Functional Characterization of a HAD Phosphatase Involved in Capsular Polysaccharide Biosynthesis in Campylobacter jejuni

Abstract

Campylobacter jejuni is a Gram-negative, pathogenic bacterium found in the intestinal tracts of chickens and many other farm animals. Infection of C. jejuni results in campylobacteriosis, which can cause nausea, diarrhea, fever, cramps, and death. The surface of the bacterium is coated with a thick layer of sugar known as the capsular polysaccharide. This highly modified polysaccharide contains an unusual d-glucuronamide moiety in serotypes HS:2 and HS:19. Previously, we have demonstrated that a phosphorylated glucuronamide intermediate is synthesized in C. jejuni NCTC 11168 (serotype HS:2) by cumulative reactions of three enzymes: Cj1441, Cj1436/Cj1437, and Cj1438. Cj1441 functions as a UDP-d-glucose dehydrogenase to make UDP-d-glucuronate, then Cj1436 or Cj1437 catalyzes the formation of ethanolamine phosphate or S-serinol phosphate, respectively, and finally Cj1438 catalyzes amide bond formation using d-glucuronate and either ethanolamine phosphate or S-serinol phosphate. Here, we investigated the final d-glucuronamide modifying enzyme, Cj1435. Cj1435 was shown to catalyze the hydrolysis of the phosphate esters from either the d-glucuronamide of ethanolamine phosphate or S-serinol phosphate. Kinetic constants for a range of substrates were determined and the stereoselectivity of the enzyme for the hydrolysis of glucuronamide of S-serinol phosphate was established using ³¹P NMR spectroscopy. A bioinformatic analysis of Cj1435 reveals it to be a member of the HAD-phosphatase superfamily with a unique DXXE catalytic motif.

Graphical Abstract

Introduction

The zoonotic pathogen Campylobacter jejuni is the leading cause of gastroenteritis worldwide (1). Infections caused by this Gram-negative bacterium are known as campylobacteriosis and have symptoms of nausea, fever, cramps, bloody diarrhea, and death (2, 3). Long term effects of campylobacteriosis includes reactive arthritis, Miller-Fischer Syndrome and Guillain-Barré Syndrome (GBS) (4, 5). It is estimated that ~40% of all GBS cases are preceded by a Campylobacter infection (6). C. jejuni is a commensal organism in chickens, and can be found in the intestinal tracts of most farm animals such as cows, goats, and dogs. The most common route of human infection is through contact with raw or improperly cooked chicken, pet feces, and unpasteurized milk (7, 8). Campylobacter infections most commonly affect the elderly and very young children. It is estimated that 25,000 children die every year from campylobacteriosis worldwide (9). The disease is most prevalent in Asia and Africa. Efforts to combat the bacterium with antibiotics have been hampered due to the development of multi-drug resistance (10–12). Currently, there are no known FDA-approved vaccines to prevent Campylobacter infections. The best vaccine candidates to date have been glycoconjugate vaccines, which enable the immune system to identify surface exposed sugars on the bacterium (13).

Surrounding C. jejuni is a thick carbohydrate layer known as the capsular polysaccharide (CPS). In C. jejuni NCTC 11168 (serotype HS:2), the CPS is composed of a repeating unit of d-glycero-l-gluco-heptose, d-glucuronate, N-acetyl-d-galactosamine, and d-ribose (14, 15). These sugars can be further modified by O-methylation, phosphoramidylation, and amidation. The d-glucuronate moiety is modified as an amide with either serinol (1) or ethanolamine (2). The gene cluster for the biosynthesis of the capsular polysaccharide in C. jejuni NCTC 11168 (serotype HS:2) is presented in Figure 1 (14, 16).

Figure 1.

(a) Gene cluster for CPS formation in C. jejuni NCTC 11168 (serotype HS:2). Shown in red are genes responsible for d-glucuronamide formation. Genes shown in green are responsible for the biosynthesis of d-glycero-l-gluco-heptose. Genes displayed in purple are used to construct the phosphoramidate modification. Blue-colored genes are annotated as sugar transferases and grey-colored genes are hypothetical/uncharacterized. Cj1439, shown in yellow is a pyranose/furanose mutase (17); wheat-colored genes are likely used for capsular polysaccharide export. (b) Serotype HS:2 capsular polysaccharide with the amide of serinol shown in red. (c) Serotype HS:2 capsular polysaccharide with the amide of ethanolamine shown in red.

We have previously shown that Cj1441 catalyzes the double NAD⁺-dependent oxidation of UDP-d-glucose to UDP-d-glucuronate (18). Cj1436 is a PLP-dependent enzyme that catalyzes the decarboxylation of l-serine phosphate to ethanolamine phosphate (19), and Cj1437 is a PLP-dependent transaminase that catalyzes the conversion of dihydroxyacetone phosphate to S-serinol phosphate using l-glutamate as the amine donor (19). The C-terminal domain of Cj1438 is an ATP-dependent amidoligase (20). This enzyme was shown to use MgATP, d-glucuronate (or the methyl glycoside of d-glucuronate) and either ethanolamine phosphate or S-serinol phosphate to synthesize the phosphorylated glucuronamide product (20). The unphosphorylated compounds are not substrates for the amidoligase enzyme (20). In this investigation, we interrogate the ability of the putative haloalkanoic acid dehalogenase (HAD) phosphatase, Cj1435, to catalyze the ultimate step in the biosynthesis of the d-glucuronamide moiety in the CPS of C. jejuni.

Cj1435 is a member of the HAD superfamily. This large and diverse superfamily is found in prokaryotes, eukaryotes, and archaea. A search of the UniProt database returns over 300,000 proteins for this superfamily. A vast majority of these enzymes catalyze the hydrolysis of phosphate esters, but some have been shown to be ATPases or phosphotransferases (21). Enzymes in this superfamily share a common catalytic scaffold that facilitates phosphate ester hydrolysis through two partial reactions (22). In the first step, the sidechain carboxylate of an active site aspartate residue attacks the phosphorus center of the substrate and forms an acyl-phosphate enzyme intermediate that is subsequently hydrolyzed by the activation of water from a second aspartate residue. Magnesium is most often used as a metal cofactor to help position the nucleophile and facilitate catalysis. Here, we demonstrate that Cj1435 is responsible for catalyzing the final step of glucuronamide biosynthesis in C. jejuni NCTC 11168 (serotype HS:2). We demonstrate that Cj1435 catalyzes the hydrolysis of phosphate from the glucuronamide of ethanolamine phosphate and of serinol phosphate. A bioinformatic analysis reveals Cj1435 to be a member of the HAD-phosphatase superfamily, with a unique DXXE catalytic motif.

Materials and Methods

Materials and Equipment.

Escherichia coli BL21 (DE3) was obtained from New England Biolabs. Nuclear magnetic resonance (NMR) spectra were recorded on a Bruker Avance III 400 MHz system equipped with a broadband probe and sample changer. Spectrophotometric data were collected on a SpectraMax₃₄₀ UV-visible plate reader. Unless otherwise noted, all compounds were purchased from commercially available sources. The phosphate colorimetric detection kit (MAK 030–1KT) was obtained from Sigma-Aldrich and used according to the manufacturer’s instructions. The chemical structures of the compound used for this investigation are presented in Figure 2. The chemical syntheses of compounds 7, R/S-8, R/S-9, 10, and 13 are presented in the Supplementary Information.

Figure 2.

Structures of compounds used in this investigation. The S- and R/S-designations indicate the stereochemistry of the serinol phosphate moiety.

Compounds S-11 and 12 were synthesized enzymatically. The reaction for the preparation of 12 contained 25 μM Cj1438, 10 mM d-glucuronate, 10 mM ATP, 20 mM MgCl₂, 5.0 mM ethanolamine phosphate (3), and 50 mM HEPES/K⁺ pH 8.0. The reaction mixture was allowed to incubate at 25 °C for 1 h before the Cj1438 enzyme was removed using a 10 kDa MWCO filter (Cytivia). This procedure was repeated for the synthesis of compound S-11 using 5.0 mM S-serinol phosphate (S-4) as the initial substrate.

Cloning, Expression, and Purification of Cj1435.

The genomic DNA of C. jejuni NCTC 11168 (ATCC 700819D-5) served as the template for amplification of the gene encoding Cj1435 (Uniprot id: Q0P8H9) by PCR using Phusion High Fidelity polymerase (New England Biolabs). Primers were designed to incorporate the NdeI and XhoI restriction sites. The amplified DNA was ligated into a pET31b expression vector. This construct contained a C-terminal hexahistidine tag and the amino acid sequence of the purified enzyme is presented in Figure S1.

The pET31b vector was used to transform BL21 (DE3) E. coli cells via electroporation. The cells harboring the pET31b-Cj1435 plasmid were cultured in lysogeny broth with 100 mg/L ampicillin. The cells were grown at 37 °C until they reached an OD₆₀₀ of ~0.8, whereupon the cells were induced with 1.0 mM isopropyl β-d-1-thiogalactopyranoside (IPTG). Cells were allowed to grow at 21 °C for 18 h before centrifugation at 15,000 rcf at 4 °C. The cell pellet was resuspended in 50 mM HEPES/K⁺ (pH 8.0), 300 mM KCl, 20 mM imidazole, and 1.0 mg DNAse from bovine liver. The cells were placed in a beaker on ice and lysed by sonication. The lysate was clarified by centrifugation at 25,000 rcf at 4 °C for 30 min. The clarified lysate was passed through a 0.45 μm filter (Whatman) and then applied to a 5.0 mL prepacked HisTrap nickel affinity purification column (Cytivia). The protein was eluted using 50 mM HEPES/K⁺, 300 mM KCl, 500 mM imidazole (pH 8.0) over a gradient of 25 column volumes. Fractions were pooled and dialyzed against 10 mM HEPES/K⁺, 200 mM KCl, pH 8.0. The dialyzed protein was concentrated using 10 kDa MWCO filters (Cytivia) to a final concentration of 5.0 mg/mL before being flash-frozen in liquid nitrogen and stored at −80 °C.

Stereoselectivity of the Reaction Catalyzed by Cj1435.

The stereoselectivity of Cj1435 was investigated with ³¹P NMR spectroscopy using the chemically synthesized glucuronamide (10 mM) with racemic serinol phosphate (R/S-9), 3.0 μM Cj1435, 10 mM MgCl₂, and 50 mM HEPES/K⁺, pH 8.0. The reaction was incubated at 25 °C for 4 h before the ³¹P NMR spectrum was obtained. The stereoselectivity of Cj1435 was also investigated by ³¹P NMR spectroscopy using the product of the reaction catalyzed by Cj1438. Cj1438 catalyzes the formation of compound S-11 using d-glucuronate, MgATP, and S-serinol phosphate (S-4). The reaction was initiated by the addition of 10 μM Cj1438, 10 mM S-serinol phosphate (S-4) with 10 mM ATP, 20 mM MgCl₂, 10 mM d-glucuronic acid (5), and 50 mM HEPES/K⁺, pH 8.0 to make the chiral glucuronamide (S-11). The reaction mixture was incubated at 25 °C for 2 h and then passed through a Vivaspin 10 kDa MWCO filter (Cytivia) to remove Cj1438. Cj1435 (3.0 μM) was added to the flowthrough and the reaction mixture was allowed to incubate for 30 min before the ³¹P NMR spectrum was obtained.

Determination of Substrate Profile by ³¹P NMR Spectroscopy.

The substrate specificity of Cj1435 was interrogated by ³¹P NMR spectroscopy. The reaction mixture contained 5.0 μM Cj1438, 5.0 mM d-glucuronate (5), 10 mM S-serinol phosphate (S-4), 10 mM ATP, 20 mM MgCl₂, and 50 mM HEPES/K⁺ pH 8.0. The reaction was allowed to incubate at 25 °C for 2 h before an NMR spectrum was obtained to ensure the complete formation of compound S-11. The amidoligase, Cj1438, was removed using a 10 kDa MWCO filter and the subsequent phosphatase reaction was initiated by the addition of 3.0 μM Cj1435 to hydrolyze S-11. The ³¹P NMR spectrum was collected after an incubation period of 15 min at 25 °C. This procedure was repeated by substituting S-serinol phosphate with ethanolamine phosphate (3) to form compound 12, which was then dephosphorylated after the addition of Cj1435.

The reaction catalyzed by Cj1435 was also investigated using ³¹P NMR spectroscopy using the N-acetyl derivatives of ethanolamine phosphate (7) and racemic serinol phosphate (R/S-8) as potential substrates. The enzyme (3.0 μM) was incubated at 25 °C with 10 mM N-acetyl ethanolamine phosphate (7), 10 mM MgCl₂, and 50 mM HEPES/K⁺, pH 8.0 for 4 h. Cj1435 (3.0 μM) was also incubated with 10 mM of racemic N-acetyl serinol phosphate (R/S-8), 10 mM MgCl₂, and 50 mM HEPES/K⁺, pH 8.0. Additionally, Cj1435 (3.0 μM) was incubated with either 10 mM ethanolamine phosphate (3) or 10 mM S-serinol phosphate (S-4), in the presence of 10 mM MgCl₂, and 50 mM HEPES/K⁺, pH 8.0. ³¹P NMR spectra were recorded after an incubation period of 4 h at 25 °C.

The reaction of Cj1435 was also tested with a phosphorylated disaccharide consisting of d-ribose and d-glucuronamide (S-14) and (15). To produce the phosphorylated amides attached to a disaccharide, Cj1438 (20 μM) was incubated with 10 mM d-glucuronate (1→2)-β−1-O-methyl ribose (13), 10 mM ATP, 2.5 mM S-serinol phosphate (S-4), 20 mM MgCl₂, and 50 mM HEPES/K⁺ pH 8.0. The enzyme was incubated at 25 °C for 30 min before the ³¹P NMR spectrum was obtained showing the reaction had gone to completion with formation of compound S-14. This process was repeated with ethanolamine phosphate (3) to obtain compound 15. Cj1438 was removed by filtration using a 10 kDa MWCO filter (Cytivia). To the flowthrough containing compound S-14, 3.0 μM Cj1435 was added and the reaction mixture allowed to incubate for 30 min at 25 °C before the ³¹P NMR spectrum was obtained. In the same fashion, compound 15 was incubated with 3.0 μM Cj1435 and allowed to incubate for 30 min at 25 °C before the ³¹P NMR spectrum was obtained.

Determination of Kinetic Constants.

The rate of product formation was investigated using ³¹P NMR spectroscopy at 25 °C. Each ³¹P NMR spectrum was obtained from 128 scans that were collected over a period of ~7 min on a Bruker Avance III 400 MHz system. The increase in inorganic phosphate concentration was monitored over the first 30 min of the reaction. The integrations were converted to concentration through use of an internal standard and plotted as a function of time. The data were fit to a linear equation and the slope was divided by enzyme concentration to obtain the first order rate constant for product formation. The following substrates were tested with Cj1435: d-glucuronamide of ethanolamine phosphate (12), glucuronamide of S-serinol phosphate (S-11), ethanolamine phosphate (3), S-serinol phosphate (S-4), N-acetyl ethanolamine phosphate (7), racemic N-acetyl R/S-serinol phosphate (R/S-8), 1-O-methyl β-glucuronamide of R/S-serinol phosphate (R/S-9), 1-O-methyl β-glucuronamide of ethanolamine phosphate (10), and O-phospho-l-serine.

For compounds 3, S-4, 7, 10 and O-phospho-l-serine, the reactions were conducted using 5.0 mM substrate, 2.0 mM MgCl₂, and 50 mM HEPES/K⁺ pH 8.0. The reactions were initiated by the addition of enzyme and the ³¹P NMR spectra were collected as a function of time. Since compounds R/S-8, and R/S-9 are racemic mixtures, 10 mM of each substrate was used with 2.0 mM MgCl₂, and 50 mM HEPES/K⁺ pH 8.0. For compounds 3, S-4 and O-phospho-l-serine, 20 μM of Cj1435 was added to the reaction mixtures. For compounds 7, R/S-8, R/S-9, and 10, 250 nM Cj1435 was added to the reaction mixtures.

For compound 7, the reaction was also monitored using a phosphate detection kit. The substrate was tested at concentrations between 0.125 mM and 10 mM. The reaction contained 2.0 mM MgCl₂, 250 nM Cj1435, a variable amount of substrate, and 50 mM HEPES/K⁺ pH 8.0 at 25 °C. The reaction was monitored for the first five min. The reaction was stopped at 30 s intervals by quenching a 10 μL aliquot with the acidic dye solution (10 μL of reaction, 30 μL dye, 210 μL of water). The absorbance was monitored at 650 nm using a SpectraMax₃₄₀ spectrophotometer. The absorbance was converted to concentration using a phosphate standard curve per the manufacturer’s instructions. The k_cat/K_m was obtained using a linear regression fit to the data.

Bioinformatic Analysis of Cj1435 and Amide Gene Cluster.

A sequence similarity network (SSN) of Cj1435 was created using the amino acid sequence downloaded from Uniprot (Q0P8H9) as the input for a BLAST search of the Uniprot database using the EFI-EST webtool (23, 24). The e-value cutoff was set to 10⁻⁵ and the maximum number of returned sequences was set to 5000. The network was generated using Cytoscape (version 3.8.2) and the clusters were created using the yFile organic layout with a cutoff value of 50% sequence identity. A genome neighborhood network (GNN) was created using the 3686 genes of the Cj1435 SSN as the input. Using the Pfam identifiers for UDP-glucose 6-dehydrogenase (PF00984), the PLP-dependent transaminase/decarboxylase (PF00155), the TupA-like amidoligase (PF14305), and the putative HAD-phosphatase (PF12710), a list of putative amide bond forming gene clusters was generated. A sequence alignment of HAD-phosphatases similar to Cj1435 was generated using Clustal Omega (25). The genes used for the alignment were identified as the phosphatase genes in the genome neighborhood network that likely participate in hydrolysis of a phosphate bond after amide bond formation.

Predicted Structure of Cj1435 using Alphafold2.

The amino acid sequence of Cj1435 was used as the input for the three-dimensional structure prediction performed by Alphafold2 through use of the ColabFold webserver developed by Mirdita et. al (26). The predicted three-dimensional structure of Cj1435 was then used to create an alignment with the closest known structurally characterized protein, the phosphoserine phosphatase from Methanocaldococcus jannaschii (PDB id: 1L7P), in Pymol (27, 28).

Results

Bioinformatic Analysis of Cj1435.

Cj1435 is 211 amino acids in length and is a member of the haloalkanoic dehalogenase (HAD) superfamily. The amino acid sequence for Cj1435 was obtained from Uniprot and used as the input for the EFI-EST BLAST retrieval option. The webtool retrieved 3686 sequences and the SSN was generated at a 50% sequence identity (Figure 3). The nodes colored blue are uncharacterized proteins. Shown as a large orange circle is a putative HAD-phosphatase from Bordetella pertussis Tohama I (PDB id: 3FVV). This enzyme shares 26% sequence identity across 89% of the query length. Cj1435 is shown as a large red circle. The yellow nodes are annotated as Campylobacter strains and tend to cluster together near Cj1435.

Figure 3.

Sequence similarity network for Cj1435. Sequence similarity network of Cj1435 and 3686 sequences at 50% sequence identity. Cj1435 is displayed as a large red circle. The large orange circle is a protein of unknown function from Bordetella pertussis Tohama I (PDB id: 3FVV). Yellow circles are genes from Campylobacter strains. Blue circles are uncharacterized proteins.

The phosphatases likely involved in the formation of the glucuronamide modification appear to cluster together away from the main clusters in the SSN (Figure 3). To investigate this further, we created a sequence alignment with all of the phosphatases identified in the genome neighborhood network shown in Figure S2. Strikingly, the catalytic motif found in the HAD-phosphatases involved in glucuronamide formation are different than what is usually found in the HAD-phosphatase superfamily. Traditionally, the catalytic motif of HAD-phosphatases is composed of a conserved DXD motif (29). In the sequence alignment of the phosphatases involved in glucuronamide phosphate hydrolysis, this motif is changed to DXXE. It is unclear why these proteins have a different catalytic motif from those previously investigated.

In an effort to more clearly understand the genes necessary for glucuronamide formation across various organisms, a genome neighborhood network (GNN) was generated using the 3686 genes identified in the SSN as the input (Figure S3). The genome neighborhood was filtered by gene clusters which contained Pfam identifiers for a UDP-glucose 6-dehydrogenase (PF00984), PLP-dependent transaminase/decarboxylase (PF00155), TupA-like amidoligase (PF14305), and the putative HAD-phosphatase (PF12710). A total of 35 gene clusters were identified to contain the genes necessary for the formation of a glucuronamide modification. All 35 gene clusters had enzymes homologous to Cj1435, Cj1438, and Cj1437/Cj1436. Only 19 of the 35 gene clusters contained an enzyme homologous to Cj1441, the UDP-glucose 6-dehydrogenase. The list of organisms contains bacteria such as Capnocytophaga canimorsus, Campylobacter avium LMG 24591, Helicobacter fennelliae, Vibrio parahaemolyticus, Helicobacter bilis WiWa, Shigella boydii, and Photobacterium phosphoreum, among others. A complete list of the Uniprot accession IDs and organism information is found in Table S1.

Alphafold2 Structure Prediction of Cj1435.

The predicted three-dimensional structure of Cj1435 was determined using Alphafold2 and is shown in Figure 4a (26). Cj1435 adopts a βαβ Rossmann fold with a type C1 capping domain as well as a C2 β-hairpin cap insertion (30). The predicted structure of Cj1435 was subsequently aligned with that of a phosphoserine phosphatase from Methanocaldococcus jannaschii (PDB id: 1L7P) shown in Figure 4b. An alignment of the two structures revealed an RMSD of 3.0 Å across 855 atoms. The two structures align well across the Rossmann fold and the active site residues, despite their low sequence identity of 23%. Where the structures differ is in their capping domains. Both contain a type C1 cap, which closes on top of the substrate upon binding (30). The phosphoserine phosphatase from M. jannaschii was co-crystallized with O-phospho-l-serine and the cap is shown in the “closed” form. The Alphafold2 predicted structure depicts the cap region of Cj1435 in an “open” conformation. The capping region of the phosphoserine phosphatase from M. jannaschii is defined by two hinge points at residues N18 and P77. An alignment of the primary sequences of Cj1435 and the structurally characterized phosphoserine phosphatase is presented in Figure S4. This comparison, along with the structural alignment, was used to estimate the C1 capping domain of Cj1435 to extend from residue N17 to N83. In addition to the C1 cap found in Cj1435, it appears Cj1435 also contains a C2 β-hairpin cap insertion extending from residue E123 to Q148. A C2 β-hairpin cap insertion is also found in the phosphoserine phosphatase of M. jannaschii extending from Y117 to E141. These β-hairpin cap insertions are commonly seen in phosphoserine phosphatases as well as pyrimidine 5’-nucleotidase families (30). The C2 β-hairpin cap insertion is not believed to be the dominant capping domain of these proteins, but is instead used to sequester the active site away from the solvent by packing against the C1 cap (30).

Figure 4.

Structural prediction of Cj1435 using Alphafold2. (a) Alphafold2 predicted structure of Cj1435. The type C1 capping domain is colored green. The type C2 capping domain is shown in orange. The Rossmann fold is colored teal. The catalytic motif is shown as yellow sticks. (b) Alignment of the predicted structure of Cj1435 and the phosphoserine phosphatase from M. jannaschii (PDB id:1L7P). Shown in teal is the Alphafold2 prediction of the structure of Cj1435. Shown in gold is the X-ray crystal structure of the phosphoserine phosphatase from M. jannaschii. The O-phospho-l-serine bound in the active site of the phosphoserine phosphatase from M. jannaschii is shown as space filing spheres occupying the active site pocket.

The catalytic motif of Cj1435 is found at the end of the β1 strand like other HAD superfamily members, however it is composed of an unusual DXXE motif. Most HAD superfamily members have a DXD catalytic motif where the first aspartate residue is used to coordinate a Mg²⁺ cofactor while the second catalytic aspartate is responsible for activating a water molecule which attacks the acyl-phosphate intermediate formed with the first aspartate residue (31–33). It is unclear why the catalytic motif for Cj1435 is different from other characterized HAD-phosphatases.

Stereoselectivity for Phosphate Hydrolysis.

We have previously demonstrated that the C-terminal domain of Cj1438 catalyzes the ATP-dependent amidation of d-glucuronate (5) and the O-methyl glycoside of d-glucuronate (6) with either S-serinol phosphate (S-4) or ethanolamine phosphate (3), but will not utilize UDP-d-glucuronate as a substrate (20). These results imply that the physiological substrate for the phosphatase, Cj1435, is the phosphorylated d-glucuronamide moiety after it has been added to the growing polysaccharide chain. To more firmly establish the substrate profile of Cj1435, the chemically synthesized amide (R/S-9) was incubated with the phosphatase Cj1435. The ³¹P NMR spectrum of the initial substrate is presented in Figure 5a and the products are shown in Figure 5b. The phosphorylated d-glucuronamide (R/S-9) resonates at 3.93 ppm, while the inorganic phosphate product resonates at 2.30 ppm. Figure 5b demonstrates that only 50% of the diastereomeric substrate (either S-9 or R-9) is capable of being dephosphorylated by Cj1435 and thus only one of the two diastereomers is a substrate for Cj1435.

Figure 5.

³¹P NMR spectra of β−1-O-methyl-d-glucuronamide of racemic serinol phosphate (R/S-9) and d-glucuronamide of S-serinol phosphate (S-11). (a) The ³¹P NMR spectrum of the chemically synthesized 1-O-methyl-glucuronamide with R/S-serinol phosphate (R/S-9). The resonance for R/S-9 is at 3.93 ppm. (b) The ³¹P NMR spectrum of the reaction catalyzed by Cj1435 and (R/S-9). The unreacted substrate resonates at 3.93 ppm while the inorganic phosphate product appears at 2.30 ppm. (c) The ³¹P NMR spectrum of the products of the reaction catalyzed by Cj1438 with R/S-serinol phosphate (R/S-4), MgATP, and d-glucuronate (5). The product (S-11) appears at 3.85 ppm, while the unreacted R-serinol phosphate (R-4) appears at 3.78 ppm. Inorganic phosphate is at 2.10 ppm. (d) The products of the reaction catalyzed by Cj1435 when incubated with the reaction mixture shown in Figure 5c. The resonance at 3.78 ppm is the unreacted R-serinol phosphate (R-4) while the phosphate resonance appears as a singlet at 2.10 ppm.

To determine the specific stereoisomer that is hydrolyzed by Cj1435 we enzymatically synthesized the phosphorylated d-glucuronamide (S-11) using R/S-serinol phosphate (R/S-4), d-glucuronate (5), and MgATP using the amidoligase Cj1438. The ³¹P NMR spectrum of the Cj1438-catalyzed products is presented in Figure 5c. The phosphorylated glucuronamide (S-11) has a chemical shift of 3.85 ppm and one equivalent of inorganic phosphate appears at 2.10 ppm. The unreacted R-serinol phosphate (R-4) resonates at 3.78 ppm. After incubation with Cj1435, all of the initial substrate (S-11) is now hydrolyzed and converted quantitatively to inorganic phosphate, while the R-serinol phosphate remains at 3.78 ppm (Figure 5d). This result clearly demonstrates that Cj1435 is specific for hydrolysis of the amide made from S-serinol phosphate (S-11). These findings are consistent with the reaction observed when Cj1435 is incubated with N-acetyl R/S-serinol phosphate (R/S-8) shown in Figure S6.

Glucuronamide Phosphate Hydrolysis.

The HAD phosphatase Cj1435 was tested as a catalyst for the hydrolysis of phosphate esters formed from the amidoligase activity of Cj1438. The amidoligase Cj1438 was incubated with MgATP, d-glucuronate (5), and S-serinol phosphate (S-4) and the ³¹P NMR spectrum of the products are shown in Figure 6a. The glucuronamide of S-serinol phosphate (S-11) appears as a singlet at 3.85 ppm while inorganic phosphate, formed from the activation of the carboxylate group with ATP appears as a singlet at 2.08 ppm. After addition of Cj1435, the resonance for the glucuronamide of S-serinol phosphate completely disappears and is replaced with the resonance for phosphate at 2.12 ppm (Figure 6b). To determine whether Cj1435 can use the glucuronamide of ethanolamine phosphate (compound 12), a separate reaction was conducted. In this experiment, Cj1438 was incubated with d-glucuronate (5), MgATP, and ethanolamine phosphate (3). The glucuronamide of ethanolamine phosphate (12) appears as a singlet at 3.56 ppm, while inorganic phosphate resonates at 2.00 ppm (Figure 6c). After Cj1435 is added, the resonance at 3.56 ppm disappears, while the inorganic phosphate resonance at 2.06 ppm has increased in size (Figure 6d).

Figure 6.

³¹P NMR spectra of the reaction products catalyzed by Cj1435. (a) The enzymatically prepared d-glucuronamide of S-serinol phosphate (S-4) appears as a singlet at 3.85 ppm and inorganic phosphate appears at 2.08 ppm. (b) The products of the reaction after the addition of Cj1435 to the enzymatically prepared S-4. The substrate was completely hydrolyzed and the phosphate appears as a singlet at 2.12 ppm. (c) The enzymatically prepared d-glucuronamide of ethanolamine phosphate (12) is present at 3.56 ppm and phosphate resonates at 2.00 ppm. (d) Products of the reaction after the hydrolysis of 12 by Cj1435. The substrate is completely converted to inorganic phosphate.

It is known from our previous investigation that Cj1438 will not accept UDP-d-glucuronate as a substrate for the formation of the phosphorylated d-glucuronamide (20). This suggests that the glycosyltransferase for polysaccharide formation likely functions before the amidation reaction. Given that UDP-d-glucuronate is activated at C1, and the CPS structure of C. jejuni HS:2 contains a d-glucuronamide (1→2)-d-ribose, we set out to determine if Cj1438 and Cj1435 would use a disaccharide of d-glucuronate and 1-O-methyl-d-ribose as a substrate for their respective reactions. Cj1438 was incubated with the disaccharide of d-glucuronate (1→2)-1-O-methyl-d-ribose (13), MgATP, and ethanolamine phosphate (3) and formation of inorganic phosphate and (15) are shown in the ³¹P NMR spectrum in Figure S5a. The products of the reaction of Cj1438 subsequently served as the substrate for Cj1435. After incubation with the HAD phosphatase, Cj1435, the resonance at 3.46 ppm disappeared and the inorganic phosphate resonance peak at 1.88 ppm increased (Figure S5b). The reaction was repeated with the substitution of S-serinol phosphate (S-4) for ethanolamine phosphate. Cj1438 catalyzed the synthesis of the phosphorylated glucuronamide disaccharide and the product (S-14) resonates at 3.65 ppm and inorganic phosphate at 1.68 ppm (Figure S5c). Upon addition of Cj1435, the resonance at 3.65 ppm completely disappeared and the inorganic phosphate peak at 1.73 ppm increased (Figure S5d).

Determination of Kinetic Constants for Cj1435.

The rate constants for the reactions catalyzed by Cj1435 were determined using ³¹P NMR spectroscopy with a single fixed concentration of substrate. The products of the reaction catalyzed by Cj1438 were used as substrates for the reaction catalyzed by Cj1435. For the reaction of Cj1435 with the d-glucuronamide of ethanolamine phosphate (12) the rate constant at a substrate concentration of 5.0 mM was determined to be 22 ± 1.0 s⁻¹. With S-serinol phosphate glucuronamide (S-11) the rate constant for substrate hydrolysis was determined to be 37 ± 5.0 s⁻¹.

To obtain a better understanding of the substrate specificity of Cj1435, the enzyme was tested against the following primary amines: ethanolamine phosphate (3), S-serinol phosphate (S-4), and O-phospho-l-serine and the rate constants were determined to be 0.060 ± 0.010 s⁻¹, 0.16 ± 0.030 s⁻¹, and 0.037 ± 0.002 s⁻¹ for the hydrolysis of 3, S-4, and O-phospho-l-serine, respectively. To address the preference of Cj1435 for substrates which contain an amide bond, the enzyme was tested using N-acetyl ethanolamine phosphate (7) where the rate constant was determined to be 7.1 ± 1.3 s⁻¹. Using a phosphate detection kit, compound 7 exhibited non-saturating kinetics up to 10 mM. The k_cat/K_m for this compound is 700 ± 30 M⁻¹ s⁻¹. The enzyme was next tested with the chemically synthesized N-acetyl R/S-serinol phosphate (R/S-8) where the rate constant was found to be 16 ± 1.0 s⁻¹.

To determine the rate constant for the hydrolysis of the chemically synthesized 1-O-methyl β-glucuronamide of (R/S)-serinol phosphate (R/S-9) and 1-O-methyl β-glucuronamide of ethanolamine phosphate (10), the substrates were incubated with Cj1435. The rate constant for the 1-O-methyl β-glucuronamide of (R/S)-serinol phosphate (R/S-9) was determined to be 4.0 ± 1.0 s⁻¹. The enzyme was tested with 1-O-methyl β-glucuronamide of ethanolamine phosphate (10) where the rate constant was observed to be 7.1 ± 1.0 s⁻¹. The values are summarized in Table 1.

Table 1:

Rate constant for the reactions catalyzed by Cj1435.

Substrate [5.0 mM]	Rate constant (s−1)
12	22 ± 1.0
11	37 ± 5.0
3	0.060 ± 0.010
4	0.16 ± 0.030
7	7.1 ± 1.3
R/S-8*	16 ± 1.0
R/S-9*	4.0 ± 1.0
10	7.1 ± 1.0

Reactions conducted at 25 °C, pH 8.0

^*10 mM substrate used

Discussion

The biosynthesis of the d-glucuronamide modification present in the CPS of C. jejuni is not completely understood. Previously, we have demonstrated that the enzymes Cj1441, Cj1436, Cj1437, and the C-terminal domain of Cj1438 from C. jejuni NCTC 11168 (HS:2) are necessary for formation of the d-glucuronamides of S-serinol phosphate and ethanolamine phosphate (18–20). The ultimate enzyme in the overall biosynthetic pathway is apparently Cj1435, which has been previously annotated as a HAD phosphatase. Here we demonstrated that Cj1435 is responsible for the hydrolysis of the phosphate group from the d-glucuronamide of either ethanolamine phosphate or S-serinol phosphate.

The catalytic substrate profile for the hydrolysis of various substrate analogs shows a marked preference of the enzyme toward substrates that contain an amide bond instead of a primary amine. For example, the rate of hydrolysis of compounds S-11 and 12 are more than two orders of magnitude faster than the hydrolysis of compounds 3 and S-4. This observation is fully consistent with the fact that the enzyme responsible for amide bond formation does not utilize ethanolamine or serinol as substrates (20). The amide bond forming enzyme Cj1438 does not utilize UDP-d-glucuronate as a substrate, but can utilize d-glucuronate (5), the methyl glycoside of d-glucuronate (6), and the disaccharide with 1-O-methyl-d-ribose (13) (20). These results demonstrate quite clearly that amide bond formation follows polysaccharide formation and that phosphate hydrolysis is the ultimate step in formation of the glucuronamide moiety in the HS:2 serotype of C. jejuni.

The glucuronamide modification is present in the capsular polysaccharides of C. jejuni serotype HS:2 and HS:19 (14, 34). Based on previous bioinformatic analysis conducted in our laboratory, we also identified serotypes HS:22 and HS:38 as likely containing a glucuronamide moiety (35). While these four serotypes contain a d-glucuronamide moiety within their CPS, it should be noted that HS:2 is the only serotype which has a d-glucuronamide moiety formed with ethanolamine. The other serotypes lack an enzyme homologous to Cj1436, the PLP-dependent l-serine phosphate decarboxylase, which is necessary to create the ethanolamine phosphate required to form the d-glucuronamide of ethanolamine phosphate.

The biosynthetic pathway begins with the dual oxidation of UDP-d-glucose to UDP-d-glucuronate catalyzed by Cj1441 (18). This is followed by the reaction of UDP-d-glucuronate with the ribosyl end of the growing capsular polysaccharide catalyzed by an unidentified glycosyltransferase to form the d-glucuronate (1→2)-d-ribosyl polysaccharide. Possible candidates for the glycosyltransferase are Cj1432, Cj1434, Cj1438, Cj1440, and Cj1442. The N-terminal domain of Cj1432 is an intriguing candidate because it contains an apparent GT4 (retaining) glycosyltransferase contained within residues 1 through 356. The other possible candidates are all GT2 inverting enzymes, which would form a glycosidic bond of the opposite stereochemistry. Investigations are ongoing to determine which enzyme is responsible for the glycosyltransferase reaction. The phosphorylated primary amines necessary for the biosynthesis of the glucuronamide are synthesized using Cj1437, a PLP-dependent transaminase, which uses DHAP and l-glutamate and produces S-serinol phosphate (19). The other primary amine is formed by Cj1436, a PLP-dependent l-serine phosphate decarboxylase, which catalyzes the formation of ethanolamine phosphate (19). These phosphorylated primary amines are then used as substrates along with MgATP and the glucuronate (1→2)-d-ribosyl end of the growing polysaccharide chain to form the phosphorylated d-glucuronamide. This molecule is dephosphorylated by Cj1435 to give the final product of the pathway. The reaction scheme is summarized in Scheme 1.

Scheme 1.

Proposed reaction pathway for the formation of the d-glucuronamide found within the capsular polysaccharide of C. jejuni NCTC 11168 (serotype HS:2)

To our knowledge, this work represents the first time a glucuronamide modification of a capsular polysaccharide has been functionally characterized. Based on known capsular polysaccharides, this modification is also present in E. coli L19, Acinetobacter baumannii G7, Shigella boydii type 8, Vibrio cholerea H11 (non-O1), Proteus mirabilis O27, and Vibrio. vulnificus strain 6353 (36–40). We propose naming the biosynthetic pathway for making this modification to the CPS of C. jejuni, the Capsule Amide Ligase pathway. Enzymes in this pathway will adopt the following naming convention: Cj1438, the glucuronate amide ligase, will be named CalA; Cj1437, the DHAP transaminase, is to be named CalB; Cj1436, the l-serine phosphate decarboxylase, will be CalC; Cj1435, the glucuronamide phosphate phosphatase, will be called CalD. Cj1441, the UDP-glucose 6-dehydrogenase, is not unique to this pathway and thus will retain its designator, KfiD.

Conclusions

Campylobacter jejuni is a pathogenic organism that is surrounded by a capsular polysaccharide. This CPS is modified with a d-glucuronoamide moiety formed from either ethanolamine or serinol. We have now shown that a phosphorylated glucuronamide moiety is formed by the combined catalytic activities of Cj1441, Cj1436, Cj1437, Cj1438 (and possibly Cj1432). We demonstrate that the final enzyme is the pathway is Cj1435 which acts to hydrolyze the phosphate ester bond. These studies were used to formulate a biosynthetic pathway for the formation of the d-glucuronamide moiety contained within the capsular polysaccharide of C. jejuni.