 2B). When the H. pullorum pglB1 gene was introduced into this strain, two additional bands of decreased mobility were produced (Fig. 2B), indicating that H. pullorum PglB1 has oligosaccharyltransferase activity that is able to transfer the C. jejuni N-linked glycan onto Cj0114. In contrast, the H. pullorum PglB2 protein was unable to N glycosylate the Cj0114 protein with the C. jejuni N-linked glycan (data not shown).

FIG. 2.

Complementation analysis of H. pullorum pglB1. (A) The C. jejuni Cj0114 gene product with four potential N-glycosylation sites (boldface type) within C. jejuni N-glycosylation sequons (italic). (B) The C. jejuni Cj0114 gene on plasmid pET0114 and the site-directed mutants indicated were expressed in E. coli cells along with either the complete C. jejuni pgl locus on plasmid pACYCpgl containing all genes required for biosynthesis and the transfer of the C. jejuni heptasaccharide (46) or a version (pACYCpglB::kan) lacking PglB function (24). The functional complementation of C. jejuni PglB function was tested by coexpressing the H. pullorum pglB1 gene on plasmid pMLHP1. The hexahistidine-tagged Cj0114 protein and glycosylated forms of decreased mobility were detected with anti-His antibodies.

To identify which two of the four Cj0114 N-glycosylation sites were glycosylated by H. pullorum PglB1, site-directed mutagenesis was used to change each of the four relevant asparagine residues to glutamine. Analysis of the H. pullorum PglB1-mediated glycosylation of each variant revealed that the sites N173 and N179 but not N101 and N155 were modified (Fig. 2B). These data demonstrate that H. pullorum PglB1 is an N-linked oligosaccharyltransferase with overlapping but distinct specificity for acceptor sequons compared to the C. jejuni PglB protein.

In order to further investigate H. pullorum pglB gene function, we attempted to inactivate both the pglB1 and pglB2 genes by employing methods used for C. jejuni and H. pylori (43). In six independent experiments, insertional knockout mutants were obtained for the pglB1 gene and analyzed as described below, but none were obtained for the pglB2 gene.

In vitro N-linked protein glycosylation in C. jejuni and H. pullorum.

A recently developed in vitro oligosaccharyltransferase assay (18) was employed to investigate the putative H. pullorum N-linked glycosylation system(s) and to compare it with the previously characterized C. jejuni N-linked glycosylation system. When a detergent-solubilized membrane preparation from C. jejuni NCTC 11168 was incubated with a 6-carboxyfluorescein (FAM)-labeled peptide (FAM-ADQNATA) containing the C. jejuni N-linked glycosylation acceptor sequon (6), two fluorescent products with reduced electrophoretic mobility compared to that of the peptide were generated (Fig. 3 A). Membranes from a C. jejuni pglB insertional knockout mutant produced only barely detectable amounts of the least mobile of these bands, confirming that this is a PglB-dependent process (Fig. 3A). The incubation of a 1:1 mixture of membranes from a pglB knockout mutant and the corresponding wild-type strain results in an almost complete conversion of the peptide to the lowest-mobility product (Fig. 3A). This is presumably because the LLO concentration is a limiting factor in the assay, and a strain lacking PglB activity accumulates LLO in the membrane. When this lowest-mobility band was excised and analyzed by MALDI-MS, it generated a peak of m/z 2,475 (data not shown), the predicted mass of a sodium ion adduct of the peptide FAM-ADQNATA modified with the characterized C. jejuni heptasaccharide (46, 51). Fluorescent peptides in which the Asn residue was replaced with a Gln or in which the charged residue Asp at the −2 position was replaced with an uncharged Ala residue were not modified (Fig. 3A), consistent with the known properties of the C. jejuni PglB-mediated protein glycosylation reaction (19, 30).

FIG. 3.

In vitro analysis of C. jejuni oligosaccharyltransferase activity. (A) Detergent-solubilized C. jejuni membrane preparations were assayed for oligosaccharyltransferase activity with peptides as indicated. Following Tricine-SDS-PAGE, the highest-mobility band in each lane is the unmodified peptide, indicated with an asterisk, and decreased-mobility bands are presumed derived glycosylated forms. The decreased mobility of the AQNAT-containing peptide relative to those of DQNAT- and DQQAT-containing peptides is due to the loss of a negatively charged residue. The C. jejuni NCTC 11168- and corresponding pglB knockout mutant-derived membrane preparations were mixed in a 1:1 ratio. (B) Products of the oligosaccharyltransferase assay with membrane preparations from wild-type strain C. jejuni NCTC 11168 and the corresponding insertional knockout mutants in the pglH, pglJ, and pglA genes.

When membranes from C. jejuni pglA, pglJ, and pglH glycosyltransferase insertional knockout mutants (23) were incubated with the peptide FAM-ADQNATA, peptide-derived products were generated, indicating the production and transfer of reduced-length oligosaccharides (Fig. 3B). The predicted glycans produced by the pglA, pglJ, and pglH knockout mutants are mono-, di-, and trisaccharides, respectively (11, 24), and this is entirely consistent with the mobility of the modified peptides produced (Fig. 3B). Interestingly, a minor product with a mobility similar to that produced by membranes from the pglJ knockout mutant was observed with membranes from the wild-type and the pglH knockout mutant strains (Fig. 3B), suggesting a pool of lipid-linked disaccharide on C. jejuni membranes.

When analyzed by using the oligosaccharyltransferase assay, an H. pullorum membrane preparation was also able to modify the FAM-ADQNAT peptide, producing a presumed glycopeptide with slightly reduced electrophoretic mobility in SDS-PAGE gels (Fig. 4). Membranes extracted from an H. pullorum pglB1 insertional knockout mutant did not generate the less electrophoretically mobile species, indicating that PglB1 is responsible for peptide modification (Fig. 4). In order to investigate whether the C. jejuni PglB enzyme was capable of transferring the H. pullorum glycan onto a peptide, a mixture of membrane preparations from an E. coli strain expressing C. jejuni pglB and an H. pullorum pglB1 knockout strain was tested for in vitro N glycosylation. The production of an intense band of reduced mobility similar to that produced by the H. pullorum wild-type strain confirmed that C. jejuni PglB is capable of transferring the H. pullorum glycan (Fig. 4). We further tested the role of the H. pullorum pglDEF gene products in the biosynthesis of the N-linked glycan by constructing an H. pullorum insertional knockout mutant lacking PglDEF function (see Materials and Methods). A membrane preparation from this strain retained the ability to produce a modified peptide with a mobility similar to that produced by the wild-type H. pullorum strain (Fig. 4). The replacement of the Asn residue with Gln abolished the production of the modified peptide by H. pullorum membranes, consistent with an N-linked glycosylation reaction (Fig. 4). The substitution of the Asp residue at the −2 position for an Ala residue also abolished the production of the less-mobile species (Fig. 4), indicating that H. pullorum PglB1, like C. jejuni PglB (19), requires a negatively charged residue at the −2 position of the acceptor sequon.

FIG. 4.

In vitro analysis of H. pullorum oligosaccharyltransferase activity. Detergent-solubilized membrane preparations from H. pullorum, H. pullorum pglB1::kan, H. pullorum pglDEF::kan, and E. coli pMAF10 (7) were assayed for oligosaccharyltransferase activity with fluorescent peptides as indicated.

Structural analysis of in vitro-generated glycopeptides.

To structurally characterize modified peptides generated by the oligosaccharyltransferase assay, a biotin-labeled fluorescent peptide (fluorescein isothiocyanate [FITC]-ADQNATAK-biotin, with a mass of 1,587 Da) was employed, and products were purified by using streptavidin-coated magnetic beads prior to mass spectrometry (see Materials and Methods). MALDI-MS analysis of the purified peptide from a reaction mixture containing C. jejuni membranes identified a peak at m/z 3,015 in the high-mass region of the spectrum that corresponds to the sodium ion of the peptide modified with the characterized C. jejuni heptasaccharide of 1,406 Da. MS/MS analysis of this peak produced a fragmentation series (Fig. 5 A) entirely consistent with the known C. jejuni heptasaccharide structure (Fig. 5C). The major peaks at m/z 2,604 and 2,626 correspond to the protonated and sodiated ions, respectively, of the glycopeptide lacking the FITC moiety, presumably due to the labile nature of FITC during MS analysis. Other major peaks corresponded to the sequential loss of sugar residues from the heptasaccharide due to cleavage at the glycosidic bond and are annotated as y ions. The peak at m/z 1,199, also present in the MS/MS fragmentation series produced by the H. pullorum-generated glycopeptide (Fig. 5B), corresponds to the y₀ ion of a biotinylated peptide lacking both glycan and FITC (Fig. 5A and C). These data confirm the utility of the in vitro oligosaccharyltransferase assay combined with tandem MALDI-MS approaches for investigating glycan structures of bacterial PglB-mediated glycosylation.

FIG. 5.

MALDI-MS/MS analysis of C. jejuni and H. pullorum N-linked glycans. (A) MS/MS spectrum of the m/z 3,015 precursor ion generated following incubation of C. jejuni membranes with a biotin-labeled fluorescent peptide (FITC-ADQNATAK-biotin, with a mass of 1,587 Da). Fragment ions resulting from the sequential loss of sugar residues are indicated in the spectrum. In both C. jejuni (A)- and H. pullorum (B)-derived spectra, a peak corresponding to the y ion of the biotinylated peptide lacking FITC was present at m/z 1,199. (B) MS/MS spectrum of the m/z 2,645 precursor ion generated following incubation of FITC-ADQNATAK-biotin with H. pullorum membranes. The peaks labeled with an asterisk (m/z 1,384 and 2,627) are likely generated by dehydration. The peak labeled with a cross at m/z 1,182, 17 Da less than the m/z 1,199 peak, is characteristic of the fragmentation of a side-chain amide bond of an N-linked glycan. a.u., arbitrary units. (C) The N-linked glycan structure of the C. jejuni heptasaccharide (51) is consistent with the fragmentation pattern of the glycopeptide as observed for A. (D) N-linked glycan structure of the H. pullorum-generated glycopeptide inferred from the fragmentation pattern in B.

Following MALDI-MS analysis of purified peptide reaction products from an in vitro oligosaccharyltransferase assay with H. pullorum membranes, a peak in the high-mass range of m/z 2,645 was observed. MS/MS analysis of this ion similarly generated a fragmentation series, as for the C. jejuni derived glycopeptide, corresponding to the sequential loss of sugars (Fig. 5B). The fragmentation pattern was consistent with the sequential loss of a 203-Da residue, almost certainly an N-acetylhexosamine (HexNAc) residue; a 217-Da residue; a second 217-Da residue; a 216-Da residue; and a second 203-Da or HexNAc residue as the reducing end sugar (Fig. 5B). This fragmentation pattern is consistent with a linear pentasaccharide attached to the peptide via a HexNAc residue (Fig. 5D). The presence of a peak corresponding to the biotinylated peptide at m/z 1,199 along with an associated peak at m/z 1,182 (Fig. 5B) is indicative of the cleavage of the side-chain amide bond of an Asn residue, confirming that the glycan is N linked. Similar spectra were generated when the modified peptide produced by the H. pullorum pglDEF knockout mutant was analyzed, confirming that the corresponding gene products are not required for the production of the N-linked pentasaccharide (data not shown).

DISCUSSION

Oligosaccharyltransferase is the key enzyme of the N-linked glycosylation process responsible for the transfer of oligosaccharide from lipid carrier to protein, and the discovery of two unlinked pglB genes per genome in three closely related Helicobacter species was therefore novel and intriguing. The attempted complementation of the C. jejuni pglB-encoded oligosaccharyltransferase activity with both H. pullorum pglB genes demonstrated that PglB1 partially complemented the production of two bands of decreased mobility compared to the four additional bands produced by C. jejuni PglB (Fig. 2B), while PglB2 did not complement (data not shown). This is convincing evidence for H. pullorum PglB1 functioning as an oligosaccharyltransferase but suggests some significant differences in activity compared to that of the C. jejuni PglB protein. Site-directed mutagenesis experiments demonstrated that of the four potential N-glycosylation sites (N101, N155, N173, and N179) glycosylated by the C. jejuni PglB enzyme, only two (N173 and N179) were glycosylated by the H. pullorum PglB1 enzyme (Fig. 2). We are currently further investigating the target specificity of the H. pullorum PglB1 enzyme.

We employed an in vitro assay to investigate N-linked glycosylation processes in both the well-characterized C. jejuni and completely uncharacterized H. pullorum systems. The PglB-mediated transfer of the C. jejuni heptasaccharide to the N-glycosylation acceptor sequon-containing peptide was observed, confirming the utility of this approach (Fig. 3 and 5). We further confirmed the requirement for an Asn residue and for a negatively charged residue (in this case an Asp) at the −2 position (Fig. 3). We have previously demonstrated that when the C. jejuni pgl locus was expressed in E. coli, versions of this locus with insertional knockouts of individual glycosyltransferases resulted in the transfer of mono-, di-, tri-, and hexasaccharides to the protein, depending on the individual roles of glycosyltransferases in heptasaccharide assembly (24). An analysis of similar glycosyltransferase knockout mutants in C. jejuni, however, did not result in the transfer of an array of glycan structures to the protein (16). Also, C. jejuni pglJ and pglH knockout mutants did not produce detectable LLOs with di- and trisaccharides, respectively (32), as might be predicted given their role in heptasaccharide assembly (11, 24). However, when membranes from C. jejuni glycosyltransferase mutants were extracted and employed in the in vitro peptide glycosylation assay described herein, fluorescent peptides of reduced electrophoretic mobility were observed, consistent with the transfer of incompletely assembled glycans to the peptide. Thus, LLOs with incomplete glycans are present in the membranes of these glycosyltransferase mutants, and PglB produced in C. jejuni is able to transfer these incomplete glycans onto the peptide. It should also be noted that the high-resolution SDS-PAGE system employed to separate glycopeptide reaction products is seemingly capable of discriminating between glycopeptides differing by only a single uncharged sugar. Thus, the peptide glycosylation assay is a simple and rapid screening system that can be applied to cultivable bacterial species for characterizing N-linked glycosylation processes using native LLO and oligosaccharyltransferase. Significantly, this approach is also suitable for investigating the role of individual glycosyltransferases in N-linked glycan assembly, enabling function to be rapidly determined.

We have demonstrated that in H. pullorum, one of the two pglB genes is required for in vitro peptide glycosylation. This reaction has some similarities to the C. jejuni N-linked glycosylation reaction in that the replacement of a negatively charged Asp residue at the −2 position of the N-glycosylation sequon with an Ala residue abolishes glycosylation. However, the structure of the H. pullorum N-linked pentasaccharide glycan is considerably different from that of the C. jejuni N-linked heptasaccharide. Notably, the H. pullorum glycan is attached to the peptide via a HexNAc residue rather than via the di-N-acetyl derivative of the uncommon diamino sugar bacillosamine (2,4-diamino-2,4,6-trideoxy-d-glucose) found in C. jejuni, although this maintains the rule that a 2′-acetamido group is present on the reducing end sugar of all N-glycans transferred by bacterial PglBs. This is despite the presence of pglD, pglE, and pglF orthologues in H. pullorum, which in C. jejuni encode enzymes responsible for bacillosamine biosynthesis from GlcNAc (31, 35), and the role of these genes in H. pullorum remains to be determined. The presence of a HexNAc at the reducing end is consistent with the ability of C. jejuni PglB to transfer the H. pullorum glycan to the peptide (Fig. 3), given that a 2′-acetamido group on the reducing end sugar is a key requirement for C. jejuni PglB activity (47). Given the observed significantly decreased electrophoretic mobility of the peptide coupled to the C. jejuni heptasaccharide (Fig. 3A), the H. pullorum PglB1-generated glycopeptide is more mobile than might be predicted for a peptide coupled to a pentasaccharide, suggesting that the glycan contains a negatively charged residue or residues. Indeed, when the biotin-labeled version of the fluorescent peptide was glycosylated in vitro, the glycopeptide was more mobile in SDS-PAGE gels than the unmodified peptide (data not shown). This negative charge is presumably due to presence of acidic sugars, with the 216- and/or 217-Da residues of the pentasaccharide as likely candidates, given that the 203-Da HexNAc residues are uncharged. Recently, a 217-Da residue was identified in the N-linked flagellar glycan of Methanococcus maripaludis as the uncharged sugar 2-acetamido-2,4,-dideoxy-5-O-methyl-α-l-erythro-hexos-5-ulo-1,5-pyranose (17), but to our knowledge, a 216-Da sugar has not been previously reported. A complete structural characterization of the H. pullorum N-linked glycan, including nuclear magnetic resonance (NMR) studies, will be required to determine the structure of the 216- and 217-Da sugar residues. The in vitro peptide glycosylation assay combined with mass spectrometry employed herein will also enable an investigation of the role for H. pullorum orthologues of C. jejuni pgl genes (Fig. 1) in the biosynthesis of the H. pullorum N-linked pentasaccharide.

The results from our study demonstrate that the in vitro N-glycosylation assay employed is a powerful approach not only to characterize novel N-glycans but also to rapidly determine the role of glycosyltransferases in N-glycan assembly. We have demonstrated the function of the H. pullorum PglB1 protein in the transfer of a pentasaccharide of novel structure to asparagine residues within the extended N-glycosylation sequon, as defined by studies of the C. jejuni N-glycosylation system. However, a number of key questions remain, in particular the role of the H. pullorum PglB2 protein. Our inability to construct an insertional knockout mutant suggests that the pglB2 gene is essential, but this will require further investigation. Other key questions include, what are the H. pullorum proteins glycosylated with the pentasaccharide by PglB1? The structural information obtained regarding the nature of the N-linked glycan may assist in identifying such proteins through lectin affinity-based purification. In summary, we have provided the first data characterizing an N-linked glycosylation pathway in a Helicobacter species, only the second bacterial N-linked glycosylation system described.