A surprising sweetener from enteropathogenic Escherichia coli

Abstract

Infections with enteropathogenic Escherichia coli (EPEC) are remarkably devoid of gut inflammation and necrotic damage compared to infections caused by invasive pathogens such as Salmonella and Shigella. Recently, we observed that EPEC blocks cell death using the type III secretion system (T3SS) effector NleB. NleB mediated post-translational modification of death domain containing adaptor proteins by the covalent attachment of N-acetylglucosamine (GlcNAc) to a conserved arginine in the death domain.  N-linked glycosylation of arginine has not previously been reported in mammalian cell biology and the precise biochemistry of this modification is not yet defined. Although the addition of a single GlcNAc to arginine is a seemingly slight alteration, the impact of NleB is considerable as arginine in this location is critical for death domain interactions and death receptor induced apoptosis. Hence, by blocking cell death, NleB promotes enterocyte survival and thereby prolongs EPEC attachment to the gut epithelium.

Glycosylation as a Key Biological Process

Protein glycosylation is the most common post-translational modification characterized to date with more than half of all human proteins estimated to contain at least one glycan chain.^1,2 Glycans associated with the cell surface and intracellular proteins and lipids influence a diverse range of biological processes in animal systems.³ For example, glycan structures contribute to the correct folding and conformational stability of a range of cellular proteins synthesized at the endoplasmic reticulum (ER).⁴ Surface glycans serve as ligands for glycan-binding proteins in cellular trafficking, adhesion and signaling events^5,6 and also mediate interactions between the vertebrate host and microbial pathogens.^7-9 Other processes influenced by glycan structures include innate immune signaling, autoimmune reactions and tumor metastases.^2,8

Glycan structure is determined by the concerted action of numerous genes that code for glycosyltransferases, glycosidases, and other enzymes that synthesize and remodel glycan chains, as well as accessory enzymes involved in the synthesis and transport of nucleotide sugars. Glycosyltransferases (GTs) constitute a family of enzymes that catalyze the biosynthesis of glycosidic linkages by the transfer of a sugar residue from an activated nucleotide sugar donor (such as UDP) to specific acceptor molecules.¹⁰ GTs can be classified into families based on their amino acid sequence similarities and their particular GT fold. There are currently 4 known GT folds. GT-A and GT-B are well characterized and are the predominant folds within known GTs, while GT-C and GT-D are more recently described.¹¹

In almost all glycoproteins, carbohydrate units are coupled to a polypeptide backbone either by N- or O-glycosidic bonds or by both types of linkage.² Alternatively glycans can be linked to the C2 position of Trp (C-mannosylation) or as a linker structure between glycophosphatidylinositol anchors to protein backbones.¹² N-linked glycosylation is the most studied type of protein glycosylation in eukaryotes and is essential for efficient protein folding and assembly in the ER and anterograde trafficking through the secretory pathway.¹² N-glycans are synthesized via the activity of oligosaccharidetransferases (OST) at the luminal side of the ER membrane. Here, OST transfers glycans from a lipid-linked oligosaccharide (LLO) precursor to a nascent polypeptide chain during translation. The glycans are linked covalently to asparagine residues within an acceptor peptide sequon, Asn-X-(Ser/Thr) (alternatively and less frequently Asn-X-Cys), where X is any amino acid except Pro.¹² This specific sequon denotes potential N-glycan sites and does not always predetermine glycosylation events.

O-linked glycosylation is a common covalent modification of mammalian glycoproteins that typically occurs on the hydroxyl moieties of Ser/Thr, and sometimes hydroxylysine (collagen) and Tyr (glycogenin) residues.¹³ O-linked glycans can be grouped into 2 categories, mucin and non-mucin glycans. Mucin O-glycans are found in all mucous secretions as transmembrane glycoproteins of cell surfaces and as secreted mucins.¹³ They begin with covalently α-linked N-acetylgalactosamine (GalNAc) molecules linked to Ser or Thr residues and can be extended with galactose, N-acetylglucosamine (GlcNAc), fucose or sialic acid. Overall mucin O-glycans are highly heterogeneous, with up to 8 possible core structures and hundreds of side chain variations. Non-mucin O-glycans include α-linked O-fucose, β-linked O-xylose, α-linked O-mannose, β-linked O-GlcNAc, and α- or β-linked O-glucose glycans. O-GlcNAc is one of the most abundant post-translational modifications (PTM) in cytoplasmic and nuclear compartments of all metazoan cells.¹⁴ In this form of O-linked glycosylation, the GlcNAc is not elongated or modified to form more complex structures. It is a highly dynamic modification that cycles rapidly on and off due to the concerted action of the cellular enzymes, O-GlcNAc transferase (OGT) and O-GlcNAcase (OGA).¹⁴ OGT is a GT that catalyzes the addition of a single GlcNAc from UDP-GlcNAc to Ser or Thr residues of target proteins, whereas OGA actively removes GlcNAc moieties. A number of consensus sites for O-GlcNAcylation have been described.¹²

The rapid cycling of O-GlcNAcylation is related to a complex and dynamic interplay between O-GlcNAc and O-phosphate, which together regulate a diverse array of proteins involved in signaling, transcription and cytoskeletal regulation. O-GlcNAc can occur in competition with Ser/Thr-O-phosphorylation, and is therefore an important regulatory mechanism in cellular processes.¹⁵ O-GlcNAc modifications are highly abundant in the nucleus, often occurring on transcriptional or translational regulatory factors, and also on cytoplasmic proteins involved in stress responses, cell signaling, energy metabolism, and cytoskeletal regulation. Thus it is not surprising that O-GlcNAc cycling is highly responsive to external stimuli, some of which include insulin, high glucose and cellular stress. In some cases, altered or dysregulated O-GlcNAcylation of proteins contributes to development of neurodegenerative diseases including Alzheimer and Parkinson disease, and also contributes to the development of diabetes in mammals.¹⁴ OGT alleles are also essential in embryogenesis.¹⁶

OGT is regulated by a number of complex mechanisms including transcriptional regulation of its expression, differential mRNA splicing, proteolytic processing, PTM, and self-oligomerisation.¹⁴ The most important factor in its regulation however is the cellular concentration of the donor substrate, UDP-GlcNAc. In a cell UDP-GlcNAc levels are highly sensitive to carbohydrate, fatty acid, energy and nitrogen metabolic fluxes, and can change rapidly.¹⁴  Upon transfer of GlcNAc, UDP release potently inhibits OGT and thus modulates further O-GlcNAcylation events.

EPEC Produces a Novel Glycosyltransferase T3SS Effector, NleB

The effect of EPEC infection on general cellular glycosylation events is unknown but recently, Gao et al.¹⁷ identified the T3SS effector NleB from A/E pathogens as a novel GlcNAc transferase. NleB and the SseK homologues from Salmonella displayed significant structural homology to the GT-8 family of GTs.¹⁷ These enzymes typically adopt a GT-A fold consisting of an α/β/α sandwich that resembles a Rossmann fold and a DxD catalytic motif which requires a divalent cation for activity, typically Mn²⁺.¹⁸ The DxD motif interacts with phosphate groups of a nucleotide sugar donor through the coordination of the divalent cation and mediates sugar transfer to target proteins.

NleB was previously implicated in the inhibition of NF-κB activation upon EPEC infection.^19,20,21 Gao et al.¹⁷ reasoned that if O-GlcNAc is involved in the regulation of signaling networks, including NF-κB²², NleB may suppress NF-κB activation through its GlcNAc transferase activity. While the study from Gao et al.¹⁷ identified GAPDH as a potential target of NleB activity, exhaustive site-directed mutagenesis of serine or threonine residues within the GAPDH sequence did not reveal a specific amino acid that was modified by NleB.

Subsequently, 2 independent studies identified host death domain (DD)-containing proteins as alternative targets for NleB.^23,24 Both groups utilised the yeast 2-hybrid system with NleB as bait to identify FADD, RIPK1²⁴ and TRADD²³ as potential mammalian binding partners for NleB. Both studies showed that NleB functioned as a GlcNAc transferase to add a single GlcNAc residue to a conserved arginine residue within the DD of these proteins. The GlcNAc addition blocked the ability of FADD, RIPK1 and TRADD to form homotypic and heterotypic DD interactions, thereby shutting down death receptor signaling pathways including inflammation, apoptosis and necroptosis.^23,24

While Li et al.²³ found that R235 within the DD of TRADD was specifically modified by NleB, our work identified the equivalent residue, R117, within the DD of FADD as the target of NleB activity.²⁴ In both studies, mutation of the NleB DxD catalytic motif or the conserved Arg residue within FADD or TRADD, abolished transfer of GlcNAc to the target proteins.^23,24 Further to this Li et al.²³ demonstrated that the DDs of TRADD, FADD and RIPK1 were efficiently modified by NleB during in EPEC infection, whereas the DDs of FAS and TNFR1 were only modified by NleB expressed ectopically and not during EPEC infection. Other DD proteins including MyD88 and IRAK1 were not modified by NleB, which is likely due to the fact that they do not contain the equivalent conserved Arg residue.²³ Importantly, the study by Li et al.²³ determined that GAPDH was not GlcNAcylated by either OGT or NleB.

NleB and Inhibition of Cell Death Signaling

Consistent with previous work,^17,19,20 Li et al.²³ also showed that ectopic expression of NleB inhibited NF-κB activation in cultured epithelial cells. Our study however, showed that NleB had no impact on production of the NF-κB-dependent inflammatory cytokine IL-8 during EPEC infection, despite the ability of NleB to modify TRADD and RIPK1, 2 key mediators of signaling via TNFR1.^23,24 Given the additional involvement of DD proteins in cell death signaling and the strong affinity of NleB for FADD,^23,24 we investigated the consequences of NleB-mediated inhibition of Fas-induced extrinsic apoptosis rather than TNFR1-induced inflammation on the outcome of infection.

nleB mutants of the murine A/E pathogen, Citrobacter rodentium (CR) are attenuated for colonization²⁵ and more recent experiments have shown that the GlcNAc transferase activity of NleB is important for this phenotype.^17,23 Given the strong inhibition of Fas-signaling by NleB in vitro through GlcNAcylation of FADD,^23,24 we postulated that attenuation of the nleB CR mutant in vivo was due to the elimination of infected enterocytes by Fas ligand (FasL) positive immune cells. Hence in the absence of NleB, infected cells would undergo apoptosis and slough off into the lumen of the gut, thereby reducing pathogen load. Consistent with this, the nleB CR mutant induced greater activation of caspase-8, a marker of FasL-induced apoptosis than wild type CR,²⁴ Moreover Fas- or FasL-deficient mice experienced greater colitis and took longer to clear CR than wild type mice. Interestingly, the reduced colonization exhibited by the nleB mutant was reversed in Fas-deficient mice, suggesting that nleB mutant attenuation was indeed due to functional Fas signaling.²⁴

The GlcNAcylation of Arg117 in the DD of FADD is a disabling modification. Arg117 is essential for death domain interactions between FADD and the DD of Fas as well as between TRADD and FADD.²⁶ Mutation of Arg117 abrogates cell death signaling through Fas or TNFR1.²⁶ The essential role of this amino acid in Fas signaling is also evident in the requirement of Arg117 in FADD for embryogenesis,²⁷ and this has been explained by structural elucidation of the Fas-FADD signalosome that puts Arg117 at the interface of adjacent FADD molecules and the DD of Fas and FADD.²⁸ Upon engagement of Fas by FasL, Fas DD binding to FADD increases the overall stability of FADD, including in the death effector domain that binds and stabilizes procaspase-8.²⁹

Conclusion

Prior to the discovery that an EPEC effector protein could mediate arginine glycosylation, there has been only one published report of the same phenomenon in a biological system. In a brief report, Singh et al.³⁰ described a protein known as amylogenin from sweet corn that autocatalytically added a single β-glucose residue to its own arginine. This self-glucosylation of amylogenin occurred in vitro in the presence of UDP-[¹⁴C]glucose, a modification that could be reversed by the addition of β-glucosidase.³⁰ The authors suggested the reaction might be important in priming starch synthesis in corn however no further reports on the activity of amylogenin or the nature of the modification have been produced. Hence, NleB remains the first example of a glycosyltransferase that can modify arginine in a mammalian cell. The unique nature of this N-linked GlcNAcylation implies that the modification is irreversible, as no mammalian enzymes exist, such as OGAs, to remove GlcNAc from arginine. However, to date this has not been investigated and more work is needed to understand the biochemical basis of this surprising activity.

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.

Funding

This work was supported by funding from the Australian National Health and Medical Research Council (NHMRC) APP1044061 awarded to ELH.