CHEMBIOCHEM
FULL PAPERS
DXP synthase was purified as previously described.[19] The protein
concentration was determined by using the Bradford assay. Porcine
pyruvate dehydrogenase was obtained from a commercial source;
its specificity activity was determined by the manufacturer. For
chemical synthesis, CH2Cl2 was distilled over calcium hydride. An-
hydrous acetonitrile was packed in Sure-Seal bottles. All reactions
were carried out under argon. NMR spectra were recorded on
a Varian 500 MHz spectrometer. Reaction progress was monitored
through 31P NMR with triphenylphosphine oxide (TPPO, d=0 ppm)
dissolved in deuterated benzene as external standard. 1H NMR
chemical shifts are reported relative to tetramethylsilane (TMS, d=
0 ppm) as internal reference. Preparative HPLC was performed on
a Beckman Gold Noveau system with a Varian Dynamax 250ꢂ
21.4 mm Microsorb C18 column.
hydrazine[19] for 20 min to ensure complete derivatization of sub-
strates and product at low concentration. The derivatization mix-
tures were analyzed by HPLC as previously described.[19] To deter-
mine the initial reaction rates in the presence of various inhibitor
concentrations, the d,l-GAP and DXP hydrazone HPLC peak areas
were measured, and the product concentration was determined as
a percentage of total peak area and plotted against reaction time.
Initial rates GraFit version 7 from Erithacus Software was used to
generate IC50 curves.
Calculating the active site volume: Coordinates for the ThDP-de-
pendent enzymes, Deinococcus radiodurans DXS (PDB ID: 2O1X),[16]
human PDHE1p (PDB ID: 3EXE),[33] and transketolase (PDB ID:
3MOS)[34] were structurally aligned in Coot[28] by using LSQ Super-
pose and residue ranges A:151–164 (2O1X), E:164–177 (3EXE), and
A:152–165 (3MOS). The choice of residues was based on their
close proximity to ThDP in order to maximize a similar orientation
of the active site region of interest. The RMS deviation, calculated
with VMD[35] between residues lining the ThDP binding site, was
1.54 ꢃ (2O1X/3EXE) or 1.01 ꢃ (2O1X/3MOS) for 16 Ca backbone
atoms. The biological assembly of transketolase (3MOS) was deter-
mined by using the PISA[36] web server. Aligned structures were up-
loaded to the Pocket-Finder[29] web server to determine the vol-
umes of the active site pockets. Cofactors ThDP or ThDP+metal
ions were treated as part of the protein, and all other molecules
were discarded for the purpose of defining the protein surface for
pocket detection. Pocket-Finder reported volumes and generated
space-filling models for the active site pocket in each structure cor-
responding to the pocket adjacent to TDP in chain A of 2O1X. An
overlay of the mesh representations with respect to the active site
cofactor and metal ion was rendered in PyMOL (The PyMOL Molec-
ular Graphics System, Version 1.5.0, Schrçdinger, LLC).
HPLC analysis of DXP synthase-catalyzed CÀN bond formation
and product characterization: Reaction mixtures containing
HEPES (100 mm, pH 8.0, 2 mm MgCl2, 5 mm NaCl), ThDP (1 mm),
BSA (1 mgmLÀ1), pyruvate (10–20 mm), DMSO (10%, v/v) and ni-
troso substrate (0.5–5 mm) were preincubated at 378C for 5 min.
Reactions were initiated with enzyme (1–5 mm). Aliquots of the
enzymatic mixture were removed at various time intervals and
quenched in an equal volume of cold methanol. Quenched mix-
tures were incubated on ice for 20 min. Precipitated proteins were
removed by centrifugation, and the supernatant was analyzed by
HPLC with UV detection under the following conditions: flow
rate=3 mLminÀ1, solvent A: NH4OAc (100 mm, pH 4.6), solvent B:
acetonitrile, method: 0–100% B over 10 min. The products were
extracted from the supernatant with ethyl acetate (3ꢂ). The com-
bined organic extracts were concentrated, and the resulting sam-
ples were dissolved in MeOH and resubjected to HPLC analysis to
confirm that product degradation had not taken place during the
extraction procedure. Products were subsequently characterized by
mass spectrometry.
Active site pocket hydrophobicity calculations with fpocket:
Fpocket[30] (Table S1) was run to detect and analyze pockets in DXP
synthase (2O1X), PDH (3EXE) and TK (3MOS). The complete coordi-
nate files for DXP synthase and PDH, and the biological assembly
for TK, were used as inputs for fpocket. The default cofactor list for
fpocket was modified to include TDP and TPP prior to program
compilation so that the ThDP cofactor would be treated as a part
of the protein as opposed to a removable ligand. The pockets cor-
responding to the active sites used for the volume calculations
with Pocket Finder were determined visually, and the parameters
were recorded.
Determination of kinetic parameters for nitroso substrates: Re-
action mixtures containing HEPES (100 mm, pH 8.0, 2 mm MgCl2,
5 mm NaCl), ThDP (1 mm), BSA (1 mgmLÀ1), pyruvate (10–20 mm),
DMSO (10%, v/v), and nitroso substrate (10–300 mm) were preincu-
bated at 378C for 5 min. Enzymatic reactions were initiated by ad-
dition of DXP synthase (0.5–2 mm; or 0.1 units per mL PDH) and
monitored spectrophotometrically by measuring the rate of disap-
pearance of the nitroso substrate at its corresponding lmax. Sub-
strate concentration as a function of time was determined from
absorbance values by using Beer’s Law. Initial reaction rates were
determined from the linear range of the reaction progress curve,
usually within 1–3 min. Data analysis to determine kcat and Km for
each alternative substrate was carried out by using GraFit version 7
from Erithacus Software.
Synthesis of BnAP (Figure S16): Benzylacetylphosphonate was
prepared starting from phosphorus trichloride in a similar manner
to butylacetylphosphonate,[18] by using standard procedures. Tri-
benzyl phosphite was generated from benzyl alcohol, diisopropyle-
thylamine, and phosphorous trichloride according to Saady et al.[37]
The spectral properties of the compound were identical to pub-
lished values. For the preparation of benzylacetylphosphonate
(BnAP), a flame-dried flask, cooled under argon, was charged with
acetyl chloride (0.32 mL, 4.5 mmol). Tribenzyl phosphite (0.46 g,
1.3 mmol) was dissolved in anhydrous CH2Cl2 (13 mL), and the re-
sulting mixture was added dropwise to acetyl chloride. The prog-
ress of the reaction was monitored through 31P NMR spectroscopy,
and complete conversion of tribenzyl phosphite (d=113 ppm) to
dibenzylacetyl-phosphonate (d=À26 ppm) was observed within
1 h. Volatiles were removed in vacuo, and the crude material was
used without further purification. Dibenzylacetylphosphonate was
dissolved in anhydrous acetonitrile (2.2 mL), and lithium bromide
(0.17 g, 0.95 mmol) was added in one portion. The reaction mixture
was heated to 508C for ~4 h. The lithium salt of benzylacetyl-
phosphonate precipitated from solution and was removed by
Evaluation of nitroso substrates as inhibitors of DXP formation:
Reaction mixtures containing HEPES (100 mm, pH 8.0, 2 mm MgCl2,
5 mm NaCl), ThDP (1 mm), BSA (1 mgmLÀ1), DMSO (10%, v/v), d-
GAP (30 mm), pyruvate (80 mm), and various concentrations of ni-
troso inhibitor were preincubated at 378C for 5 min. Enzyme reac-
tions were initiated by addition of DXP synthase (0.1 mm). Aliquots
(150 mL) of the enzymatic mixture were removed between 0.5 and
3 min and quenched in ice-cold methanol (150 mL). Precipitated
protein was removed by centrifugation, and the supernatant was
diluted in an equal volume of water. The nitroso substrate was re-
moved by extraction into acetonitrile (3ꢂ) by using a previously
described freeze-extraction technique.[32] The aqueous layer main-
tained a constant ratio of d-GAP and DXP during the extraction,
and was subjected to derivatization conditions to produce the cor-
responding hydrazones with a fivefold excess of 2,4-dinitrophenyl-
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2013, 14, 1309 – 1315 1314