Genetic engineering of E. coli K-12 for heterologous carbohydrate antigen production

Abstract

Background: Carbohydrate-based vaccines have made a remarkable impact on public health over the past three decades. Efficient production of carbohydrate antigens is a crucial prerequisite for the development of such vaccines. The enzymes involved in the synthesis of bacterial surface carbohydrate antigens are usually encoded by large, uninterrupted gene clusters. Non-pathogenic E. coli glycoengineering starts with the genetic manipulation of these clusters. Heterologous gene cluster recombination through an expression plasmid has several drawbacks, including continuous antibiotic selection pressure, genetic instability, and metabolic burdens. In contrast, chromosome-level gene cluster expression can minimize the metabolic effects on the host and reduce industrial costs.

Results: In this study, we employed the suicide vector-mediated allelic exchange method to directly replace the native polysaccharide gene clusters in E. coli with heterologous ones. Unlike previously strategies, this method does not rely on I-SceI endonuclease or CRISPR/Cas system to release the linearized DNA insert and λ-red recombinase to promote its homologous recombination. Meanwhile, the vectors could be conveniently constructed by assembling multiple large DNA fragments in order in vitro. The scarless chromosomal insertions were confirmed by whole-genome sequencing and the polysaccharide phenotypes of all glycoengineered E. coli mutants were evaluated through growth curves, silver staining, western blot, and flow cytometry. The data indicated that there was no obvious metabolic burden associated with the insertion of large gene clusters into the E. coli W3110 O-antigen locus, and the glycoengineered E. coli can produce LPS with a recovery rate around 1% of the bacterial dry weight. Moreover, the immunogenicity of the heterologously expressed carbohydrate antigens was analyzed by mice immunization experiments. The ELISA data demonstrated the successful induction of anti-polysaccharide IgM or IgG antibodies.

Conclusions: We have provided a convenient and reliable genomic glycoengineering method to produce efficacious, durable, and cost-effective carbohydrate antigens in non-pathogenic E. coli. Non-pathogenic E. coli glycoengineering has great potential for the highly efficient synthesis of heterologous polysaccharides and can serve as a versatile platform to produce next-generation biomedical agents, including glycoconjugate vaccines, glycoengineered minicells or outer membrane vesicles (OMVs), polysaccharide-based diagnostic reagents, and more.

Keywords: E. coli K-12; S. enterica; DNA Long fragment editing; Glycoengineering; Polysaccharide biosynthesis.

Fig. 1

Schematic map of suicide vectors. A All suicide vectors were derived from pHY093 (pRE112-Cm-GFP) or pHY094 (pRE112-Kan-GFP). B Two homologous arms, i.e., HA_up and HA_down were cloned from E. coli W3110 genome and fused together by fusion PCR. The fused homologous arms replaced the GFP cassette in pHY093 (pRE112-Cm-GFP), resulting in pHY095 (pRE112-Cm-ΔOAg). C the viaB_locus was cloned from S. Typhi genome. The linearized pHY095 (pRE112-Cm-ΔOAg) vector and viaB_locus were assembled in vitro, resulting in pHY129 (pRE112-Cm-ΔOAg::viaB locus). D A cassette of FRT-flanked chloramphenicol resistance gene was cloned from pCP3, and cloned into the middle of homologous arms, resulting in pHY099 (pRE112-Km-ΔOAg::Cm_FRT2). E The OAg^ST O-antigen gene cluster was cloned from S. Typhimurium genome via Long-range PCR. The linearized pHY099 (pRE112-Km-ΔOAg::Cm_FRT2) vector and OAg^ST were assembly in vitro, resulting in pHY100 (pRE112-Km-ΔOAg::OAg^ST-Cm_FRT2). F The OAg^SE O-antigen gene cluster was cloned from S. Enteritidis genome via Long-range PCR. The linearized pHY099 (pRE112-Km-ΔOAg::Cm_FRT2) vector and OAg^SE were assembly in vitro, resulting in pHY101 (pRE112-Km-ΔOAg::OAg^SE-Cm_FRT2). G pHY102 (pRE112-Km-ΔOAg::OAg^SA-Cm_FRT2) was derived from pHY101 by deliberately deleting the tyv gene

Fig. 2

General outline of the Chromosomal-level gene cluster insertion mutations. A Scarless insertion mutation. Successful mutations are achieved by consecutive single-crossovers. Step 1: The Suicide plasmid integrates into the chromosome of E. coli W3110 by a first cross-over recombination on either side of HA_up or HA_down. The schematic diagram only takes the cross-over on HA_up side as an example, while cross-over on HA_down side is similar. Plasmid integrated intermediate strains are selected on chromosomal plates. Step 2: The suicide plasmid is excised from the chromosome by a second cross-over recombination on HA_down side. Notably, cross-over recombination on HA_up side will lead to reversion allelic exchanges. All plasmid excised derivatives, i.e., either successful or reversed allelic exchange, are selected on sucrose plates. Positive mutants are screened by colony PCR and later sequenced by whole genome sequencing. B Insertion mutation with a single FRT scar. A cassette of FRT-flanked chloramphenicol resistance gene is fused immediately after the heterologous gene cluster to facilitate the selection process and increase the positive mutant rate. At step 1, plasmid integrated intermediate strains are selected on kanamycin plates. At step 3, plasmid excised derivatives are selected on sucrose plus chloramphenicol plates. The chloramphenicol will eliminate all reversion allelic exchange derivatives, and the chromosomal resistance gene could be later deleted by FLP recombinase, leaving a single FRT scar on the genome

Fig. 3

The O-antigen polysaccharide (O-PS) chain-length distribution. The LPS profiles of wild-type S. Paratyphi A S356 (A), S. Typhimurium S100 (B), S. Enteritidis S246 (C), and E. coli W3110 ΔwecA derivatives are revealed by silver staining (upper panel) and western bot (lower panel). The loading samples and loading sequence are the same in each vertical group. The wild-type S. enterica strains show a relatively wide distribution of O-PS lengths, while no apparent O-PS pattern was observed in E. coli W3110 K062 (ΔwecA), L0137 (ΔwecA ΔOAg::OAg^SA), L0136 (ΔwecA ΔOAg::OAg^ST) and L0133 (ΔwecA ΔOAg::OAg^SE). Interestingly, E. coli W3110 ΔwecA derivatives harboring either pET9a-wzy^ST or pET9a- (wzy-wzzB)^ST show a tight modal chain-length distribution (marked by left brackets) with no appreciable amounts of shorter O-PS. Meanwhile, a higher-molecular-weight modal cluster of bands is observed in pET9a- (wzy-wzzB)^ST derivatives L0169, L0170, and L0171, when compared to their pET9a-wzy^ST counterpart L0166, L0167, and L0168, respectively

Fig. 4

Vi capsular phenotype. Western blot results of Vi capsular. Comparable amounts of Vi capsular were detected in glycoengineered E. coli W3110 L0149 (ΔOAg::viaB_locus)

Fig. 5

Antibody responses in mice sera determined by ELISA. Anti-S. Paratyphi A LPS (A, E), Anti-S. Typhimurium LPS (B, F) and anti-S. Enteritidis LPS (C, G) and anti-S. Typhi (D, H) sera antibody concentrations in non-adjuvant (A-D) and CFA/IFA (E–H) vaccinated groups. Responses that differed from the PBS control group are noted by asterisks (*P < 0.05, **P < 0.01, ***P < 0.001, ns: non-significant difference). Antibody concentrations were calculated using a standard curve and all the measured sample concentrations were within the standard curve range. The error bars represent the standard deviations (SD) of the means calculated by the GraphPad Prism software

Fig. 6

Survival curves after orally or intraperitoneally challenged by wild-type S. enterica virulent strains. Six weeks after vaccination, BALB/c mice were challenged by about 100 times the LD50 of wild-type virulent S. Paratyphi A (A), S. Typhimurium (B), S. Enteritidis (C), and S. Typhi (D). *P < 0.05, **P < 0.01, ns: non-significant difference, for all marked groups versus the PBS control