Characterization of the Enzymes Encoded by the Anthrose Biosynthetic Operon of Bacillus anthracis

Abstract

Bacillus anthracis spores, the etiological agents of anthrax, possess a loosely fitting outer layer called the exosporium that is composed of a basal layer and an external hairlike nap. The filaments of the nap are formed by trimers of the collagenlike glycoprotein BclA. Multiple pentasaccharide and trisaccharide side chains are O linked to BclA. The nonreducing terminal residue of the pentasaccharide side chain is the unusual sugar anthrose. A plausible biosynthetic pathway for anthrose biosynthesis has been proposed, and an antABCD operon encoding four putative anthrose biosynthetic enzymes has been identified. In this study, we genetically and biochemically characterized the activities of these enzymes. We also used mutant B. anthracis strains to determine the effects on BclA glycosylation of individually inactivating the genes of the anthrose operon. The inactivation of antA resulted in the appearance of BclA pentasaccharides containing anthrose analogs possessing shorter side chains linked to the amino group of the sugar. The inactivation of antB resulted in BclA being replaced with only trisaccharides, suggesting that the enzyme encoded by the gene is a dTDP-β-l-rhamnose α-1,3-l-rhamnosyl transferase that attaches the fourth residue of the pentasaccharide side chain. The inactivation of antC and antD resulted in the disappearance of BclA pentasaccharides and the appearance of a tetrasaccharide lacking anthrose. These phenotypes are entirely consistent with the proposed roles for the antABCD-encoded enzymes in anthrose biosynthesis. Purified AntA was then shown to exhibit β-methylcrotonyl-coenzyme A (CoA) hydratase activity, as we predicted. Similarly, we confirmed that purified AntC had aminotransferase activity and that purified AntD displayed N-acyltransferase activity.

Bacillus anthracis, the causative agent of anthrax, is a Gram-positive, rod-shaped soil bacterium that forms spores when deprived of essential nutrients (15). Spore formation begins with an asymmetric septation that divides the developing cell into a forespore compartment and a larger mother cell compartment, each of which contains a copy of the genome. The mother cell then engulfs the forespore and surrounds it with three protective layers: a cortex composed of peptidoglycan, a closely apposed proteinaceous coat, and a loosely fitting exosporium (10). Mother cell lysis releases the mature spore, which is dormant and capable of surviving in harsh environments for many years (17). When spores encounter an aqueous environment containing nutrients, they can germinate and grow as vegetative cells (21).

Recently, interest in B. anthracis spores has intensified in response to their use as agents of bioterrorism. Of particular interest has been the outermost layer of the spore, the exosporium, which serves as a semipermeable barrier to potentially harmful macromolecules (8, 25) and as the vital first point of contact with the immune system of an infected host (11, 18, 30). The exosporium of B. anthracis and of closely related species, such as Bacillus cereus and Bacillus thuringiensis, is comprised of a paracrystalline basal layer and an external hairlike nap (1). The basal layer contains approximately 20 different proteins (20, 23), while the filaments of the nap are formed by trimers of a single collagenlike glycoprotein called BclA (2, 26). The central region of BclA contains a large number of GXX repeats, and the region varies in length in naturally occurring strains of B. anthracis, resulting in hairlike naps of differing lengths (22, 27). Most of the GXX repeats are GPT, and many of the threonine residues are glycosylated. Two major oligosaccharide side chains are present, a pentasaccharide and a trisaccharide, and both are linked to the protein through reducing terminal N-acetylgalactosamine (GalNAc) residues (3). Several studies have demonstrated that the oligosaccharides are antigenic and are exposed on the surface of Bacillus anthracis spores (14, 29). This makes them prime targets for both detection devices and immunoprophylaxis.

We previously reported our use of hydrazinolysis to release BclA oligosaccharides from exosporium preparations (3). The primary product was a tetrasaccharide that formed as a result of the undesirable loss of the reducing terminal GalNAc residue of the pentasaccharide, a process called “peeling.” We determined that the oligosaccharide consisted of a linear chain of three rhamnose residues with a novel deoxyamino sugar at its nonreducing terminus. This unusual sugar, 2-O-methyl-4-(3-hydroxy-3-methylbutamido)-4,6-dideoxy-d-glucose, was given the trivial name anthrose.

Rhamnose is the major sugar present in both the trisaccharide and the pentasaccharide, and a four-gene rhamnose biosynthetic operon was previously identified (22). Previously, we proposed a pathway for anthrose biosynthesis (Fig. 1) and identified a four-gene operon (Fig. 2) that is essential for its biosynthesis (5). An in-frame deletion of the first gene of the operon reduced the amount of anthrose by approximately 50%, whereas the deletion of any one of the other three genes totally abolished anthrose synthesis. Here, we describe the characterization of the altered oligosaccharide side chains of the four deletion mutants. We also cloned several genes that we predicted are involved in anthrose biosynthesis and demonstrated that the gene products possessed the expected biochemical activities.

FIG. 1.

Proposed biosynthetic pathway of anthrose. The pathway utilizes dTDP-4-keto-6-deoxy-α-d-glucose, an intermediate in rhamnose biosynthesis, and methylcrotonyl-CoA, derived from leucine catabolism. (Modified from reference 5.)

FIG. 2.

Anthrose operon and flanking genes. The four genes of the anthrose operon are antA (BAS3322), antB (BAS3321), antC (BAS3320), and antD (BAS3319). The operon is flanked by genes that encode a putative collagenase (BAS3323) and a putative methyltransferase (BAS3318). (Modified from reference 5.)

MATERIALS AND METHODS

Materials.

Isopropyl β-d-thiogalactopyranoside (IPTG) and 5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside (X-Gal) were purchased from Fisher Scientific (Pittsburgh, PA). Other reagents used in this study, including sugars and cofactors, were obtained from Sigma-Aldrich Corp. (St. Louis, MO). Restriction enzymes were obtained from Promega (Promega, Madison, WI).

Chemical synthesis of 3-hydroxy-3-methylbutyryl-CoA.

3-Hydroxy-3-methylbutyryl-coenzyme A (CoA) was prepared from 3-hydroxy-3-methylbutyric acid by the mixed anhydride method (9). To a solution of 3-hydroxy-3-methylbutyric acid (54 ml, 0.5 mmol) in 3 ml of diethyl ether cooled in an ice-water bath, we added triethylamine (73 ml, 0.53 mmol), followed by chloro ethylformate (72 ml, 0.75 mmol). The reaction mixture was stirred at 0°C for 3 h and then filtered and concentrated to afford the desired mixed anhydride as a colorless liquid. A portion of the prepared mixed anhydride (1.9 mg, 10 mmol) was added to coenzyme A trilithium dihydrate (4 mg, 5 mmol) dissolved in 0.1 ml of a mixture of sodium carbonate (50 mM), ethanol, and ethyl acetate (1:1:1 [vol/vol/vol]). The reaction mixture was stirred at room temperature for 5 h and concentrated in vacuo. The residue was purified by reverse-phase high-performance liquid chromatography (RP-HPLC) with a C₁₈ analytical column eluted with a gradient of 5-to-50% methanol in 10 mM aqueous ammonium acetate over 30 min. The desired product had a retention time of 11.5 min. The combined product fractions from multiple injections of the HPLC reaction mixture were concentrated in vacuo to produce 3-hydroxy-3-methylbutyryl-CoA (2 mg) in a 47% yield. Analytical data: negative-ion electrospray mass spectrum ([M]⁻) calculated for C₂₆H₄₄N₇O₁₈P₃S was 867.2, observed was 867.4; positive-ion electrospray mass spectrum ([M + H]⁺) calculated for C₂₆H₄₅N₇O₁₈P₃S was 868.2, observed was 869.0.

Bacterial strains and plasmids.

The Sterne 34F2 veterinary vaccine strain of B. anthracis used as the wild-type parental strain in this and previous studies was obtained from the U.S. Army Medical Research Institute of Infectious Diseases, Fort Detrick, MD. Spores were prepared, quantitated, and stored as described previously (4). Plasmids pMAD, pCLT1242, and pCLT1476 and the shuttle plasmid pCLT1376 have been described previously (5). In-frame deletion mutants for genes BAS3318, BAS3319, BAS3320, BAS3321, and BAS3322 were made as described previously (5). Escherichia coli strains INVαF′, GM1684, and BL21(DE3) were routinely grown at 37°C with shaking in Luria-Bertani (LB) broth. For protein expression, cultures were grown at room temperature.

Cloning and expression of the genes rmlA, rmlB, antD, antC, and antA.

The BAS1135 (rmlA), BAS1137 (rmlB), BAS3319 (antD), BAS3320 (antC), and BAS3322 (antA) genes were all PCR amplified using chromosomal DNA of the Sterne strain of B. anthracis as a template and the primers listed in Table S1 in the supplemental material. The PCR products were cloned into the pET21a vector (Novagen), which gives a C-terminally His-tagged fusion protein, and transformed into E. coli strain BL21(DE3) for enzyme expression. The plasmids were sequenced to confirm their correct construction. E. coli BL21(DE3) cells containing the recombinant plasmids were grown overnight in LB medium containing 100 μg/ml ampicillin at 37°C with shaking at 250 rpm. The cultures were diluted with 20 volumes of fresh medium (1 liter) and grown at 37°C until reaching an optical density at 600 nm (OD₆₀₀) of 0.6. IPTG was added to yield a final concentration of 0.2 mM. After an additional 16 h of growth at room temperature, the cells were harvested by centrifugation and washed three times with cold (4°C) lysis buffer (50 mM NaH₂PO₄, 300 mM NaCl, 10 mM imidazole, pH 8.0). Pellets were resuspended in 30 ml of cold lysis buffer and sonicated in an ice bath for a total of 5 min at 60% maximum power using a model 300 Fisher sonic dismembrator. Cell debris was removed by centrifugation, and the resulting supernatant passed through a 2-ml Ni-nitrilotriacetic acid affinity column (Qiagen). The columns were washed and eluted according to the manufacturer's instructions. Imidazole was removed from the samples by dialysis against 20 mM Tris HCl, 1% glycerol, pH 8.0.

Preparation and analysis of oligosaccharides.

The preparation and analysis of oligosaccharides were carried out as described previously (4). Briefly, this process involves extracting spores with hot protein sample loading buffer, fractionating the extracted proteins by SDS-PAGE, blotting the gel to a polyvinylidene difluoride (PVDF) membrane, excising the BclA band, and releasing the oligosaccharides from the bound proteins by hydrazinolysis. Oligosaccharide hydrazides were analyzed using electrospray-ionization mass spectrometry (ESI-MS) on a Micromass Q-TOF 2 mass spectrometer (Micromass Ltd., Manchester, United Kingdom). Injections were made into a stream of 1:1 acetonitrile-water (vol/vol) containing 0.1% formic acid at a flow rate of 1 μl/min. The mass spectral data were processed using the Max-Ent3 module of MassLynx 3.5.

HPLC analyses.

The sugar nucleotides in the enzymatic reaction mixtures were analyzed using ion exchange high-pressure liquid chromatography (HPLC) as described previously (4).

Confirmation of CoA hydratase activity.

The reaction mixtures for confirmation of CoA hydratase activity consisted of 0.1 mM unsaturated CoA thioester (either β-methylcrotonyl-CoA or crotonyl-CoA) and 0.1 μg freshly prepared enoyl-CoA hydratase (antA gene product) in 20 mM Tris-HCl, pH 8.0, in a total volume of 50 μl. The reactions were performed at 37°C for 2 h and were quenched by boiling for 5 min. The control reaction mixtures contained enzyme that had been heated to 100°C for 5 min. The reaction mixtures were passed through a 3,000-molecular-weight (MW)-cutoff spin filter (Millipore, Bedford, MA) and analyzed by reverse-phase HPLC. This was done using a Vydac 218TP54 column (4 by 250 mm) connected to a Dynamax pump system and UV detector (Rainin Instrument Company, Inc.). The column was equilibrated for 2.5 min with 50 mM potassium phosphate, pH 7, plus 10% methanol by volume before the samples were applied. CoA thioesters were separated by eluting the column with a linear methanol gradient (10% to 100%) over 25 min. The flow rate was 0.7 ml/min, and the absorbance at 263 nm of the effluent was recorded. Product peaks were analyzed by ESI-MS.

CoA hydratase assay.

The CoA hydratase reactions were monitored by following the decrease in absorbance at 263 nm of a solution of β-methylcrotonyl-CoA resulting from the hydration of its double bond. This was measured in temperature-regulated quartz cuvettes (1-cm light path) in a Cary 50 spectrophotometer (Varian). To determine the optimal reaction temperature of the enzyme, 0.05 μg of the purified enzyme (BAS3322) was added to 500 μl 20 mM Tris-HCl, 0.04% (wt/vol) bovine serum albumin, pH 8.0, containing 40 μM β-methylcrotonyl-CoA at temperatures ranging from 10° to 50°C. The influence of pH was determined using the following buffers: 50 mM sodium acetate (pH 4.0 to 5.0), 50 mM sodium phosphate (pH 6.0 to 8.0), and 50 mM N-cyclohexyl-3-aminopropanesulfonic acid (CAPS) (pH 9.0 to 11.0). To determine K_m and k_cat values, we used initial β-methylcrotonyl-CoA or crotonyl-CoA concentrations of 10, 15, 20, 30, 50, 75, and 100 μM. Enzyme (50 ng) was added to 500 μl of 20 mM Tris-HCl, 0.04% (wt/vol) bovine serum albumin, pH 8.0, containing the unsaturated CoA substrate at 25°C. A unit of enzyme activity is defined as the amount of enzyme that hydrates 1.0 μmol of crotonyl-CoA per minute at pH 7.5 and at 25°C. A molar extinction coefficient of 6.7 × 10³ M⁻¹ cm⁻¹ is used in the calculations (24). Experimental data were analyzed using the EZ-FIT program (19).

Preparation of dTDP-α-d-glucose.

The reaction mixture for the preparation of dTDP-α-d-glucose was composed of 100 μl of 10 mM dTTP, 20 mM glucose-1-phosphate, 10 mM MgCl₂, and 50 mM Tris-HCl, pH 7.8, plus 20 μg of glucose-1-phosphate thymidyl transferase (BAS1135, the rmlA gene product). The mixture was incubated at 37°C for 2 h and then passed through a centrifugal ultrafilter with a nominal 3,000-MW cutoff. The filtrate was diluted 1:5 with water and analyzed by anion exchange HPLC. The pooled-product peak was then desalted on an Extract-Clean solid-phase extraction (SPE) carbograph (150 mg/4 ml) column (Alltech). Prior to sample application, the column was activated by washing with 4 ml of 80% acetonitrile, 0.1% trifluoroacetic acid, followed by 10 ml of water. The HPLC-purified sample was then applied to the column, followed by 10 ml of water. Sugar nucleotides were eluted with 4 ml of 25% acetonitrile containing 0.1% trifluoroacetic acid. The effluent was freeze-dried using a SpeedVac and stored at −80°C prior to analysis by ESI-MS. A control reaction mixture contained heat-inactivated enzyme.

Preparation of dTDP-4-amino-6-deoxy-α-d-glucose.

Since dTDP-α-d-glucose and dTDP-4-keto-6-deoxy-α-d-glucose were difficult to adequately separate by HPLC, a concerted reaction using both the dehydratase (BAS1137, the rmlB gene product) and the aminotransferase (antC gene product) was carried out to make the amino sugar nucleotide. The reaction mixture was composed of 4 mM dTDP-α-d-glucose (prepared as described above), 1 mM NAD⁺, 0.6 mM pyridoxal phosphate, 0.45 mM glutamine (or glutamic acid), 50 mM Tris-HCl, pH 7.8, 5 μg dTDP-d-glucose-4,6-dehydratase (rmlB gene product), and 5 μg of the aminotransferase (antC gene product). The mixture was incubated at 37°C for 2 h, ultrafiltered, and subjected to anion exchange HPLC. The product peak was then desalted on an Extract-Clean SPE carbograph column as described above and analyzed by ESI-MS. A control reaction mixture contained heat-inactivated enzymes.

Preparation of dTDP-4-(3-hydroxy-3-methylbutamido)-4,6-dideoxy-d-glucose.

The reaction mixture for the preparation of dTDP-4-(3-hydroxy-3-methylbutamido)-4,6-dideoxy-d-glucose was composed of the HPLC-purified amino sugar nucleotide prepared as described above, 0.4 mM 3-hydroxy-3-methylbutyryl-CoA (either chemically synthesized or prepared by the enzymatic hydration of methylcrotonyl-CoA as described above) or 0.4 mM acetyl-CoA, and 5 μg of the acyltransferase (antD gene product). Once again, the reaction mixtures were incubated at 37°C for 2 h, ultrafiltered, and fractionated by ion exchange HPLC. The product peak was purified as described above and analyzed by ESI-MS. The control reaction mixtures contained heat-inactivated enzyme.

Site-directed mutagenesis of enoyl-CoA hydratase.

Site-directed mutagenesis of enoyl-CoA hydratase was performed using a QuikChange site-directed mutagenesis kit (Stratagene). The cloned antA gene was used as the PCR template. The mutagenesis forward primers were as follows: E117Q, 5′ GCA CTA GGC GGA GGT TGT CAA TTA GCA CTT GCT TGT GAT 3′, and E137Q, 5′ GCC TTA ATT GGA TTA CCA CAA ATT ACG TTA GGT CTA TTT C 3′. The reverse primers were the exact complements of the forward primers. After PCR amplification, the parental DNA was digested with 10 units of DpnI restriction enzyme. A 2-μl aliquot of the DpnI-treated PCR product was used to transform E. coli INVαF′ competent cells (Invitrogen). The transformed cells were plated on LB agar containing 50 μg ampicillin ml⁻¹. Colonies were picked, and minipreps (Qiagen) were prepared from the clones. The desired nucleotide sequence changes were confirmed by sequencing. Recombinant proteins were made as described above.

RESULTS AND DISCUSSION

Characterization of the oligosaccharides of anthrose operon mutants.

We previously described the construction of B. anthracis mutants containing in-frame deletions of the four genes of the anthrose operon (5). The deletion of antA reduced the anthrose-to-rhamnose ratio of the mutant spores to about 50% of that seen in wild-type spores. However, the deletion of any one of the other three anthrose operon genes, antB, antC, and antD, completely abolished anthrose production. In these early studies, the anthrose levels were measured by a procedure that was highly specific for the sugar, but we did not determine the effects of the gene deletions on the BclA oligosaccharide structure. However, we recently devised a new microscale procedure that made it possible to study the oligosaccharides. As described in the previous section, the oligosaccharides of BclA on PVDF membrane blots were released by hydrazinolysis and analyzed using ESI-MS. The oligosaccharide hydrazides obtained from the four different anthrose operon gene deletion mutants were compared to those liberated from wild-type spores. The mass spectrometry results are shown in Fig. 3. It can be seen that two prominent oligosaccharide hydrazides, a pentasaccharide (m/z 933) and a trisaccharide (m/z 542) with a methylated rhamnose, are obtained from wild-type spores (Fig. 3A). The m/z 730 peak results from a partial “peeling” of GalNAc from the reducing terminus of the pentasaccharide during hydrazinolysis. Figure 4 shows the structures of these oligosaccharides.

FIG. 3.

Positive-ion electrospray mass spectra of BclA oligosaccharides from wild-type B. anthracis (A) and four deletion mutants with mutations ΔantA (B), ΔantB (C), ΔantC (D), and ΔantD (E). Prominent peaks in the spectrum of wild-type oligosaccharides correspond to those of a pentasaccharide (m/z 933) and a methylated trisaccharide (m/z 542). The m/z 730 peak is the product of the peeling of a GalNAc residue from the pentasaccharide. See the text for additional interpretation of the spectra.

FIG. 4.

Structures of some of the oligosaccharide hydrazides obtained from wild-type and mutant spores. The pentasaccharide hydrazide of m/z 933 (A) was derived from wild-type and ΔantA mutant spores. The tetrasaccharide of m/z 730 (B) results from the loss of the reducing terminal GalNAc from the pentasaccharides of the above two strains. The tetrasaccharide hydrazide of m/z 688 (C) is derived from ΔantC and ΔantD mutant spores. The methylated trisaccharide hydrazide of m/z 542 (D) was obtained from the wild-type and all four mutant strains.

We predicted that the first gene of the anthrose operon, antA, annotated as an enoyl-CoA hydratase, encoded an enzyme involved in the biosynthesis of the five-carbon acyl side chain of anthrose. The mass spectrum of the oligosaccharide hydrazides released from the antA mutant (Fig. 3B) differed from the wild-type spores in that it showed two new pentasaccharide peaks, m/z 919 and m/z 875, in addition to the m/z 933 peak in wild-type spores. The three pentasaccharide peaks were individually fragmented using tandem mass spectroscopy (MS-MS). When this was performed on the m/z 933 ion, a fragment peak of m/z 260 was observed, corresponding to an anthrose residue (Fig. 5A). However, MS-MS of the m/z 919 peak yielded a fragment of m/z 246, corresponding to an anthrose analog with a four-carbon side chain, a 3-hydroxybutyryl group (Fig. 5B). MS-MS of the m/z 875 peak gave an m/z 202 fragment, consistent with it being an acetyl analog of anthrose (Fig. 5C). These two anthrose analogs were not present in the wild-type spores. The peaks of m/z 730, 716, and 672 are products of GalNAc peeling of their parent pentasaccharides. It is not clear why any anthrose at all is made by the ΔantA mutant. However, several genes in the B. anthracis genome have been annotated as enoyl-CoA hydratases, so it is possible that the ΔantA mutant is complemented by another enoyl-CoA hydratase. The presence of anthrose analogs in the mutant might be the result of a less stringent CoA thioester requirement of the complementing enzyme.

FIG. 5.

Results of tandem positive-ion electrospray mass spectrometry (MS-MS) of the three pentasaccharide fragment ions derived from the ΔantA mutant shown in Fig. 3B. The structures of the major fragment ions are also shown.

As can be seen in Fig. 3C, the mass spectrum of the antB mutant oligosaccharides shows only the trisaccharide (m/z 542) and a small amount of a disaccharide artifact (m/z 339) that is due to peeling. No pentasaccharides or tetrasaccharides are present. The antB gene product (BAS3321) was annotated as a glycosyl transferase. Our results strongly suggest that it is a dTDP-β-l-rhamnose α-1,3-l-rhamnosyl transferase that is capable of catalyzing the addition of a rhamnose residue to the 3 position of another rhamnose residue. This would explain the total absence of tetrasaccharides and pentasaccharides in the antB mutant.

We predicted that the antC and antD gene products are involved in the biosynthesis of the acyl side chains of anthrose. The almost identical mass spectra obtained for spores of the antC mutant (Fig. 3D) and the antD mutant (Fig. 3E) confirmed that no anthrose was present. In addition to the usual trisaccharide, these mutants also possessed a new tetrasaccharide. MS-MS revealed that the nonreducing terminal rhamnose residues of both the trisaccharide and tetrasaccharide were O methylated (data not shown). The function of this methylation is not known, but it might serve to block further oligosaccharide chain extension.

Biochemical characterization of anthrose operon gene products.

The first gene of the anthrose operon, antA, was cloned and expressed with a His tag, and the product was Ni affinity column purified. We predicted that this gene encoded a hydratase that catalyzes the addition of water to the double bond of β-methylcrotonyl-CoA to form 3-hydroxy-3-methylbutyryl-CoA, a final step in the biosynthesis of the acyl side chain of anthrose (Fig. 1). The recombinant enzyme gave a single intense band upon SDS-polyacrylamide gel electrophoresis of approximately 30 kDa, consistent with a predicted molecular mass of 29.7 kDa. The absorbance at 260 nm of a solution of β-methylcrotonyl-CoA rapidly decreased after the addition of the recombinant enzyme, consistent with the hydration of a double bond in the substrate. The enzyme exhibited maximal activity at 35°C and had a pH optimum of about 8. The HPLC results illustrated in Fig. 6A and B show that the antA gene product caused the conversion of β-methylcrotonyl-CoA to a compound with a shorter retention time. An identical retention time was also observed for a sample of 3-hydroxyl-3-methylbutyryl-CoA that we chemically synthesized. These results indicate that the enzyme does possess the predicted β-methylcrotonyl hydratase activity, yielding 3-hydroxy-3-methylbutyryl-CoA. The enzyme also catalyzed the hydration of a related compound, crotonyl-CoA (Fig. 6C and D). The product, 3-hydroxybutyryl-CoA, was chromatographically identical to a commercially obtained sample of the compound. The kinetic constants of the enzyme for both of the substrates were determined. Crotonyl-CoA and β-methylcrotonyl-CoA had similar k_cat values, 86.3 s⁻¹ and 82.9 s⁻¹, respectively, but the K_m of the enzyme acting on crotonyl-CoA (29 μM) was higher than when β-methylcrotonyl-CoA was the substrate (5.6 μM). Thus, the k_cat/K_m ratio is 1.5 × 10⁷ M⁻¹ s⁻¹ for methylcrotonyl-CoA and 3.0 × 10⁶ M⁻¹ s⁻¹ for crotonyl-CoA, making methylcrotonyl-CoA the better substrate. This would explain why an anthrose analog with a 3-hydroxylbutyryl-CoA side chain was not seen in wild-type spores. However, the fact that this anthrose analog is seen in ΔantA mutant spores indicates that the crotonyl-CoA substrate is available during spore formation.

FIG. 6.

Reverse-phase HPLC analysis of CoA derivatives. The antA gene product catalyzed the conversion of β-methylcrotonyl-CoA (A) to 3-hydroxy-3-methylbutyryl-CoA (B). The enzyme also catalyzed the conversion of crotonyl-CoA (C) to 3-hydroxybutyryl-CoA (D).

We used site-directed mutagenesis to identify amino acid residues essential for activity in the hydratase. In the distantly related rat enoyl-CoA hydratase, two glutamate residues (E144 and E164) were found to play an important role in the catalytic hydration of substrate (6, 12, 16). The analogous glutamates in the antA gene product are E117 and E137. We changed each individual glutamate residue to a glutamine residue. Both of the purified recombinant mutant enzymes gave single strong bands around 30 kDa on SDS-PAGE, at the same position as for the wild-type enzyme.

The specific activity of wild-type hydratase was measured as described in Materials and Methods and was found to be 157 U/mg protein. The hydration reaction, which we measured by monitoring the decrease in absorbance at 263 nm, went to completion in less than 3 min when 20 ng of wild-type BAS3322 enzyme was used. However, neither mutant exhibited any detectable hydratase activity, even when 1,000 times more protein (20 μg) was tested. This showed that the two glutamate residues were essential for enzyme activity.

We were not able to biochemically confirm that the antB gene encodes a rhamnosyl transferase that catalyzes the attachment of a rhamnose residue to the 3 position of another rhamnose because we did not have the necessary substrate.

We cloned and expressed the third gene of the anthrose operon, antC, encoding a putative aminotransferase. We predicted that the enzyme encoded by this gene catalyzes the addition of an amino group to dTDP-4-keto-6-deoxy-α-d-glucose, an intermediate in rhamnose biosynthesis. Since this sugar nucleotide is not available commercially, we had to make it. We therefore cloned and expressed the first two genes of the rhamnose biosynthetic pathway, rmlA and rmlB, for use in synthesizing a substrate for the aminotransferase. α-d-Glucose-1-phosphate was first converted to dTDP-α-d-glucose using the rmlA gene product, glucose-1-phosphate thymidyl transferase. The product was purified by HPLC (Fig. 7A), and its identity confirmed by ESI-MS in the negative-ion mode (Fig. 8A). The dTDP-α-d-glucose was then converted to dTDP-4-keto-6-deoxy-α-d-glucose using the rmlB gene product, dTDP-d-glucose-4,6-dehydratase. The keto sugar product was partially purified by HPLC (Fig. 7B), and its structure confirmed by ESI-MS (Fig. 8B). A small amount of dTDP-α-d-glucose is present in the keto sugar mass spectrum because we were unable to adequately separate the two sugars by HPLC. Because of the difficulty in purifying the keto sugar, we used a concerted reaction involving both the dehydratase (rmlB gene product) and the aminotransferase (antC gene product) to make the amino sugar nucleotide. The reactions were carried out in the presence of pyridoxal phosphate and either glutamine or glutamic acid as the amino group donor. As shown in Fig. 7C, a product with an HPLC retention time of about 10 min was obtained. ESI-MS confirmed that it was the amino sugar nucleotide (Fig. 8C). The amino sugar nucleotide peak was not present when heat-inactivated enzyme was used (not shown).

FIG. 7.

Anion exchange HPLC analysis of the sugar nucleotide products of the rmlA gene product acting on glucose-1-phosphate (A), the rmlB gene product acting on dTDP-α-d-glucose (B), and the antC gene product acting on dTDP-4-keto-6-deoxy-α-d-glucose (C). Samples were analyzed using a Dionex CarboPac PA-1 column (4 by 250 mm).

FIG. 8.

Negative-ion electrospray mass spectra of the sugar nucleotides whose HPLC separations are illustrated in Fig. 7.

The fourth gene of the anthrose operon, antD, is annotated as an acetyl-CoA transferase gene, but we predicted that it really codes for a 3-hydroxyl-3-methylbutyryl-CoA transferase. In order to confirm this, we needed to demonstrate that the affinity-purified gene product catalyzes the addition of a 3-hydroxyl-3-methylbutyryl group from its CoA derivative to dTDP-4-amino-4,6-dideoxy-α-d-glucose. We used chemically synthesized 3-hydroxyl-3-methylbutyryl-CoA as our substrate and analyzed the reaction mixture by HPLC (Fig. 9B). The expected product of the enzymatic reaction was an anthrose nucleotide without an O-methyl group at the 2 position of the sugar ring. Its production was demonstrated by negative-ion mass spectrometry (Fig. 10) and confirmed by tandem mass spectrometry of the molecular ion (m/z 646) (see the supplemental material). We also tested the ability of the enzyme to utilize other acyl substrates. When 3-hydroxybutyryl-CoA was used as the substrate, an anthrose analog with a β-hydroxylbutyryl side chain was made (Fig. 9C). It gave a molecular ion of m/z 632 upon ESI-MS in the negative-ion mode. Using acetyl-CoA as the substrate yielded an acetyl analog of anthrose (Fig. 9D) that gave a molecular ion of m/z 588.

FIG. 9.

Anion exchange HPLC analysis of anthrose analog nucleotides having different acyl side chains. Samples were analyzed on a Dionex CarboPac PA-1 column (4 by 250 mm). The control (A) contained heat-inactivated acyltransferase and the dTDP-4-amino sugar. The small peak at 14.8 min is dTMP, which results from the partial breakdown of some of the sugar nucleotide. The reaction mixtures yielding the chromatograms shown in panels B, C, and D contained acyltransferase and the indicated CoA derivative and gave the products whose structures are shown on the right.

FIG. 10.

Negative-ion electrospray mass spectrum of the product of antD enzyme acting on 3-hydroxy-3-methylbutyryl-CoA and the amino sugar nucleotide product of antC. The base peak of m/z 646 shows that the acyl group has been transferred to the amino sugar nucleotide, making an anthrose analog with a five-carbon side chain. The results of further fragmentation of the m/z 646 ion by tandem mass spectrometry and their interpretation are in the supplemental material.

Anthrose is O methylated at the 2 position of the sugar, and we suspected initially that a methyltransferase encoded by a gene adjacent to the anthrose operon, BAS3318, catalyzed the methylation. However, as we previously reported, the deletion of this gene did not abolish anthrose biosynthesis (5). It is still possible that the methyltransferase is involved in the normal methylation of anthrose but that, in its absence, another O-methyltransferase can complement it.

Since our structural characterization of the anthrax oligosaccharide, several strikingly similar oligosaccharides have been found in a variety of other organisms (Fig. 11). Its similarity to saccharides of certain Pseudomonas and Shewanella strains has been pointed out previously (13). The anthrose oligosaccharide is also very similar to an oligosaccharide made by a mycobacterium (7). As can be seen by the diagrams in Fig. 11, the nonreducing terminal sugar in all four cases is a 4- amino-4,6-dideoxy glucose that is amide linked to a side chain containing a penultimate hydroxyl group. In three of the oligosaccharides, the nonreducing terminal deoxyamino sugar is β linked to the 3 position of a rhamnose and followed by one or more rhamnose residues, the last of which is linked at the 2 position. These remarkably similar structures suggest that similar biosynthetic pathways are likely to be utilized by all four organisms. It is of interest that the anthrose analog with the hydroxybutyryl side chain made by the antA mutant is identical to the anthrose analog made by Pseudomonas syringae pv. tabaci 6605 (28).

FIG. 11.

Anthrose oligosaccharide of B. anthracis (A) and structurally similar saccharides from Mycobacterium (B), Pseudomonas (C), and Shewanella (D) strains.

Our results provide strong biochemical evidence supporting our proposed anthrose biosynthetic pathway (Fig. 1). Several enzymes in the pathway and the genes that encode them have yet to be identified. For example, we have not been able to identify the methyltransferase responsible for adding the methyl group to the 2 position of anthrose or the methyltransferase responsible for capping terminal rhamnose residues. The identification of these and other genes involved in anthrose biosynthesis, including those involved in the assembly of the pentasaccharide, is the goal of future work.