Olefin Isomerization Regiochemistries during Tandem Action of BacA and BacB on Prephenate in Bacilysin Biosynthesis

Abstract

BacA and BacB, the first two enzymes of the bacilysin pathway, convert prephenate to an exocylic regioisomer of dihydrohydroxyphenylpyruvate (ex-H₂HPP) on the way to the epoxycyclohexanone warhead in the dipeptide antibiotic, bacilysin. BacA decarboxylates prephenate without aromatization, converting the 1,4-diene in prephenate to the endocyclic 1,3 diene in Δ⁴Δ⁸-dihydrohydroxyphenylpyruvate (en-H₂HPP). BacB then performs an allylic isomerization to bring the diene into conjugation with the 2-ketone in the product Δ³Δ⁵-dihydrohydroxyphenylpyruvate (ex-H₂HPP). To prove that BacA acts regiospecifically on one of the two prochiral olefins in prephenate, we generated 1,5,8-[¹³C]-chorismate from bacterial fermentation of 5-[¹³C]-glucose and in turn produced 2,4,6-[¹³C]-prephenate via chorismate mutase. Tandem action of BacA and BacB gave 2,4,8-[¹³C]-7R-ex-H₂HPP, showing that BacA isomerizes only the pro-R double bond in prephenate. Nonenzymatic isomerization of the BacA product into conjugation gives only the Δ³ E-geometric isomer of Δ³Δ⁵-ex-H₂HPP. On the other hand, acceleration of the allylic isomerization by BacB gives a mixture of the E- and Z-geometric isomers of the 7R-product, indicating some rerouting of the flux, likely through dienolate geometric isomers.

Bacilysin, a dipeptide antibiotic (1, 2, 3) produced by Bacillus subtilis strains, contains L-Ala and the nonproteinogenic anticapsin, bearing the antibiotic warhead. Anticapsin (Figure 1A) has an expoxycyclohexanone moiety attached to C₃ of alanine. Once taken up by the dipeptide permeases of susceptible bacterial or fungal cells (4, 5, 6), bacilysin is subjected to peptidase action to release the free anticapsin as an analog of glutamine (7, 8). Anticapsin can then bind to glutaminase domain active sites. The epoxyketone acts as an electrophile to capture the nucleophilic cysteine thiolate in the glutaminase domain of glucosamine-6-phosphate synthetase. Inactivation of both fungal and bacterial glucosamine-6-phosphate synthetases thereby blocks supply of this key amino-sugar for cell wall peptidoglycan assembly and leads to the cell lysis that explains the name bacilysin (8, 9).

Figure 1.

(A) Proposed pathway for the formation of bacilysin from prephenate. Important stereochemical questions are highlighted. (B) Possible enantiomers of en-H₂HPP formed from the enzymatic decarboxylation of prephenate by BacA.

The epoxycyclohexanone ring in the anticapsin amino acid moiety of bacilysin is an unusual modification. It was presumed to arise by shunting some of the flux from the chorismate pathway (3, 10) that would normally yield the aromatic amino acids Phe and Tyr (11, 12). Genetic analysis of bacilysin-producing strains of B. subtilis identified a bac gene cluster (13), that was originally identified as five contiguous genes but has since been expanded by biochemical studies to the seven genes bacA-E and ywfGH. In previous work we have shown that the purified BacAB and YwfGH enzymes can convert prephenate to a 2S-tetrahydrotyrosine diastereomer (Figure 1A); there are three stereogenic centers in H₄Tyr, with the C₄ and C₇ stereochemistries previously unassigned (14). Most notably, the first two enzymes in the pathway, BacA and BacB, act to divert prephenate away from typical aromatization fates to the exocyclic diene Δ³Δ⁵-dihydrohydroxyphenylpyruvate (ex-H₂HPP) (Figure 1A).

The BacA enzyme and homologs we have subsequently characterized in other bacterial pathways to nonproteinogenic dihydro- and tetrahydroamino acid antimetabolites (15, 16) catalyze a novel decarboxylative transformation on prephenate. When canonical prephenate decarboxylases liberate CO₂ from prephenate, the electrons that flow into the 1,4-cyclohexadiene ring lead to aromatization as the C₇-OH is ejected and phenylpyruvate is formed (17, 18). By contrast, the BacA subfamily enzymes, during the comparable decarboxylation, capture the electrons by protonation at one end of the starting 1,4-diene (Figure 1B). This leaves the C₇-OH group intact, resulting in formation of a methylene group at C₈ and a net isomerization of the starting 1,4-diene of prephenate into the 1,3 endocyclic diene of the product Δ⁴, Δ⁸-dihydrohydroxyphenylpyruvate (en-H₂HPP). Instead of aromatization of the cyclohexadiene, the ring remains at the dihydroaromatic oxidation state (14).

BacB acts next and accelerates isomerization to the Δ³ Δ⁵-H₂HPP product. The Δ⁴ double bond has migrated from being endocyclic to the exocyclic Δ³ position (the other double bond has not moved but the numbering priority has changed (see Figure 1A)). This is a thermodynamically favored allylic isomerization (Δ⁴ to Δ³) bringing the diene into conjugation with the 2-ketone (and generating a yellow chromophore (14)). The rate of isomerization is accelerated about 10³ fold by BacB (19). Subsequent conjugate addition of a hydride equivalent from NADH at C₄ by YwfH and transamination by YwfG yields tetrahydrotyrosine, presumably two steps (C₇-alcohol to ketone oxidation and epoxidation of the remaining double bond) away from anticapsin.

To assist in understanding the novel mechanism by which the BacA subfamily separates decarboxylation from aromatization, we have undertaken examination of its stereochemical specificity. The 1,4-diene system in prephenate is symmetric and prochiral, containing pro-R and pro-S olefins. As shown in path 1 of Figure 1B, decarboxylation and isomerization of the 5,6-olefin (pro-R) with protonative quenching at C₆ would yield the 7R-OH product (carbon numbering schemes can be viewed in Figure S4 of Supporting Information). On the other hand, isomerization of the pro-S olefin (path 2) would yield the 7S-OH product. This stereochemistry may have consequences for the action of subsequent enzymes in the pathway. Analogously, as the allylic isomerization occurs from en-H₂HPP to ex-H₂HPP, moving the double bond from the endocyclic Δ⁴ to the exocyclic Δ³ position, two possible olefin isomers, the Z- or E-geometric isomers (Figure 2), could arise for either the 7R or 7S alcohol generated by BacA action (3Z-7R, 3E-7R, 3Z-7S, and 3E-7S: Figure 2). In turn, the configuration at those carbon centers could affect recognition and processing by downstream enzymes in this and related pathways which generate distinct chemical outcomes, such as the biosynthesis of the antimetabolite 2,5-dihydrophenylalanine (16) or 2-carboxy-6-hydroxy-octahydroindole (Choi) in aeruginosin assembly (20).

Figure 2.

Stereoisomers of ex-H₂HPP that could possibly be obtained through the tandem action of BacA decarboxylation of prephenate followed by nonenzymatic or BacB enzymatic isomerization of the internal diene.

This paper describes how we have addressed these two stereochemical questions in the bacilysin pathway. With BacA we desired a source of asymmetrically labeled [¹³C]-prephenate to assess if the BacAB ex-H₂HPP product was ¹³C-enriched at an sp² carbon (C₆) (7S-alcohol) or sp³ carbon (C₈) (7R-alcohol) (Figure 3). In turn, because of the problematic symmetry of prephenate, we decided to make a labeled [¹³C]-chorismate as precursor to asymmetric prephenate, convertible via chorismate mutase action. For determination of the E-/Z-geometric isomer content of the ex-H₂HPP from BacB action, we were able to separate by HPLC and assign by NMR spectroscopy the two geometric isomers of the Δ³, Δ⁵-exocyclic diene product.

Figure 3.

Strategy of positional trace of ¹³C-label from 5-[¹³C]-glucose to ex-H₂HPP to reveal which en-H₂HPP (and thereby ex-H₂HPP) enantiomer(s) is being formed via BacA action on prephenate.

MATERIALS AND METHODS

Materials and Instrumentation

Prephenic acid barium salt and unlabeled chorismic acid were purchased from Sigma-Aldrich. DNA primers were purchased from Integrated DNA Technologies. NMR solvent (D₂O) and [¹³C]-glucoses were purchased from Cambridge Isotope Laboratories. B. subtilis sp. 168 genomic DNA was purchased from ATCC. Herculase II DNA polymerase was purchased from Agilent. Restriction enzymes and T4 DNA ligase were purchased from New England Biolabs. Competent cells were purchased from Invitrogen. Vector (pET-28a) was purchased from Novagen. DNA purification was performed with kits purchased from Qiagen. DNA sequencing was performed by Genewiz. 5 mL His-Trap columns were purchased from GE Healthcare. Any kD^™ SDS-PAGE gels were purchased from Bio-Rad. Protein was dialyzed using 10,000 MWCO SnakeSkin Pleated Dialysis tubing from Thermo Scientific and protein was concentrated using Amicon Ultra 10,000 MWCO centrifugal filters from Millipore. Restriction-grade thrombin was purchased from EMD Biosciences. Hypercarb 5 μm columns were purchased from Thermo Scientific.

¹H and 2D-NMR spectra were collected on a Varian VNMRS 600 MHz spectrometer equipped with a triple-resonance probe. ¹³C NMR spectra were recorded on a Varian MR 400 MHz spectrometer (100.497 MHz for ¹³C) equipped with a OneNMR probe. NMR data were analyzed with ACD/Labs software. High-resolution LC/MS data were collected on an Agilent Technologies 6520 Accurate-Mass Q-TOF LC/MS and analyzed with its accompanying software. HPLC was performed on a Beckman Coulter System Gold instrument. Protein purification was performed on an Amersham Pharmacia Biotech AKTA FPLC.

Cloning, expression and purification of chorismate mutase and BacAB

The chorismate mutase (CM) gene (aroH) (21) was amplified from B. subtilis sp. 168 genomic DNA via PCR using primers encoded with NdeI and BamHI restriction sites (5′-GGCAGCCATATGATGATTCGCGGAATTCGCGGAG-3′; 5′-GAATTCGGATCCTTACAATTCAGTATTTTTTGTCAATGATAAATCGGGCCTC- 3′, respectively). The amplified gene was ligated into vector pET-28a and transformed into chemically-competent TOP10 E. coli cells (Invitrogen). Proper gene insertion was confirmed by DNA sequencing of the purified plasmid DNA. The sequence-confirmed plasmid was then transformed into chemically-competent BL21(DE3) E. coli cells for protein expression. The cloning of BacA and BacB and their transformation into BL21(DE3) E. coli cells has been previously described (14). For CM, BacA, and BacB expression, transformed cells were grown at 37 °C in Luria broth media supplemented with 50 μg/mL kanamycin until the OD_600nm was ~ 0.2. The temperature was then reduced to 15 °C for 30 min and protein expression was induced via addition of 1 mM IPTG. Protein was allowed to express overnight (~ 16 h) at 15 °C before cells were harvested by centrifugation (4000g for 20 min). Cells/protein were then kept at 4 °C or on ice for all remaining purification steps. Pelleted cells were resuspended in cold lysis buffer (50 mM potassium phosphate pH 8, 500 mM NaCl, 5 mM β mercaptoethanol, 15 mM imidazole, 5% glycerol) and lysed by two passes of exposure to 5000–15000 psi in an Avestin EmulsiFlex-C5 homogenizer. The cell lysate was clarified via centrifugation (50,000g for 35 min) and the supernatant was filtered through a 0.45 μm PES syringe filter before being loaded onto a 5 mL His-Trap HP column at 1.5 mL/min via an AKTA FPLC. Bound protein was eluted with a linearly increasing gradient of lysis buffer containing 500 mM imidazole. Fractions were analyzed via SDS-PAGE with visualization by coomassie blue staining. Proteins were then dialyzed into S-75 buffer (50 mM potassium phosphate pH 8.0, 150 mM NaCl, 1 mM dithiothreitol, and 5% glycerol) using a dialysis membrane. CM aliquots were flash frozen at this point and stored @ −80 °C. CM concentration was determined using an extinction coefficient of 8480 M⁻¹cm⁻¹ at 280 nm. BacA and BacB were subjected to further purification by gel filtration on a Sephadex 75 26/60 HiLoad column. Fractions were analyzed as described above, and BacA aliquots were flash frozen at this point and stored at −80°C. BacA concentration was determined using an extinction coefficient of 17420 M⁻¹cm⁻¹ at 280 nm.

The N-terminal hexa-histidine tag of BacB was then removed via thrombin treatment at 16 °C. (The His₆-tag was removed to prevent any interactions between the hexa-histidine tag and the bound divalent metals of BacB (19).) Thrombin cleavage was monitored via SDS-PAGE with coomassie blue visualization. After cleavage was complete (~ 16 h), BacB was once again subjected to gel filtration chromatography. The fractions were analyzed by SDS-PAGE and aliquots of BacB were flash frozen and stored at −80 °C. BacB concentration was determined using an extinction coefficient of 24410 M⁻¹ cm⁻¹ at 280 nm.

Assignment of ¹H and ¹³C spectra of unlabeled chorismate and prephenate

Separately, 3 mg of chorismic acid and potassium prephenate (converted from barium prephenate as previously described (14)) were dissolved in 500 μL of 10 mM potassium phosphate buffer pH 8.0 and purified by loading onto a 100 × 10 mm Hypercarb HPLC column equilibrated in 10 mM potassium phosphate buffer pH 8.0. Compounds were eluted with a linearly increasing gradient of acetonitrile. Chorismate elution was monitored via UV absorbance at 275 nm. Prephenate elution was monitored by LC/MS run in positive ion detection mode with water + 0.1% formic acid as the mobile phase and acetonitrile + 0.1% formic acid as the eluent. HPLC fractions containing the desired compound (total volume of 3 mL) were frozen and lyophilized to dryness. Dried compound and associated potassium phosphate salt were dissolved in 300 μL of D₂O and placed in a 5 mm D₂O matched Shigemi tube for NMR analysis. Spectral assignments were made via analyses of ¹H, ¹³C, ¹H-¹H-gCOSY, ¹H-¹³C-gHSQC, and ¹H-¹³C-gHMBC spectra. Proton spectra were referenced to residual H₂O (4.79 ppm) while ¹³C spectra were referenced to the methyl carbon of acetonitrile at (1.47 ppm) (22). (Acetonitrile was spiked into the NMR sample at 0.3% v/v after all necessary spectra were acquired and an additional ¹³C experiment was performed to obtain the reference so as not to contaminate the original spectra.) Unless otherwise stated, water suppression (via presaturation) was utilized in all proton spectra collected.

Production of [¹³C]-chorismate from [¹³C]-glucose

Aerobacter aerogenes 62-1 (also known as Klebsiella pneumoniae 62-1) was obtained from our lab’s strain inventory from ~ 1990 (23) (also available from ATCC #25306). This strain (which lacks chorismate mutase activity) and the associated method for the over-production of chorismate were developed by Gibson and colleagues (11, 24). Medium A (for the growth of A. aerogenes 62-1) and medium B (for the production of chorismate) were made exactly as previously described, except glucose was not added to medium B (24). 500 mL of medium A was inoculated with a single colony of A. aerogenes 62-1 in a 2800 mL wide-mouth baffled flask. The culture was shaken at 200 rpm at 30 °C until the OD_600nm measured 1.0. The cells were pelleted at 3,000g at 4 °C for 15 min. The supernatant was discarded and the cell pellet was gently washed with 250 mL of medium B lacking glucose. The cells were again pelleted via centrifugation as described above. The supernatant was once again discarded and the cell pellet was resuspended in 250 mL of fresh medium B along with 2 g of either 1-[¹³C]-glucose or 5-[¹³C]-glucose (99% enrichment) that had been dissolved in 10 mL of medium B and filtered through a 0.22 μm PES syringe filter. The culture was transferred to a 2800 mL wide-mouth baffled flask and incubated at 30 °C for 15 h while shaking at 200 rpm. The cells were pelleted by centrifugation and discarded.

The supernatant containing the excreted chorismate was filtered through a 0.22 μm PES filter and loaded in 50 mL aliquots onto a 100 × 21.2 mm Hypercarb HPLC column equilibrated in 10 mM potassium phosphate buffer pH 8.0. The bound compound was eluted via a linearly increasing gradient of acetonitrile. UV absorption at 275 nm was monitored to detect chorismate elution. Fractions containing pure chorismate were initially indentified by a UV-absorbance wavelength scan (λ_max = 275 nm, data not shown) and then verified by ¹H NMR spectroscopy (Supporting Information Figures S2B and S4A). Fractions were pooled, frozen, lyophilized to dryness, and stored at −80 °C. Chorismate concentrations were determined by UV absorbance (ε_275nm = 2630 M⁻¹ cm⁻¹) (10). The ¹³C enrichment of chorismate was determined by integration of ¹H NMR spectra (satellite peaks of ¹³C-linked protons) and inverse-gated decoupled ¹³C NMR spectra collected with an interscan delay time of 150 s to allow full relaxation of the quaternary carbons (25). [¹³C]-Chorismate samples were prepped for NMR exactly as described above for unlabeled chorismate.

Conversion of [¹³C]-chorismate to [¹³C]-prephenate and NMR assignment

2.1 mg of [¹³C]-chorismate was incubated with 145 μg of chorismate mutase in 3 mL of 50 mM potassium phosphate buffer pH 8.0 for 15 h at 21 °C. Reaction progress was monitored by the disappearance of the UV absorbance of chorismate. The reaction was quenched via addition of acetonitrile to 30% v/v and vortexing. The quenched reaction was frozen and lyophilized to dryness. Lyophilized components were resuspended in 500 μL of 10 mM potassium phosphate buffer pH 8.0 and precipitated protein was pelleted via centrifugation at 16,000g for 10 min. The supernatant containing the [¹³C]-prephenate was removed and purified as described above for unlabelled prephenate. 3 mL of prephenate containing fractions were frozen and lyophilized to dryness. The lyophilized prephenate and salt were resuspended in 300 μL of D₂O and placed in a 5 mm D₂O matched Shigemi tube for NMR analysis. ¹H and ¹³C NMR data were referenced as described above.

Conversion of [¹³C]-prephenate to [¹³C]-ex-H₂HPP and NMR assignment

To produce [¹³C]-ex-H₂HPP via nonenzymatic isomerization of [¹³C]-en-H₂HPP, 2 mg of [¹³C]-prephenate (purified as described above) was incubated with 10 μM BacA in 3 mL of 50 mM potassium phosphate buffer pH 8.0 for 24 h at 21 °C. The conversion of prephenate (no UV absorbance) to en-H₂HPP (λ_max = 258 nm) and then to ex-H₂HPP (λ_max = 295 nm) was monitored by UV absorbance (14). The completed reaction was quenched, purified, and NMR data acquired exactly as described above for prephenate, except UV absorbance at 295 nm (rather than LC/MS) was used to identify ex-H₂HPP elution in the purification.

Production and purification of [¹³C]-ex-H₂HPP isomers #1 and #2 generated from enzymatic BacB isomerization of [¹³C]-en-H₂HPP was performed as described above for the nonenzymatic isomerization reaction, except 5 mg of [¹³C]-prephenate was used in a reaction volume of 5 mL and 10 μM BacB was added to the reaction. NMR data was acquired exactly as described above for prephenate.

Production, purification, and NMR assignments of unlabeled ex-H₂HPP isomers

The production and purification procedure described above for [¹³C]-ex-H₂HPP isomers was repeated with unlabeled prephenate to obtain samples of unlabeled ex-H₂HPP isomers #1 and #2 for NMR data collection. Full NMR assignments were made using the same methods as described above for unlabeled chorismate and prephenate. Nuclear Overhauser effect (NOE) data reporting on the spatial distance between protons were obtained by acquiring a ¹H-¹H-NOESY with a mixing time of 500 ms. Mass spectra of the isomers were obtained via LC/MS in negative ion detection mode using H₂O + 0.1% ammonium hydroxide as the mobile phase and acetonitrile + 0.1% ammonium hydroxide as the eluent.

Reactions comparing the production of ex-H₂HPP isomers #1 and #2 between BacA alone (with nonenzymatic diene isomerization) and BacAB in tandem were set up to contain 500 μM prephenate and 15 μM of each enzyme (incubated for 24 h at 21 °C). Reactions were quenched by addition of acetonitrile to 30% v/v, then frozen and lyophilized to dryness. Dried mixtures were resuspended in 10 mM potassium phosphate buffer pH 8.0 and loaded onto a 100 × 2.1 mm analytical Hypercarb column equilibrated in 10 mM potassium phosphate buffer pH 8.0. Compounds were eluted with a linearly increasing gradient of acetonitrile and elution monitored via UV absorbance at 295 nm. The 295 nm HPLC traces were integrated using the Karat 32 HPLC software from Beckman Coulter.

Kinetics of ex-H₂HPP isomer formation by BacAB

Kinetics of ex-H₂HPP isomer production from tandem BacAB incubation with prephenate were measured from a reaction containing 1 mM prephenate and 5 μM BacAB. The reaction was incubated at 21 °C and reaction time points were taken by quenching 25 μL reaction aliquots with 50 μL of acetonitrile and vortexing. The quenched reactions were stored at −80 °C until all time points were quenched. Quenched aliquots were dried in vacuo and resuspended in 175 μL of 10 mM potassium phosphate buffer pH 8.0. The resuspended aliquots were analyzed by HPLC identically to the BacA vs. BacAB reactions described in the previous paragraph.

Equilibration of 3E-7R- and 3Z-7R-ex-H₂HPP with BacB (and BacA)

Production and purification of unlabeled ex-H₂HPP isomers was accomplished as described above for the production of [¹³C]-ex-H₂HPP-isomers via tandem BacAB incubation, except 5 mg of unlabeled potassium prephenate was used as the starting material. The concentration of each ex-H₂HPP isomer was determined by UV absorbance (ε_295nm = 15300 M⁻¹ cm⁻¹ (14)). The reaction mixture for the equilibration time course consisted of 3 μM BacB (or BacA) in 50 mM potassium phosphate buffer pH 8.0. The reaction was initiated via addition of 400 μM ex-H₂HPP (either purified isomer #1 or #2). A control reaction was also setup containing no BacB to monitor nonenzymatic equilibration. Time points were taken by quenching 25 μL aliquots of the reaction with 50 μL of acetonitrile and vortexing. Quenched time points were stored at −80 °C until aliquots for all time points were collected. Samples were then dried in vacuo and resuspended in 75 μL of 10 mM potassium phosphate buffer pH 8.0. Resuspended samples were analyzed by HPLC exactly as described above for the BacA vs. BacAB comparison reactions. The fraction of isomer #2 was calculated by dividing the integral value of the peak of isomer #2 by the sum of the integral values of the peaks of isomers #1 and #2 for each run. The data were fit to a first-order exponential decay equation using GraphPad Prism.

RESULTS AND DISCUSSION

Generation of [¹³C]-Chorismates from [¹³C]-Glucose isotopomers

Our approach to determine which prochiral double bond in prephenate is isomerized during non-aromatizing decarboxylation by BacA was to use carbon magnetic resonance to distinguish between sp³ and sp² carbon centers in the BacA product Δ⁴, Δ⁸-en-H₂HPP, since large shifts in the resonance positions were formerly seen in the ¹³C NMR spectrum (for example between C₆ (142.62 ppm) and C₈ (30.5 ppm) (14)). In turn, coupling of the BacA product to BacB action to generate the more stable Δ³, Δ⁵-ex-H₂HPP product would also allow the same carbon NMR approach. The acceleration of conversion of BacA–generated Δ⁴, Δ⁸-endocyclic-H₂HPP to the thermodynamically favored Δ³, Δ⁵-exocyclic-H₂HPP by BacB also prevents nonenzymatic breakdown of the endo isomer to phenylpyruvate and so provides quantitative flux to Δ³, Δ⁵-ex-H₂HPP.

For example, if one had a sample of 6-[¹³C]-prephenate as substrate for BacA, (and then tandem processing by BacB) isomerization of the pro-R double bond would yield an sp³-labeled carbon in Δ³, Δ⁵-ex-H₂HPP while isomerization at the pro-S double bond (Figures 1B and 3) would instead yield the ¹³C-enrichment at an sp² center, readily distinguishable by NMR spectroscopy. In case the BacA enzyme is not completely enantioselective, the ratio of [¹³C] enrichment at the sp³/sp² carbons in the BacB product would provide the degree of selectivity in the double bond isomerization. Because the double bonds in the 1,4-diene ring system of prephenate are prochiral, a chemical synthesis to introduce [¹³C] at one of the double bonds (but not the other) with absolute stereochemical control seemed daunting.

In contrast, the immediate metabolic precursor chorismate, convertible to prephenate by the 3,3-electrocyclic rearrangement catalyzed by chorismate mutase (26), presented a more appealing target. While chemical synthesis of chorismate labeled at particular carbons with ¹³C could be feasible (27), we turned instead to a microbial biosynthetic route.

The biosynthetic path of glucose carbons into chorismate has been known for decades starting with the pioneering studies of Sprinson and colleagues (28, 29); randomization at the level of the C₃ triose phosphates during glycolysis gives doubling of label in shikimate and then tripling of label via the enolpyruvyl group found in chorismate even from singly-labeled forms of glucose (25). As a control to establish the methodology in our group, we fermented the readily available and relatively inexpensive 1-[¹³C]-glucose with the classical Aerobacter aerogenes 62-1 blocked mutant strain (24) that accumulates chorismate in the medium. We saw enrichment at carbons 2, 6, and 9 in the NMR spectrum of chorismate as previously reported (25) (Supporting Information Figure S2), validating the approach. However, this labeling pattern would not solve the BacA regiochemistry question as it would go on to label C₅ and C_5′ of prephenate equally (and then carbons 3, 5, and 9 of Δ³Δ⁵-ex-H₂HPP) (see Supporting Information Figure S4 for chorismate and prephenate numbering).

Instead, we reasoned that 5-[¹³C]-glucose (a limiting reagent at the commercial price of $1950/g) would be a useful starting material to get to an appropriately labeled sample of prephenate by carrying out such fermentations. From 2 g of 5-[¹³C]-glucose in a 250 mL fermentation, we obtained 56 mg of chorismate after purification on a preparative HPLC column. As anticipated, the chorismate sample had three ¹³C atom enrichments of 57%, 70%, and 56% in the chorismate carbon NMR spectrum (Figures 3 and 4A), at carbons 1 (135.0 ppm), 5 (130.5 ppm), and 8 (153.9 ppm), respectively (see Figure S5A and Table S1 of Supporting Information for full chorismate assignments). The carbon spectra were acquired with inverse-gated decoupling and a sufficiently long interscan delay time (150 s with no decoupler) to allow full carbon relaxation to achieve the full intensities of each of the carbon lines (25). ¹³C enrichment was calculated from a combination of the ¹³C-linked proton satellite peaks of C₅ in the ¹H spectrum (Supporting Information Figure S4A) and signals from the inverse-gated decoupled carbon spectrum (Figure 4A).

Figure 4.

Inverse-gated decoupled carbon spectra of (A) 1, 5, 8-[¹³C]-chorismate, (B) 2, 4, 6-[¹³C]-prephenate, (C) 2, 4, 8-[¹³C]-ex-H₂HPP isomer #1, and (D) 2, 4, 8-[¹³C]-ex-H₂HPP isomer #2. 1, 5, 8-[¹³C]-chorismate was obtained from fermentation of 5-[¹³C]-glucose with A. aerogenes 62-1; 2, 4, 6-[¹³C]-prephenate from CM action on 1, 5, 8-[¹³C]-chorismate; and 2, 4, 8-[¹³C]-ex-H₂HPP isomers #1 and #2 from tandem BacAB action on 2, 4, 6-[¹³C]-prephenate.

Conversion of 1,5,8-[¹³C]-Chorismate to 2,4,6-[¹³C]-Prephenate

To convert this labeled [¹³C]-chorismate sample to prephenate we turned to chorismate mutase, which we cloned and overproduced from the B. subtilis bacilysin producer. Trial studies with unlabeled chorismate established that quantitative rearrangement of the chorismate scaffold to the prephenate scaffold occurred on a scale useful for carbon NMR and the prephenate could be purified, isolated and characterized (data not shown). Subsequent incubation of 2.1 mg of the 1, 5, 8 [¹³C]-chorismate with 145 μg of pure chorismate mutase gave prephenate with the indicated carbon NMR resonances (Figure 4B). During the course of the 3,3-rearrangement, C₁ of chorismate becomes C₄ of prephenate, C₅ becomes C₆, and C₈ becomes the C₂ ketone. Figure 4B validates that the prephenate product has ¹³C enrichment at carbons 2 (204.4 ppm), 6 (128.2 ppm), and 4 (49.2 ppm) (see Figure S5B and Table S2 of Supporting Information for full prephenate NMR assignments).

Because of the symmetry of the cyclohexadiene ring of prephenate, carbons 6 and 6′ cannot be resolved in the NMR. Thus, without knowing the origin of this triply labeled ¹³C-prephenate sample, one could not tell from the carbon NMR spectrum that only one of the two prochiral carbons, C₆ and C_6′, contained the ¹³C enrichment. That differentiation can only be revealed by subsequent reactions that can distinguish the two prochiral olefins, (e.g. the nonaromatizing decarboxylase/isomerase BacA).

Processing of 2,4,6-[¹³C]-Prephenate by BacA

Assignment of Product Stereochemistry at C₇ from action of BacA. Tandem incubation of the triply labeled [¹³C]-prephenate with purified B. subtilis enzyme BacA followed by nonenzymatic isomerization allowed HPLC purification of Δ³, Δ⁵-ex-H₂HPP. The carbon NMR spectrum (Supporting Information Figure S3A) of this compound, which turns out to be the 3E-olefinic isomer as detailed in the following paragraphs, shows three peaks, at 30.9 (C₈), 156.8 (C₄) and 196.1 (C₂) ppm vs. a CH₃CN standard. By comparison with our assignment of the carbon NMR spectrum of unlabeled material (Figure S5D and Table S4 of Supporting Information), these resonances correspond to carbons 2, 4, and 8 of Δ³, Δ⁵-ex-H₂HPP. The diagnostic resonance is the ¹³C enrichment at the sp³ carbon 8 of ex-H₂HPP. There is no detectable enrichment at the sp² C₆.

Therefore, it is clear that BacA is an enantioselective decarboxylation and isomerization catalyst. As decarboxylation proceeds, the electrons released in the C₄-COO bond cleavage move into the ring, creating the new double bond, and the electrons in the original pro-R double bond are used to pick up a proton (Path 1 of Figure 1B) at C₆ of prephenate. From the perspective of C₇ this is the R-isomer. No 7S-OH product appears to be generated.

Detection and NMR assignment of Geometric Isomers of Δ³, Δ⁵-H₂HPP from Action of BacB

When BacB was added to the BacA incubations with the 2,4,6-[¹³C]-prephenate to accelerate the allylic isomerization of the en-H₂HPP to the more stable ex-H₂HPP product, in contrast to the single product peak detected with BacA and nonenzymatic isomerization, the BacAB tandem incubation gave rise to two peaks separable by HPLC as shown in Figure 5A. Because of the substantial polarity of the H₂HPP isomers, a Hypercarb column proved to be the support on which isomer separation could be achieved. These are labeled as isomer #1 and isomer #2 because they have the same molecular mass (observed m/z 181.0503; calculated m/z 181.0506) by LC/MS analysis. Also as seen in Figure 4C and 4D they have comparable carbon NMR spectra, indicating the pro-R double bond has moved (BacA action as expected) and both isomers are 7R-alcohol forms of ex-H₂HPP.

Figure 5.

(A) HPLC separation (295 nm UV trace) of ex-H₂HPP isomers #1 and #2 produced via action of BacA alone (5%/95% of isomers #1/#2) or tandem BacAB action (26%/74% of isomers #1/#2) on prephenate after 24 h incubation at 21 °C. (B) ¹H NMR of unlabeled ex-H₂HPP isomer #1. (C) ¹H NMR of unlabeled ex-H₂HPP isomer #2. ex-H₂HPP isomers in (B) and (C) were produced by the tandem action of BacAB on prephenate.

It seemed likely that isomer #1 and #2 might be the 3E/3Z geometric isomers around the exocyclic double bond in ex-H₂HPP. The first indication confirming this was the proton NMR spectra (Figures 5B and C). Isomer #1 and isomer #2 have equivalent resonances for alcoholic H7 and the methylene hydrogens 8a and 8b. The dispersion for protons 3, 5, and 6 are radically different and the splitting pattern for the methylene hydrogens 9a and 9b are also distinct, consistent with different geometries at the 3,4-olefin. NOESY spectra are shown in Figure 6A for isomer #1 and Figure 6B for isomer #2. Most diagnostic is the NOE cross peak between the olefinic H3 and one of the H9 resonances (indistinguishable due to identical chemical shifts) for isomer #1 that is absent in isomer #2. Thus, isomer #1 is assigned as the 3Z-7R diastereomer and isomer #2 the 3E-7R diastereomer of ex-H₂HPP. (Complete NMR spectral data for isomer #1 is presented in Figure S5C and Table S3 of Supporting Information.)

Figure 6.

¹H-¹H-NOESY spectra of ex-H₂HPP isomer #1 (A) and isomer #2 (B) produced from the tandem action of BacAB on prephenate. (A) NOE interaction between protons 3 and 9 of isomer #1 (cross-peak is boxed) identify the Δ³-olefin as possessing “Z” geometry. (B) NOE interaction between protons 3 and 5 could not be distinguished due to overlapping resonances. However, a NOE cross-peak between protons 3 and 9 is absent indicating “E” geometry for the Δ³-olefin.

Although both 3Z- and 3E-ex-H₂HPP isomers are present after long incubations containing BacA and BacB, kinetic analysis (Figure 7A) indicates that the 3E isomer is formed first (at about a 50/1 E/Z ratio) and the 3Z-geometric isomer grows in slowly. The initial product was confirmed as ex-H₂HPP (λ_max = 295 nm) and not en-H₂HPP (λ_max = 258 nm) by UV absorption (14) (data not shown). The 3Z isomer then grows in over a period of ~ 17 hours under the particular experimental conditions. Comparing the 17 and 40 hour time points suggest an equilibrium has been approached with a ratio of ~ 3/1 of the E/Z- isomers of ex-H₂HPP.

Figure 7.

(A) HPLC traces (295 nm) showing the kinetics of ex-H₂HPP isomer #1 and #2 formation from the tandem action of 5 μM BacAB on 1 mM prephenate at 21 °C. (B) HPLC traces (295 nm) showing the equilibration of 400 μM ex-H₂HPP isomer #1 with 3 μM BacB at 21 °C. (C) HPLC traces (295 nm) showing the equilibration of 400 μM ex-H₂HPP isomer #2 with 3 μM BacB at 21 °C. (D) Kinetics of equilibration of ex-H₂HPP isomers #1 and #2 with BacB. Data were fit to a first-order exponential decay equation yielding t_1/2 values of 3.6 h and 10.4 h for the equilibration of isomers #1 and #2, respectively.

BacB-mediated equilibration of 3E-7R- and 3Z-7R-ex-H₂HPP

From the trace in Figure 7A, it appeared that BacB could in fact catalyze equilibration between the initially formed 3E and the late-appearing 3Z isomers. As shown in Figure 7B starting from 85% pure Z-isomer, BacB alone catalyzed the conversion of Z- to E-isomer until equilibrium was reached. (4% nonenzymatic conversion of the Z-isomer was observed as background during the 24-hour interval.) Going in the opposite direction, when BacB was added to 85% pure E-isomer it was converted to the Z-isomer until equilibrium was reached (Figure 7C). (0.1% nonenzymatic conversion of the E-isomer was observed as background during the 24-hour interval). For the incubation of BacB with both the Z- (isomer 1) and E-isomer (isomer 2) of ex-H₂HPP, equilibrium was obtained at the same 3/1 ratio of E/Z geometric isomers (Figure 7D) as was observed for the reaction from prephenate with BacAB (Figure 7A). Pure BacA has no ability to interconvert the 3E and 3Z-ex-H₂HPP isomers (data not shown).

A possible route from the 3E-ex-H₂HPP initial product to the 3Z-olefin regioisomer is shown in Figure 8. BacB must be able to act on the 3E product isomer by abstraction of one of the acidic C₉ methylene hydrogens as a proton to yield the C₉ carbanion, which has the indicated dienolate as a resonance contributor. The dienolate can rotate around the C₃-C₄ single bond to give a mixture of E- and Z-dienolates. Protonation at C₉ would yield either the 3E- or 3Z-ex-H₂HPP conjugated products, depending on dienolate composition. At 3 μM [BacB] and 400 μM [ex-H₂HPP], the half-time for equilibration was found to be 3.6 hours in the Z- to E-direction. The 133/1 ratio of substrate to enzyme suggests about 0.3 equilibration events per minute under the given conditions.

Figure 8.

Tandem action of BacAB with prephenate initially yields only the E-isomer of ex-H₂HPP. BacB is then able to equilibrate between the E- and Z-isomers of ex-H₂HPP via a proposed dienolate intermediate generated by abstraction of a proton from C₉.

From BacA incubations in the absence of BacB and at initial time points with BacB the 3E-/3Z-isomer ratio is about 50/1, indicating a kinetically favored path to 3E. The question arises why BacB is retained in the biosynthetic pathway, accelerating the flux by 10³ to the 3E-ex-H₂HPP regioisomer. It may be that the subsequent equilibration of up to 25% of the 3Z-7R-ex-H₂HPP product by BacB suggests that one or more of the downstream enzymes in the bacilysin (or related biosynthetic pathways) pathway may care about the E/Z olefin geometry. That is the subject of future efforts now that the isomers have been detected, structures assigned, and kinetics of equilibration determined.

CONCLUSIONS

This study establishes that BacA is indeed an enantioselective pro-R olefin isomerase during prephenate non-aromatizing decarboxylation. It is likely that the selective isomerization of the pro-R olefin in prephenate by BacA reflects an active site orientation where the enzyme-substrate complex with bound prephenate brings a side chain conjugate acid (BH⁺) adjacent to C₆ of prephenate in the active site. Determination of the 7R-stereochemistry in Δ³, Δ⁵-ex-H₂HPP, and by extension its predecessor Δ⁴, Δ⁸-en-H₂HPP with no detectable 7S-isomer, indicates a high degree of olefin selectivity. This occurs as BacA concomitantly and surprisingly suppresses the aromatization fate that is characteristic of canonical prephenate decarboyxlases in biosynthetic pathways to phenylalanine.

BacB, the next enzyme in the bacilysin biosynthetic pathway, takes the Δ⁴, Δ⁸-endocyclic-H₂HPP and first generates the 3E-geometric isomer. BacB will then catalyze the formation of the 3Z- from the 3E-olefin and proceed to equilibrate between the 3E and 3Z isomers around the 3,4-exocyclic double bond. We anticipate this reflects the reversible ability to generate the dienolate intermediates in the BacB active site and reprotonate at C₉. BacB speeds up the 1,3-allylic isomerization some 10³ over the nonenzymatic rate, which otherwise yields the E-olefin at about a 50/1 kinetic ratio over the Z-isomer, far away from the 3/1 thermodynamic ratio.

These findings on BacA and BacB set the stage for evaluation of subsequent issues of stereochemistry and mechanism in these non-aromatizing pathways arising from decarboxylation of prephenate to anticapsin/bacilysin and other dihydroaromatic amino acids. As prephenate is processed by BacA and BacB, the C₅H=C₆H (pro-R) olefin gets saturated to a CH₂-CH₂ pair of carbon atoms. Protonation occurs first at C₈ of en-H₂HPP and then at C₉ as ex-H₂HPP is formed. Because both C₈ and C₉ are methylene groups in the BacB product, in tetrahydrotyrosine, and then in anticapsin, the stereochemistry is cryptic when incubations are conducted in H₂O. In D₂O it should next be possible to determine chirality and thus orientation of the active site BD⁺ groups relative to the plane of the cyclohexadiene ring of bound substrates in BacA and BacB, respectively, as constraints on mechanism and comparisons to how canonical prephenate aromatizing decarboxylases work.

The mix of E- and Z-geometries in the Δ³, Δ⁵-ex-H₂HPP generated by BacB is also cryptic in the overall pathway since that olefin gets reduced in the next step by delivery of a hydride ion from NADH to C₄. However, that E- vs. Z-olefin geometry could influence the stereochemistry at C₄ in the reductive step catalyzed by YwfH (Figure 1A) and evaluation of that outcome will also be a subject of future efforts. Currently, only natural products with tetrahydrotyrosine of the 4S-configuration have been assigned (20, 30). However, other natural products containing tetrahydrotyrosine with unassigned stereochemistry at C₄ are known (31, 32), and how stereochemistry is controlled at that center during biosynthesis is a mystery.

Ultimately stereochemical and mechanistic studies need to be coupled with BacA structural studies to give insights into how electrons released into the cyclohexadiene ring on decarboxylation of prephenate are diverted out of the canonical aromatization manifold and quenched instead by regioselective protonation at C₆ in this newly appreciated enzyme subfamily.

Abbreviations

B. subtilis

Bacillus subtilis

E. coli

Escherichia coli

A. aerogenes

Aerobacter aerogenes

H₂HPP

dihydro-4-hydroxyphenylpyruvate

H₄HPP

tetrahydro-4-hydroxyphenylpyruvate

H₄Tyr

tetrahydrotyrosine

IPTG

isopropyl-β-D-galactopyranoside

SDS-PAGE

sodium dodecyl sulfate-polyacrylamide gel electrophoresis

LC/MS

liquid chromatography/mass spectrometry

NMR

nuclear magnetic resonance

gCOSY

gradient homonuclear correlation spectroscopy

gHSQC

gradient heteronuclear single-quantum coherence

gHMBC

gradient heteronuclear multiple bond coherence

NOESY

nuclear Overhauser effect spectroscopy

FPLC

fast protein liquid chromatography

HPLC

high performance liquid chromatography

PCR

polymerase chain reaction

NADH

reduced nicotinamide adenine dinucleotide

CM

chorismate mutase

D₂O

deuterium oxide