- Split View
-
Views
-
Cite
Cite
John G. Riley, Mohammed Menggad, Pedro J. Montoya-Peleaz, Walter A. Szarek, Cristina L. Marolda, Miguel A. Valvano, John S. Schutzbach, Inka Brockhausen, The wbbD gene of E. coli strain VW187 (O7:K1) encodes a UDP-Gal: GlcNAcα-pyrophosphate-R β1,3-galactosyltransferase involved in the biosynthesis of O7-specific lipopolysaccharide, Glycobiology, Volume 15, Issue 6, June 2005, Pages 605–613, https://doi.org/10.1093/glycob/cwi038
- Share Icon Share
Abstract
In this work, we demonstrate that the wbbD gene of the O7 lipopolysaccharide (LPS) biosynthesis cluster in Escherichia coli strain VW187 (O7:K1) encodes a galactosyltransferase involved in the synthesis of the O7-polysaccharide repeating unit. The galactosyltransferase catalyzed the transfer of Gal from UDP-Gal to the GlcNAc residue of a GlcNAc-pyrophosphate-lipid acceptor. A mutant strain with a defective wbbD gene was unable to form O7 LPS and lacked this specific galactosyltransferase activity. The normal phenotype was restored by complementing the mutant with the cloned wbbD gene. To characterize the WbbD galactosyltransferase, we used a novel acceptor substrate containing GlcNAcα-pyrophosphate covalently bound to a hydrophobic phenoxyundecyl moiety (GlcNAc α-O-PO3-PO3-(CH2)11-O-phenyl). The WbbD galactosyltransferase had optimal activity at pH 7 in the presence of 2.5 mM MnCl2. Detergents in the assay did not increase glycosyl transfer. Digestion of enzyme product by highly purified bovine testicular β-galactosidase demonstrated a β-linkage. Cleavage of product by pyrophosphatase and phosphatase, followed by HPLC and NMR analyses, revealed a disaccharide with the structure Gal β1-3GlcNAc. Our results conclusively demonstrate that WbbD is a UDP-Gal: GlcNAcα-pyrophosphate-R β1,3-galactosyltransferase and suggest that the novel synthetic glycolipid acceptor may be generally applicable to characterize other bacterial glycosyltransferases.
Introduction
The O-specific polysaccharide (O antigen) is the outermost component of the lipopolysaccharide (LPS), a major constituent of the outer membrane in Gram-negative bacteria. The genetics and biochemistry of the pathways for the biosynthesis and assembly of O antigens have been well defined over the past few decades using model systems (Raetz and Whitfield, 2002; Samuel and Reeves, 2003; Valvano, 2003). One of these models is the O7 polysaccharide of the Escherichia coli strain VW187 (O7:K1), which arises from the polymerization of a pentasaccharide O repeating unit consisting of 3-VioNAc β1-2 [Rha α1-3]Man α1-4Gal β1-3 GlcNAc α1- (L’vov et al., 1984). The synthesis of the O7 polysaccharide involves the assembly of the repeating unit onto undecaprenol-phosphate (Und-P) (Alexander and Valvano, 1994). The initiation reaction is catalyzed by the integral membrane protein WecA, a UDP-GlcNAc:Und-P GlcNAc-1-phosphate transferase that belongs to the family of polyisoprenyl-phosphate N-acetylhexosamine-1-phosphate transferases (Valvano, 2003). This family comprises proteins that are present both in prokaryotes and in eukaryotes and share conserved residues involved in catalysis (Amer and Valvano, 2001, 2002).
As with other O antigens, the assembly of GlcNAc-pyrophosphate-Und (GlcNAc-PP-Und)-linked O antigen subunits is thought to occur at the cytosolic face of the plasma membrane (Whitfield and Valvano, 1993). Four additional sugars are subsequently added to the GlcNAc-PP-Und intermediate to complete the formation of the O7 subunits. These reactions are catalyzed by specific glycosyltransferases, which are either soluble enzymes or associated with the plasma membrane. The candidate enzymes for the completion of the O7 subunit are predicted to be encoded by the genes wbbABCD, which map within the O7 LPS biosynthesis gene cluster (Marolda et al., 1999). The lipid-linked pentasaccharide is subsequently translocated across the inner membrane and then polymerized to form a lipid-linked polysaccharide by transfer of repeating units to the reducing end of the growing polymer, a process that differs from biosynthetic mechanisms of oligosaccharides in mammalian systems where the growth occurs at the nonreducing end of the glycan chains (Brockhausen et al., 1998). This is followed by the transfer of the completed O7-specific polysaccharide to the lipid A-core moiety of the LPS molecule (Feldman et al., 1999; Marolda et al., 1999).
From previous studies, the WbbD protein was implicated in the glycosyl transfer of the second sugar, galactose, to the nascent O7 antigen repeat (Marolda et al., 1999). In the study described in this article, we used a synthetic exogenous acceptor substrate containing GlcNAcα-pyrophosphate bound to a phenoxyundecyl moiety in [GlcNAc α1-O-PO2-O-PO2-O-(CH2)11-O-phenyl]2− (GlcNAc-PP-PhU) to characterize the galactosyltransferase from the E. coli strain VW187. Using a combination of genetic and biochemical approaches, we conclusively demonstrated that the wbbD gene encodes a UDP-Gal: GlcNAc-R β1,3-galactosyltransferase (Gal-transferase) activity involved in O7 LPS synthesis.
Results
Construction and characterization of a wbbD mutant in the strain VW187
In a previous work, we reported the genetic organization of the wbEcO7 biosynthesis cluster (Figure 1) and tentatively assigned wbbD as a gene encoding a galactosyltransferase involved in the extension of the O7 antigen subunit by adding Gal to GlcNAc-PP-Und (Marolda et al., 1999). The gene wbbD encodes a polypeptide that belongs to the glycosyltransferase family 2 (http://afmb.cnrs-mrs.fr/~cazy/cazy/index.html). This family includes a large group of inverting glycosyltransferases, such as cellulose synthase, chitin synthase, dolichyl-phosphate β-D-mannosyltransferase, dolichyl-phosphate β-glucosyltransferase, GlcNAc transferase, N-acetylgalactosaminyltransferase, hyaluronan synthase, β1,3-glucan synthase, and others acting on unknown substrates (Coutinho et al., 2003).
.
To determine whether wbbD is required for O7 LPS synthesis, we constructed a mutant derivative of strain VW187 using a homologous recombination strategy that resulted in the replacement of the wild-type wbbD gene by wbbD::aph. The aph gene cassette encodes resistance to the antibiotic kanamycin, and it is oriented in such manner that the transcription of the wbEcO7 genes located downstream of wbbD is driven by the aph promoter (Figure 1). The LPS from the resulting mutant strain, MB1, was analyzed by silver staining of LPS preparations after electrophoresis in sodium dodecyl sulfate (SDS)/N-[tris(hydroxymethyl)methyl]glycine gels. Figure 2 shows that the LPS prepared from the parental strain VW187 displays a typical ladder-like banding pattern with a band distribution that is characteristic of the O7-specific polysaccharide (Marolda et al., 1999). In contrast, the wbbD::aph mutant MB1 only shows a rapidly migrating band corresponding to lipid A-core oligosaccharide (Figure 2). This band is also present in LPS prepared from strain MV501 (Figure 2), which carries a wecA::Tn10 insertion that precludes the initiation of the synthesis of the O7 repeat (Alexander and Valvano, 1994). Neither MV501 nor MB1 bacteria gave positive slide agglutination in the presence of O7-specific rabbit antiserum. The ability of MB1 to form O7 polysaccharide was restored by the introduction of pCM227, which encodes a glutathione transferase (GST)-WbbD chimeric protein (Figure 1), confirming that the insertion of the aph cassette has no effect on the expression of the downstream genes of the wbEcO7 cluster (Figure 1). Also, this experiment confirms that the fusion protein encoded by pCM227 is functional as an enzyme involved in O-chain synthesis. MB1(pCM227) and VW187 gave a positive slide agglutination with the O7-specific antibody, demonstrating that the O polysaccharide produced in the complemented MB1 mutant has the same immunoreactivity as the parental O7 LPS.
Additional experiments with the Ffm bacteriophage showed that mutants MB1 and MV501 were sensitive to lysis by Ffm, whereas the parental strain VW187 and the complemented MB1(pCM227) strain were resistant, suggesting that there is sufficient surface-exposed O7 LPS in the complemented strains to mask the Ffm receptor in the core oligosaccharide and prevent the entry of the phage DNA into the bacterial cells. Altogether, we conclude from these experiments that WbbD carries an essential function for the synthesis of O7 LPS, which is consistent with its predicted Gal-transferase activity.
The MB1 strain lacks a UDP-GlcNAc-dependent galactosyltransferase
Gal is the second sugar transferred during synthesis of the O7 subunit and is transferred to the GlcNAc-PP-Und endogenous acceptor. Thus the incorporation of radioactive Gal to endogenous acceptor should be dependent on the presence of UDP-GlcNAc in the assay. We first determined that the UDP-GlcNAc:Und-P GlcNAc-1-phosphotransferase in the homogenates prepared from the wild-type strain VW187 was active toward endogenous substrate by demonstrating that up to 0.13 nmol [3H]GlcNAc/assay were incorporated into a chloroform/methanol-extractable product. Gal-transferase assays were then carried out using crude bacterial homogenates as the source of the enzyme and endogenous acceptor in the presence of an excess of UDP-GlcNAc and radiolabeled UDP-Gal. The synthesis of the Gal-transferase product was followed by quantifying the incorporation of [3H]Gal into a chloroform/methanol-extractable fraction.
Optimal assay conditions were determined with respect to metal ion cofactors and buffer requirements. Optimal activity of homogenates from the parental strain VW187 (75 pmol of Gal transferred per assay, 0.36 nmol/h/mg protein) was found with 0.1 M Tris/acetate buffer, pH 8.5, and 20 mM Mg2+. Using these conditions, the incorporation of Gal was linear with time up to 30 min and proportional to enzyme concentration (data not shown). Table I shows that the homogenate prepared from the wecA::Tn10 mutant MV501 had a low endogenous Gal-transferase activity (0.07 nmol/h/mg), relative to the activity from the parental strain VW187 (0.36 nmol/h/mg). This is consistent with the lack of GlcNAc-PP-Und synthesis in the absence of a functional WecA protein (Amer and Valvano, 2001, 2002). The homogenate prepared from the wbbD::aph mutant MB1 had 25% endogenous Gal-transferase activity relative to the activity in homogenates from the parental strain. However, introduction of pCM227 into MB1 increased the activity in the homogenate 2.9-fold (Table I). Homogenates from E. coli DH5α carrying the plasmid pCM227 also showed a relatively high activity (2.6-fold higher than the activity in the parental strain homogenate). These results demonstrate that the presence of an intact wbbD gene is associated with a Gal-transferase activity that is dependent on the availability of endogenous GlcNAc-PP-Und glycolipid acceptor.
. | Relative activity (%)a . | . | |
---|---|---|---|
Extracts prepared from . | Endogenous substrate . | GlcNAc-PP-PhU substrate . | |
VW187 | 100 | 100 | |
MB1 | 25 | 4 | |
MB1 (pCM227) | 72 | 173 | |
MV501 | 19 | 107 | |
MV501 (pCM227) | ND | 759 | |
DH5a (pCM227) | 261 | 136 |
. | Relative activity (%)a . | . | |
---|---|---|---|
Extracts prepared from . | Endogenous substrate . | GlcNAc-PP-PhU substrate . | |
VW187 | 100 | 100 | |
MB1 | 25 | 4 | |
MB1 (pCM227) | 72 | 173 | |
MV501 | 19 | 107 | |
MV501 (pCM227) | ND | 759 | |
DH5a (pCM227) | 261 | 136 |
Gal-transferase activity was assayed as described in Materials and methods, using endogenous substrate or exogenous substrate GlcNAc-PP-PhU. The endogenous substrate is of unknown structure and concentration, and the enzyme product(s) or the specific Gal-transferase activity measured is undefined. The use of exogenous substrate allows assaying a defined activity, UDP-Gal: GlcNAc-PP-PhU β3-Gal-transferase. The activity in homogenates of the wild–type bacteria was set to 100% (0.13 nmol/h/mg for the endogenous activity and 758 nmol/h/mg [0.0126 mmol/min/mg] for the activity using GlcNAc-PP-PhU substrate). Assays were carried out in at least duplicate determinations; assay results varied by < 10%. ND, not done.
. | Relative activity (%)a . | . | |
---|---|---|---|
Extracts prepared from . | Endogenous substrate . | GlcNAc-PP-PhU substrate . | |
VW187 | 100 | 100 | |
MB1 | 25 | 4 | |
MB1 (pCM227) | 72 | 173 | |
MV501 | 19 | 107 | |
MV501 (pCM227) | ND | 759 | |
DH5a (pCM227) | 261 | 136 |
. | Relative activity (%)a . | . | |
---|---|---|---|
Extracts prepared from . | Endogenous substrate . | GlcNAc-PP-PhU substrate . | |
VW187 | 100 | 100 | |
MB1 | 25 | 4 | |
MB1 (pCM227) | 72 | 173 | |
MV501 | 19 | 107 | |
MV501 (pCM227) | ND | 759 | |
DH5a (pCM227) | 261 | 136 |
Gal-transferase activity was assayed as described in Materials and methods, using endogenous substrate or exogenous substrate GlcNAc-PP-PhU. The endogenous substrate is of unknown structure and concentration, and the enzyme product(s) or the specific Gal-transferase activity measured is undefined. The use of exogenous substrate allows assaying a defined activity, UDP-Gal: GlcNAc-PP-PhU β3-Gal-transferase. The activity in homogenates of the wild–type bacteria was set to 100% (0.13 nmol/h/mg for the endogenous activity and 758 nmol/h/mg [0.0126 mmol/min/mg] for the activity using GlcNAc-PP-PhU substrate). Assays were carried out in at least duplicate determinations; assay results varied by < 10%. ND, not done.
Homogenates of the wild-type strain VW187 were centrifuged for 10 min at 10,000 × g. Essentially all of the activity stayed in the pellet fraction, suggesting that the native enzyme was associated with membranes. The recombinant WbbD enzyme protein, expressed as a GST-fusion protein encoded by pCM227 in DH5α, was solubilized in 0.1% Triton-X100, and purified on glutathione-Sepharose. Coomassie blue stains of the fusion protein separated by SDS–polyacrylamide gel electrophoresis (PAGE) showed a major band at 57 kDa, suggesting a high level of expression. Surprisingly, however, the purified enzyme protein had very little activity (less than 1% of activity of the homogenate from DH5α carrying pCM227).
Characterization of the galactosyltransferase activity using a novel synthetic acceptor substrate, GlcNAc-PP-PhU
In previous experiments, the rate and extent of galactosyl transfer to endogenous acceptor was relatively low, and the reaction could not be characterized because of the dependence on the activity of the prior enzyme in the pathway. Also, because endogenous acceptor was limiting, it was difficult to isolate sufficient quantities of the resulting product for a detailed characterization. Therefore, we used a new acceptor substrate, GlcNAc-PP-PhU (Montoya-Peleaz et al., 2005), to directly assay for the presence of the Gal-transferase and to characterize both the enzymatic reaction and the enzyme product.
The Gal-transferase activity of strain VW187 catalyzed galactosyl transfer from UDP-Gal to GlcNAc-PP-PhU as the acceptor in standard assays (Table I). At 30°C incubation temperature, Gal-transferase activity was reduced by 28%. The enzyme had a broad pH optimum between 6.0 and 8.0. Galactosyl transfer was dependent on the presence of divalent cations, and 2.5 mM Mn2+ supported maximal activity. MgCl2, CoCl2, and ZnCl2 at 5 mM concentration yielded 11%, 38%, and 9% activity, respectively, compared with 5 mM MnCl2. In the presence of NiCl2, ethylenediamine tetraacetic acid (EDTA) or with no addition, enzyme activity was < 4.5%. The Km for GlcNAc-PP-PhU was 0.08 mM (Vmax 0.027 mmol/min/mg) and the Km for UDP-Gal was 1.2 mM (Vmax 0.042 mmol/min/mg).
The activities of membrane-bound glycosyltransferases are often stimulated by detergents. However, octylglucoside at concentrations of 0.125% and 0.25% in the assay decreased Gal-transferase activity by 30% and 77%, respectively, and the activity was diminished in the presence of CHAPS at concentrations of 0.125–0.6%. Gal-transferase retained up to 50% activity in the presence of Triton X-100 at concentrations between 0.05% and 1%.
Crude homogenates prepared from the E. coli MB1 wbbD mutant and the parental strain VW187, as well as from E. coli MV501 mutant and DH5α containing pCM227 (wbbD+) were assayed with exogenously added GlcNAc-PP-PhU. The results were similar to those obtained from assays using endogenous substrate (Table I). The extracts from the MB1 mutant, which was incapable of synthesizing O-chain due to a disruption of the wbbD gene, had very low (4%) activity relative to that of the extracts from the parental strain VW187 (Table I). On transformation of MB1 with pCM227, the activity was increased to 173%, most likely due to overexpression of the WbbD protein. Similarly, extracts from E. coli DH5α(pCM227) that do not synthesize O7 LPS but expressed the GST-WbbD construct showed a high Gal-transferase activity (136%). Extracts from mutant MV501, which lacks the GlcNAc-P-transferase activity provided by the WecA protein, had similar values of Gal-transferase activity relative to those of VW187, although this mutant was incapable of synthesizing O-chain. Complementation of MV501 with pCM227, as expected, resulted in extracts with high Gal-transferase activity (759%) which was due to both endogenous and plasmid-derived Gal-transferase. These results clearly show an association of Gal-transferase activity with the presence of a functional wbbD gene.
Product identification
Enzyme product from assays using VW187 homogenates containing UDP-[3H]Gal as the donor substrate and GlcNAc-PP-PhU as the acceptor substrate was isolated by high-performance liquid chromatography (HPLC) for structural analysis. We previously showed by mass spectrometry that the radioactive product was a disaccharide-PP-PhU (Montoya-Peleaz et al., 2005). When the radioactive product was incubated with both nucleotide pyrophosphatase and alkaline phosphatase, the radioactivity was no longer retained by Sep-Pak C18, and >95% of the radioactivity eluted with the water fractions. This indicated cleavage of the pyrophosphate bond between the GlcNAc residue and the lipid tail. The resulting radioactive disaccharide did not bind to AG1x8, indicating that >95% of the phosphate residues were cleaved, and free disaccharide was produced. The radioactive compounds were further separated on HPLC using an amine column and acetonitrile/water (90/10) as the mobile phase. More than 95% of the radioactivity eluted later than free [3H]Gal but earlier than Gal β1-4GlcNAc (Figure 3) and coeluted with Galβ1-3GlcNAc, showing that the linkage between Gal and GlcNAc was not β1-4.
Galactosidases were used as a tool to determine the anomeric linkage between Gal and GlcNAc in the Gal-GlcNAc-PP-PhU enzyme product. The released free Gal was monitored by measuring the radioactivity of the water fractions of Sep-Pak eluates, whereas the intact product was measured similarly in the methanol eluates. The enzyme product was resistant to jack bean (β1-4-specific) β-galactosidase (>96%) and coffee bean α-galactosidase (>92%). However, treatment with bovine testicular β-galactosidase, which cleaves Galβ1-3,-4, and -6 linkages, resulted in 90% of the radioactivity from enzyme product eluting as [3H]Gal on HPLC. Thus, we concluded that Gal was β-linked in the enzyme product.
The 600 MHz proton–nuclear magnetic resonance (NMR) spectra of the Gal′β1-3GlcNAc standard (in CD3OD) showed characteristic signals of the Gal H-1′ residue at 4.34 ppm and the H-1 of GlcNAcα at 3.85 ppm (Table II). Because the H-3 signal was obscured by other signals, it was further determined by 2D NMR, that is, correlation spectroscopy (COSY) (proton connectivity), heteronuclear multiple-quantum coherence spectroscopy (HMQC) (carbon connectivity), and nuclear Overhauser effect spectroscopy (NOESY) (through space interaction with H-1′). Thus, the proton and carbon chemical shifts of H-1, H-2, H-3, and H-1′, as well as C-1, C-2, C-3, and C-1′, are shown in Table II. More than 1 mmol of enzyme product was produced using the highly active homogenate from the MV501(pCM227) strain. Enzyme product was purified by C18 Sep-Pak and HPLC and analyzed by NMR. An aliquot of the enzyme product was cleaved with pyrophosphatase and phosphatase, and the resulting disaccharide was purified by anion exchange and HPLC.
. | Standard Gal′ β(1-3) GlcNAc (α) . | . | Product . | |
---|---|---|---|---|
Nuclei position . | D2Oa . | CD3OD . | CD3OD . | |
H-1 | 4.58 (d, J = 3.0 Hz) | 5.08 (d, J = 3.6 Hz) | 5.58 (br s) | |
H-2 | 5.80 | 3.99(dd, J = 3.2, 10.6 Hz) | 4.14 (br d, J = 9.7 Hz) | |
H-3 | NA | 3.85b | 3.84b | |
H-1′ | 5.28 (d, J = 7.0 Hz) | 4.34 (d, J = 7.6 Hz); | 4.35 (d, J = 7.5 Hz) | |
C-1 | 91.7 | 91.2c | 94.8c | |
C-2 | 53.5 | 53.1c | 52.4c | |
C-3 | 80.9 | 80.5c | 80.3c | |
C-1′ | 104.0 | 103.7c | 104.0c |
. | Standard Gal′ β(1-3) GlcNAc (α) . | . | Product . | |
---|---|---|---|---|
Nuclei position . | D2Oa . | CD3OD . | CD3OD . | |
H-1 | 4.58 (d, J = 3.0 Hz) | 5.08 (d, J = 3.6 Hz) | 5.58 (br s) | |
H-2 | 5.80 | 3.99(dd, J = 3.2, 10.6 Hz) | 4.14 (br d, J = 9.7 Hz) | |
H-3 | NA | 3.85b | 3.84b | |
H-1′ | 5.28 (d, J = 7.0 Hz) | 4.34 (d, J = 7.6 Hz); | 4.35 (d, J = 7.5 Hz) | |
C-1 | 91.7 | 91.2c | 94.8c | |
C-2 | 53.5 | 53.1c | 52.4c | |
C-3 | 80.9 | 80.5c | 80.3c | |
C-1′ | 104.0 | 103.7c | 104.0c |
Literature data (Lemieux and Driguez, 1975) of standard Gal β(1-3) GlcNAc (α) in deuterium oxide. The standard was also run in deuterated methanol for better correlation with the product.
Determined by COSY.
Determined by HSQC.
. | Standard Gal′ β(1-3) GlcNAc (α) . | . | Product . | |
---|---|---|---|---|
Nuclei position . | D2Oa . | CD3OD . | CD3OD . | |
H-1 | 4.58 (d, J = 3.0 Hz) | 5.08 (d, J = 3.6 Hz) | 5.58 (br s) | |
H-2 | 5.80 | 3.99(dd, J = 3.2, 10.6 Hz) | 4.14 (br d, J = 9.7 Hz) | |
H-3 | NA | 3.85b | 3.84b | |
H-1′ | 5.28 (d, J = 7.0 Hz) | 4.34 (d, J = 7.6 Hz); | 4.35 (d, J = 7.5 Hz) | |
C-1 | 91.7 | 91.2c | 94.8c | |
C-2 | 53.5 | 53.1c | 52.4c | |
C-3 | 80.9 | 80.5c | 80.3c | |
C-1′ | 104.0 | 103.7c | 104.0c |
. | Standard Gal′ β(1-3) GlcNAc (α) . | . | Product . | |
---|---|---|---|---|
Nuclei position . | D2Oa . | CD3OD . | CD3OD . | |
H-1 | 4.58 (d, J = 3.0 Hz) | 5.08 (d, J = 3.6 Hz) | 5.58 (br s) | |
H-2 | 5.80 | 3.99(dd, J = 3.2, 10.6 Hz) | 4.14 (br d, J = 9.7 Hz) | |
H-3 | NA | 3.85b | 3.84b | |
H-1′ | 5.28 (d, J = 7.0 Hz) | 4.34 (d, J = 7.6 Hz); | 4.35 (d, J = 7.5 Hz) | |
C-1 | 91.7 | 91.2c | 94.8c | |
C-2 | 53.5 | 53.1c | 52.4c | |
C-3 | 80.9 | 80.5c | 80.3c | |
C-1′ | 104.0 | 103.7c | 104.0c |
Literature data (Lemieux and Driguez, 1975) of standard Gal β(1-3) GlcNAc (α) in deuterium oxide. The standard was also run in deuterated methanol for better correlation with the product.
Determined by COSY.
Determined by HSQC.
The 600 MHz proton NMR spectrum of 200 nmol of the purified disaccharide released from the enzyme product showed signals identical to those for Galβ1-3GlcNAc standard. The 1D NMR spectrum of the intact enzyme product Gal-GlcNAc-PP-PhU measured in deuterated methanol is shown in Figure 4. The signals were assigned using the same 2D NMR procedures as used for the standard (Table II). The shifts of GlcNAc-H-1 and H-2 show minor differences due to the linkage of GlcNAc to pyrophosphate. However the characteristic shifts for Gal H-1′ and GlcNAc H-3 are similar to those of the disaccharide (Table II; Figure 4). Thus, 1D NMR fingerprint as well as the 2D NMR analysis of the disaccharide and intact enzyme product conclusively show that the sugar linkage between Gal and GlcNAc is β1-3.
Discussion
The O7 polysaccharide of the E. coli strain VW187 has been used as a model system to understand the biosynthesis of the O antigen repeating subunit structure (Valvano, 2003; Marolda et al., 1990, 1999; Marolda and Valvano, 1993, 1995, 1998). The O7 antigen consists of a repeating pentasaccharide with the structure VioNAc α1-2 [Rha α1-3] Man α1-4Gal β1-3 GlcNAc α1-3, which is assembled on a C55 polyisoprenoid alcohol with subsequent polymerization to the final polysaccharide. The initial reaction in the synthesis pathway involves the transfer of GlcNAc-P from UDP-GlcNAc to Und-P, resulting in the formation of GlcNAc-PP-Und, a reaction mediated by the WecA protein (Alexander and Valvano, 1994; Amer and Valvano, 2002; Marolda et al.,1999). The addition of the subsequent sugars presumably requires the products of additional glycosyltransferase genes. Initial studies of Gal-transferase activities using endogenous acceptor substrate that is present in crude homogenates from the wbbD mutant MB1 demonstrated a low enzyme activity in the presence of UDP-GlcNAc, despite the fact that the GlcNAc-P-transferase reaction occurred at normal levels. This was consistent with the lack of O-chain formation and the high sensitivity of the mutant to the Ffm phage. This phenotype was complemented by transforming MB1 with a functional wbbD gene provided by plasmid pCM227, indicating that the Gal-transferase activity was encoded by wbbD.
The inclusion of the synthetic GlcNAc-PP-PhU in the crude cell homogenates obviated the need to add UDP-GlcNAc to the assay mixtures, as demonstrated by high Gal-transferase activities toward GlcNAc-PP-PhU found in extracts of the parental strain VW187 and the GlcNAc-P-transferase mutant MV501. Therefore we confirmed that GlcNAc-PP-PhU salts are excellent substrates for the WbbD enzyme and that GlcNAc linked through pyrophosphate to either Und or to a smaller hydrophobic chain can serve as substrate for the Gal-transferase. Our results also indicate that both the naturally occurring Gal-transferase as well as the GST-Gal-transferase construct are both active toward the new acceptor substrate GlcNAc-PP-PhU.
Enzyme product analysis unequivocally demonstrated that the newly assayed Gal-transferase activity transfers Gal from UDP-Gal to GlcNAc-PP-PhU in β1-3 linkage. The Gal-transferase activity was low in the presence of EDTA and was stimulated primarily by divalent metal ions Mn2+, Mg2+, and Co2+. This property of Gal-transferase is similar to that of mammalian β3- and β4-Gal-transferases. These metal ions may be involved in binding and stabilizing UDP-Gal in the donor binding site of the enzyme (Boeggeman and Qasba, 2002).
Pyrophosphate-containing lipids with a phenoxyundecyl group may also be acceptor substrates for other glycosyltransferases that catalyze the subsequent steps of O7 chain synthesis and for glycosyltransferases of other strains of Gram-negative bacteria. GlcNAc-PP-PhU is an excellent substrate, possibly because, in contrast to the natural Und-P substrate, its relatively short hydrophobic tail may not be inserted in the membrane, thus rendering the GlcNAc residue accessible to Gal-transferase. The Gal-transferase activity was lost on purification of the protein in the absence of membranes. The detergents used in this study appeared to be unfavorable for mimicking the membrane environment or producing appropriate micelle structures to support enzyme activity. These combined results suggest that the topology and membrane association not only of the enzymes themselves but also of substrates are important factors for the sugar transfer reaction in bacteria. Alternatively, it is possible that additional proteins are required for full enzymatic activity as a part of a putative protein complex.
We have devised a synthetic acceptor molecule that will help in studying the biochemical glycosyl transfer reactions in bacterial systems. The utility of the new compound was (1) to allow the development of a sensitive and accurate assay system for a specific O-chain glycosyltransferase; (2) to characterize the crude endogenous enzyme, and to show that the enzyme was active as a fusion protein in the presence of bacterial membranes; (3) to follow enzyme purification; and (4) to produce substrate for other enzymes subsequently acting in the O-chain synthesis pathway. Current experiments are under way in our laboratories to characterize these additional reactions for the biosynthesis of the complete O7 subunit and to determine the appropriate conditions for the characterization of purified glycosyltransferases that will enable us to perform structure–function studies.
Materials and methods
Reagents
All reagents were of the highest grade available and were from Sigma Chemical (St. Louis, MO) unless otherwise indicated. 5-Bromo-4-chloro-3-indolyl β-D-galactopyranoside (X-Gal), isopropyl β-D-thiogalactopyranoside (IPTG), T4 DNA ligase, and DNA polymerases were purchased from Roche Diagnostics (Laval, Quebec). Plasmid preparation kits were from Qiagen (Mississauga, Ontario). Oligonucleotide primers were purchased from Life Technologies and Cortec (Kingston, Ontario). pGEM-T vector was from Promega (Madison, WI). Tris-glycine SDS–PAGE gels were purchased from Novex (San Diego, CA). Protein concentrations were determined with the BioRad (Hercules, CA; Bradford) protein assay using bovine serum albumin as a standard. GlcNAc-PP-PhU was synthesized as described (Montoya-Peleaz et al., 2005)
Bacterial strains and growth conditions
E. coli DH5α was used as the initial host for the cloning experiments. Strains VW187 (E. coli O7:K1) and its wecA:: Tn10 mutant, MV501, have been previously described (Alexander and Valvano, 1994; Valvano and Crosa, 1989). Strain MB1 is a wbbD mutant of VW187 constructed as described shortly. Bacteria were grown at 37°C in Luria Bertani (LB) medium consisting of 10 g NaCl, 5 g yeast extract, and 10 g tryptone per liter, which was supplemented with ampicillin (50 mg/ml), tetracycline (20 mg/ml), kanamycin (50 µg/ml), and chloramphenicol (20 µg/ml), as appropriate. For some experiments, cultures were plated on LB agar plates supplemented with 0.2% (w/v) X-Gal and 0.4 mM IPTG. For selection against sacB, LB agar plates were supplemented with sucrose to a final concentration of 5% (w/v). Spot assays with the rough-specific bacteriophage Ffm (Wilkinson et al., 1972), which lyses bacteria lacking the O-chain, were carried out as described (Expert and Toussaint, 1985). O antigen reactivity was determined as described (Valvano and Crosa, 1989).
DNA methods
Restriction enzymes, T4 DNA ligase, and Klenow DNA polymerase were used according to the conditions recommended by the supplier. Recombinant plasmids were introduced into E. coli strains by CaCl2 transformation or electroporation (2500 V, 5 ms) using an Eppendorf Multiporator. DNA fragments for cloning were isolated from agarose gels using the Qiaquick kit (Qiagen, Valencia, CA). Polymerase chain reaction (PCR) amplifications were carried out with PwoI or Taq DNA polymerase as recommended by the manufacturer. Amplification was done in an Eppendorf thermocycler after denaturation at 94°C for 1 min, followed by 35 cycles consisting of 30 s at 94°C (denaturation), 30 s at 55°C (annealing), and 2 min at 72°C for (extension), and a final extension cycle of 10 min at 72°C. PCR amplification products were visualized on 1% (w/v) agarose gels. PCR products were recovered by ethanol precipitation and ligated into the appropriate cloning vectors as described next.
Cloning of wbbD and construction of a wbbD::aph mutant of VW187 (strain MB1)
A 2.7-kb fragment containing the wbbD gene and the surrounding region (Figure 1) was amplified by PCR, using the PCR conditions just described, and the primers 5′-GGATCCTATTCGATGGGATTGATTGC-3′ and 5′-GGATCCATTCCCAAAGCGAAGACCAT-3′ (underscoring indicated BamHI recognition sites not present in the original DNA sequence that were introduced into the primers to facilitate future cloning steps). Chromosomal DNA from the strain VW187 served as a template. The PCR product was ligated into the thymidylated EcoRV site of pGEM-T vector to yield plasmid pMB1 (Figure 1). The cloned DNA fragment in pMB1 carries a BglII site located in the middle of wbbD that was used to insert the aminoglycoside 3′-phosphotransferase gene (aph) from Tn903, which encodes kanamycin resistance. To facilitate this construction, we first cloned aph from pUC4K into the unique BamHI site of pKO3, resulting in the recombinant pMB2, which encodes both chloramphenicol and kanamycin resistance. After this step, the ligation mixture of pMB1 linearized with BglII and pMB2 linearized with BamHI was transformed into E. coli DH5α and transformants selected on plates with kanamycin and ampicillin. One of these recombinants contained the plasmid pMB3, which corresponded to pMB1 with the aph gene inserted into wbbD. The plasmid pMB3 was digested with BamHI and ligated into the temperature-sensitive suicide plasmid pKO3 that carries the sacB gene for counterselection of double crossover recombinants. This experiment resulted in the isolation of the mutagenic plasmid pMB4, which was transformed by electroporation into E. coli O7 strain VW187. One of the selected transformants was incubated at 42°C to isolate derivatives where the pMB4 plasmid was integrated by homologous recombination into the VW187 chromosome. To isolate mutants with a second crossover that resulted in the replacement of the parental wbbD by the mutated wbbD::aph gene, purified colonies were streaked on plates supplemented with 5% (w/v) sucrose. The loss of chloramphenicol resistance in sucrose-resistant, kanamycin-resistant colonies confirmed the loss of pMB4. One of these isolates, designated MB1, was used for additional experiments to confirm by PCR analysis that the correct gene replacement took place (data not shown).
The coding region of the wbbD gene was amplified by PCR with primers as described and cloned into the unique BamHI site of the vector pGEX-2T. The forward primers were designed in a way such that wbbD was translationally fused to the distal portion of the GST gene that is present in pGEX-2T. The resulting plasmid pCM227 encoded a chimeric protein of ~ 57 kDa in mass (with a 27-kDa N-terminal region contributed by the GST, followed by a thrombin cleavage site, and a 30-kDa C-terminal region contributed by WbbD).
LPS analysis
LPS was extracted as previously described, and analyzed by SDS–PAGE (14% gels) and silver staining (Marolda et al., 1990).
Enzyme preparation
Bacterial homogenates were prepared as the enzyme source as described (Montoya-Peleaz et al., 2005). Briefly, bacteria were grown in Luria broth containing 100 mg/ml ampicillin and sedimented by centrifugation. Bacterial cells were sonically ruptured using a Sonic Dismembrator Model 100 (Fisher Scientific, Silver Spring, MD) for two pulses of 15 s with 2-min intervals to allow for cooling on ice. Protein concentrations ranged between 2 and 12 mg/ml. For purification of the fusion protein GST–WbbD, encoded by plasmid pCM227, bacterial homogenates in phosphate buffered saline containing 0.1% Triton X-100 were incubated with glutathione Sepharose 4B. Bound proteins were eluted with 10 mM reduced glutathione/50 mM Tris–Cl, pH 8.0 and stored in phosphate buffered saline. The fusion protein was analyzed by SDS–PAGE (8% gels) and western blots using anti-GST antibody.
Enzyme assays for GlcNAc-phospho transfer and Gal transfer to endogenous acceptors
The GlcNAc-phospho transfer from UDP-GlcNAc to endogenous Und-P was assayed in a total volume of 0.08 ml of 50 mM 2-N-morpholinosulfonate (MES)/acetate, pH 7.0, containing 1.5 mM UDP-[3H]GlcNAc (1270 cpm/nmol), 25 mM MgCl2, 0.8 mM EDTA, and 0.04 ml of freshly prepared enzyme homogenate (400 mg protein). After 30 min at 37°C the reactions were quenched by the addition of 0.7 ml CHCl3/CH3OH (2:1) and thorough mixing on a vortex mixer. After standing 15 min at room temperature, 0.6 ml of pure solvent upper phase (a mixture of 15 ml CHCl3, 240 ml methanol, 1.83 g KCl in 235 ml water), were added to the solution, the mixtures were vigorously stirred on a vortex mixer and subjected to centrifugation at 8000 rpm for 1.5 min to separate the two phases; the aqueous upper phase and interface were aspirated and discarded. The organic phase was washed two times with 0.5 ml pure solvent upper phase and then transferred to scintillation vials, taken to dryness, and counted in scintillation fluid. All assays were carried out at least in duplicate.
Galactosyl transfer to endogenous acceptor was carried out in reaction mixtures containing 2 mM UDP-GlcNAc (for the synthesis of GlcNAc-PP-Und as a galactosyl acceptor), 75 mM MES, pH 7.0, 10 mM MgCl2, and 0.9 mM [3H]-UDP-Gal (10,650 cpm/nmol). Reactions were incubated for 30 min at 37°C and quenched by the addition of 0.7 ml CHCl3/CH3OH (2:1). The formation of radioactive product was measured as already described. Enzyme homogenate that was heated for 10 min at 100°C was used as a negative control.
Enzyme assays for Gal transfer to exogenous acceptor
Standard assays for galactosyl transfer to exogenously added substrate were carried out as described (Montoya-Peleaz et al., 2005). Briefly, 20 ml enzyme homogenate (3–12 mg protein) were incubated in reaction mixtures of 40 ml total volume, containing 0.5 mM GlcNAc-PP-PhU, 5 mM MnCl2, 75 mM MES buffer, pH 7, and 0.5 mM UDP-[3H]Gal (800–4400 cpm/nmol). Enzyme product was isolated using a C18 Sep-Pak column. For further analysis, enzyme product was isolated by reverse-phase HPLC (Montoya-Peleaz et al., 2005). Enzyme kinetics were determined with acceptor substrate concentrations ranging from 0.05 to 1 mM (with 2 mM UDP-Gal), and 0.1 to 6 mM UDP-Gal (with 1 mM acceptor). Km and Vmax values were established with double-reciprocal Lineweaver-Burk plots.
Galactosidase digestion
The anomeric configuration of the linkage formed in the product was determined by digestion with specific galactosidases. Aliquots of pooled radioactive enzyme product from VW187 bacterial homogenate (750 cpm ) were treated in a total volume of 100 ml with 25 ml MacIlvaine buffer (0.1 M citric acid/0.2 M Na-phosphate), pH 4.3, 10 ml of 0.1% bovine serum albumin, and either 4 ml (0.04 U) jack bean β-galactosidase or 10 ml (0.01 U) of bovine testicular β-galactosidase, or 10 ml (0.54 U) green coffee bean α-galactosidase. Mixtures were incubated for 1 h at 37°C, diluted with 800 ml water, and applied to C18 SepPak columns. Released [3H]Gal was eluted with 5 ml water, whereas unreacted enzyme product was eluted with 5 ml methanol. The 1-ml fractions were counted in 5 ml scintillation fluid.
Analysis of enzyme product by NMR
To prepare large amounts of enzyme product for NMR, Gal-transferase assays were carried out as follows. The incubation mixtures (a total of 8 ml) contained: 4 ml bacterial homogenate (MV501 strain complemented with plasmid pCM227), 4 mmol GlcNAc-PP-PhU, 8.08 mmol UDP-[3H]Gal (60 cpm/nmol), 600 µmol HEPES buffer, pH 7, and 400 mmol MnCl2. After incubation for 30 min at 37°C, 8 ml of cold water was added and the mixtures were applied to 10 C18 Sep-Pak columns. Each column was washed with 4 ml water, and the product was eluted with 4 ml methanol. The methanol fractions were pooled, flash evaporated, and redissolved in 500 µl methanol. Aliquots of the first methanol fraction obtained from the Sep-Pak column were purified by HPLC, using a C18 column and acetonitrile/water (6:94) as the mobile phase, as described (Brockhausen et al., 2002). Enzyme product was dried, exchanged three times with 99.96% D2O and CH3OD, and analyzed by 600 MHz proton NMR spectroscopy.
For digestion with phosphatases, 20 standard assays were carried out, and the product was isolated by Sep-Pak, dried, and redissolved in 500 ml methanol, 2 ml water, and 2 ml 0.5 M HEPES, pH 7.5. Nucleotide pyrophosphatase from Crotalus adamenteus venom (200 ml of 20 U) in 35% Tris(hydroxymethyl)aminomethane and alkaline phosphatase from intestinal mucosa (50 ml of 1700 U) in 3 M NaCl, 1 mM MgCl2, 0.1 mM ZnCl2, 30 mM triethanolamine, pH 7.6, were added. Mixtures were incubated for 1 h at 37°C and applied to a 1-ml AG1x8 column (Cl− form). The column was washed with 10 ml water. Eluate was lyophilized. Released disaccharide product was separated by HPLC using an amine column and acetonitrile/water (90:10). Fractions containing radioactivity were combined, flash evaporated, lyophilized, exchanged three times with 99.96% D2O, and analyzed in CD3OD with a 600 MHz Bruker spectrometer, using 1D proton NMR methods without water suppression. COSY, NOESY, and HMQC methods were also applied to analyze both the intact enzyme product as well as the released disaccharide. Control substances were Galβ1-3GlcNAc (Gefco-Chemicals, Israel) and Galβ1-4GlcNAc.
We thank Yi Li for growing bacteria and enzyme assays. This work was supported by grants from the Natural Science and Engineering Research Council of Canada (to I.B., W.A.S., and M.A.V.) and a Research Scientist Award from the Arthritis Society of Canada; M.A.V. holds a Canada Research Chair in Infectious Diseases and Microbial Pathogenesis.
References
Montoya-Peleaz, P.J., Riley, J.G., Szarek, W.A., Valvano, M.A., Schutzbach, J.S., Brockhausen, I. (2005) Identification of UDP-Gal: GlcNAc-R galactosyltrasferase activity in Escherichia coli VW187.
Author notes
2Department of Medicine, Department of Biochemistry, The Arthritis Centre and Human Mobility Research Centre, Queen’s University, Kingston General Hospital, Kingston, Ontario K7L 2V7, Canada; 3Department of Chemistry, Queen’s University, Kingston, Ontario K7L 3N6, Canada; 4Department of Microbiology and Immunology, University of Western Ontario, London, Ontario N6A 5C1, Canada