Structural insights from random mutagenesis of Campylobacter jejunioligosaccharyltransferase PglB
- Julian Ihssen†1,
- Michael Kowarik†2,
- Luzia Wiesli1,
- Renate Reiss1,
- Michael Wacker2 and
- Linda Thöny-Meyer1Email author
© Ihssen et al.; licensee BioMed Central Ltd. 2012
Received: 12 June 2012
Accepted: 13 September 2012
Published: 24 September 2012
Protein glycosylation is of fundamental importance in many biological systems. The discovery of N-glycosylation in bacteria and the functional expression of the N-oligosaccharyltransferase PglB of Campylobacter jejuni in Escherichia coli enabled the production of engineered glycoproteins and the study of the underlying molecular mechanisms. A particularly promising application for protein glycosylation in recombinant bacteria is the production of potent conjugate vaccines where polysaccharide antigens of pathogenic bacteria are covalently bound to immunogenic carrier proteins.
In this study capsular polysaccharides of the clinically relevant pathogen Staphylococcus aureus serotype 5 (CP5) were expressed in Escherichia coli and linked in vivo to a detoxified version of Pseudomonas aeruginosa exotoxin (EPA). We investigated which amino acids of the periplasmic domain of PglB are crucial for the glycosylation reaction using a newly established 96-well screening system enabling the relative quantification of glycoproteins by enzyme-linked immunosorbent assay. A random mutant library was generated by error-prone PCR and screened for inactivating amino acid substitutions. In addition to 15 inactive variants with amino acid changes within the previously known, strictly conserved WWDYG motif of N-oligosaccharyltransferases, 8 inactivating mutations mapped to a flexible loop in close vicinity of the amide nitrogen atom of the acceptor asparagine as revealed in the crystal structure of the homologous enzyme C. lari PglB. The importance of the conserved loop residue H479 for glycosylation was confirmed by site directed mutagenesis, while a change to alanine of the adjacent, non-conserved L480 had no effect. In addition, we investigated functional requirements in the so-called MIV motif of bacterial N-oligosaccharyltransferases. Amino acid residues I571 and V575, which had been postulated to interact with the acceptor peptide, were subjected to cassette saturation mutagenesis. With the exception of I571C only hydrophobic residues were found in active variants. Variant I571V performed equally well as the wild type, cysteine at the same position reduced glycoprotein yield slightly, while a change to phenylalanine reduced activity by a factor of three.
This study provides novel structure-function relationships for the periplasmic domain of the Campylobacter jejuni N-oligosaccharyltransferase PglB and describes procedures for generating and screening oligosaccharyltransferase mutant libraries in an engineered E. coli system.
Keywordsglycosylation oligosaccharyltransferase PglB Campylobacter jejuni random mutagenesis screening ELISA directed evolution conjugate vaccine
N-linked glycosylation is of fundamental importance for the biological activity of many eukaryotic proteins. Bacterial oligosaccharyltransferases (OST) perform a similar type of posttranslational modification in prokaryotic cells and thus hold great promise for producing proteins with defined glycosylation patterns in simple and cost-efficient expression systems such as Escherichia coli. The OST of Campylobacter jejuni (PglB Cj ) has been functionally expressed in E. coli and has been shown to have a relaxed specificity towards its lipid-linked substrate. Therefore it can be exploited, e.g., for the production of conjugate vaccines where polysaccharide antigens of bacterial pathogens are covalently coupled to carrier proteins [3, 4]. Polysaccharide-protein conjugates (conjugate vaccines) belong to the most effective measures for combating life-threatening infectious diseases . Diverse O-specific polysaccharides of the outer lipopolysaccharide layer of Gram-negative bacteria can be efficiently coupled to carrier proteins containing a surface-exposed D/E-X1-N-X2-T/S acceptor sequon by PglB Cj , provided that they contain an N-acetyl sugar at the reducing end . Examples are the O7, O9a, O16 and O157 antigens of E. coli, the O1 antigen of Shigella dysenteriae as well as the O11 antigen of Pseudomonas aeruginosa[3, 4, 6]. Production of conjugate vaccines by in vivo conjugation in E. coli is much simpler than with conventional chemical conjugation technologies, and it has already been demonstrated that glycoconjugates can be produced at a larger scale in fed-batch culture with yields of 18–24 mg L-1 of purified product . Recently, it was found that capsular polysaccharides of medically important Gram-positive bacterial pathogens such as Staphylococcus aureus can also be transferred to acceptor proteins by PglB Cj (M. Wacker and J. Lee, in preparation).
PglB Cj is a monomeric protein of 713 amino acids and comprises 11 predicted transmembrane helixes in the N-terminal two thirds of the polypeptide and a globular domain at the C-terminus which is soluble and exposed to the periplasm . The latter domain of PglB Cj was crystallized by Maita et al. . It harbors the highly conserved W457WDYG461 motif which is also found in the STT3 subunit of eukaryotic OST [2, 8]. When W458 and D459 were both replaced by alanine residues, PglB Cj was rendered inactive . The crystal structure of the soluble domain of PglB Cj revealed a kinked helix opposite of the WWDYG motif which contained the conserved residues M568, I571 and V575, subsequently termed “MIV” motif of bacterial OST [8, 9]. Mutational studies confirmed the importance of I571 for activity . PglB of Campylobacter lari, a bacterial OST with 56% amino acid identity to PglB Cj , also glycosylates recombinant proteins in the periplasm when over-expressed in E. coli. The crystal structure of full-length PglB C.lari was elucidated recently . The structure contained a divalent metal co-factor as well as a bound synthetic D-Q-N-A-T-p-nitro phenylalanine acceptor peptide. It revealed that both the transmembrane and the soluble periplasmic domains of PglB C.lari participate in substrate binding and catalysis . The strictly conserved residues W463-W464-D465 of PglB C.lari were shown to form a strong network of hydrogen bonds with the hydroxyl group of the threonine residue at position +2 relative to the glycosylated asparagine (N) of the acceptor peptide . This explains the strict requirement for either S or T in the D/E-X1-N-X2-S/T (X2: any amino acid except proline) consensus sequence of bacterial N-glycosylation sites [12, 13]. Furthermore, the crystal structure indicated that conserved residues D56, R147, D154, D156, E319, R331 and R375 of PglB C.lari are ligands of the acceptor peptide and/or the metal cofactor. Amino acid changes at D56, D154, E319 of PglB C.lari lead to partial or complete loss of activity . The relevance of D54 (equivalent to PglB C.lari D56) within the transmembrane domain of PglB Cj had been demonstrated previously (by alanine replacement) . A role of the MIV motif in substrate binding is also supported by the PglB C.lari structure because the position of the side chain of I572 (equivalent to PglB Cj I571) is in agreement with a hydrophobic interaction with the methyl group of the +2 T residue of the acceptor peptide, presumably leading to enhanced affinity for D-X1-N-X2-T acceptor sequons compared to D-X1-N-X2-S as previously shown by Chen et al. . From the full-length structure of PglB C.lari it was deduced that binding of acceptor proteins occurs from one side of the protein while undecaprenyl-pyrophosphate-linked oligosaccharide precursors access the active site from the other side . An elongated, flexible loop of the transmembrane domain containing the essential amino acid E319 was suggested to play a central role in catalysis . An in silico glycan docking experiment indicated that conserved Y468 (PglB Cj Y462) may interact with the N-acetyl group of the innermost sugar of the transferred oligosaccharide .
In this study we aimed at elucidating residues of the periplasmic domain of PglB Cj which are important for the formation of engineered glycoconjugates. Therefore, we established methodologies for the construction and screening of OST variant libraries in E. coli. Two mutant libraries generated by either error-prone PCR mutagenesis of the periplasmic domain or cassette mutagenesis of specific active site residues were analyzed for inactive, but PglB polypeptide producing variants. The obtained results were discussed in the context of the recently published crystal structure of the C. lari OST.
ELISA detection of glycoconjugates
N-linked glycosylation in recombinant E. coli is a powerful method for obtaining conjugates of polysaccharide antigens and immunogenic carrier proteins that can be used as safe and efficient vaccines against bacterial pathogens. A method for detection of glycoproteins in the periplasm of E.coli which is suitable for medium throughput screening has not yet been described, but is required for the analysis of pglB mutant libraries. Therefore, we developed an enzyme-linked immunosorbent assay (ELISA) in 96-well format and compared the new analytical method to the commonly used Western blot analysis.
Plasmids used in this study
Medium copy number vector for IPTG-inducible expression, lacI, P tac , ori: pACYC184
Kanamycin resistance cassette inserted into chloramphenicol resistance gene of pACT3
IPTG-inducible expression of PglB containing a hemagglutinin epitope tag (HA) at the C-terminal end, low copy number plasmid, lacI, P tac , ori: IncW
Similar to pGVXN114, but inactive PglB-HA W458A-D459A (PglB mut)
IPTG-inducible expression of PglB containing an HA tag at the C-terminal end, template plasmid for mutant libraries, medium copy number
Similar to pGVXN112Kan, but inactive PglB-HA W458A-D459A (PglB mut)
L-arabinose inducible expression of periplasmic, hexahistidine-tagged Pseudomonas aeruginosa exotoxoid A (EPA) with 2 engineered N-glycosylation sites, araC, P BAD , high copy number plasmid, ori: pBR322
Constitutive expression of CP5 capsular polysaccharides of Staphyloccus aureus, native promoter, low copy number plasmid, ori: IncPa
M. Wacker and J. Lee, in preparation
An alternative medium copy number expression vector for PglB Cj , pGVXN112Kan (Table 1) also enabled glycosylation of EPA, even though glycoprotein levels were lower (Figure 1A). When periplasmic proteins of control strains harboring the same combinations of EPA and CP5 producing plasmids and expressing mutated, inactive PglB (PglB mut: W458A-D459A) were analyzed, no specific bands were detected (Figure 1A). However, strong bands with the expected size of unglycosylated EPA (MW ≈ 70 kDa) were detected in periplasmic extracts of both wild-type PglB and mutated (mut) PglB strains when blots were analyzed with anti-EPA antibodies (Figure 1A, right panel). A ladder of proteins of a similar size range as the ladder observed in anti-CP5 blots was also seen for wild-type PglB samples on this blot (80–200 kDa). Anti-CP5 specific bands below the size of EPA were most likely due to glycosylated degradation products of EPA. Such degradation products were also observed when engineered EPA was glycosylated with Shigella O1 polysaccharides .
Next we tested whether glycosylation using this system can also be reliably detected by ELISA, as this method could be used for the analysis of a larger number of clones. The ELISA setup with the best signal-to-noise ratio for periplamic extracts of E. coli comprised anti-EPA capturing antibodies and anti-CP5 detection antibodies (sandwich ELISA). ELISA absorbance values were reproducibly and significantly higher for samples from wild-type PglB strains compared to samples from isogenic control strains expressing inactive PglB (Figure 1B). Background signals (PglB mut) were higher when the low copy number pglB expression plasmid was used than when the medium copy number, pACT3 derived plasmid was used (Figure 1B). The latter plasmid was used for the construction of pglB mutant libraries. In later experiments where mutants were re-analyzed in replicates, a second batch of anti-CP5 antibodies with a higher degree of purity was used, which lead to a reduction of background signals to ± 0.20 (measured against air, i.e., including absorbance of the polystyrene plate and buffer), independent of the type of pglB expression plasmid.
The ELISA results for shake flask cultures indicated that the method is suitable for detecting and quantifying glycoproteins in E. coli clones expressing different variants of PglB. The coefficient of variation of ELISA absorbance values was below 5% of the mean value for periplasmic extracts of strains expressing either wild-type or inactive PglB. The lower levels of glycoproteins produced by the strain with medium copy number plasmid pGVXN112Kan were reflected in lower values of ELISA absorbance (Figure 1B).
Screening system for OST mutant libraries
Based on the positive results with the sandwich ELISA assay for shake flask samples, we developed a 96-well cultivation and screening method for glycoprotein production in E. coli. Screening of library clones was performed in 96-well format using deep well plates (DWP) for cultivation followed by sucrose-lysozyme extraction of periplasmic proteins and relative quantification of EPA-CP5 glycoconjugates by sandwich ELISA. A test plate was inoculated with 90 clones of the EPA-CP5 producing, wild-type PglB strain (pGVXN112Kan) and with 6 clones of the negative control strain expressing inactive PglB mut (pGVXN113Kan). The ELISA signals of all wells containing the wild-type PglB strain were clearly separated from those wells which contained the inactive control strain (Additional file 1). The average absorbance (coefficient of variation) was 0.402 (5.1%) and 0.275 (4.5%) for wild-type and inactive PglB, respectively. The reproducibility and sensitivity of the DWP-ELISA screening assay was considered sufficiently high for detecting inactivating mutations in pglB mutant libraries.
Random mutagenesis of PglB Cj by error-prone PCR
Amino acid substitutions identified in inactive clones
The epPCR experiment did not yield any variants with reproducibly increased glycosylation efficiency; however, numerous variants with neutral single or double mutations were found, e.g., V470M, D705Y and K603E-S664R all of which exhibited EPA-CP5 ELISA signals similar to wild-type PglB. In contrast to residues mutated in inactive variants, only non-conserved residues were mutated in the latter cases.
Cassette mutagenesis of I571 and V575 (MIV motif)
Mutations tolerated at the MIV motif
Abs. 450 nm
% of wt
anti-HA, % of wt
The relative levels of EPA-CP5 and PglB were analyzed for each type of PglB variant in replicate DWP mini-cultures after re-transformation of the sequenced variant plasmids (Table 3). No significant difference to wild-type PglB was found for variants I571V and I571V-V575I, respectively (Table 3). In the case of I571V -V575L, a moderate, but significant reduction of EPA-CP5 biosynthesis was observed which was not due to reduced expression of PglB (Table 3). The level of EPA-CP5 detected by ELISA was reduced by a factor of three in variant I571F-V575L without any reduction of PglB expression, indicating that an F at position 571 has a negative effect on OST activity (Table 3). The double mutant I571C-V575I was characterized by a parallel, moderate reduction in the levels of both EPA-CP5 and PglB (Table 3). The alignment of a set of diverse, non-redundant amino acid sequences of bacterial PglB homologues indicates that I571 and V575 are strictly conserved in ε-Proteobacteria, but not in other species (Figure 6A). However, also in non-ε-Proteobacteria mostly hydrophobic amino acids occur at the corresponding positions (Figure 6A). Isoleucine and leucine at position PglB Cj V575 were found in active variants of the cassette mutagenesis experiment and also occur in natural PglB sequences (Figure 6A). The spatial orientation of the equivalent residues I572 and V576 of PglB C. lari towards the bound acceptor peptide is shown in Figure 6B. The aliphatic side chains of both amino acids are in close proximity to the methyl group of the threonine residue at the N + 2 position of the D-Q-N-A-T-p-nitro phenylalanine acceptor peptide (Figure 6B).
The use of bacterial OST in recombinant E. coli facilitates in vivo production of conjugates of polysaccharide antigens of pathogenic bacteria and immunogenic carrier proteins [3, 4, 18]. Recombinant E. coli equipped with the oligosaccharyltransferase PglB of C. jejuni is the most advanced system for the biotechnological production of N-linked conjugate vaccines.
In order to gain a better understanding of residues of PglB Cj required for the glycosylation reaction we studied the effect of random mutations on the formation of glycoconjugates composed of Pseudomonas aeruginosa exotoxoid and Staphylococcus aureus capsular polysaccharide CP5. Therefore, we established methods for constructing epPCR mutant libraries of recombinant PglB and for the screening of glycoprotein-producing E. coli in 96-well format. Growth and induction of pglB library clones in deep well plates followed by sandwich ELISA analysis of the periplasmic extracts allowed a reliable distinction of active and inactive OST variants.
The screening of an epPCR library where the periplasmic domain of PglB Cj had been mutagenized facilitated the identification of previously unknown residues which are relevant for OST activity. Apart from confirming the importance of the well-known, strictly conserved W457WDYG461 residues of PglB Cj , the epPCR experiment revealed that the region V464 to G481 immediately adjacent to this motif is also highly important (Table 2). In the partial crystal structure of PglB Cj and the recently revealed full-length structure of PglB C.lari , these amino acids form a flexible loop in close proximity to the side chains of tryptophan and aspartic acid residues of the WWDYG motif [8, 11]. Due to the spatial orientation of the bound acceptor peptide and the assumed reaction mechanism of PglB , this loop is believed to interact with sugar moieties of the glycan substrate. The highly conserved residue G481, which was hit several times in inactive epPCR mutants, could function as a helix breaker before the ascending helix starting at K482, thereby providing the loop D475-G481 (C. lari D481-G487) with sufficient flexibility for interaction with the oligo-/polysaccharide (Figure 5B). The inactivating mutations at G481 identified in our epPCR experiment presumably induce structural changes which disturb the interaction of loop residues such as H479 and D475 with the glycan substrate. In most homologous sequences which do not bear a glycine at this position, a helix-breaking glycine or proline residue is present at an adjacent position (Figure 5A). Our results further confirm the importance of the strictly conserved aspartic acid residue at position 475 (C. lari D481) for activity of PglB Cj as proposed by Maita et al. . In a study using radio-labeled undecaprenyl-pyrophosphate linked oligosacharides of C. jejuni at limiting concentrations and a synthetic Ac-D-Q-N-A-T-p-nitro phenylalanine acceptor peptide in excess, mutations D475E and D475A decreased glycosylation rates of PglB Cj by a factor of four and six, respectively . The negative effect of the mutations was strongest at low glycan substrate concentrations . Together with the reduced activity of a K478A mutant reported by the same authors these results support the importance of the flexible loop D475-G481 and its probable involvement in glycan binding as suggested by our data. Judged from its distance to the acceptor nitrogen atom (Figure 5B), the carboxylate group of D475 might be involved in a hydrogen bond network coordinating the C2 N-acetyl group of the innermost sugar of the transferred glycan. Similarly to eukaryotic OST, PglB Cj is believed to require an N-acetyl sugar at the reducing end for efficient transfer of poly- and oligosaccharides to acceptor proteins . The side chain of the strictly conserved tyrosine residue at position 462 (C. lari Y468) is in close proximity to D475 (C. lari D481) (Figure 5B). For C. lari Y468 an interaction with the N-acetyl group of the innermost sugar has been suggested by others based on in silico glycan docking . The importance of the loop D475-G481 is further backed by the phenotype of the H479Q variant generated in this study. The side chain of the equivalent residue H485 in PglB C.lari is spatially very close to the nucleophilic amide nitrogen of the bound D-Q-N-A-T-p-nitro phenylalanine acceptor peptide (Figure 5B). With few exceptions glutamine is the only other amino acid occurring at this position in bacterial PglB homologues (Figure 5A). The side chain of glutamine has a similar length as that of histidine and also contains an amino functionality. In PglB Cj glutamine seems to be able to substitute for the function of histidine at position 479 only to a limited degree. Interestingly, removal of the aliphatic side chain of the adjacent, non-conserved residue L481 by mutation to alanine had no effect on activity at all. This is in agreement with the spatial orientation of the side chain of the equivalent residue L486 in PglB C.lari pointing away from the acceptor peptide (Figure 5B).
In only one inactive clone with strong expression of pglB an amino acid change was found which mapped to a region outside of W457 to G481. The mutation V432F strongly reduced the glycoprotein specific ELISA signals, but the variant was not completely inactive as EPA-CP5 formation was still detectable. Judged from the position of the equivalent residue V438 in the crystal structure of PglB C.lari , the bulkier and less flexible phenyl side chain in the variant enzyme might interfere with the binding of D-X1-N-X2-T/S acceptor peptides. In particular, steric hindrance of the side chain of the glutamine residue at the −1 position in D-Q-N-X-T/S sequons could play a role (Figure 6B). In the engineered EPA used in this study as acceptor protein two glycosylation sites had been introduced in surface exposed loops, D-N-N-N-S and D-Q-N-R-T. It would be interesting to study whether glycosylation of the D-Q-N-R-T site is more affected than that of the other site and whether a change of −1 Q to amino acids with shorter side chains could compensate for the negative effect of PglB Cj V432F. V432 is not strictly conserved in PglB homologues; however, in ε-Proteobacteria a short “VF” motif is present at this position in most species (Figure 6A).
The so-called MIV motif of bacterial N-oligosaccharyltransferases (M568, I571, V575 in PglB Cj , Figure 6A) which is assumed to interact with the acceptor peptide was hit only once in the epPCR experiment. The side chains of the equivalent residues I572 and V576 of PglB C.lari are very close to the side chains of the WWDYG motif in the crystal structure (Figure 6B). Therefore, we specifically randomized these residues in PglB Cj using degenerate primers and screened the resulting focused library. Our results indicate that only few other, chemically similar amino acids are tolerated at these positions. Wild-type isoleucine and valine at position 571 exhibited the same phenotype. Presumably, both side chains allow for sufficient hydrophobic interaction with the methyl group of +2 T of the acceptor sequon as suggested for PglB C.lari . The strong preference for hydrophobic residues at I571 as well as V575 in active variants indicates that a hydrophobic surface promotes binding of the polypeptide chain of the acceptor protein. In addition to I572 and V576, numerous other surface-exposed amino acids of the acceptor peptide binding pocket are also hydrophobic in PglB C.lari : I317, M318, Y433, V438, Y466, Y468 and V575 . With the exception of M318 and V575 these residues are conserved in PglB Cj (data not shown). An elongated hydrophobic patch containing the WWDYG motif in its center was suggested to interact with acceptor polypeptides in the archaeal N-oligosaccharyltransferase of Pyrococcus furiosus.
Directed evolution, comprising multiple rounds of random mutagenesis, randomization of active site residues and recombination of beneficial mutations, has proven to be a powerful method for tuning enzymes to specific needs, provided that suitable screening systems are available [21–23]. Theoretically, it should be possible to evolve bacterial OSTs for enhanced activity towards non-native polysaccharide substrates, although no such studies have yet been reported in the literature. We did not detect any PglB variants with increased yields of glycoprotein in our libraries. Most likely, the number of screened epPCR clones was too low to find improved variants, which are expected to be very rare. Amino acid substitutions at the MIV motif were at best neutral, which indicates that it is difficult to further optimize interactions of D-X1-N-X2-S/T acceptor peptides with PglB. However, we cannot exclude the possibility that protein glycosylation in the applied E. coli system was limited by other factors than OST activity, e.g., biosynthesis of undecaprenyl-pyrophosphate linked polysaccharide precursors. Furthermore, the crystal structure of full-length PglB C.lari strongly suggests that residues of the membrane domain are also involved in the catalytic mechanism . It might well be that an increase in catalytic activity is not possible without simultaneous changes both in the periplasmic and membrane domains of PglBCj.
A general problem of enzyme engineering is the destabilizing effect of introduced mutations. In the case of PglB Cj this was reflected in a high proportion of amino acid substitutions having detrimental effects on production and stable maintenance of the PglB polypeptide. As conserved positions of PglB Cj such as I571 and V575 were found to be particularly sensitive to changes, future mutagenesis strategies should focus on non-conserved residues in the vicinity of the active site. In addition, randomization at specific positions might be restricted to amino acids found in natural sequences as proposed by Jochens and Bornscheuer , thereby reducing the number of unwanted instable variants and thus the screening effort.
In summary, this study adds important information to structure-function relationships of bacterial OST and provides methods for generating and screening OST mutant libraries in E. coli strains capable of producing engineered glycoproteins.
Bacterial strains and plasmids
Escherichia coli CLM24 (O polysaccharide ligase WaaL negative derivative of W3110, ) was used as host strain for all in vivo glycosylation experiments. For construction of plasmid libraries by cassette mutagenesis, E. coli XL10 Gold cells were used (Agilent-Stratagene, La Jolla, CA, USA).
Plasmids used in this study are listed in Table 1. Expression vector pACT3Kan was constructed by insertion of a Kanamycin resistance cassette into the cat chloramphenicol resistance gene of pACT3  by homologous recombination. Plasmids pGVXN112Kan and pGVXN113Kan were constructed by PCR-sub cloning of the sequence coding for wt PglB with C terminal HA tag and inactive PglB-HA W458A-D459A from pGVXN114 and pGVXN115, respectively, into pACT3Kan using restriction sites KpnI and BamHI. Construction of plasmid pGVXN345 for expression of the CP5 polysaccharide biosynthesis cluster of S. aureus will be described elsewhere (GlycoVaxyn and Jean Lee, manuscript in preparation).
Expression of N-glycoproteins in shake flasks
Glycoprotein producing and control strains were cultivated in 100 mL Erlenmeyer flasks with 70 mL LB medium (5 g L-1 yeast extract, 10 g L-1 casein-based tryptone, 5 g L-1 NaCl) supplemented with 100 mg L-1 ampicillin, 30 mg L-1 kanamycin and 20 mg L-1 tetracycline. Flasks were inoculated 1:100 from overnight tube cultures and incubated at 37°C and 160 rpm. When an OD600 of ≈ 0.5 was reached 1 mM IPTG and 2 g L-1 L-arabinose were added and 4 h later again 2 g L-1 L-arabinose. Cells were harvested after overnight incubation by centrifugation (10’000 x g, 15 min, 4°C). Periplasmic proteins were extracted by sucrose-lysozyme treatment . Cell pellets were resuspended to an OD600 of 20 in lysis buffer containing 20% w/v sucrose, 30 mM Tris·HCl (pH 8.0), 1 mM EDTA, 1 mg mL-1 lysozyme, 1 tablet per 80 mL of CompleteTM protease inhibitor mix (Roche, Basel, Switzerland) and incubated for 1 h on ice with gentle magnetic stirring. Spheroblasts with cytoplasmic and membrane proteins were separated by centrifugation (5000 x g, 15 min, 4°C) and the supernatant was used for ELISA and Western blot analysis.
Western blot analysis
Glycoproteins and oligosaccharyltransferase expressed by recombinant E. coli were analyzed by SDS-PAGE and subsequent Western blot following standard electrophoresis, blotting, immunoreaction and washing procedures. For analysis of EPA-CP5, extracts of periplasmic proteins were supplemented with similar volumes of 2 x SDS PAGE sample buffer, denatured for 5 min at 95°C and similar volumes were loaded on acrylamide gels. EPA and EPA-CP5 glycoproteins were detected with anti-EPA (rabbit, Sigma-Aldrich, Buchs, Switzerland) and anti-CP5 antibodies (rabbit, GlycoVaxyn AG), respectively.
For Western blot analysis of membrane bound PglB-HA, total cell protein samples were prepared by resuspending cell pellets obtained from overnight induced 1 mL mini-cultures in deep-well plates (see below) in 100 μL of 1 x SDS-PAGE sample buffer followed by denaturation at 60°C for 30 min. In order to facilitate relative quantification, similar volumes of protein samples were loaded per well (5 μL). PglB-HA protein bands were visualized by hybridization with anti-HA antibodies (rabbit, Sigma-Aldrich).
Western blots were generally developed by hybridization with secondary anti-rabbit IgG-HRP antibodies (goat, Bio-Rad) followed by reaction with Super Signal Dura West HRP substrate according to the manufacturer’s instructions (Thermo-Scientific-Pierce, Rockford, IL, USA). Chemiluminescence signals were recorded with a ChemiDoc-It imaging system (UVP, Upland, CA, USA). Relative intensities of bands specific for full-length PglB-HA (upper bands in Figure 3) were quantified by using band area and relative Gray values calculated with the image analysis software ImageJ (http://rsbweb.nih.gov/ij/).
Enzyme linked immunosorbent assay (ELISA)
The relative amount of glycosylated proteins in different periplasmic extract samples was analyzed by sandwich ELISA using transparent polystyrol 96-well plates with high capacity for protein binding (F96 MaxiSorp, Nunc). For coating with the capture antibody, wells were filled with 60 μL of protein G purified goat-anti-EPA antiserum (diluted in PBS to ≈ 10 μg mL-1) (US Biological/Lucerna Chem AG, Lucerne, Switzerland) and incubated overnight at 4°C without shaking. After four washing steps of 2 min with 300 μL PBS containing 0.05% v/v Tween-20, wells were blocked for at least 2 h at room temperature with 300 μL PBS containing 10% w/v dry milk. Following another washing step of 2 min with 300 μL PBS-Tween, 50 μL of periplasmic extract was added per well. For hybridization of the analyte with the capture antibodies, micro plates were incubated overnight at 4°C. Unbound EPA/EPA-CP5 was removed by washing the plate four times for 2 min with 300 μL PBS-Tween per well. Secondary antibodies (rabbit anti-CP5 antiserum, GlycoVaxyn AG) were added as 1:100 dilution in PBS + 1% dry milk (50 μL per well), followed by incubation for at least 1 h at room temperature and 10 rpm. Excess antibodies were removed by four washing steps as described above. Detection antibodies (goat-anti-rabbit IgG-HRP, Bio-Rad, Reinach, Switzerland) were added as 1:20’000 dilution in PBS + 1% dry milk, followed by incubation for at least 1 h (room temperature, 10 rpm) and four final washing steps. Light-protected ELISA plates were developed with Ultra-TMB-ELISA HRP substrate (Thermo-Scientific/Pierce) according to the manufacturer’s instructions. ELISA signals were recorded at 450 nm using a Cary 50 MPR microplate reader.
Random mutagenesis methods
Random mutations were introduced into the pglB gene by error-prone PCR (epPCR) using mutagenic Mutazyme II DNA polymerase (Agilent-Stratagene). Due to the very low transformation effiency of spectinomycin-selectable, low copy number expression vector pGVXN114, an alternative, kanamycin-selectable, medium copy number expression vector (pGVXN112Kan) was used as template for the construction of pglB mutant libraries. The soluble, periplasmic domain of PglB (residues H423 to I713) was amplified in 25 PCR cycles using forward primer 5’-GCAACTATTTTGACTTTAGCTCCAGTATTTATC-3’ and reverse primer 5’-CTAGAGGATCCTTAAGCGTAATCTGGAACATCGTATGGGTA-3’. The concentration of the template plasmid was 17 ng μL-1 and the annealing temperature was set to 55°C. PCR products with the expected a size of ≈ 900 bp were purified with a gel extraction kit (Sigma, Buchs, Switzerland). Purified epPCR product was used in 50-fold molar excess as mega primer in a quick change-like PCR reaction (25 cycles) using high fidelity DNA polymerase (PhusionTM, New England Biolabs, Ipswich, MA, USA) and 1 ng μL-1 of pGVXN112Kan as template. Methylated wt plasmid DNA was removed by DpnI digest. Salts and reagents were removed with a PCR clean-up kit (elution with ddH2O). Electro competent Escherichia coli CLM24 (pGVXN150, pGVXN345) were prepared by standard procedures and transformed with 2 μL of purified mega primer-quick change product using electroporation (1 mm gap width cuvettes, 5 ms pulses of 1.8 kV). Transformants were selected on LB Agar supplemented with 100 mg L-1 ampicillin, 30 mg L-1 kanamycin and 20 mg L-1 tetracycline.
Cassette mutagenesis of I571 and V575 was performed by performing quick change PCR with PhusionTM polymerase (New England Biolabs). Wild-type plasmid DNA (pGVXN112Kan, 1 ng μL-1) was amplified with mutagenic primers in 18 cycles, the annealing temperature was 55°C. The sequences of forward and reverse primers were 5'-T TAT ATG CCC GCT AGA ATG TCT TTG NDT TTC TCT ACG NDT GCT AGT TTT TCT TTT TAT TAA TTT AG-3' and 5'-CT AAA TTA ATA AAA AGA AAA ACT AGC AHN CGT AGA GAA AHN CAA AGA CAT TCT AGC GGG CAT ATA A-3', respectively (mutagenized codons and silent control mutation TTT > TTC underlined). After DpnI digestion the PCR product was directly used for heat shock transformation of commercial ultra-competent E. coli XL10 Gold (Agilent-Stratagene). Transformants were selected on LB-Kanamycin plates. After confirming the quality of the library by sequencing of randomly picked XL10 Gold clones, several thousand XL10 Gold colonies were resuspended in phosphate buffered saline (PBS) and mixed. The plasmid library was isolated with a mini-prep kit and transformed into the expression strain E. coli CLM24 (pGVXN150, pGVXN345).
For mutant screening, individual E. coli clones were cultivated in 96-deep well plates (DWP) with a total volume of 2.2 mL per well (riplateTM sw 2 mL, Ritter GmbH, Schwabmünchen, Germany). For overnight pre-cultures, wells were filled with 0.5 mL of buffered medium containing 5 g L-1 yeast extract, 10 g L-1 tryptone, 12.8 g L-1 Na2HPO4·7H2O, 3.0 g L-1 KH2PO4, 0.5 g L-1 NaCl, 1.0 g L-1 NH4Cl, 2 mM MgSO4·7H2O and 0.1 mM CaCl2, 100 mg L-1 ampicillin, 30 mg L-1 kanamycin and 20 mg L-1 tetracycline.
DWPs were sealed with perforated rubber-metal lids with thin sheets of PTFE and microfiber filter material as sterility barrier (SMCR1296 sandwich cover, system Duetz, Kuhner AG, Birsfelden, Switzerland). Preculture plates were inoculated with single colonies from library transformation plates and incubated at 37°C and 500 rpm for 16–20 h in a microplate shaker (Infors AG, Bottmingen, Switzerland). Aliquots of pre-culture plates (100 μL) were transferred to sterile 96-well microtiter plates (Nunc, Roskilde, Denmark), supplemented with an equal volume of medium containing 50% v/v glycerol and stored at −80°C.
Expression DWPs contained 1.3 mL of the medium described above per well and were inoculated 1:50 from pre-culture plates. Plates were incubated at 37°C and 500 rpm. OD600 was determined after 4–5 h and then every hour with a microplate reader (Cary 50 MPR, Varian, Palo Alto, CA, USA). At an OD600 of ≥ 0.4 expression of PglB and carrier protein was induced by adding 1 mM IPTG and 4 g L-1 L-arabinose, respectively, and agitation was reduced to 100 rpm. After overnight incubation (induction time 15–20 h) 1 mL of culture fluid per well was transferred to a new DWP and cells were separated by centrifugation (4000 rpm, 15 min, 4°C). Cell pellets in wells were resuspended in 200 μL ice-cold lysozyme-sucrose lysis buffer (see above). DWPs were incubated without agitation for 1 h at 4°C and cell suspensions were then transferred to a transparent microtiter plate and centrifuged at 3000 rpm and 4°C for 15 min. The supernatant (extract of periplasmic proteins) was used for ELISA analysis in 96-well plates (see above).
DNA and protein sequence analysis
For analysis of library quality, plasmid mini-preps of random clones of epPCR and cassette mutagenesis experiments were sequenced using primers binding within the second half of pglB (sequencing service: GATC Biotech GmbH, Constance, Germany). Inactive clones with positive response on PglB blots were collected in new 96-well glycerol stock libraries and sequenced using automated 96-well plasmid extraction (GATC Biotech). DNA and protein sequences were compared to the wild-type sequence using LasergeneTM SeqbuilderTM and MegAlignTM programs (DNASTAR, Madison, WI, USA). Mutations were manually checked for lack of ambiguity in sequencing chromatograms. Non-redundant protein sequences homologous to PglB Cj were retrieved from a protein BLAST search on the NCBI website and aligned using the ClustalW algorithm of MegAlignTM. For visualisation of individual residues and spatial distances in the 3D-structure of PglB, the crystal structure file 3RCE  was retrieved from Protein Data Bank and displayed using the program Discovery Studio Visualizer 3.1 (Accelrys, Cambridge, UK).
Staphylococcus aureus capsular polysaccharide serotype 5
Pseudomonas aeruginusa exotoxin (toxoid form)
deep well plate
enzyme-linked immunosorbent assay
error-prone polymerase chain reaction
Inactive variant W458A-D459A
sodium dodecyl sulfate–polyacrylamide gel electrophoresis
phosphate buffered saline
We thank Dr. Carmen Momeu for important scientific input at the beginning of this study. This work was supported by the Swiss Federal Department of Economic Affairs, Commission for Technology and Innovation CTI [grant No. CTI 10094.1 PFLS-LS].
- Nothaft H, Szymanski CM: Protein glycosylation in bacteria: sweeter than ever. Nat Rev Microbiol. 2010, 8 (11): 765-778. 10.1038/nrmicro2383.View ArticleGoogle Scholar
- Wacker M, Linton D, Hitchen PG, Nita-Lazar M, Haslam SM, North SJ, Panico M, Morris HR, Dell A, Wren BW, et al: N-linked glycosylation in Campylobacter jejuni and its functional transfer into E. coli. Science. 2002, 298 (5599): 1790-1793. 10.1126/science.298.5599.1790.View ArticleGoogle Scholar
- Feldman MF, Wacker M, Hernandez M, Hitchen PG, Marolda CL, Kowarik M, Morris HR, Dell A, Valvano MA, Aebi M: Engineering N-linked protein glycosylation with diverse O antigen lipopolysaccharide structures in Escherichia coli. Proc Natl Acad Sci USA. 2005, 102 (8): 3016-3021. 10.1073/pnas.0500044102.View ArticleGoogle Scholar
- Ihssen J, Kowarik M, Dilettoso S, Tanner C, Wacker M, Thöny-Meyer L: Production of glycoprotein vaccines in Escherichia coli. Microb Cell Fact. 2010, 9: 61-10.1186/1475-2859-9-61.View ArticleGoogle Scholar
- Lockhart S: Conjugate vaccines. Expert Rev Vaccines. 2003, 2 (5): 633-648. 10.1586/14760522.214.171.1243.View ArticleGoogle Scholar
- Wacker M, Feldman MF, Callewaert N, Kowarik M, Clarke BR, Pohl NL, Hernandez M, Vines ED, Valvano MA, Whitfield C, et al: Substrate specificity of bacterial oligosaccharyltransferase suggests a common transfer mechanism for the bacterial and eukaryotic systems. Proc Natl Acad Sci USA. 2006, 103 (18): 7088-7093. 10.1073/pnas.0509207103.View ArticleGoogle Scholar
- Li L, Woodward R, Ding Y, Liu XW, Yi W, Bhatt VS, Chen M, Zhang LW, Wang PG: Overexpression and topology of bacterial oligosaccharyltransferase PglB. Biochem Biophys Res Commun. 2010, 394 (4): 1069-1074. 10.1016/j.bbrc.2010.03.126.View ArticleGoogle Scholar
- Maita N, Nyirenda J, Igura M, Kamishikiryo J, Kohda D: Comparative structural biology of eubacterial and archaeal oligosaccharyltransferases. J Biol Chem. 2010, 285 (7): 4941-4950. 10.1074/jbc.M109.081752.View ArticleGoogle Scholar
- Matsumoto S, Igura M, Nyirenda J, Matsumoto M, Yuzawa S, Noda N, Inagaki F, Kohda D: Crystal structure of the C-terminal globular domain of oligosaccharyltransferase from Archaeoglobus fulgidus at 1.75 A resolution. Biochemistry. 2012, 51 (20): 4157-4166. 10.1021/bi300076u.View ArticleGoogle Scholar
- Schwarz F, Lizak C, Fan YY, Fleurkens S, Kowarik M, Aebi M: Relaxed acceptor site specificity of bacterial oligosaccharyltransferase in vivo. Glycobiology. 2011, 21 (1): 45-54. 10.1093/glycob/cwq130.View ArticleGoogle Scholar
- Lizak C, Gerber S, Numao S, Aebi M, Locher KP: X-ray structure of a bacterial oligosaccharyltransferase. Nature. 2011, 474 (7351): 350-355. 10.1038/nature10151.View ArticleGoogle Scholar
- Kowarik M, Young NM, Numao S, Schulz BL, Hug I, Callewaert N, Mills DC, Watson DC, Hernandez M, Kelly JF, et al: Definition of the bacterial N-glycosylation site consensus sequence. EMBO J. 2006, 25 (9): 1957-1966. 10.1038/sj.emboj.7601087.View ArticleGoogle Scholar
- Nita-Lazar M, Wacker M, Schegg B, Amber S, Aebi M: The N-X-S/T consensus sequence is required but not sufficient for bacterial N-linked protein glycosylation. Glycobiology. 2005, 15 (4): 361-367.View ArticleGoogle Scholar
- Chen MM, Glover KJ, Imperiali B: From peptide to protein: comparative analysis of the substrate specificity of N-linked glycosylation in C. jejuni. Biochemistry. 2007, 46 (18): 5579-5585. 10.1021/bi602633n.View ArticleGoogle Scholar
- Raetz CRH, Whitfield C: Lipopolysaccharide endotoxins. Annu Rev Biochem. 2002, 71: 635-700. 10.1146/annurev.biochem.71.110601.135414.View ArticleGoogle Scholar
- Rath A, Glibowicka M, Nadeau VG, Chen G, Deber CM: Detergent binding explains anomalous SDS-PAGE migration of membrane proteins. Proc Natl Acad Sci USA. 2009, 106 (6): 1760-1765. 10.1073/pnas.0813167106.View ArticleGoogle Scholar
- Reetz MT, Kahakeaw D, Lohmer R: Addressing the numbers problem in directed evolution. ChemBioChem. 2008, 9 (11): 1797-1804. 10.1002/cbic.200800298.View ArticleGoogle Scholar
- Faridmoayer A, Fentabil MA, Haurat MF, Yi W, Woodward R, Wang PG, Feldman MF: Extreme substrate promiscuity of the Neisseria oligosaccharyl transferase involved in protein O-glycosylation. J Biol Chem. 2008, 283 (50): 34596-34604. 10.1074/jbc.M807113200.View ArticleGoogle Scholar
- Jaffee MB, Imperiali B: Exploiting topological constraints to reveal buried sequence motifs in the membrane-bound N-linked oligosaccharyl transferases. Biochemistry. 2011, 50 (35): 7557-7567. 10.1021/bi201018d.View ArticleGoogle Scholar
- Igura M, Maita N, Kamishikiryo J, Yamada M, Obita T, Maenaka K, Kohda D: Structure-guided identification of a new catalytic motif of oligosaccharyltransferase. EMBO J. 2008, 27 (1): 234-243. 10.1038/sj.emboj.7601940.View ArticleGoogle Scholar
- Arnold FH, Wintrode PL, Miyazaki K, Gershenson A: How enzymes adapt: lessons from directed evolution. Trends Biochem Sci. 2001, 26 (2): 100-106. 10.1016/S0968-0004(00)01755-2.View ArticleGoogle Scholar
- Stemmer WP: Rapid evolution of a protein in vitro by DNA shuffling. Nature. 1994, 370 (6488): 389-391. 10.1038/370389a0.View ArticleGoogle Scholar
- Reetz MT, Wang LW, Bocola M: Directed evolution of enantioselective enzymes: Iterative cycles of CASTing for probing protein-sequence space. Angew Chem Int Edit. 2006, 45 (8): 1236-1241. 10.1002/anie.200502746.View ArticleGoogle Scholar
- Jochens H, Bornscheuer UT: Natural diversity to guide focused directed evolution. ChemBioChem. 2010, 11 (13): 1861-1866. 10.1002/cbic.201000284.View ArticleGoogle Scholar
- Dykxhoorn DM, StPierre R, Linn T: A set of compatible tac promoter expression vectors. Gene. 1996, 177 (1–2): 133-136.View ArticleGoogle Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.