Stereochemical outcome at four stereogenic centers during conversion of prephenate to tetrahydrotyrosine by BacABGF in the bacilysin pathway

Article abstract of DOI:10.1021/bi3006362

The first four enzymes of the bacilysin antibiotic pathway, BacABGF, convert prephenate to a tetrahydrotyrosine (H₄Tyr) diastereomer on the way to the anticapsin warhead of the dipeptide antibiotic. BacB takes the BacA product endocyclic-Δ⁴,Δ⁸-7R- dihydrohydroxyphenylpyruvate (en-H₂HPP) and generates a mixture of 3E- and 3Z-olefins of the exocyclic-Δ³,Δ⁵- dihydrohydroxyphenylpyruvate (ex-H₂HPP). The NADH-utilizing BacG then catalyzes a conjugate reduction, adding a pro-S hydride equivalent to C ₄ to yield tetrahydrohydroxyphenylpyruvate (H₄HPP), a transamination away (via BacF) from 2S-H₄Tyr. Incubations of the pathway enzymes in D₂O yield deuterium incorporation at C₈ from BacA and then C₉ from BacB action. By ¹H NMR analysis of samples of H₄Tyr, the stereochemistry at C₄, C₈, and C₉ can be assigned. BacG (followed by BacF) converts 3E-ex-H₂HPP to 2S,4R,7R-H₄Tyr. The 3Z isomer is instead reduced and transaminated to the opposite diastereomer at C₄, 2S,4S,7R-H₄Tyr. Given that bacilysin has the 2S,4S stereochemistry in its anticapsin moiety, it is likely that the 2S,4S-H₄Tyr is the diastereomer "on pathway". NMR determination of the stereochemistry of the CHD samples at C₈ and C₉ allows assignment of all stereogenic centers (except C₃) in this unusual tetrahydro-aromatic amino acid building block, giving insights into and constraints on the BacA, BacB, and BacG mechanisms.

Full text of DOI:10.1021/bi3006362

Article
pubs.acs.org/biochemistry
Stereochemical Outcome at Four Stereogenic Centers during
Conversion of Prephenate to Tetrahydrotyrosine by BacABGF in the
Bacilysin Pathway
Jared B. Parker and Christopher T. Walsh*
Department of Biological Chemistry and Molecular Pharmacology, Harvard Medical School, 240 Longwood Avenue, Boston,
Massachusetts 02115, United States
S
* Supporting Information
ABSTRACT: The first four enzymes of the bacilysin antibiotic pathway,
BacABGF, convert prephenate to a tetrahydrotyrosine (H4Tyr) diastereomer
on the way to the anticapsin warhead of the dipeptide antibiotic. BacB takes
the BacA product endocyclic-Δ4,Δ8-7R-dihydrohydroxyphenylpyruvate (en-
H2HPP) and generates a mixture of 3E- and 3Z-olefins of the exocyclic-
Δ3,Δ5-dihydrohydroxyphenylpyruvate (ex-H2HPP). The NADH-utilizing
BacG then catalyzes a conjugate reduction, adding a pro-S hydride
equivalent to C4 to yield tetrahydrohydroxyphenylpyruvate (H4HPP), a
transamination away (via BacF) from 2S-H4Tyr. Incubations of the pathway
enzymes in D2O yield deuterium incorporation at C8 from BacA and then C9
1
from BacB action. By H NMR analysis of samples of H4Tyr, the stereochemistry at C4, C8, and C9 can be assigned. BacG
(followed by BacF) converts 3E-ex-H2HPP to 2S,4R,7R-H4Tyr. The 3Z isomer is instead reduced and transaminated to the
opposite diastereomer at C4, 2S,4S,7R-H4Tyr. Given that bacilysin has the 2S,4S stereochemistry in its anticapsin moiety, it is
likely that the 2S,4S-H4Tyr is the diastereomer “on pathway”. NMR determination of the stereochemistry of the CHD samples at
C8 and C9 allows assignment of all stereogenic centers (except C3) in this unusual tetrahydro-aromatic amino acid building block,
giving insights into and constraints on the BacA, BacB, and BacG mechanisms.
he dipeptide antibiotic bacilysin, produced by strains of
Bacillus subtilis, consists of L-alanine in amide linkage with
and into the anticapsin pathway, yielding tetrahydrotyrosine
(H4Tyr) as an intermediate.7
T
the nonproteinogenic amino acid anticapsin.1 Anticapsin is 3-
epoxycyclohexanonyl-alanine and bears the reactive epoxyketone
warhead that covalently captures the active site cysteine thiolate
in the glutaminase domain of the enzyme glucosamine-6-
phosphate synthase.2 Blockade of this enzyme shuts down
production of GlcNAc units required for both bacterial and
fungal cell wall assembly.3,4 The producing B. subtilis ligates
anticapsin to L-Ala to protect itself and then exports the resultant
bacilysin as a prodrug. Uptake by a dipeptide permease
transporter of a neighboring microbial cell followed by
dipeptidase action liberates anticapsin in a neighboring victim
cell (Figure 1).
We have previously determined that BacA decarboxylates
prephenate without aromatization7 and in the process specifically
isomerizes the pro-R double bond in the 1,4-diene system to
yield endocyclic-Δ4,Δ8-7R-dihydrohydroxyphenylpyruvate (en-
H2HPP).8 The next enzyme, BacB, conducts an allylic
isomerization to move the double bond into conjugation at the
exocyclic Δ3-position (ex-H2HPP). In so doing, it generates a 3/
1 mixture of the E- and Z-geometric isomers at the Δ3-olefin
upon equilibration. The third enzyme, BacG, uses NADH to
reduce that conjugated exocyclic double bond and yield an
H4HPP product.7 The fourth enzyme, BacF, is a transaminase
and uses L-Phe as an amino donor to reductively aminate the 2-
keto group of H4HPP and yields an H4Tyr diastereomer (Figure
2).
The H4Tyr is presumably two steps away, epoxidation of the
olefin and oxidation of the C7-OH group to the ketone, from
anticapsin. H4Tyr, an unusual noncanonical amino acid, has three
stereocenters: C2, C4, and C7. As shown below, BacF action, with
L-Phe as the cosubstrate, sets the S-stereochemistry at C2, and we
have previously established8 that BacA sets R-stereochemistry at
C7, which is carried through to H4Tyr.
The biosynthetic pathway to anticapsin, and then bacilysin, has
been studied genetically to reveal a cluster of five bacA−E
genes,5,6 and more recently at the enzyme level to decipher the
molecular logic for construction of the epoxycyclohexanone
moiety of this unusual amino acid.7 The biochemical studies have
shown that ywf GH, the latter transcribed on the opposite DNA
strand from bacA−E, encode an NADH-dependent reductase
(ywf H) and an L-specific transaminase (ywf G) that carry out
steps 3 and 4 of the pathway. Therefore, we rename the genes
bacF (ywf G) and bacG (ywfH) and use that nomenclature
throughout this work (Figure 2). In sum, the set of four enzymes
BacABGF, in that order, act in tandem to divert some of the
cellular flux of prephenate away from Phe and Tyr biosynthesis
Received: May 16, 2012
Revised: June 21, 2012
Published: July 5, 2012
© 2012 American Chemical Society
5622
dx.doi.org/10.1021/bi3006362 | Biochemistry 2012, 51, 5622−5632

Biochemistry
Article
Figure 1. Ligation of L-alanine and anticapsin yields the dipeptide antibiotic bacilysin. The host cell further self-protects from anticapsin and bacilysin
toxicity by transporting it into the extracellular environment, where it can be imported by neighboring cells. Peptide bond cleavage of bacilysin releases
the anticapsin warhead, which blocks cell wall biosynthesis via inhibition of glucosamine-6-phosphate synthase.
Figure 2. Proposed pathway for the biosynthesis of anticapsin from prephenate. YwfG and YwfH are hereby renamed BacF and BacG, respectively, as
noted in the inset.
At issue then is the stereochemistry at C4, presumably
established by the NADH-utilizing BacG. Also of note is the fact
that while the protio version of H4Tyr has the three previously
mentioned stereogenic centers, if the BacA, BacB, and BacG
incubations starting from prephenate are conducted in D2O
buffers, then one deuterium (D = 2H) is incorporated at C8 via
the action of BacA, one at C9 from the action of BacB, and then
another incorporated at C3 by BacG catalysis (the hydride from
NADH is added at C4).7 Thus, the C3, C8, and C9 methylene
groups (-CH2-) become chiral by virtue of deuterium
substitution (-CHD-). Therefore, in 2H3-, 2H8-, and 2H9-
containing samples, H4Tyr produced via the BacABGF pathway
possesses six chiral centers: C2−C4 and C7−C9. In this study, we
report delineation of stereochemistry at the four centers (C2, C4,
C8, and C9) that provide mechanistic insights and constraints
into the action of BacF, BacG, BacA, and BacB, respectively.
1H and two-dimensional NMR spectra were recorded at 25 °C
on a Varian VNMRS 600 MHz spectrometer equipped with a
triple-resonance probe. 13C one-dimensional 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 and integrated with ACD/Laboratories software. High-
resolution LC−MS data were collected on an Agilent
Technologies 6520 Accurate-Mass Q-TOF LC−MS system
and analyzed using its accompanying software. HPLC was
performed on a Beckman Coulter System Gold instrument, and
peaks were analyzed using the accompanying Karat 32 software.
UV−vis measurements were collected using a Cary 50 BIO UV−
vis spectrometer and analyzed with the accompanying software.
Cloning, Expression, and Purification of the Enzymes
BacABGF. The cloning of plasmids for production of BacABGF
has been previously described.7 The expression and purification
of BacA and BacB were also performed as previously described.8
The expression and purification of BacG and BacF were identical
to the procedure for BacA. Protein concentrations of BacABGF
were determined by UV−vis absorbance using the following
extinction coefficients (ε280) calculated from the protein primary
sequence using the ExPASy Bioinformatics Research Portal:
17420 M−1 cm−1 for BacA, 24410 M−1 cm−1 for BacB, 9970 M−1
cm−1 for BacG, and 41830 M−1 cm−1 for BacF.
MATERIALS AND METHODS
■
Materials and Instrumentation. Prephenic acid barium
salt, L- and D-tyrosine (Tyr), L- and D-phenylalanine (Phe), 4-
hydroxyphenylpyruvic acid (HPP), β-nicotinamide adenine
dinucleotide hydrate (NAD+), and β-nicotinamide adenine
dinucleotide reduced disodium salt hydrate (NADH) were
purchased from Sigma-Aldrich. NMR solvent (D2O), D-glucose
(1-D, 98%), and ethanol-d6 (D, 99%) were purchased from
Cambridge Isotope Laboratories. The enzymes glucose dehy-
drogenase from Pseudomonas sp., alcohol dehydrogenase from
Saccharomyces cerevisiae, and aldehyde dehydrogenase from S.
cerevisiae were also purchased from Sigma-Aldrich.
Generation and Purification of Protonated 3Z-7R- and
3E-7R-ex-H2HPP Isomers for Use as Substrates for NMR
and Kinetic Analyses of BacG Action. Conversion of barium
prephenate to potassium prephenate was performed as
previously described.7 Protonated 3E-ex-H2HPP was generated
5623
dx.doi.org/10.1021/bi3006362 | Biochemistry 2012, 51, 5622−5632

Biochemistry
Article
in reaction mixtures containing 12 mM potassium prephenate, 4
μM BacA, and 1 μM BacB in 2 mL of potassium phosphate buffer
(pH 8.0). The reaction mixtures were incubated at ∼20 °C for 16
h. The reactions were quenched via addition of acetonitrile to a
concentration of 30% (v/v), and the mixtures were frozen and
lyophilized in preparation for purification (discussed after
generation of 3Z-ex-H2HPP).
phosphate buffer (pH 8.0) in a volume of 250 μL. NADH (2
mM) was found to be >90% saturating for the reactions of both
isomers. Reactions were initiated via addition of BacG, and the
progress of the reaction was monitored by observing the
oxidation of NADH to NAD+ (UV absorbance at 370 nm) in a 1
mm cuvette. Reaction velocities (kobs) in units of absorbance per
minute were determined by calculating the slope of the initial,
linear region of the 370 nm time course of each reaction. The
change in NADH absorbance at 370 nm was converted to
concentration using the experimentally determined extinction
coefficient ε370 of 2690 M−1 cm−1 (referenced to NADH
concentration determined from the ε340 of 6220 M−1 cm−1 10).
Data were fit to the Michaelis−Menten equation (eq 1) using
GraphPad Prism
Production and purification of large amounts of 3Z-ex-H2HPP
from BacAB reactions were found to be very laborious because of
a maximum of only 25% of the total product being the 3Z
isomer.8 However, NMR and kinetic experiments with the
enzymes AerD and AerE (BacA and BacB homologues,
respectively, from Planktothrix agardhii9) showed that these
enzymes produce products identical to those of BacAB action
when acting on prephenate substrate, except the 3Z/3E isomer
ratio is 9/1 (J. B. Parker and C. T. Walsh, unpublished data) as
opposed to 1/3 for BacAB. Thus, fully protonated 3Z-ex-H2HPP
was generated from AerDE reaction mixtures containing 12 mM
potassium prephenate, 2 μM AerD, and 4 μM AerE in 4 mL of 50
mM potassium phosphate buffer (pH 8.0). {It is important to
note that formation of the 3Z isomer from either AerE or BacB
action was severely inhibited by high concentrations of salt
[approximately >250 mM (data not shown)].} The reaction
mixtures were incubated at ∼20 °C for 16 h; the reactions were
quenched via addition of acetonitrile to a final concentration of
30% (v/v), and the mixtures were frozen and lyophilized to
dryness.
For purification of the ex-H2HPP isomers, the lyophilized
reaction mixtures of BacAB (for 3E isomer formation) and
AerDE (for 3Z isomer formation) were resuspended (separately)
in 750 μL of 10 mM potassium phosphate buffer (pH 8.0) and
loaded onto a 100 mm × 21.2 mm Hypercarb column (Thermo
Scientific) equilibrated in the resuspension buffer. Bound
compounds were eluted (3Z and 3E isomers separated) by
exposure to a shallow linear gradient of acetonitrile. Compound
elution was monitored by UV absorbance at 295 nm. Fractions
containing the appropriate isomers (normally ∼30 mL) were
pooled, frozen, and lyophilized to dryness. Pure ex-H2HPP
isomers were then moderately desalted by being resuspended in
500 μL of 10 mM potassium phosphate buffer (pH 8.0) and
loaded onto a 100 mm × 10 mm Hypercarb column equilibrated
in the same buffer. Bound ex-H2HPP was eluted with a sharp
linear gradient of acetonitrile (0 to 90% over 25 min), and 4.5 mL
of fractions containing ex-H2HPP were pooled and lyophilized to
dryness. The lyophilized compound and residual potassium
phosphate buffer were resuspended in 300 μL of ddH2O, and the
kcat[S]
Km + [S]
kobs
=
(1)
where kcat is the maximal velocity, [S] is the substrate
concentration, and Km is the substrate concentration that yields
k
obs = 1/2kcat.
All reactions were performed in triplicate except the reaction
that contained 8 mM 3E-ex-H2HPP, which was performed in
duplicate because of a lack of material.
Generation and Purification of Protonated 4S- and 4R-
H4HPP and 4S- and 4R-H4Tyr Isomers for NMR Analysis.
4S-H4HPP was generated in a reaction mixture containing 2 mM
3Z-ex-H2HPP, 5 mM NADH, and 20 μM BacG incubated in 1
mL of 50 mM potassium phosphate buffer (pH 8.0). 4S-H4Tyr
was generated in a reaction mixture containing ∼2 mM 4S-
H4HPP (described in the previous sentence), 10 mM L-Phe, 0.1
mM PLP, and 20 μM BacF in 1 mL of 50 mM potassium
phosphate buffer (pH 8.0). 4R-H4HPP was generated in a
reaction mixture containing 2 mM 3E-ex-H2HPP, 5 mM NADH,
and 20 μM BacG in 1 mL of 50 mM potassium phosphate buffer
(pH 8.0). 4R-H4Tyr was generated in a reaction mixture
containing ∼2 mM 4R-H4HPP (described in the previous
sentence), 10 mM L-Phe, 0.1 mM PLP, and 20 μM BacF in 1 mL
of 50 mM potassium phosphate buffer (pH 8.0). All reaction
mixtures were incubated at ∼20 °C for 16 h.
All of the reactions described above were quenched via
addition of acetonitrile to a final concentration of 30% (v/v), and
the mixtures were frozen and lyophilized to dryness. Dried
reaction mixtures were resuspended in 500 μL of 10 mM
potassium phosphate buffer (pH 8.0) and purified as described
above for the ex-H2HPP isomers except using a 100 mm × 10
mm Hypercarb column. As H4HPP and H4Tyr do not have a
characteristic UV−vis absorbance, their elution was monitored
via LC−MS in negative detection mode (0.1% NH4OH as the
solvent additive) using a 50 mm × 2.1 mm Hypercarb column
with ddH2O as the mobile phase and acetonitrile as the eluent. A
mass of 183.0668 was observed for the H4HPP isomers (calcd,
183.0663) and 184.0986 (calcd, 184.0979) for the H4Tyr
isomers. Three milliliters of fractions containing the desired
product were frozen and lyophilized to dryness, resuspended in
300 μL of 99.99% D2O, and placed in a 5 mm D2O matched
Shigemi tube for NMR analysis.
concentration was determined by UV−vis absorbance [ε295
=
15300 M−1 cm−1 for the 3E isomer,7 and ε295 = 13700 M−1 cm−1
for the 3Z isomer (Figure S10 of the Supporting Information)].
Isomer purity was analyzed via analytical HPLC using a 100 mm
× 2.1 mm Hypercarb column equilibrated in 10 mM potassium
phosphate buffer (pH 8.0). Bound isomers were eluted using a
linear gradient of acetonitrile, and elution was monitored by
UV−vis absorbance at 295 nm (Figure S1 of the Supporting
Information). Isomers were stored at −80 °C to prevent
nonenzymatic isomer equilibration when not in use.
Kinetics of the Action of BacG on 3Z- and 3E-ex-H2HPP.
Kinetic reaction mixtures for the analysis of the action of BacG on
3Z-ex-H2HPP contained 1 μM BacG, 2 mM NADH, 0−1.1 mM
3Z-ex-H2HPP, and 50 mM potassium phosphate buffer (pH 8.0)
in a volume of 250 μL. Kinetic reaction mixtures for the analysis
of the action of BacG on 3E-ex-H2HPP contained 5 μM BacG, 2
mM NADH, 0−8 mM 3E-ex-H2HPP, and 50 mM potassium
NMR spectral assignments were made via analyses of 1H, 13C,
1H−1H gCOSY, 1H−13C gHSQC, 1H−13C gHMBC, and
1H−1H NOESY (500 ms mixing time) spectra. Proton spectra
were referenced to residual H2O (4.79 ppm), while 13C spectra
were referenced to the methyl carbon of acetonitrile (1.47
ppm).11 [Acetonitrile was spiked into the NMR sample at 0.3%
5624
dx.doi.org/10.1021/bi3006362 | Biochemistry 2012, 51, 5622−5632

Biochemistry
Article
(v/v) after all necessary spectra had been acquired, and an
additional 13C experiment was performed to obtain the reference
so as not to contaminate the original spectra with the acetonitrile
signal.] Unless otherwise stated, water suppression (via
presaturation) was utilized in all proton spectra that were
collected.
All deuterated H4HPP and H4Tyr production reaction
mixtures were incubated at ∼20 °C for 16 h and then reactions
quenched and mixtures purified; appropriate fractions were
identified by LC−MS as for the protonated H4HPP and H4Tyr
reactions described above. H4HPP masses obtained were
185.0816, 186.0876, 187.0938, and 188.0993 (calcd, 185.0788,
186.0851, 187.0914, and 188.0977). H4Tyr masses obtained
were 186.1159, 187.1215, 188.1279, and 189.1339 (calcd,
186.1105, 187.1167, 188.1230, 189.1293). These masses
correspond to compounds with incorporation of two, three,
four, and five deuteriums, respectively.
Generation and Purification of Deuterated 3Z-ex-
H2HPP and 3E-ex-H2HPP from the Action of BacA and
BacB and Deuterated 3E-ex-H2HPP from the Action of
Only BacA. The reaction mixtures for the production of
deuterated ex-H2HPP isomers from the action of BacAB
contained 5 mM prephenate, 10 μM BacA, and 40 μM BacB in
1 mL of 95% D2O containing 50 mM potassium phosphate (pD
8.0). The reaction mixtures were incubated at 37 °C for 24 h; the
reactions were quenched, and the mixtures were purified and
desalted as described above for the protonated ex-H2HPP
isomers. (BacAB reactions in D2O were very inefficient at
producing 3Z-ex-H2HPP. Therefore, it took many of these
reactions to obtain enough deuterated compound for NMR
analysis.) After desalting and lyophilization, the purified
deuterated 3Z- and 3E-ex-H2HPP isomers were resuspended in
D2O, and 1H, 1H−13C gHSQC, and 1H−1H NOESY data were
collected and referenced as described above for the protonated
H4HPP isomers.
Enzymatic Synthesis of 4R-2H- and 4S-2H-NADH. 4R-2H-
NADH (also known as A-side NADD) was synthesized following
a modified protocol of Viola et al.12 Ethanol-d6 (25 mM), 10 mM
NAD+, 300 units of alcohol dehydrogenase, and 32 units of
aldehyde dehydrogenase were mixed in 10 mL of 25 mM Taps
buffer (pH 9.0). The pH of the buffer was readjusted to 9.0 after
addition of the NAD+ but before addition of enzymes. The
reaction mixture was incubated at ∼20 °C, and reaction progress
was monitored by following the appearance of NADH (UV
absorbance at 340 nm). After the mixture had been incubated for
45 min, the reaction was quenched via protein removal using a
3500 MWCO centrifugal filtration device (Amicon).
4S-2H-NADH (also known as B-side NADD) was synthesized
according to a slightly modified protocol of Ottolina et al.13
Glucose-d1 (30 mM), 15 mM NAD+, and 100 units of glucose
dehydrogenase were mixed in 6 mL of 50 mM potassium
phosphate buffer (pH 7.5). The pH of the buffer was readjusted
to 7.5 before enzyme addition. Reaction progress was monitored,
and the reaction was quenched (after an 80 min reaction time)
exactly as for the A-side NADD synthesis reaction described
above.
Production of deuterated 3E-ex-H2HPP from BacA action
(without BacB) in tandem with nonenzymatic diene isomer-
ization was accomplished in a reaction mixture containing 10
mM prephenate and 10 μM BacA in 2 mL of 98% D2O
containing 50 mM potassium phosphate buffer (pD 8.0). The
reaction mixture was incubated at 37 °C for 144 h. The progress
of the extremely lethargic nonenzymatic isomerization in D2O
was monitored by UV absorbance (λmax of en-H2HPP = 258 nm;
λmax of ex-H2HPP = 295 nm7). The reaction was then quenched;
The processing of both NADD reactions was identical for the
remaining steps. The postfiltration flow-through containing the
NADD was frozen and lyophilized to dryness. The dried reaction
components were resuspended in 1 mL of H2O. NADD at this
stage was used directly in the BacG enzymatic reactions after the
pH had been corrected to 8.0. The NADD concentration was
determined by UV absorbance (ε340 = 6220 M−1 cm−1 10). To
prepare NADD samples for NMR analysis of the deuterium
content at C4, NADD was crudely purified by being loaded onto
a 250 mm × 10 mm Luna C18 5 μm HPLC column
(Phenomenex) equilibrated in H2O and 0.01% NH4OH.
Bound NADD was eluted using a linear gradient of acetonitrile
spiked with 0.01% NH4OH. Fractions containing NADD were
identified by the characteristic absorbance maxima at 260 and
340 nm. The observed 260 nm/340 nm ratio was 2.3 for A-side
and 2.5 for B-side NADD (previously reported as ∼2.2 − 2.3 for
pure NADH12). Fractions containing NADD were pooled and
lyophilized to dryness. Samples (2 mM) of both A-side and B-
side NADD were prepared in 99.99% D2O and placed in 5 mm
D2O matched Shigemi tubes for 1H NMR analysis (Figure S2 of
the Supporting Information). Pro-R and pro-S hydride peaks
were identified on the basis of previous assignments.13,14
Enzymatic Synthesis of H4HPP Isomers from BacG
Action with 4R-2H-NADH and 4S-2H-NADH as Cosub-
strates. Reaction mixtures for the generation of 4S- and 4R-
H4HPP from protonated 3Z- and 3E-ex-H2HPP, respectively,
contained 1 mM ex-H2HPP, 2 mM 2H-NADH, and 30 μM BacG
in 1 mL of H2O containing 50 mM potassium phosphate buffer
(pH 8.0). The reaction mixtures were incubated, the reactions
quenched, the mixtures purified, and the fractions identified
exactly as described above for protonated H4HPP isomers
1
1
the mixture was purified and desalted, and H and H−13C
gHSQC NMR data were collected as described above for the
BacAB-generated deuterated ex-H2HPP isomers.
Enzymatic Synthesis of Deuterated H4HPP Isomers
and H4Tyr Isomers in D2O Solvent. Two forms of deuterated
4R-H4HPP were produced in two distinct reactions. The first
reaction mixture contained 1 mM deuterated 3E-ex-H2HPP
produced from the deuterated BacAB reaction (described
above), 4 mM NADH, and 30 μM BacG in 1 mL of 97% D2O
containing 50 mM potassium phosphate buffer (pD 8.0). The
second reaction mixture contained 5 mM potassium prephenate
and 20 mM NADH and was simultaneously incubated with 20
μM BacABG in 95% D2O containing 50 mM potassium
phosphate buffer (pD 8.0). Likewise, two forms of deuterated
4R-H4Tyr were produced in two distinct reactions. The first
reaction mixture contained 2 mM deuterated 3E-ex-H2HPP
produced from BacAB action (described above), 8 mM NADH,
0.1 mM PLP, 10 mM L-Phe, and 15 μM BacGF in 1 mL of 97%
D2O containing 50 mM potassium phosphate buffer (pD 8.0).
The second reaction mixture contained 5 mM potassium
prephenate, 20 mM NADH, 20 mM L-Phe, 0.1 mM PLP, and
15 μM BacABGF in 95% D2O containing 50 mM potassium
phosphate buffer (pD 8.0).
Only one form of deuterated 4S-H4HPP and 4S-H4Tyr could
be produced. The reactions for the production of deuterated 4S-
H4HPP and 4S-H4Tyr were identical to the first set of reactions
used to produce deuterated 4R-H4HPP and 4R-H4Tyr
(described in the previous paragraph) except that deuterated
3Z-ex-H2HPP (produced from the deuterated BacAB reaction
described above) was used in place of 3E-ex-H2HPP.
5625
dx.doi.org/10.1021/bi3006362 | Biochemistry 2012, 51, 5622−5632

Biochemistry
Article
Figure 3. Michaelis−Menten kinetic traces for the reductase action of BacG on (A) 3Z-ex-H2HPP (kcat = 71.3 min−1; Km = 0.09 mM) and (B) 3E-ex-
H2HPP (kcat = 33.2 min−1; Km = 2.7 mM) in the presence of a saturating level of NADH (2 mM).
Figure 4. 1H NMR traces of purified (A) 4S-H4HPP from the action of BacG on 3Z-ex-H2HPP, (B) 4S-H4Tyr from the action of BacF on panel A, (C)
4R-H4HPP from the action of BacG on 3E-ex-H2HPP, and (D) 4R-H4Tyr from the action of BacF on panel C.
generated from protonated NADH. The detected LC−MS mass
of H4HPP produced from 4R-2H-NADH was 183.0670 (calcd,
183.0663). A major mass (184.0726) and a minor mass
(183.0666) were both detected from H4HPP produced from
4S-2H-NADH (calcd, 184.0726 and 183.0663, respectively).
Preparation of Tyrosine from the Transaminase Action
of BacF on 4-Hydroxyphenylpyruvate for the Determi-
nation of Amino Acid Chirality. HPP (2 mM), 10 mM Phe
(both L and D enantiomers in separate reactions), 0.5 mM PLP,
and 10 μM BacF were mixed in 1 mL of 50 mM potassium
phosphate buffer (pH 8.0). Reaction mixtures were incubated at
∼20 °C for 16 h, reactions quenched via addition of acetonitrile
to a final concentration of 30% (v/v), and mixtures lyophilized to
1
1
Collection of H and H−13C gHSQC NMR data was also
performed as described above (Figure S3 of the Supporting
Information).
5626
dx.doi.org/10.1021/bi3006362 | Biochemistry 2012, 51, 5622−5632

Biochemistry
Article
dryness and resuspended in 500 μL of H2O. The expected
transamination product (tyrosine) was identified from the BacF
reaction mixture containing L-Phe by LC−MS in positive
detection mode (0.1% formic acid as a solvent additive) using
a 50 mm × 2.1 mm Hypercarb column with water as the mobile
phase and acetonitrile as the eluent. No tyrosine was observed
from the BacF reaction mixture containing D-Phe as the
cosubstrate.
efficiencies of two substrates for the same enzyme, the 3Z,7R-ex-
H2HPP isomer is a 64-fold better substrate than the
corresponding E isomer. Thus, although the 3Z geometric
isomer is kinetically disfavored during BacB action, and
essentially absent from the nonenzymatic isomerization of en-
H2HPP to the thermodynamically favored ex-H2HPP,8 it is very
much the preferred substrate for the conjugate reduction
reaction of BacG.
Prior to chiral HPLC analysis of the BacF-produced tyrosine, it
was necessary to purify the nascent tyrosine from the other
reaction components to avoid spectral contamination. The
resuspended reaction mixture was loaded onto a 100 mm × 10
mm Hypercarb column equilibrated in 10 mM potassium
phosphate buffer (pH 8.0). Bound compounds were eluted via a
linearly increasing gradient of acetonitrile. Fractions containing
tyrosine were initially located by their UV−vis spectra and later
confirmed by LC−MS analysis as described above. The
appropriate fractions were pooled and lyophilized to dryness.
The amino acid chirality of the BacF-generated tyrosine was
determined by comparing its elution profile from a 150 mm × 4.6
mm Chirex 3126 column (Phenomenex) equilibrated in 2 mM
copper(II) sulfate in an 84/16 (v/v) water/methanol mixture
with that of authentic L- and D-tyrosine standards. Tyrosine
elution was monitored by UV−vis absorbance at 280 nm (Figure
S4 of the Supporting Information).
Panels A and C of Figure 4 show the proton NMR spectra of
the H4HPP samples generated from the action of BacG on 3Z-ex-
H2HPP and 3E-ex-H2HPP substrates, respectively. It is clear
from examination of the H8 and H9 resonances that the products
are distinct, most likely because of distinct diastereomers at C4,
the anticipated site of hydride addition by BacG. On subsequent
BacF-mediated transamination of each of the H4HPP keto acid
samples to the corresponding 2S-H4Tyr samples, the proton
NMR spectra in panels B (derived from the Figure 4A sample)
and D (derived from the Figure 4C sample) of Figure 4 also gave
distinct diastereomers as evidenced by the distinct splitting
patterns from the methylene hydrogens at C3, C8, and C9. The
assignments of the indicated stereochemistries as 4S (from the
3Z substrate) and 4R (from the 3E substrate) are detailed in a
subsequent section.
Proof that BacG acted by conjugate addition of an NADH-
derived hydride to C4 of the ex-H2HPP substrates was explicitly
tested with 4R-D1-NADH and 4S-D1-NADH samples as
1
RESULTS
reductants. H NMR analysis of the two chirally deuterated
■
NADH samples established that 4R-D1-NADH was 40% D1
while 4S-D1-NADH was 62% D1 (Figure S2 of the Supporting
Information).
BacG Acts on 3Z,7R- and 3E,7R-ex-H2HPP Geometric
Isomers To Give H4HPP Products That Are Diastereo-
meric at C4. In initial studies of the Bac pathway enzymes for the
conversion of prephenate to H4Tyr, we conducted incubations of
either BacAGF or BacABGF in parallel, based on the observation
that the BacA-derived en-H2HPP will isomerize nonenzymati-
cally and quantitatively into conjugation to give the exocyclic ex-
H2HPP product.7 BacB accelerates the allylic isomerization by up
to 103 and so was of use in shortening the kinetics in the four
enzyme incubations.15 These incubations could be conducted in
buffered H2O or buffered D2O, to yield H4Tyr with one
deuterium each at C8 and C9 [as well as at C2 and C3 (vide infra)].
A deeper analysis of the BacB-mediated isomerization recently
led us to the finding that while the E/Z ratio of ex-H2HPP from
BacA followed by nonenzymatic allylic isomerization was 50/1,
in BacB-containing equilibrated incubations the E/Z ratio was 3/
1. In turn, that led us to separation and structural assignment of
the 3E and 3Z isomers on Hypercarb HPLC columns and a
proposed mechanism for geometric isomer equilibration via
dienolate intermediates.8
NMR analysis of the H4HPP products from 4R-D1-NADH
incubations on both the 3E- and 3Z-ex-H2HPP substrates shows
the transfer of the diastereotopic 4-pro-S hydride to both the 3E
and 3Z substrate isomers. Figure S3 of the Supporting
Information shows the 3Z substrate had 0.94 equiv of 1H at C4
while the 3E substrate possessed 1.0 equiv of 1H, referenced to
the single hydrogen at C7 in the product H4HPPs. One could
worry that the lack of deuterium reflected a kinetic isotope effect
between NADH and 4R-D1-NADH molecules in the NADH
reductant pool, but the complementary results with the transfer
of deuteride from the 4S-D1-NADH sample ruled out that
possible ambiguity. The 4S-D1-NADH donor sample (62%
deuterium content) gave H4HPP with 68% D at C4 when the 3Z
substrate was reduced and 58% D at C4 when the 3E substrate
was reduced. These results clearly indicate that BacG transfers
the pro-S hydride of NADH during ex-H2HPP reduction.
These results validate several points. First, BacG is acting as a
regiospecific hydride transfer catalyst to C4 for each of the 3E-
and 3Z-ex-H2HPP substrate isomers. Second, the pro-S hydride
of NADH is transferred to each of the substrate geometric
isomers. Third, from the NMR spectra in panels A and C of
Figure 4 noted above, the H4HPP products are likely to be
diastereomers at C4.
The BacF Transaminase Is 2S-Specific. Given that the
BacF transaminase processes L-Phe (2S) as an amino donor to
cosubstrate H4HPP, we anticipated it would show the same
specificity toward partner keto acids that become reductively
aminated. To that end, we have now demonstrated that while L-
Phe is an effective amino donor, D-Phe is not (data not shown).
To validate that the keto acid partner undergoing amination gives
a 2S amino acid product, we used p-hydroxyphenylpyruvate as
the acceptor with L-Phe as a donor. As anticipated, tyrosine was
generated (LC−MS mass of 182.0805; calcd, 182.0812) as L-Phe
The presumption from the above results was that each of the
3E- and 3Z,7R-ex-H2HPP geometric isomers could be a substrate
for the NADH-utilizing BacG. This enzyme is assayed by loss of
the 340 nm chromophore of the reduced nicotinamide ring of
NADH as ex-H2HPP is reduced to H4HPP (and NADH is
oxidized to NAD+). Because of the partially overlapping
chromophores of the ex-H2HPP substrate (λmax = 295 nm)
and cosubstrate NADH, measurements of the reaction were
taken away from the absorbance maximum of NADH, at 370 nm,
where a sufficient signal-to-noise ratio persisted. Indeed, the
separated geometric 3E (96% purity) and 3Z [97% purity (Figure
S1 of the Supporting Information)] isomers were each saturable
substrates (Figure 3). The kcat for the 3Z isomer at 71.3 min−1
was approximately twice that for the 3E isomer (33.2 min−1).
The 3Z isomer also had a 30-fold lower Km value (0.09 mM vs 2.7
mM). By kcat/Km ratios, which compare the relative catalytic
5627
dx.doi.org/10.1021/bi3006362 | Biochemistry 2012, 51, 5622−5632

Biochemistry
Article
Figure 5. 1H−1H NOESY spectra and accompanying NOE correlations used to determine the C4 stereochemistry of (A) 4S-H4Tyr and (B) 4R-H4Tyr.
Because of the spectral overlap of H3a, H8a, and H9b in the spectra of (B) 4R-H4Tyr, NOE correlations of (C) 4R-H4Tyr-d4 were also analyzed to allow
analysis of correlations to and from H9b to complete the stereochemical determination. Black cross-peaks in the NOESY spectra indicate positive
resonances, while red cross-peaks indicate negative resonances. Key NOE interactions have been noted on the structures with dashed lines, and the
correlating cross-peaks are boxed in the NOESY spectra.
was oxidatively deaminated in the first half of the BacF reaction.
The chirality of the nascent Tyr was determined to be “L” (2S) by
comparison with authentic D- and L-Tyr samples by chiral HPLC
(Figure S4 of the Supporting Information). By analogy, we assign
the H4Tyr from reductive amination of H4HPP by BacF as the 2S
diastereomer. At this point, with the 2S stereochemistry assigned,
5628
dx.doi.org/10.1021/bi3006362 | Biochemistry 2012, 51, 5622−5632

Biochemistry
Article
we turned to analysis of the absolute stereochemistry at C4 of the
H4Tyr samples generated from tandem action of BacABGF to
obtain the absolute stereochemistry of the H4Tyr emerging from
the four enzymes in the bacilysin pathway.
Figure 5B by the strong correlation between H8b and H9a).
1
Additional support for these analyses comes from the H−1H
NOESY spectra of protonated 4R-H4HPP spectra (Figure S9 of
the Supporting Information), where H8a is a distinct resonance.
In this spectrum, strong correlations are seen between H8a and
the protons on C4 and C7 (same face) while little or no
correlations are seen between H9a and either H4 or H7 (opposite
face).
Stereochemical Insights at C8 and C9 in Deuterated
Samples of H4HPP and H4Tyr. A bonus from NMR analyses of
the deuterated 4R- and 4S-H4Tyr samples (as seen in Figure 5C
and Figures S5B, S6C, and S7C of the Supporting Information)
is information about the placement of deuterium at C8 (from
BacA action) and C9 (from BacB action). We know the
stereochemistry at C2 is 2S from BacF transaminase action data
cited above. We have not tried to determine the stereochemistry
at the C3-HD center. One deuterium is picked up at that site
during BacG action, but there is also solvent exchange at C3 when
BacB conducts the equilibration between the 3E and 3Z isomers
of ex-H2HPP, via dienolate intermediates.8 The extent of that
exchange may well have differed during different incubations,
leading to partial deuterium incorporations at C3.
At C8, the deuterium in both the 4R- and 4S-H4Tyr products is
on the same face as H7; therefore, the stereochemistry in the C8-
HD samples is 8S. The stereochemical outcome at C9 in the C9-
HD samples is more complex. As shown in Figure 5 and Figure
S5B of the Supporting Information in the 4R-H4Tyr sample from
simultaneous BacABGF D2O incubations with prephenate, the
deuterium at C9 is on the ring face opposite both H4 and H7 and
therefore is the 9S monodeutero isomer.
The results with 4S-H4Tyr samples from D2O incubations
were less clear-cut. Under some conditions (Figure S7C of the
Supporting Information), with BacB present, there was partial
deuterium substitution at both the 9S and 9R positions. In
experiments where the BacA product en-H2HPP was allowed to
isomerize nonenzymatically (in solutions estimated at 98% D2O
and then carried through to the BacGF reactions in D2O), there
was very little deuterium at C9 (Figure S8A of the Supporting
Information), perhaps reflecting a large deuterium isotope effect
in the nonenzymic isomerization of en-H2HPP solely to the 3E-
ex-H2HPP isomer. Because of these vagaries due to BacB-
mediated exchange and the nonenzymatic isomerization giving
solely the 3E-ex-H2HPP product,8 we have not conducted a full
interpretation of the C9 results in the 4S-H4Tyr diastereomer.
Interpretation of NOE Correlations in Determining the
Stereochemistry of 4S-H4Tyr and 4R-H4Tyr at C4. To
determine the stereochemistry at C4 of H4Tyr (and by inference
H4HPP) derived from BacG reduction of 3Z-ex-H2HPP, we
collected 1H−1H NOESY data reporting on the relative distances
between protons on a fully protonated sample of H4Tyr (Figure
5A). Because the proton resonances needed to make the
appropriate correlations did not overlap in this sample,
stereochemical assignment was straightforward. We began
linking correlations using H7 as the reference, because the
stereochemistry at C7 had been previously assigned in our earlier
study as R stereochemistry.8 From inspection of the NOE cross-
peaks, it is seen that H7 shows a strong correlation with H8a and
H9a, but not H8b and H9b. This indicates that H7, H8a, and H9a are
on the same face of the cyclohexenol ring, while H8b and H9b are
on the opposite face (Figure 5A). We then moved our attention
to the H4 proton and noticed that H4 correlates strongly with H8b
and H9b, but not H8a and H9a, indicating that H4, H8b, and H9b are
on the same face of the cyclohexenol ring and thus on the
opposite face from H7. These correlations establish that H4Tyr
(and thus H4HPP, as shown in Figure 4) originating from BacG
reduction of 3Z-ex-H2HPP possesses 4S stereochemistry (Figure
5A).
Determination of the C4 stereochemistry of H4Tyr (and
H4HPP) derived from BacG reduction of 3E-ex-H2HPP was not
so straightforward because of the overlapping cross-peaks of H3a,
1
H8a, and H9b in the H−1H NOESY data of protonated H4Tyr
(Figure 5B). However, two important NOE correlations could
be clearly seen in these data: a strong cross-peak between H8b and
H9a and a weak cross-peak between H4 and H7. While the strong
cross-peak between H8b and H9a indicates they are most likely on
the same face of the cyclohexenol ring, the NOE correlation
between H4 and H7 confirms that H4 and H7 are on the same face
of the ring. This shows that 4R-H4Tyr (and thus 4R-H4HPP) is
derived from the action of BacG on 3E-ex-H2HPP (Figure 5B).
Although the stereochemistry at C4 of 4R-H4Tyr could be
1
interpreted despite the overlapped resonances in the H−1H
NOESY spectrum (Figure 5B), NOE analysis of 4R-H4Tyr-d4
(Figure 5C) was necessary to determine the relative positions of
the nonequivalent protons of C8 and C9. As observed in Figure
5C and in greater detail in Figure S5B of the Supporting
Information, simultaneous incubation of BacABGF with
prephenate in 95% D2O solvent incorporates one deuterium
each at C2, C3, C8, and C9 of 4R-H4Tyr-d4 and fortuitously yields
a 1H NMR spectra without overlapped resonances. Inspection of
DISCUSSION
■
Prephenate is the immediate precursor to both phenylpyruvate
(via prephenate dehydratase16,17) and hydroxyphenylpyruvate
(via prephenate dehydrogenase18,19) on the way to L-Phe, L-Tyr,
and many phenylpropanoid metabolites.20,21 In the anticapsin/
bacilysin pathway, prephenate is instead shunted down a newly
appreciated metabolic branch. BacA, the first enzyme in this and
related pathways to dihydro- and tetrahydro-aromatic metabolite
scaffolds,9 is a remarkable catalyst, engineering the decarbox-
ylation without aromatization of the 1,4-cyclohexadiene ring of
prephenate.7 It is a regiospecific isomerization catalyst, working
only on the pro-R, not the pro-S, double bond of the prephenate
diene and delivering a proton only to C6, not C6′, a carbon that
becomes C8 in the en-H2HPP product.8 Three enzymes later, by
tandem action of BacBGF, H4Tyr is formed.
1
the H−1H NOESY data of 4R-H4Tyr-d4 shows an especially
strong cross-peak between H4 and H9b, indicating they are on the
same face of the cyclohexenol ring. (For comparison, the cross-
peak between H4 and H9a in Figure 5B is weaker. The cross-peak
between H4 and H7 in panels B and C of Figure 5 can be used to
normalize the spectra to each other.) Also in Figure 5C, almost
equivalent cross-peak intensities are seen between H7 and H8b
and between H7 and H9b. For this to be possible, H8b would need
to be on the opposite face of H7 while H9b would need to be on
the same face as H7. This correlation agrees with the
determination made above that H4 and H9b are on the same
face of the cyclohexenol ring (Figure 5C). Combined, these
determinations imply that H4, H7, H8a, and H9b are on one ring
face while H8b and H9a are on the opposing face (as suggested in
Unlike its familiar aromatic congener L-tyrosine that has one
chiral center (2S), H4Tyr has three chiral centers: C2, the ring
junction C4, and the original alcoholic carbon of prephenate, C7.
5629
dx.doi.org/10.1021/bi3006362 | Biochemistry 2012, 51, 5622−5632

Biochemistry
Article
Figure 6. BacG transfers the pro-S hydride of NADH to opposite faces of (A) 3E-ex-H2HPP and (B) 3Z-ex-H2HPP to produce 4R-H4HPP and 4S-
H4HPP, respectively. (C) BacA decarboxylation of prephenate performed in D2O solvent produces en-H2HPP with “S” stereochemistry at C6. This
indicates that the incoming proton is delivered to the face of the cyclohexadiene ring opposite the departing CO2. (D) Action of BacABGF on
prephenate produces 2S,4S,7R-H4Tyr and 2S,4R,7R-H4Tyr, with the relative amounts produced being determined by the action of BacB.
One expects enzymes to act as chiral catalysts such that of the
eight possible H4Tyr diastereomers, only one is anticipated. The
2S,4S diastereomer of H4Tyr is indeed a known nonproteino-
genic building block of the cyanobacterial nonribosomal peptides
aeruginoside 126A and aeruginoside 126B.22 In the initial
structural assignment of anticapsin, the C4 stereochemistry was
initially incorrectly assigned1 but corrected subsequent to total
synthesis23 as the 2S,4S diastereomer, consistent with 2S,4S-
H4Tyr being on pathway. The stereochemistry at C7 of anticapsin
has become moot because of the oxidation of the C7 alcohol in
prephenate to the ketone in anticapsin. Given that we have
recently proven that BacA, by isomerizing the pro-R olefin in
prephenate, generates the 7R-Δ3,Δ5-diene, we therefore
anticipated the H4Tyr emerging from tandem action of
BacABGF would thus be the 2S,4S,7R diastereomer.
reduction by BacG with delivery of a (pro-S) hydride to C4 and a
solvent-derived hydrogen at C3. At the level of these H4HPP keto
acid products, and after they had been swept through to the more
stable H4-tyrosines via BacF, it is clear that opposite chirality
results at C4. The 3Z isomer yields 2S,4S,7R-H4Tyr from BacG
action, while the 3E isomer gives 2S,4R,7R-H4Tyr (Figure 6D).
We expect the 4S diastereomer is on pathway to anticapsin (4S)
but not the 4R diastereomer. However, this proposition must
await study of the last two missing enzymatic steps in anticapsin
biogenesis.
The relative orientation of the dihydronicotinamide ring of the
hydride donor NADH and the conjugated dienyl keto acid of the
ex-H2HPP must switch between the 3Z and 3E isomers as they
are recognized as substrates. If we assume that the NADH
orientation stays the same within its BacG binding site, as it
transfers the pro-S hydrogen to both geometric isomers of
substrate, the different faces of the olefin would be presented to
the incoming hydride as shown in panels A and B of Figure 6,
assuming the C3−C1 chain of the geometric ex-H2HPP isomers
stays in the same orientation.
As we have reported here, this expectation is met, but the
2S,4R,7R diastereomer also emerges from the tandem action of
the four Bac enzymes. Perhaps most intriguing is the action of the
second enzyme, BacB, to generate a mix of 3Z and 3E isomers of
ex-H2HPP. Both are then substrates for conjugate hydride
5630
dx.doi.org/10.1021/bi3006362 | Biochemistry 2012, 51, 5622−5632

Biochemistry
Article
The formation of C4 diastereomers of H4HPP (and then of
H4Tyr from subsequent BacF action) gives further insight into
the possibly corrective role of BacB in the pathway. We have
previously noted that the 7R-en-H2HPP product of BacA can
undergo nonenzymatic conversion to the more conjugated 7R-
ex-H2HPP isomer. We wondered why the producing cell would
expend energy to make a protein (BacB) that would accelerate a
reaction that would otherwise occur. Certainly, matching the
kinetics of metabolite flux and prevention of off-pathway
decomposition and aromatization are reasonable rationales, but
most telling is the finding that the nonenzymatic route gives 98%
3E geometric isomer. The work reported here shows that the E
isomer will lead to, via the action of BacG as a reductase, the 4R
diastereomer of H4Tyr, which is presumably the wrong chirality
to go on to the anticapsin antibiotic. BacB instead equilibrates the
more stable 3E and less stable 3Z geometric isomers to give up to
25% of the 3Z isomer, which on subsequent reduction by BacG
gives the desired 4S stereochemistry. If there is enough BacB in
producer cells coupled to the BacG catalytic efficiency preference
of 3Z/E (64/1 found in this study), then flux could be routed
selectively to the productive 4S-cyclohexenol scaffold. It may be
that the 4R stereochemistry is useful for one or more alternate
downstream metabolites in this or related microbes [i.e., the diepi
form of 2-carboxy-6-hydroxyoctahydroindole (diepi-Choi)
(2R,4R stereochemistry) residues found as constituents of
nonribosomal peptides in cyanobacteria].24
to obtain the anticapsin warhead: 7R-alcohol oxidation and olefin
epoxidation.
ASSOCIATED CONTENT
■
S
* Supporting Information
HPLC analysis of ex-H2HPP isomer purity; 1H NMR spectra of
1
4R- and 4S-2H-NADH; H NMR spectra of H4HPP isomers
2
synthesized from H-NADH; chiral HPLC analysis of BacF-
generated tyrosine; additional NMR spectra of deuterated forms
of ex-H2HPP, H4HPP, and H4Tyr isomers; and tabulated NMR
spectral data for H4HPP and H4Tyr isomers. This material is
available free of charge via the Internet at http://pubs.acs.org.
AUTHOR INFORMATION
Corresponding Author
*E-mail: christopher_walsh@hms.harvard.edu. Phone: (617)
432-1715. Fax: (617) 432-0483.
■
Funding
This work was supported in part by National Institutes of Health
Grants AI042738 and GM49338 (C.T.W.).
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
We thank Drs. Sarah A. Mahlstedt, Stuart W. Haynes, Timothy A.
Wencewicz, and Steven J. Malcolmson for helpful advice and
discussions.
While a main objective of this study was to ascertain the C4
stereochemistry from the action of the NADH-utilizing BacG at
the stage of reduction of ex-H2HPP to H4HPP, we have also
focused on the stereochemical outcomes at C8 and C9 from
consecutive action of BacA and BacB in deuterated buffers. While
C8 and C9 sequentially become CH2 groups in en-H2HPP and ex-
H2HPP, respectively, and the stereochemistry is thus cryptic, in
principle the stereochemical course of each enzyme at those
carbons could be revealed with deuterium in place of hydrogen.
We have previously found7 that one deuteron from solvent is
fixed at C8 and one at C9, and in this study, we have shown that
BacA introduces the solvent derived (H+/D+) from above the
plane of the diene in prephenate, yielding 6S-en-H2HPP-d1.
Because of the numbering priority changes, this compound
becomes 8S-ex-H2HPP upon BacB action (Figure 6C). In turn,
as gathered from the 4R-H4Tyr result, BacB instead transfers a
solvent-derived D+ from below the plane at C9 (H9a) and yields a
9R-deutero product, swept through to H4Tyr. (BacB also
removes one of the methylene hydrogens at C3 as it equilibrates
the E and Z geometric isomers of ex-H2HPP.8). Interpretation of
ABBREVIATIONS
■
H2HPP, dihydro-4-hydroxyphenylpyruvate; H4HPP, tetrahydro-
4-hydroxyphenylpyruvate; H4Tyr, tetrahydrotyrosine; HPP, 4-
hydroxyphenylpyruvate; 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; HPLC, high-perform-
ance liquid chromatography; D2O, deuterium oxide; GlcNAc, N-
acetylglucosamine.
REFERENCES
■
(1) Walker, J., and Abraham, E. (1970) Structure of bacilysin and other
products of Bacillus subtilis. Biochem. J. 118, 563−570.
(2) Milewski, S., Chmara, H., and Borowski, E. (1986) Anticapsin: An
active-site directed inhibitor of glucosamine-6-phosphate synthetase
from Candida albicans. Drugs Exp. Clin. Res. 12, 577−583.
(3) Chmara, H. (1985) Inhibition of glucosamine synthase by bacilysin
and anticapsin. J. Gen. Microbiol. 131, 265−271.
1
the H−1H NOESY spectrum of protonated 3E-ex-H2HPP
(Figure S8B of the Supporting Information) established that H9a
is on the ring face opposite of H7.
(4) Chmara, H., Zahner, H., and Borowski, E. (1984) Anticapsin, an
active-site directed irreversible inhibitor of glucosamine-6-phosphate
synthetase from Escherichia coli. J. Antibiot. 37, 1038−1043.
(5) Steinborn, G., Hajirezaei, M., and Hofemeister, J. (2005) Bac genes
for recombinant bacilysin and anticapsin production in Bacillus host
strains. Arch. Microbiol. 183, 71−79.
(6) Inaoka, T., Takahashi, K., Ohnishi-Kameyama, M., Yoshida, M.,
and Ochi, K. (2003) Guanine nucleotides guanosine 5′-diphosphate 3′-
diphosphate and GTP co-operatively regulate the production of an
antibiotic bacilysin in Bacillus subtilis. J. Biol. Chem. 278, 2169−2176.
(7) Mahlstedt, S. A., and Walsh, C. T. (2010) Investigation of
anticapsin biosynthesis reveals a four-enzyme pathway to tetrahydrotyr-
osine in Bacillus subtilis. Biochemistry 49, 912−923.
While it will take structural studies of BacA in complex with
substrate and/or product to identify candidate side chains of
active site residues such as BH+, it must be a very efficient process
to derail what has been viewed as the default outcome in other
members of the prephenate dehydratase superfamily.16 That
default option is cleavage of the C7−O bond as the cyclo-
hexadiene ring aromatizes, and there has been substantial debate
over whether decarboxylation and loss of water is a concerted
process.25,26
Knowledge of the configuration of the H4Tyr diastereomers
generated by the first four enzymes of the Bac pathway, and the
realization that 4R and 4S products emerge, will provide the
appropriate substrates for evaluation and identification of the
enzymes responsible for what should be the two remaining steps
(8) Parker, J. B., and Walsh, C. T. (2012) Olefin isomerization
regiochemistries during tandem action of BacA and BacB on prephenate
in bacilysin biosynthesis. Biochemistry 51, 3241−3251.
5631
dx.doi.org/10.1021/bi3006362 | Biochemistry 2012, 51, 5622−5632

Biochemistry
Article
(9) Mahlstedt, S., Fielding, E. N., Moore, B. S., and Walsh, C. T. (2010)
Prephenate decarboxylases: A new prephenate-utilizing enzyme family
that performs nonaromatizing decarboxylation en route to diverse
secondary metabolites. Biochemistry 49, 9021−9023.
(10) Horecker, B., and Kornberg, A. (1948) The extinction coefficients
of the reduced band of pyridine nucleotides. J. Biol. Chem. 175, 385−390.
(11) Gottlieb, H., Kotlyar, V., and Nudelman, A. (1997) NMR
chemical shifts of common laboratory solvents as trace impurities. J. Org.
Chem. 62, 7512−7515.
(12) Viola, R. E., Cook, P. F., and Cleland, W. W. (1979)
Stereoselective preparation of deuterated reduced nicotinamide
adenine-nucleotides and substrates by enzymatic-synthesis. Anal.
Biochem. 96, 334−340.
(13) Ottolina, G., Riva, S., Carrea, G., Danieli, B., and Buckmann, A.
(1989) Enzymatic-synthesis of [4R-H2]NAD(P)H and [4S-H2]NAD-
(P)H and determination of the stereospecificity of 7α-hydroxysteroid
and 12α-hydroxysteroid dehydrogenase. Biochim. Biophys. Acta 998,
173−178.
(14) Mostad, S., and Glasfeld, A. (1993) Using high-field NMR to
determine dehydrogenase stereospecificity with respect to NADH: An
undergraduate biochemistry lab. J. Chem. Educ. 70, 504−506.
(15) Rajavel, M., Mitra, A., and Gopal, B. (2009) Role of Bacillus subtilis
BacB in the synthesis of bacilysin. J. Biol. Chem. 284, 31882−31892.
(16) Kleeb, A. C., Kast, P., and Hilvert, D. (2006) A monofunctional
and thermostable prephenate dehydratase from the archaeon
Methanocaldococcus jannaschii. Biochemistry 45, 14101−14110.
(17) Schmit, J. C., and Zalkin, H. (1969) Chorismate mutase-
prephenate dehydratase. Partial purification and properties of enzyme
from Salmonella typhimurium. Biochemistry 8, 174−181.
(18) Koch, G. L. E., Shaw, D. C., and Gibson, F. (1970) Tyrosine
biosynthesis in Aerobacter aerogenes: Purification and properties of
chorismate mutase-prephenate dehydrogenase. Biochim. Biophys. Acta
212, 375−386.
(19) Hermes, J. D., Tipton, P. A., Fisher, M. A., Oleary, M. H.,
Morrison, J. F., and Cleland, W. W. (1984) Mechanisms of enzymatic
and acid-catalyzed decarboxylations of prephenate. Biochemistry 23,
6263−6275.
(20) Vogt, T. (2010) Phenylpropanoid biosynthesis. Mol. Plant. 3, 2−
20.
(21) Cotton, R., and Gibson, F. (1965) Biosynthesis of phenylalanine
and tyrosine: Enzymes converting chorismic acid into prephenic acid
and their relationships to prephenate dehydratase and prephenate
dehydrogenase. Biochim. Biophys. Acta 100, 76−88.
(22) Ishida, K., Christiansen, G., Yoshida, W. Y., Kurmayer, R., Welker,
M., Valls, N., Bonjoch, J., Hertweck, C., Boerner, T., Hemscheidt, T., and
Dittmann, E. (2007) Biosynthesis and structure of aeruginoside 126A
and 126B, cyanobacterial peptide glycosides bearing a 2-carboxy-6-
hydroxyoctahydroindole moiety. Chem. Biol. 14, 565−576.
(23) Baldwin, J., Adlington, R., and Mitchell, M. (1993) Stereo-
controlled enantiospecific synthesis of anticapsin: Revision of the
configuration. J. Chem. Soc., Chem. Commun.,, 1332−1335.
(24) Lifshits, M., and Carmeli, S. (2012) Metabolites of Microcystis
aeruginosa bloom material from Lake Kinneret, Israel. J. Nat. Prod. 75,
209−219.
(25) Van Vleet, J., Kleeb, A., Kast, P., Hilvert, D., and Cleland, W. W.
(2010) 13C isotope effect on the reaction catalyzed by prephenate
dehydratase. Biochim. Biophys. Acta 1804, 752−754.
(26) Zhang, S., Wilson, D. B., and Ganem, B. (2000) Probing the
catalytic mechanism of prephenate dehydratase by site-directed
mutagenesis of the Escherichia coli P-protein dehydratase domain.
Biochemistry 39, 4722−4728.
5632
dx.doi.org/10.1021/bi3006362 | Biochemistry 2012, 51, 5622−5632