|Home | About | Journals | Submit | Contact Us | Français|
Colitose, also known as 3,6-dideoxy-l-galactose or 3-deoxy-l-fucose, is one of only five naturally occurring 3,6-dideoxyhexoses. Colitose was found in lipopolysaccharide of a number of infectious bacteria, including Escherichia coli O55 & O111 and Vibrio cholera O22 & O139. To date, no colitosyltransferase (ColT) has been characterized, probably due to the inaccessibility of the sugar donor, GDP-colitose. In this study, starting with chemically prepared colitose, 94.6 mg of GDP-colitose was prepared via a facile and efficient one-pot two-enzyme system involving an l-fucokinase/GDP-l-Fuc pyrophosphorylase and an inorganic pyrophosphatase (EcPpA). WbgN, a putative ColT from E. coli O55:H5 was then cloned, overexpressed, purified and biochemically characterized by using GDP-colitose as a sugar donor. Activity assay and structural identification of the synthetic product clearly demonstrated that wbgN encodes an α1,2-ColT. Biophysical study showed that WbgN does not require metal ion, and is highly active at pH 7.5–9.0. In addition, acceptor specificity study indicated that WbgN exclusively recognizes lacto-N-biose (Galβ1,3-GlcNAc). Most interestingly, it was found that WbgN exhibits similar activity toward GDP-l-Fuc (kcat/Km = 9.2 min−1 mM−1) as that toward GDP-colitose (kcat/Km = 12 min−1 mM−1). Finally, taking advantage of this, type 1 H-antigen was successfully synthesized in preparative scale.
Deoxysugars are widely found in natural products of plant and fungi, and secondary metabolites of bacteria, which are common sources for antibiotics or anticancer agents, e.g. anthracyclines, avermectins, enediynes and angucyclines. A great number of bacterial lipopolysaccharide (LPS) and capsular polysaccharides also contains diverse deoxysugars, for example, l-rhamnose (6-deoxy-l-mannose, found in nearly one-third of Escherichia coli LPS), 6-deoxy-l-talose, l-fucose (6-deoxy-l-galactose) and d-FucNAc (6-deoxy-GalNAc). Deoxysugars are not only structural components of these bioactive molecules, but in many cases are critical for the biological recognition of the mother compounds, serving as ligands for cell–cell interaction or as targets for toxins, antibodies, and microorganisms (Weymouth-Wilson 1997).
Among the large number of monosaccharide residues found as components of O-antigen of bacteria LPS, 3,6-dideoxyhexoses have drawn special attentions due to their highly immunogenic characteristics. Of the eight possible stereoisomers of 3,6-dideoxyhexoses, only five exist naturally, including ascarylose, abequose, paratose, tyvelose, and colitose. In particular, colitose was found in a number of pathogenic bacteria, for example, E. coli O55 & O111, Vibrio cholera O22 & O139, Salmonella enterica O35 & O50, Yersinia pseudotuberculosis O6 and O7 & O10 (Knirel et al. 1996, 1998; Senchenkova et al. 1997; Ovchinnikova et al. 2007; Kenyon et al. 2011). In these cases, colitose locates at the termini of O-antigens, serving as a special antigenic component of LPS (Luderitz et al. 1958, 1960; Westphal et al. 1959). The presence of colitose in pathogenic E. coli may enhance the lipophilic character of endotoxin LPS (Medearis et al. 1968). Colitose was also considered an essential component of specific ligands of certain LPS-binding lectins, e.g. houseshoe crab tachylectin-4 (specifically binds to E. coli O111 LPS) (Saito et al. 1997). In addition, aerobic marine heterotrophic bacteria of the genera Alteromonas, Echinicola and Pseudoalteromonas also contain colitose in their O-antigens (Gorshkova et al. 2001; Muldoon et al. 2001; Silipo et al. 2005; Tomshich et al. 2015). Given the emerging importance, chemical synthesis had been performed to prepare colitose and colitose-containing oligosaccharides for vaccine development (Senchenkova et al. 1997; Ruttens and Kovac 2004; Turek et al. 2006; Calin et al. 2012, 2013). In contrast, there are limited studies on the biosynthesis of colitose-containing O-antigen, especially on characterization of colitosyltransferases (ColT) that are responsible for the transfer of colitose. The lack of efficient approaches for the access of sufficient amounts of the sugar donor, GDP-colitose, has been a roadblock.
Different from the other four 3,6-dideoxyhexoses (ascarylose, abequose, paratose and tyvelose) that are derived from CDP-glucose, the biosynthetic pathway of colitose (also known as 3,6-dideoxy-l-galactose or 3-deoxy-l-fucose) begins from mannose-1-phosphate, and share a few enzyme-catalyzed steps with that of GDP-l-galactose and GDP-l-Fuc. This is understandable when concerning structural similarities between colitose and l-galactose/l-fucose (l-Fuc) (Figure 1). Functional assignment and biochemical characterization (Beyer et al. 2003; Alam et al. 2004) of putative enzymes in GDP-colitose biosynthesis gene clusters revealed a similar pathway as that of other 3,6-dideoxyhexoses. As depicted in Figure 1, the biosynthesis of GDP-colitose begins with GDP-mannose pyrophosphorylase (or ColE)-catalyzed pyrophosphorylation of mannose-1-phosphate to form GDP-mannose, followed by GDP-mannose 4,6-dehydratase (or ColB)-catalyzed intramolecular oxidoreduction of GDP-mannose to afford GDP-4-keto-6-deoxy-mannose (GKDM), an key intermediate for biosynthesis of both GDP-colitose and GDP-l-Fuc. Subsequently, GKDM was converted into GDP-4-keto-3,6-dideoxy-mannose (GKDDM) by GDP-4-keto-6-deoxy-mannose-3-dehydrase-catalyzed C-3 deoxygenation in the presence of l-glutamate and cofactor pyridoxal-5′-phosphate (Alam et al. 2004). Finally, GDP-colitose is generated by GDP-colitose synthase, a bifunctional enzyme catalyzing the C-5 epimerization of GKDDM and the subsequent C-4 keto reduction (Alam et al. 2004). The last step requires equal moles of NAD(P)H as that of GKDDM. Given the involvement of multiple steps and cofactors, and the structural similarities between final product and intermediates, one can imagine it must be complicated to produce and purify milligrams of GDP-colitose via such a biosynthetic pathway (Heath and Elbein 1962; Isshiki et al. 1996). A facile and efficient approach is thus in demand for large-scale preparation of GDP-colitose.
Escherichia coli O55:H7 belongs to enteropathogenic E. coli, a group of E. coli isolates among the major causes of acute and prolonged infantile diarrhea in developing countries (Ochoa and Contreras 2011). Genomic analysis revealed the E. coli O55:H7 is closely related to the most notorious E. coli isolate O157:H7, which accounts for majority of E. coli infections in North America every year, leading to severe acute hemorrhagic diarrhea, complication hemolytic uremic syndrome, even death (Sodha et al. 2015). The O-antigen repeating unit of E. coli O55:H7 contains five monosaccharide residues, with a structure of Colα1,2-Galβ1,3-GlcNAcβ1,3-Galα1,3GalNAc (Figure 2) (Lindberg et al. 1981). The O-antigen biosynthesis gene cluster has been sequenced (Wang et al. 2002), unveiling four putative glycosyltransferase genes (wbgM, wbgN, wbgO and wbgP) for the assembly of the repeating unit (Figure 2). Among these four genes, we have biochemically demonstrated that wbgO encodes an β1,3-galactosyltransferase (Liu et al. 2009). WbgN shares 23% sequence identity and 47% sequence similarity with human FUT2 (α1,2-fucosyltransferase), and was thought to the ColT. In this study, we cloned the wbgN gene, overexpressed and purified the recombinant protein, and performed biochemical characterization using chemoenzymatically prepared GDP-colitose as sugar donor.
De novo biosynthetic pathways of GDP-l-Fuc and GDP-colitose share the same initial steps including generation of intermediates mannose-1-phosphate, GDP-mannose and GKDM (Figure 1). The complexity of the pathways and lack of cost-effective approaches to access intermediates greatly restricted large-scale preparation of these valuable sugar donors. In efforts to develop facile and cost-effective approaches for large-scale synthesis of GDP-l-Fuc and derivatives, we and others exploited l-fucokinase/GDP-l-Fuc pyrophosphorylase (FKP) (Wang et al. 2009; Yi et al. 2009; Zhao et al. 2010), a bifunctional enzyme found in the fucosylation salvage pathway of a human intestinal bacterium, Bacteroides fragilis 9343 (Coyne et al. 2005). As depicted in Figure 3, the salvage pathway involves (i) l-fucokinase activity of FKP catalyzed formation of l-Fuc-1-phosphate (l-Fuc-1-P) from l-Fuc in presence of ATP and (ii) GDP-l-Fuc pyrophosphorylase activity of FKP catalyzed generation of GDP-l-Fuc and byproduct pyrophosphate in the presence of GTP. Using a one-pot two-enzyme system which contains FKP and an inorganic pyrophosphatase (drive reaction forward by digesting byproduct pyrophosphate), GDP-l-Fuc and a number of C-5 modified derivatives including GDP-l-galactose were successfully prepared in large scale (Wang et al. 2009; Yi et al. 2009). In another example, a C-2 modified derivative, GDP-2-deoxy-2-fluoro-l-fucose (GDP-Fuc2F), was also efficiently synthesized (Rillahan et al. 2012). These reports indicated that FKP is very promiscuous towards l-Fuc and derivatives.
To explore whether FKP could tolerate colitose (3-deoxy-l-fucose) as a substrate, we chemically prepared colitose starting with l-Fuc as previously reported (Ruttens and Kovac 2004). Activity assays of FKP towards colitose were carried out in presence of ATP (for fucokinase activity) or both ATP and GTP (for pyrophosphorylase activity), with positive controls using l-Fuc as substrate instead of colitose. As shown in Figure 4, after incubation with FKP, both spots corresponding to colitose and ATP weakened, whereas two new spots appeared (Figure 4, lane 5). Mass spectroscopy (MS) analysis of the reaction mixture (Supplementary data, S2) showed two peaks at m/z 227.0290 and 426.0151, corresponding to colitose-1-phoshate and ADP. In addition, when GTP was added to above reaction mixture, the spot corresponding to colitose-1-phoshate disappeared, yielding a new spot (Figure 4, lane 6) of GDP-colitose, which is later confirmed by MS analysis with a m/z peak of 572.0705 [M−H]−. These results clearly indicate that FKP can well accept colitose and efficiently catalyze the formation of GDP-colitose. To confirm the structure, large-scale synthesis of GDP-colitose was carried out using the one-pot two-enzyme system as illustrated in Figure 3. After P2 purification, 94.6 mg of GDP-colitose was obtained, with a total yield of 83%. The chemical structure was confirmed by high-resolution MS and NMR (see Supplementary data, S3 for spectra). Such a one-pot two-enzyme system provides a facile, efficient and cost-affordable approach for the access of large scales of GDP-colitose.
According to sequence similarities, WbgN was classified into glycosyltransferase family 11 (www.cazy.org, accessed 01 February 2016), a group of enzymes with known activities of α1,2-l-fucosyltransferase (α1,2-FucT) and α1,3-l-fucosyltransferase (α1,3-FucT). In addition, amino acid sequence of WbgN shares 37% identity and 57% similarity with a recently identified α1,2-FucT, WbgL, which is involved in the biosynthesis of E. coli O126 O-antigen (Engels and Elling 2014). Considering the structure similarities between O-antigen of E. coli O55:H5 (contains Colα1,2-Galβ1,3-GlcNAc motif) and O126 (contains l-Fucα1,2-Galβ1,3-GlcNAc motif), it is assumed that wbgN encodes the α1,2-ColT that is responsible for transfer of colitose from GDP-colitose to the Gal residue of Galβ1,3-GlcNAc via an α1,2-linkage. To test this hypothesis, the wbgN gene was codon optimized (host E. coli), synthesized and cloned into pQE80 L for expression. Even though induction was performed under a low temperature (16°C) with minimum amounts of inducers (0.1 mM IPTG), majority of WbgN proteins tend to precipitate (Figure 5, lane 4), further optimization to increase soluble form proteins failed. Such phenomena were found common among O-antigen synthesis enzymes (especially fucosyltransferases) (Li, Liu, et al. 2008; Li, Shen, et al. 2008; Engels and Elling 2014), which are thought membrane associated and with low solubility (Raetz and Whitfield 2002). Nevertheless, 1.2 mg of soluble WbgN was purified from 1 L of cell culture. As shown in Figure 5 (Lane 7), the N-terminal His-tagged WbgN has an apparent MW of 30 kDa on SDS–PAGE, slightly lower than the calculated value (35 kDa). This is normal for membrane associated hydrophobic proteins (Rath et al. 2009). Western blot using an anti-His antibody as a primary antibody confirmed the result. Alternatively, fusion protein strategy has been extensively used to increase the solubility of α1,2-FucTs (Li, Liu, et al. 2008; Li, Shen, et al. 2008; Engels and Elling 2014), which usually resulted in greatly decreased activity. Therefore, it was not applied in this study.
To reveal the function of WbgN, a panel of sugar nucleotides (GDP-colitose, UDP-Gal, UDP-GlcNAc, and UDP-GalNAc) were tested using lacto-N-biose (LNB, Galβ1,3-GlcNAc) as an acceptor. MS analysis of the reaction with GDP-colitose as donor found a new peak at m/z of 536.1933 (Supplementary data, S2), corresponding to the trisaccharide product Colα1,2-Galβ1,3-GlcNAc [M + Na]. No new peaks were found when using other sugar nucleotides as donors. Thus, the donor specificity study indicated that wbgN encodes ColT, consistent with bioinformatics analysis. To investigate whether WbgN catalyzes the formation of the α1,2-linkage, large-scale preparation of the trisaccharide were performed. Briefly, a 5 mL reaction system was set up containing 30 mg of LNB (16 mM), 1.25 equiv. of GDP-colitose, WbgN (20 µg) and 50 mM Tris–HCl (pH 8.0) buffer. After LNB was totally converted (monitored by TLC), the trisaccharide product was isolated by gel filtration chromatography with a total yield of 92% (37 mg), The structure of trisaccharide was confirmed by NMR spectroscopy including 1H, 1H–1H COSY, 13C, HSQC and HMBC (Supplementary data, S4). The α-linkage of Col with Gal was identified by the coupling constant (JH1−H2 = 4.8 Hz) of colitose. Moreover, HMBC revealed a good correlation of the anomeric carbon (C-1) of colitose with H-2 of Gal, which indicated 1,2-linked colitose with Gal. Collectively, these results provide solid evidences that wbgN encodes an α1,2-ColT that transfers colitose onto LNB via a α1,2-linkage. To the best of our knowledge, WbgN is the first biochemically characterized ColT.
Study on pH optimum with LNB as an acceptor revealed that WbgN is highly active with in a broad pH spectrum ranging from pH 7.5 to 9.0, and exhibits ~60% activity at a higher basic condition (pH 10.2) (Figure 6A). Sequence analysis showed that WbgN does not contain a DXD motif, a conserved amino acid sequence exists in all glycosyltransferase superfamily A (GT-A) members (Breton et al. 2006). Such DXD motifs play a role in coordinating the binding of sugar nucleotides in the active site with the requirement of certain divalent metal cation. This is consistent with activity assay results showing metal ion is not necessary for WbgN. To further investigate the metal ion dependence, a panel of divalent metal cations and EDTA were tested with positive (no additive) and negative (no enzyme) controls. As shown in Figure 6B, WbgN exhibited highest activities in the presence or the absence of EDTA, indicating that a metal ion is not required for the activity of WbgN. In contrast, the addition of metal ions dramatically decreased its activity to 0–40%. It is thus assumed that instead of GT-A, WbgN belongs to GT-B superfamily, which is in accordance with similar fucosyltransferases WbgL (Engels and Elling 2014), WbsJ (Li, Liu, et al. 2008) and WbiQ (Pettit et al. 2010).
Concerning the sequence similarities between WbgN and a number of α1,2-FucT (e.g. WbgL from E. coli O126, WbsJ from E. coli O128 and WbwK from E. coli O86), as well as the structural similarities between sugar donors GDP-colitose and GDP-l-Fuc, it is of great interest to reveal whether WbgN could recognize GDP-Fuc. To this end, reaction was conducted using GDP-Fuc as the sugar donor and LNB as the acceptor. Briefly, a 50 µL reaction was set up containing 50 mM Tris–HCl (pH 8.0), 5 mM acceptor Galβ1,3-GlcNAc, 5 mM of acceptor GDP-l-Fuc and 50 µg of recombinant WbgN. The reaction was allowed to proceed at 37°C for 2 h before subjecting to MS analysis. Two new peaks at m/z of 530.2071 and 552.1888 were found on MS profile, corresponding to the trisaccharide product l-Fucα1,2-Galβ1,3-GlcNAc [M + H]+, and [M + Na]+. In addition, MS peak corresponding to the acceptor LNB disappeared (Supplementary data, S2). This result clearly indicated that WbgN can employ GDP-l-Fuc as a sugar donor efficiently. To further investigate the donor preference of WbgN, kinetic characterization was performed. As listed in Table I, WbgN has a lower Km value toward GDP-colitose (0.76 ± 0.18 mM) than that toward GDP-l-Fuc (1.3 ± 0.42 mM), implying that GDP-colitose has a higher enzyme affinity. In contrast, a higher turnover rate was found toward GDP-l-Fuc (kcat = 12 ± 1.5 min−1) than that toward GDP-colitose (kcat = 8.8 ± 0.8 min−1), indicating that WbgN-catalyzed reaction is more favorable towards the trisaccharide product containing l-Fuc. Collectively, WbgN exhibits comparable activities toward both sugar nucleotide donors, with a kcat/Km value of 12 min−1 mM−1 toward GDP-colitose, and that of 9.2 min−1 mM−1 toward GDP-l-Fuc.
Among all colitose-containing bacterial polysaccharide identified so far (http://csdb.glycoscience.ru/bacterial/), none contains l-Fuc residue(s). This could be caused by (i) the lack of sugar donor GDP-l-Fuc or/and (ii) the lack of corresponding FucT(s). Genetic analysis revealed that GDP-l-Fuc synthase encoding gene (fcl) is always missing or inactivated in gene clusters responsible for the biosynthesis of these bacterial polysaccharides (Bastin and Reeves 1995; Yamasaki et al. 1999; Wang and Reeves 2000; Wang et al. 2002; Cunneen et al. 2011; Kenyon et al. 2011), supporting the first assumption. On the other hand, our results clearly showed that, in E. coli O55:H7, ColT can also function as FucT (or possibly is intrinsic FucT), excluding the second assumption. This phenomenon could be common among other colitose-containing bacteria, as putative ColTs usually share significant sequence similarities with FucTs (Bastin and Reeves 1995; Yamasaki et al. 1999; Wang et al. 2002; Wang and Reeves 2000; Cunneen et al. 2011; Kenyon et al. 2011). Biochemical characterizations are needed to confirm such hypothesis.
The purified WbgN was used to investigate its substrate specificity towards a group of mono- and disaccharide acceptors. As shown in Table II, WbgN is highly specific for LNB in the presence of either GDP-colitose or GDP-l-Fuc, and cannot tolerate any other acceptors tested, including galacto-N-biose (GNB, Galβ1,3-GalNAc), LacNAc (Galβ1,4-GlcNAc) and lactose. These results indicated that not only the terminal Gal but also the adjacent sugar residue and the way they coupled together are involved in the enzyme-acceptor binding. Such a strict acceptor preference can be applied specifically for the preparation of type 1 H-antigen (l-Fucα1,2-Galβ1,3-GlcNAc), or complex glycans that contain the antigen. In a test trial, 40 mg (95% yields) of type 1 H-antigen was synthesized in a WbgN-catalyzed reaction, starting with LNB and GDP-l-Fuc (see Supplementary data, S5 for NMR characterization). Taken together, the acceptor pattern shown in the assay is in agreement with the proposed function of WbgN in the biosynthesis of the E. coli O55:H7 O-antigen.
In summary, taken advantage of relaxed substrate specificity of FKP, we have developed a facile and cost-efficient system for enzymatic synthesis of rare sugar nucleotide GDP-colitose in preparative scale. Using the synthesized GDP-colitose, we have biochemically characterized the first ColT (WbgN) from E. coli O55:H7. WbgN was found high active in a pH range of 7.5–9.0, and a metal ion is not required for its activity. Interestingly, WbgN exhibited similar activity towards both GDP-colitose and GDP-l-Fuc, taking together with the fact that ColTs and FucTs share significant sequence similarities, it is possible that many ColTs may have FucT activities. Acceptor substrate specificity study revealed that WbgN is specific for lacto-N-biose, which may found great use in the preparation of type 1 H-antigen-containing complex glycans.
E. coli DH5α competent cells were purchased from Life Technologies (Carlsbad, CA), E. coli BL21(DE3) competent cells and Nitrocellulose membrane were purchased from New England BioLabs (Ipswich, MA). Vector pQE80L and Ni-NTA agarose were from Qiagen (Valencia, CA). Anti-His monoclonal antibody (from mouse) was from Sigma (St. Louis, MO). HRP-linked anti-mouse IgG, and ECL plus western blot detection reagents were purchased from GE healthcare (Piscataway, NJ). Fast Digest restriction enzymes were purchased from Fermentas (Glen Burnie, MD). Shrimp alkaline phosphatase was from New England Biolabs (Ipswich, MA). GDP-Fuc, Galβ1,3-GlcNAc, Galβ1,3-GalNAc, Galβ1,4-GlcNAc (LacNAc), FKP and EcPpA were prepare previously in in our group (Zhao et al. 2010; Li et al. 2013, 2015). All other chemicals and solvents were purchased from Sigma-Aldrich.
Full-length putative α1,2-l-ColT gene (wbgN) from E. coli O55:H7 (GeneBank: 18266395) was codon optimized (target host, E. coli) and synthesized (Qinglan Bio, China) together with 5′-terminal BamHI restriction sites and 3′-terminal HindIII restriction sites (see Supplementary data, S1 for optimized gene sequence). The gene was then digested with BamHI and HindIII, and inserted into pQE80L pre-treated with same restriction enzymes, yielding an expression plasmid pQE-WbgN. After DNA sequencing, the plasmid was transformed into E. coli BL21(DE3) for overexpression.
Escherichia coli BL21 (DE3) harboring the pQE-WbgN was grown at 37°C in 1 L of LB medium (Lennox) with 100 µg/mL Ampicillin until OD600 reached 0.6–0.8. After cooling the culture on ice for 20 min, isopropyl 1-thio-β-d-galactopyranoside (IPTG) was added to a final concentration of 0.1 mM. Expression was allowed to proceed at 16°C for 20 h. Cells were harvested by brief centrifugation and stored at −20°C until use.
To purify soluble recombinant enzymes, cells were resuspended in buffer A (20 mM Tris–HCl, pH 8.0, 0.3 M NaCl, 5 mM imidazole, 0.2% Triton X-100 and 10% glycerol), and disrupted by a Microfluidics M-110P Homogenizer. The lysate was cleared by centrifugation (15,000 × g, 30 min, 4°C) and loaded onto a 3 mL Ni-NTA gravity column preequilibrated with buffer B (20 mM Tris–HCl, pH 8.0, 0.3 M NaCl, 5 mM imidazole and 10% glycerol). The column was washed with 200 mL of buffer B and 200 mL of buffer C (20 mM Tris–HCl, pH 8.0, 0.3 M NaCl, 25 mM imidazole and 10% glycerol). The protein was finally eluted with buffer D (20 mM Tris–HCl, pH 8.0, 0.3 M NaCl, 250 mM imidazole, 10% glycerol) and desalted against buffer E (50 mM Tris–HCl, pH 8.0, 0.1 M NaCl, 20% glycerol) for long-term storage at −20°C.
To test the fucokinase activity of FKP towards colitose, reactions were carried out in a 50 µL system, containing 100 mM of Tris–HCl (pH 8.0), 20 mM of colitose or l-Fuc (as positive control), 20 mM of ATP, 20 mM of MgCl2 and 10 µg of purified FKP. To test the GDP-l-Fuc pyrophosphorylase activity, 20 mM of GTP and 2 µg of EcPpA were added to the above reaction mixtures. Reactions were incubated at 37°C for 1 h, and analyzed by MS and TLC (developing solvent: isopropanol/ammonium acetate (1 M)/Acetic acid = 7:3:0.1 (v/v/v)). The compounds on TLC plates were stained with anisaldehyde/acetic acid/H2SO4/H2O = 7:3:10:27 (v/v/v/v), and visualized by brief heating. For large-scale GDP-colitose synthesis, a one-pot two-enzyme system similar as described before was used (Zhao et al. 2010). Briefly, a reaction was carried out at 37°C for 5 h in a total volume of 10 mL in 100 mM Tris–HCl (pH 8.0) buffer, containing 20 mM colitose, 22 mM ATP, 22 mM GTP, 20 mM MgCl2, 20 µg/mL FKP and 2 µg/mL EcPpA. The reaction was monitored by TLC. To purify GDP-colitose, the reaction was quenched by the addition of equal volumes of cold ethanol. The mixture was placed on ice for 30 min, and centrifuged to remove debris. The supernatant was then concentrated and loaded onto bio-gel P2 for separation. The fractions contain pure GDP-colitose (detected by TLC) were then concentrated and lyophilized for MS and NMR analysis. The fractions contain impure GDP-colitose were also collected and treated with shrimp alkaline phosphatase (New England Biolabs) to digest ADP byproduct following instructions (100 U, 30°C for 5 h), and then subject to bio-gel P2 for further separation.
GDP-l-Colitose: 1H NMR (400 MHz, D2O) δ: 8.00 (s, 1 H), 5.83 (d, J = 6.0 Hz,1 H), 4.82 (t, J = 8.0 Hz, 1 H), 4.43 (t, J = 4.0 Hz, 1 H), 4.25 (s,1 H), 4.10 (d, J = 4.4 Hz,1 H), 3.56–3.76 (m, 4 H), 2.07–2.11 (m, 1 H), 1.56–1.63 (m, 1 H), 1.07 (d, J = 6.4 Hz, 3 H); 13C NMR (100 MHz, D2O) δ:172.89, 158.97, 153.89, 140.84, 116.23, 100.34,100.28, 86.73, 83.83, 83.74, 74.70, 73.45, 70.42, 67.93, 36.24, 15.57; ESI HRMS: [M − H]− calcd for 572.0946, found: 572.0941.
Briefly, a 50 µL reaction was set up containing 50 mM Tris–HCl (pH 8.0), 5 mM acceptor Galβ1,3-GlcNAc, 5 mM of acceptor GDP-colitose and 20 µg of recombinant WbgN. The reaction was performed at 37°C for 1 h, and quenched by adding equal volume of ice-cold ethanol. After brief centrifugation, the supernatant was analyzed by mass spectrometry.
To study the kinetics of WbgN towards GDP-colitose and GDP-l-Fuc, reactions were performed at 37°C for 10 min in a reaction buffer containing 50 mM Tris–HCl (pH 8.0), 5 mM Galβ1,3-GlcNAc, 5.25 µM recombinant WbgN and varied concentrations of GDP-colitose or GDP-l-Fuc (0.25, 0.5, 0.75, 1.0, 2.0 and 4.0 mM). To study the kinetics of WbgN towards an acceptor Galβ1,3-GlcNAc, reactions were performed at 37°C for 10 min in a reaction buffer containing 20 mM Tris–HCl (pH 8.0), 5 mM GDP-colitose, 18 µg recombinant WbgN and varied concentrations of Galβ1,3-GlcNAc (0.25, 0.5, 0.75, 1.0, 2.0 and 4.0 mM). Reactions were quenched immediately by addition of equal volume of ice-cold ethanol and subject to capillary electrophoresis (CE) analysis using Beckman Coulter P/ACE MDQ Capillary electrophoresis to detect substrate GDP-colitose or GDP-Fuc and byproduct GDP. CE conditions were as follows: 75 µm i.d. capillary, 25 KV/170 µA, 5 s vacuum injection, with monitoring at 254 nm and 50 mM sodium tetraborate running buffer, pH 9.4. The parameters Km and Vmax were obtained by Lineweaver-Burk plots of substrate concentration–initial velocity.
To study pH dependence of WbgN, reactions were carried out in a total volume of 50 µL, containing 20 µg of recombinant WbgN, 5 mM GDP-colitose, 5 mM Galβ1,3-GlcNAc and 50 mM of different buffer (PBS, pH 6.5; Tris–HCl, pH 7.5; Tris–HCl, pH 7.5; Glycine–NaOH, pH 9.0; Glycine–NaOH, 10.2). To study metal ion dependence, reactions were carried out in a total volume of 50 µL, containing 50 mM Tris–HCl, pH 8.0, 10 µg of recombinant WbgN, 5 mM GDP-colitose, 5 mM Galβ1,3-GlcNAc and 10 mM of different additions (EDTA, None, MgCl2, MnCl2, CoCl2, CaCl2, ZnCl2 or CuCl2). A negative control was also carried out containing EDTA but no acceptor Galβ1,3-GalNAc. All reactions were performed in triplicate. Reactions were incubated at 37°C for 60 min, quenched and analyzed by CE as described in Kinetics study.
Reactions were carried out in a total volume of 50 µL, containing 50 mM Tris–HCl, pH 8.0, 10 µg of recombinant WbgN, 5 mM GDP-colitose and 2.5 mM of different acceptors, including galactose (Gal), Galβ1,3GlcNAc, Galβ1,4GlcNAc (LacNAc), Galβ1,3GalNAc and Galβ1,4Glc (Lac). All reactions were performed in triplicate. Reactions were incubated at 37°C for 60 min, quenched and analyzed by CE as described above.
In a 15-mL centrifuge tube, reactions were carried out in a total volume of 5 mL, containing 50 mM Tris–HCl, pH 8.0, 300 µg of recombinant WbgN, 20 mM GDP-colitose (or GDP-l-Fuc for the synthesis of type 1 H-antigen), and 16 mM of Galβ1,3-GlcNAc. The reactions were allowed to proceed at 37°C and monitored by TLC (developing solvent: isopropanol: NH4OH: H2O = 7: 3: 2, v/v/v). After Galβ1,3-GlcNAc was totally converted to products, reactions were quenched by the addition of equal volumes of ice-cold ethanol, followed by centrifugation at 4°C for 10 min (10,000 × g). The supernatant was 10 concentrated and subject for separation by P2 gel filtration. The fractions contain pure products were pooled, lyophilized and stored at −20°C until chemical characterization. The fractions contain impure products were also collected and treated with shrimp alkaline phosphatase (New England Biolabs) to digest GDP byproduct following instructions (50 U, 30°C for 5 h), and then subject to bio-gel P2 for further separation. NMR and ESI HRMS data of trisaccharides (see Supplementary data, S4 and S5 for spectra):
Colα1,2-Galβ1,3-GlcNAc: 1H NMR (D2O, 400 MHz): δ 1.16 (t, J = 5.6 Hz 3 H, CH3), 1.76–1.87 (m, 1 H, H″-3), 1.93–1.96 (m, 1 H, H″-3), 2.07 (s, 3 H, NHCOCH3), 3.52–3.57 (m, 1 H, H′-2), 3.63 (dd, J = 4.8 Hz, J = 6.8 Hz,1 H, H″-2), 3.70–3.71(m, 1 H, H-2), 3.73–3.76 (m, 2 H), 3.79–3.83 (m, 4 H), 3.90–3.93 (m, 2 H), 3.99–4.03 (m, 1 H), 4.15–4.28 (m, 2 H), 4.63–4.71 (m, 2.2 H, H″-5, H′-1, H-1-β), 5.12 (m, 1.8 H, H-1-α, H″-1); 13C NMR (D2O, 100 MHz): δ 15.25 (CH3), 21.79 (NHCOCH3), 32.74 (C″-3), 53.63, 60.43, 60.93, 63.03, 66.05, 68.29, 68.59, 69.01, 71.14 (C″-2), 73.48 (C-2), 74.86, 74.98, 76.41 (C′-2), 90.68 (C-1-α), 95.46 (C-1-β), 98.73 (C″-1), 100.12 (C′-1), 173.55 (NHCOCH3); ESI HRMS: m/z calcd for [C20H35NO14+Na]+, 536.1955, found: 536.1933.
Fucα1,2-Galβ1,3-GlcNAc: 1H NMR (D2O, 400 MHz): δ 1.23 (t, J = 5.6 Hz, 3 H, CH3), 2.08 (s, 3 H, NHCOCH3), 3.52–3.65 (m, 2 H, H′-2, H″-2), 3.68–3.90 (m, 9 H), 4.14–4.33 (m, 2 H), 4.63–4.70 (m, 2.3 H, H″-5, H′-1, H-1-β), 5.12 (d, J = 2.8 Hz, 0.7 H, H-1-α), 5.20 (d, J = 4.0 Hz, 1 H, H″-1); 13C NMR (D2O, 100 MHz): δ 14.41 (CH3), 21.06 (NHCOCH3), 52.87, 59.69, 60.18, 60.23, 65.60, 67.22, 67.85, 68.27, 68.66, 70.41, 70.97, 72.68, 74.14, 75.76, 89.94 (C-1-α), 98.58 (C″-1), 99.25 (C-1-β), 99.34 (C′-1), 172.79 (NHCOCH3); ESI HRMS: m/z calcd for [C20H35NO15 +Na]+ 552.1904, found: 552.1888.
This work was supported by National Institute of General Medical Sciences (U01GM116263 to P.G.W. and L.L., R01GM085267 to P.G.W.) and National Institute of Allergy and Infectious Diseases (R01AI083754 to P.G.W.).
ADP, adenine 5′-diphosphate; ATP, adenine 5′-triphosphate; CDP, cytosine 5′-diphosphate; Col, colitose; ColC, GDP-colitose synthase; ColD, GDP-4-keto-6-deoxy-mannose-3-dehydratase; Gal, galactose; GalNAc, N-acetylgalactosamine; GDP, guanosine 5′-diphosphate; GDP-l-Fuc, GDP-l-fucose; GFS, GDP-l-Fuc synthase; GKDM, GDP-4-keto-6-deoxy-mannose; GKDDM, GDP-4-keto-3,6-dideoxy-mannose; Glc, glucose; GlcNAc, N-acetylglucosamine; GMD, GDP-mannose 4,6-dehydratase; GME, GDP-mannose 3,5-epimerase; GMP, GDP-mannose pyrophosphorylase; GNB, galacto-N-biose; GTP, guanosine 5′-triphosphate; l-Fuc, l-fucose; LNB, lacto-N-biose; PLP, pyridoxal-5′-phosphate.