Many pathogenic bacteria rely on motility during different stages of their infection process (Figure 1B). Especially, flagellar motility has been shown to play a critical role in successful infection in many organisms, as it contributes to bacterial movement, adhesion, and biofilm formation. In addition, the glycosylation of flagella is crucial for the proper assembly of flagellar structures and their motility function (Logan, 2006; Merino and Tomás, 2014).

Flagellar glycans feature diverse structures and often incorporate bacteria-specific monosaccharides. For instance, in the gastric pathogens H. pylori and C. jejuni the unique bacterial carbohydrates pseudaminic acid (Pse), legionaminic acid (Leg), and derivatives containing acetamidino and methylglycerol moieties are required for the proper assembly of flagella (Ud-Din and Roujeinikova, 2018). Biosynthesis of Pse is a multi-step process that relies on several enzymes (PseB-PseI), as shown in Figure 2A (Ud-Din and Roujeinikova, 2018). H. pylori and C. jejuni strains expressing non-functional Pse biosynthesis genes show defects in the formation of flagella and are thus non-motile and less virulent (Linton et al., 2000; Goon et al., 2003; Schirm et al., 2003; Guerry et al., 2006; Schoenhofen et al., 2006; Hopf et al., 2011; Javed et al., 2015a). Consequently, the inhibition of the Pse biosynthesis in these bacterial species is a promising antibacterial strategy.

Small-molecule inhibitors of Pse biosynthesis enzymes of H. pylori were identified using high-throughput screening (HTS) and virtual screening (VS) approaches in combination with kinetic studies and structure–activity relationship (SAR) analysis (Ménard et al., 2014). Ultimately, three PseB inhibitors were identified with a conserved N-phenyl-2-pyrrolidone core featuring different substitution patterns on the phenyl groups (Figure 2B). These three PseB inhibitors exhibited IC₅₀ values of ~14μM in vitro on purified PseB enzymes. Importantly, the inhibitors were also able to penetrate the bacterial cell wall and inhibit flagellin production in C. jejuni in a dose-dependent manner as determined by whole cell ELISA. The relatively low IC₅₀ and the ability to cross the bacterial cell wall make these compounds interesting molecules for further development into clinical antibacterial drugs.

In addition to the O-linked flagellin glycosylation with unmodified Pse, C. jejuni also decorates its flagellin with Pse variants, mainly Pse derivative 7-acetamidino-Pse (Pse5Ac7Am; Figure 2B) in which an acetamido group has been substituted for an acetamidino moiety (Thibault et al., 2001; Schirm et al., 2005; Logan et al., 2009). Interestingly, the phage protein FlaGrab [previously called Gp047 (Sacher et al., 2020)] of the Campylobacter phage NCTC 12673 specifically binds to Pse5Ac7Am-modified flagellins of C. jejuni, resulting in reduced motility and partially inhibition of cell growth (Javed et al., 2015a,b). The C-terminal flagellin binding domain of FlaGrab has only been used for the detection of C. jejuni and C. coli so far (Singh et al., 2011, 2012; Javed et al., 2013). While a therapeutic use of FlaGrab against C. jejuni infections remains attractive, it has not yet been further explored. Notably, the C. jejuni strains 12661 and 12664 show reduced binding by FlaGrab due to strain-specific glycan remodeling mechanisms (Sacher et al., 2020). Still, the example of FlaGrab points out the promises of phage proteins as potential therapeutic agents against pathogenic bacteria specifically targeting glycans or other (glycosylated) bacterial structures.

The motility of C. difficile, an opportunistic Gram-positive pathogen, is dependent on O-glycosylation of its flagellum protein FliC. C. difficile strains display two different glycan structures (Twine et al., 2009; Faulds-Pain et al., 2014), with the core N-acetyl-β-glucosamine (β-GlcNAc) as the only conserved residue (Figure 3). NMR studies revealed that the type A O-glycan of the C. difficile 630 (Figure 3A) is composed of the core β-GlcNAc residue modified with an N-methylated Thr via a phosphate at the C3 position (Faulds-Pain et al., 2014). The more complex type B flagellar glycosylation (in strains BI-I, NAP-I, and ribotype 027) is composed of a Ser/Thr-linked β-GlcNAc, elongated with two rhamnose residues (O-methylated at the C3 position, Figure 3B; Bouché et al., 2016). An alternative structure featured an additional 3-amino-3-deoxy-d-fucose (Fuc3N in Figure 3B) modified with a sulfopeptide (Gly-Ala-taurine) at the C3-amino group (Bouché et al., 2016). The glycosyltransferases involved in the synthesis of the type B glycan include GT1 (core GlcNAc transfer onto Ser/Thr), bifunctional GT2 (Rha transfer and Rha methylation), and GT3 (partially involved in the synthesis of the sulfopeptide; Valiente et al., 2016). C. difficile knockout strains of GT1 and GT2 both resulted in decreased motility of the bacterium, whereas a GT1 deletion mutant showed only reduced adherence. These enzymes are interesting targets for antivirulence strategies, as the type B glycan is increasingly associated with the emerging hypervirulent and more aggressive strains of C. difficile (e.g. RT027, RT023).

Capsular polysaccharides (CPS; Figure 1B) of Gram-negative and Gram-positive bacteria are constituents of the bacterial glycocalyx, providing protection to immune system recognition. Interestingly, the CPS of bacterial pathogens are often found to contain carbohydrate epitopes that mimic those of human cells which help to evade the immune system and promote infection (Cress et al., 2014). Consequently, encapsulated bacterial pathogens tend to be more virulent as they are less susceptible to immune system recognition and penetration of the antibiotics. Rendering bacterial pathogens non-encapsulated is an attractive prospect, as it would make the bacteria vulnerable to the innate immune response or resensitize the resistant strains to antibacterial treatments.

The CPS of pathogenic E. coli (so-called K capsules or K antigens) display highly diverse glycan structures, and ~80 different CPS are reported and classified into four groups depending on their assembly and export machinery (Whitfield, 2006). An interactive overview of the E. coli K antigens, their structures, and 3D modeling can be found in the E. coli K antigen 3D structure Database (EK3D; Kunduru et al., 2016).¹

For example, the K15 antigen (Figure 4A) of enterotoxigenic E. coli O6:K15 is a polymer containing α-GlcNAc(1➔5)-α-Kdo(2➔4) disaccharide repeating units (Azurmendi et al., 2020), the K1 capsule (Figure 4A) of uropathogenic E. coli (UPEC) is composed of α-2,8-Neu5Ac repeats, and the K5 capsule is a polymer of α-GlcNAc(1➔4)-β-GlcA(1➔4) repeats (Figure 4A). The highly acidic capsule polysaccharides enhance bacterial survival by sequestering the antimicrobial peptides produced by the host immune system. A capsule-specific phage screen was used to identify inhibitors of CPS synthesis of UPEC K1 and K5 capsules (Goller and Seed, 2010). The most potent compound was 2-(4-phenylphenyl)benzo[g]quionoline-4-carboxylic acid (also called “C7,” Figure 4A), which showed an IC₅₀ value of 12.5–25μM with UPEC K1 UTI89. C7 was evaluated in a variety of biochemical studies and was shown to specifically disrupt the oligomerization of the E. coli K1 antigen leading to its absence on the outside of the cell. Importantly, upon treatment with C7, the E. coli cells were more susceptible to human serum and the compound proved to be active also on clinical E. coli isolates. In a follow-up study, compounds DU003 and DU011 (Figure 4A) were identified from a structurally diverse set of small-molecule inhibitors, because they improved pharmacological properties (IC₅₀, solubility, toxicity, permeability and plasma stability; Goller et al., 2014). Compound DU011 was later shown to attenuate CPS production in E. coli via interaction with the multi-drug efflux pump transcriptional regulator MprA (Arshad et al., 2016). Notably, this mode of inhibition was found to be antivirulent in nature, as it did not lead to the development of antibiotic resistance (Arshad et al., 2016).

A different strategy to de-encapsulate E. coli is based on phage-derived polysaccharide depolymerases (Lin et al., 2017). Several depolymerases were tested for their in vitro and in vivo (mouse models) activity with depolymerase enzyme K5 displaying the highest efficacy and consequent survival of mice. Importantly, the enzymes tested in this study (K1E, K1F, K1H, K5, and K30) were not toxic when injected in animals (based on survival, behavior and body weight monitoring for 5days). Furthermore, when E. coli was tested in a serum sensitivity assay, in the presence of depolymerases, the viability of the cells was reduced significantly, with K5 depolymerase displaying the most pronounced effect. The high specificity of depolymerases toward certain CPS structures represents a potential novel narrow-spectrum treatment.

The capsules of Acinetobacter baumannii are the main virulence factor of these bacterial species (Harding et al., 2017). Their CPS structures feature an impressive diversity of monosaccharides in repeating units and linkages, all of which complicate the development of treatments, especially vaccines (Singh et al., 2019). For example, a study of the association of different A. baumannii capsule types with carbapenem resistance revealed four main serotypes that contribute to resistance (KL2, KL10, KL22, and KL52), indicative of the importance of capsule structure in infection (Hsieh et al., 2020).

The repeating K units of A. baumannii CPS typically consist of 2–6 monosaccharide units and feature glucose, galactose, glucuronic acid, and nonulosonic acid, among others, also with acetyl or acyl modifications (K83 and K11; Figure 4B; Singh et al., 2019). Interestingly, several clinical isolates of A. baumannii (strains KL106, KL112, 48-1789, MAR24) were found to contain the bacterial monosaccharides 6-deoxy-l-talose and l-rhamnose (Kenyon et al., 2017; Kasimova et al., 2021). Importantly, dTDP-6-deoxy-l-talose is produced either from dTDP-l-Rha by the action of C4-epimerase or from dTDP-6-deoxy-l-lyxo-hexos-4-ulose by C4-reductase (Figure 4C; Kenyon et al., 2017). Consequently, the disruption of the dTDP-l-Rha biosynthesis pathway (rmlABCD cluster) can potentially abolish the synthesis of both rare monosaccharides in A. baumannii.

Lipooligosaccharides (LOS) are a major family of glycolipids presented on the outer membrane of the Gram-negative bacteria, and they play central roles in the virulence of many pathogens such as Neisseria gonorrhoeae and C. jejuni. Among other functions, LOS may aid pathogens in evading the host immune system or conferring immune resistance (Preston et al., 1996; Harvey et al., 2000; Song et al., 2000).

Several N. gonorrhoeae strains express the tetrasaccharide lacto-N-neotetraose (LNnT) at their LOS termini which mimics the terminal glycan structure of the human glycosphingolipid precursor paragloboside (Mandrell et al., 1988; Tsai and Civin, 1991). The LNnT termini of LOS can be sialylated with N-acetylneuraminic acid (Neu5Ac) by the N. gonorrhoeae sialyltransferase LsT (Figure 5Aa; Mandrell et al., 1990, 1993; Gulati et al., 2005; Packiam et al., 2006). The sialylated LNnT motif confers resistance to the bactericidal effect of the complement system, a trait also denoted as “serum resistance,” and enables immune evasion (Mandrell et al., 1990; Wetzler et al., 1992; Ram et al., 1998, 2017; Gulati et al., 2005; Ricklin et al., 2010). Due to its key role in the establishment and maintenance of an infection by immune evasion, LsT of N. gonorrhoeae is therefore an attractive target for antivirulence intervention. In recent studies (Gulati et al., 2015), several CMP-nonulosonate analogues were identified which partly inhibited serum resistance and relieved the burden of gonococcal infection in mice models. Of the identified CMP-nonulosonate analogues, CMP-Leg5,7Ac₂ and CMP-ketodeoxynonulosonate (CMP-Kdn; Figure 5B) were found to be most promising as future therapeutics (Gulati et al., 2020). Both compounds were stable in an acidic environment mimicking the human vaginal site of infection. Furthermore, they effectively treated infection with multi-drug-resistant gonococci in mice models presenting a humanized sialome or expressing a humanized complement system. Thus, CMP-Leg5,7Ac₂ and CMP-Kdn are promising candidates for future therapeutics against multi-drug resistant N. gonorrhoeae strains. Interestingly, their mode of action follows the mechanism of metabolic oligosaccharide engineering (MOE), as will be discussed in section “Promising Strategies to Abolish Bacterial Glycosylation Systems”. Interestingly, a recent study identified an alternative terminal epitope of LNnT, featuring a Kdo residue that was transferred by the sialyltransLsT (Jen et al., 2021). This specific LOS structure was identified in the clinical isolates of N. gonorrhoeae and shown to be recognized by anti-Kdo monoclonal antibody 6E4 with potential for the future vaccine development.

Biosynthesis of the LOS core of Campylobacter jejuni is performed by a series of carbohydrate biosynthesis and glycosyltransferase enzymes. Whereas the inner LOS core of C. jejuni (which contains two heptose and two glucose moieties) is conserved (Klena et al., 1998; Gilbert et al., 2002; Kanipes et al., 2004, 2006), the outer core LOS is highly variable among C. jejuni strains. For example, the outer LOS core of C. jejuni strains CCUG 10938 and 10,950 contains the monosaccharides Gal, GalNAc, and Neu5Ac, which together resemble the terminal saccharides of host gangliosides GM1 or GD1a (Figures 5Ab-e; Yuki et al., 2004; Goodfellow et al., 2005; Godschalk et al., 2007; Janssen et al., 2008; Jasti et al., 2016). Bacterial strains that express the enzymes to produce these host-mimicking epitopes have been linked to the development of the autoimmune Guillain-Barré syndrome (GBS) wherein autoantibodies induce damage to nerve gangliosides (Nachamkin et al., 2002; Godschalk et al., 2004; Mortensen et al., 2009; Poole et al., 2018). Counteracting the molecular mimicry of C. jejuni is therefore of interest for reducing immune evasion and severity of GBS following a C. jejuni infection. Inhibition of the glycosyltransferases that are required to build the core LOS glycans of C. jejuni would be an effective way to preclude the immune evasion caused by molecular mimicry of C. jejuni.

The Gram-negative bacterial pathogen P. aeruginosa produces two main types of the lipopolysaccharides: common polysaccharide antigen (CPA) with d-Rha repeats as an outer core (Figure 5Af) and O-specific antigen (OSA) with varied structures across 20 serotypes (heteropolymer of 2–4 monosaccharides: GalNAc, GlcNAc, ManNAc, QuiNAc, GulNAc, FucNAc, l-Rha, Xyl, among others; Lam et al., 2011). Both types of LPS are important for the pathogenesis and survival of P. aeruginosa, as they confer serum resistance, prevent phagocytosis, and promote swimming and swarming motility and biofilm formation (Huszczynski et al., 2020).

Due to the importance of rhamnose in the infection strategies of P. aeruginosa, the dTDP-l-Rha donor synthesis pathway is an attractive antibiotic target. Several studies explored substrate mimics as inhibitors of the TDP-l-Rha biosynthesis. For example, thymidine-based allosteric inhibitors have been developed for RmlA, the first enzyme in the TDP-Rha synthesis that performs condensation of the glucose-1-phosphate (G1P) with thymidine triphosphate (dTTP; Alphey et al., 2013). The library screen and subsequent optimization of the lead compounds yielded a potent inhibitor (called “8a,” Figure 5B) with IC₅₀ of 0.073μM, as determined in vitro. Importantly, via X-ray structure determination and SPR analyses, the authors deduced that 8a acts as a competitive allosteric inhibitor of glucose-1-phosphate binding. The allosteric site of the RmlA plays a role of inducing a negative feedback loop of the TDP-Rha synthesis upon binding the TDP-Rha. It is hypothesized that binding of 8a locks the tetrameric enzyme in a fixed conformation, which prevents G1P binding in the active site.

In a different study, a panel of l-rhamnose 1C-phosphonate and (fluorinated) ketosephosphonate compounds was prepared and evaluated as inhibitors of the TDP-l-Rha biosynthesis enzymes from P. aeruginosa and Streptococcus pneumoniae (Loranger et al., 2013). l-rhamnose 1C-phosphonate (called “11,” Figure 5B) was determined to be the best with an IC₅₀ of 5.7mM. Compound 11 is expected to behave as a competitive inhibitor of the G1P binding in the active site of Cps2L of S. pneumoniae (RmlA in P. aeruginosa), and addition of a thymidine moiety may improve the potency of the inhibitor.

One of the best-studied examples of bacterial cytotoxins is TcdA and TcdB of the opportunistic pathogen C. difficile, with TcdB expressed predominantly in hypervirulent strains (Figure 8A). These clostridial toxins are the main determinants of bacterial pathogenesis, as they form pores in the host cells and modulate cell death, thereby spreading the infection. TcdA/TcdB toxins are composed of four domains, namely, transporter domain, receptor-binding domain, cysteine protease domain (CPD), and an N-terminal glycosyltransferase domain (GT; Di Bella et al., 2016; Aktories et al., 2017). Upon acidic endocytosis into the host cell, the toxins are translocated through the membrane where the CDP domain catalyzes cleavage and release of the N-terminal GT domain (Figure 8A; Di Bella et al., 2016). Subsequently, the GT domain transfers d-glucose onto threonine residues of host cell Rho-guanosine triphosphatases (Rho-GTPase; Just et al., 1995a,b; Kuehne et al., 2010). This leads to necrosis characterized by cell rounding, membrane blobbing, and finally, cell death (Sehr et al., 1998; Genth et al., 1999; Voth and Ballard, 2005).

Several HTS studies were performed to identify small-molecule inhibitors of C. difficile toxins. A screen of six million compounds, followed by extensive optimization of the lead compounds via chemical synthesis and SAR analysis, yielded compounds “13j” and “13l” (Figure 8B). These compounds share a benzodiazepinedione core and displayed potent inhibitory activity against TcdB (low nM IC₅₀ in vitro, low μM in a cell assay; Letourneau et al., 2018). Interestingly, the compounds were not bactericidal to C. difficile or gut bacteria. However, all potent compounds demonstrated low mouse plasma stability and rapid clearance. The same research group also reported the biological evaluation of compound VB-82252 (Figure 8B), which exhibited low plasma stability, but high oral bioavailability (Stroke et al., 2018). Compound VB-82252 was found to be a potent inhibitor (IC₅₀ of 32nM) of UDP-Glc hydrolysis by TcdB (used as a measure of TcdB activity), as determined in an in vitro assay (Stroke et al., 2018). The compound was effective in preventing CHO cells rounding in an assay with several strains of C. difficile. The therapeutic efficiency of VB-82252 was further evaluated in a mouse and hamster C. difficile disease model, where it was effective in sustaining body weight and prolonging the survival of the animals.

In addition, the cell rounding assay was also employed to quantify the effects of various approved therapeutics on TcdB toxins inhibition (Tam et al., 2018). From this library screen assay, the drug niclosamide (Figure 8B), originally developed to treat GI parasites, was most potent (EC₅₀~0.5 μM) in protecting the human cells from rounding. Interestingly, niclosamide does not inhibit the TcdB toxin directly, but instead increases the pH of the host endosomes which disrupts the toxin uptake into the cells. Treatment with niclosamide was effective in a murine model of infection while it exhibited no bactericidal effect on C. difficile or beneficial gut bacteria (determined with an MIC assay and diversity monitoring, respectively).

In an alternative approach to TcdA/B inhibition, small-molecule inhibitors of the cysteine protease domain (CPD) of the C. difficile toxin were identified (Bender et al., 2015). By utilizing fluorescence-polarization HTS of compound libraries with clinically safe drugs (e.g., LOPAC library), multiple inhibitors were identified, with the selenium-containing drug ebselen (Figure 8B) exhibiting the highest potency (IC₅₀ 6.9nM). Notably, the drug could preclude the GT domain release and cell rounding. The effects of the drug were confirmed to be due to the prevention of Rho-GTPases glucosylation, and it was shown to be effective in a murine model of C. difficile infection. Importantly, ebselen is a developed drug in late clinical trials for the treatment of tinnitus, hearing loss, and bipolar disorder and has been proven safe for use in humans.

Besides conventional antibiotics to treat a C. difficile infection, the TcdB-neutralizing antibody bezlotoxumab is an FDA-approved therapy against recurring C. difficile (Zinplava, Merck; Navalkele and Chopra, 2018). Bezlotoxumab binds the N-terminal part of the receptor-binding domain of the TcdB toxin, preventing toxin binding and entry into host cells (Orth et al., 2014). It was also effective against hypervirulent C. difficile strains (NAP1, BI, 027). Currently, bezlotoxumab is only used in combination with antibiotic treatments and is not a stand-alone therapy against C. difficile.

Adhesion is one of the first step in the bacterial colonization of the host. It is mediated by various adhesion factors presented on the surface of the bacterium that recognize and bind to the host cell receptors (Chagnot et al., 2013; Poole et al., 2018). Adhesin proteins in particular are often (hyper)glycosylated, and the presence of glycans often plays a vital role for their stability and proper function (Lu et al., 2015). Therefore, glycosylation of the adhesion factors is an attractive target for the development of novel anti-adhesive therapies.

Diffusely adhering E. coli (DAEC), enterotoxigenic E. coli (ETEC), and the murine pathogen Citrobacter rodentium share a common adhesion mechanism to host cells. These bacteria rely on a type Va secretion system, which is also called a self-associating autotransporter (SAAT) system (Lu et al., 2014). Autotransporter proteins consist of a C-terminal β-barrel domain that forms a transport channel in the outer membrane and a passenger domain which is translocated through this channel and fulfils the effector adhesion function (Leyton et al., 2012). In DAEC, ETEC, and C. rodentium, the passenger domains of autotransporters AIDA-I, TibA, and CARC, respectively, are O-hyperglycosylated with bacteria-specific d-glycero-d-manno-heptose (ddManHep Figure 1) by a cognate GT belonging to the bacterial autotransporter heptosyltransferase (BAHT) family (Lu et al., 2014, 2015). Hyperglycosylation is important for the successful adherence of AIDA-I to HeLa cells (Benz and Schmidt, 2001) and was later found to enhance protein stability (Charbonneau et al., 2007). Similarly, TibA is the SAAT of enterotoxigenic E. coli and depends on hyperheptosylation for stability as it was found to mediate its (re)folding and subsequently influence adherence function (Côté et al., 2013). Interestingly, heptose residues also constitute the LPS core of Gram-negative bacteria, and the synthesis pathway of ADP-l-glycero-β-d-manno-heptose (Kneidinger et al., 2002) is considered a promising target for inhibitors. Several studies have already identified inhibitors with IC₅₀ values in the milli/micromolar range (De Leon et al., 2006; Kim et al., 2021a,b). It would be interesting to investigate whether these inhibitors indeed abolish SAAT hyperheptosylation and subsequent adherence of the bacterial cells.

Non-typeable Haemophilus influenzae (NTHi) utilizes a type Vb secretion system (also called two-partner secretion (TPS) pathway) to transport and present high molecular weight (HMW) adhesin proteins on the surface as a first step in host colonization (St Geme et al., 1993; Grass and St Geme, 2000). Stability and efficient surface tethering of HMW adhesins is dependent on N-hyperglycosylation on asparagine with simple mono- and disaccharides of glucose (Glc; Grass et al., 2003). A total of 31 glycosylation sites have been identified at asparagine residues in the Asn-X-Ser/Thr consensus sequence of HMW1A (Grass et al., 2003, 2010; Gross et al., 2008), modified by the action of the associated glycosyltransferase HMW1C (Grass et al., 2010). Interestingly, the glycosylation of HMW1A by HMW1C follows an unconventional OTase-independent N-glycosylation pathway, wherein cytoplasmic HMW1C transfers single nucleotide-activated carbohydrates to the acceptor protein HMW1A. Upon deletion of the genes encoding for HMW1C and UDP-Glc biosynthesis, hmw1c or galU, respectively, HMW1A surface presentation as well as adhesion to epithelial cells was abolished in vitro (Grass et al., 2010). Interestingly, we recently revealed that hyperglycosylation is established through a semiprocessive mechanism in vitro (Yakovlieva et al., 2021). Homologues of the HMW1C glycosyltransferase have been identified in Kingella kingae (HMW1C_Kk) and Aggregatibacter aphrophilus (HMW1C_Aa) where they perform the glycosylation of cognate trimeric autotransporters Knh and EmaA, respectively (Rempe et al., 2015). Analogously to the H. influenzae HMW1A, abolishing glycosylation of Knh and EmaA was shown to inhibit the bacterial aggregation and adherence to the host cells.

Multiple members of the serine-rich repeat proteins (SRRPs) of clinically relevant Gram-positive bacteria are found to be (hyper) O-glycosylated with carbohydrates that influence stability and adhesive function. Examples include fimbriae-associated protein Fap1 of Streptococcus parasanguinis, GspB of Streptococcus gordonii, SraP of S. aureus, PsrP of S. pneumoniae, and others, which are reviewed elsewhere (Zhou and Wu, 2009; Lizcano et al., 2012). Glycosyltransferases termed Gtf1-Gtf2 (GtfA-GtfB) are responsible for the core GlcNAc modification on the Ser/Thr residues of SRRPs. These enzyme pairs operate in tandem with Gtf1 performing the glycosylation reaction and Gtf2 acting as a chaperone and substrate-binding domain (Wu and Wu, 2011; Chen et al., 2016; Zhao et al., 2018). After attachment of the initial GlcNAc, the character of the glycan modifications varies between different SRRPs and features GlcNAc/Glc (Srr1, GspB; Bensing et al., 2004; Chaze et al., 2014), Glc/GlcNAc/Rha (Fap1; Zhu et al., 2016), and GlcNAc (SraP; Li et al., 2014). The presence of the multiple glycans on the SRRPs is crucial for their function in conferring adhesion and biofilm formation of the Gram-positive pathogens and therefore constitutes an interesting antibacterial target.

Neisseria gonorrhoeae produces type IV pili (TFP; Patel et al., 1991) that are required for effective adhesion to epithelial cells in an initial stage of the infection (Swanson, 1973; McGee and Stephens, 1984; Virji and Heckels, 1984; Craig et al., 2004). The glycans of TFP interact with complement receptor 3 (CR3), an innate pattern recognition receptor expressed on human cervical cells. The PilE subunits that make up the TFP feature glycans containing either a N,N′-diactetylbacillosamine (diNAcBac) or a galactose-modified diNAcBac (Gal(α1-3)diNAcBac) linked to serine residues (Jennings et al., 1998, 2011; Power et al., 2003; Hegge et al., 2004; Hartley et al., 2011). Only N. gonorrhoeae cells carrying the disaccharide Gal(α1-3)diNAcBac on their PilE proteins survive infection of the primary human cervical (Pex) cells, while TFP decorated with a single diNAcBac die within the cervical cells, even though they were found to be hyperinvasive. Glycosylation of PilE follows an OTase-dependent O-glycosylation pathway by multiple pilin glycosylation genes (pgl), which encode enzymes for the synthesis and attachment of diNAcBac to an intermediate lipid carrier, as well as GTs that attach galactose or glucose to diBacNAc (Hartley et al., 2011). Considering the importance of PilE glycosylation in the colonization capacity of N. gonorrhoeae, inhibitors of the pgl enzymes are attractive antibacterial agents. However, no inhibitors of N. gonorrhoeae pgl enzymes have been reported to date. Nonetheless, there are alternative approaches for targeting the interaction between the glycosylated TFP and CR3. Recently, two clinically approved drugs have been identified that inhibit the interaction of glycosylated PilE of N. gonorrhoeae with the l-domain of CR3. The drugs carbamazepine and methyldopa act as competitive inhibitors of CR3 binding and thereby efficiently blocked N. gonorrhoeae infection in Pex cells. Importantly, both drugs were also effective against multi-drug resistant gonococci and did not lead to development of resistance.

Efflux pumps are membrane proteins involved in the transport of various molecules (Alcalde-Rico et al., 2016). In bacterial pathogens, efflux pumps are often responsible for ejecting antibiotics from bacterial cells. They are especially prominent in the multi-drug resistance species and are an attractive drug target, especially in combination therapies (Ferrer-Espada et al., 2019; Marshall et al., 2020; Rodrigues et al., 2020).

In C. jejuni, the CmeABC complex is the main multi-drug efflux pump that confers resistance to various antibiotics. Together, CmeA, CmeB, and CmeC form a superstructure that spans the inner membrane, periplasmic space and creates a pore in the outer membrane of the bacterial cell. It was previously reported to be N-glycosylated with complex C. jejuni N-glycans (Abouelhadid et al., 2020). In a recent study, the importance of N-glycosylation for CmeABC efflux pump function was revealed (Dubb et al., 2020). Abolishing glycosylation of CmeA, which spans the periplasmic space, led to the increased accumulation of ethidium bromide and significant increase in antibiotic susceptibility, both indicating the impaired functioning of the efflux pump machinery. Additionally, the loss of CmeA glycosylation resulted in the loss of colonization ability of the chicken ceca.

With the increasing knowledge of bacterial glycosylation systems, a variety of strategies to inhibit the enzymes involved has been developed. For instance, inhibitors of the enzymes involved in production of the carbohydrate-nucleotide donors, such as dTDP-Rha (vide supra), have been developed and described elsewhere (Alphey et al., 2013; Loranger et al., 2013; Van Der Beek et al., 2019). In addition, several classes of compounds have been designed as inhibitors for glycosyltransferases (Compain and Martin, 2001; Kajimoto and Node, 2009; Tedaldi and Wagner, 2014; Ema et al., 2018; Conforti and Marra, 2021). For this review, we decided to focus on the technique of MOE, as a promising strategy to interfere with bacterial glycosylation systems in a specific manner.

The technique of MOE, as originally developed by Reutter (Kayser et al., 1992) and Bertozzi (Mahal et al., 1997; Bertozzi and Saxon, 2000), relies on hijacking the cell’s own metabolism to introduce carbohydrate variants with altered properties. In this way, carbohydrate precursors carrying bioorthogonal handles can be introduced into the native glycans by permissive enzymes, allowing the subsequent attachment of reporter groups to detect carbohydrate incorporation and glycan production. While the technique was originally developed on eukaryotic cells, the interest in applying MOE to bacterial cells is steeply rising, and several bacterial glycans have now been targeted with unnatural carbohydrates (Tra and Dube, 2014; Clark et al., 2016).

In addition to the promising application of labeling bacterial glycans for visualization, the MOE technique can also be used to introduce monosaccharide analogues that inhibit the proper assembly of bacterial glycans. To this end, both substrate decoys, which act as surrogate glycan acceptor sites (Dimitroff et al., 2003; Metastasis et al., 2009; Gloster and Vocadlo, 2012; Rillahan et al., 2012; Villalobos et al., 2015), and chain-terminating carbohydrate analogues, which lack a specific hydroxyl group for elongation (Li et al., 2016) have been developed for different bacterial strains. In a recent study, analogues of DiNAcBac, FucNAc, and DATDG were employed both as substrate decoys and inhibitors (Figure 9) to perturb glycan synthesis in H. pylori (Williams et al., 2020). The benzyl glycoside analogues BnBac, BnFucNAc, and BnDAT were synthesized as decoy substrates, and fluoro analogues F-Bac, F-FucNAc, and F-DAT were designed as chain-terminating inhibitors. Interestingly, treatment of H. pylori with BnBac, BnFucNAc, and F-DAT resulted in reduction of glycoprotein synthesis and defects in growth, biofilm formation, and motility. These functional defects could be largely reproduced in an isogenic H. pylori ∆GT mutant lacking a functional glycosylation system, proving that the MOE approach indeed has potential to be an antivirulence strategy. In addition, the analogues under study here also revealed bacteria-specific effects. In C. jejuni, none of the carbohydrate analogues impacted glycan biosynthesis or fitness, and only subtle changes were observed in the commensal Bacteroides fragilis. It will be interesting to test these carbohydrate analogues in animal models of infection and to understand their potential as narrow-spectrum antivirulence compounds.