Mechanistic studies on pyridoxal phosphate synthase: The reaction pathway leading to a chromophoric intermediate

Article abstract of DOI:10.1021/ja076604l

Two routes for the de novo biosynthesis of pyridoxal-5′-phosphate (PLP) have been discovered and reconstituted in vitro. The most common pathway that organisms use is dependent upon the activity of just two enzymes, known as Pdx1 (YaaD) and Pdx2 (YaaE) in

Full text of DOI:10.1021/ja076604l

Published on Web 02/14/2008
Mechanistic Studies on Pyridoxal Phosphate Synthase: The
Reaction Pathway Leading to a Chromophoric Intermediate
Jeremiah W. Hanes,† Kristin E. Burns,† David G. Hilmey,† Abhishek Chatterjee,†
Pieter C. Dorrestein,‡ and Tadhg P. Begley*,†
Department of Chemistry and Chemical Biology, 120 Baker Laboratory, Cornell UniVersity,
Ithaca, New York 14853, and Skaggs School of Pharmacy and Pharmaceutical Sciences,
UniVersity of California at San Diego, La Jolla, California 92093
Received September 1, 2007; E-mail: tpb2@cornell.edu
Abstract: Two routes for the de novo biosynthesis of pyridoxal-5′-phosphate (PLP) have been discovered
and reconstituted in vitro. The most common pathway that organisms use is dependent upon the activity
of just two enzymes, known as Pdx1 (YaaD) and Pdx2 (YaaE) in bacteria. Pdx2 has been shown to have
glutaminase activity and most likely channels ammonia to the active site of the PLP synthase subunit,
Pdx1, where ribose-5-phosphate (R5P), glyceraldehyde-3-phosphate (G3P), and ammonia are condensed
in a complex series of reactions. In this report we investigated the early steps in the mechanism of PLP
formation. Under pre-steady-state conditions, a chromophoric intermediate (I320) is observed that accumulates
upon addition of only two of the substrates, R5P and glutamine. The intermediate is covalently bound to
the protein. We synthesized C5 monodeuterio (pro-R, pro-S) and dideuterio R5P and showed that there is
a primary kinetic isotope effect on the formation of this intermediate using the pro-R but not the pro-S
labeled isomer. Furthermore, it was shown that the phosphate unit of R5P is eliminated rather than
hydrolyzed in route to intermediate formation and also that there is an observed C5-deuterium kinetic isotope
effect on this elimination step. Interestingly, it was observed that the formation of the intermediate could be
triggered in the absence of Pdx2 by the addition of high concentrations of NH4Cl to a preincubated solution
of Pdx1 and R5P. The formation of I320 was also monitored using high-resolution electrospray ionization
Fourier transform mass spectrometry and revealed a species of mass 34 304 Da (Pdx1 + 95 Da). These
results allow us to narrow the mechanistic possibilities for the complex series of reactions involved in PLP
formation.
Introduction
This pathway is found in a relatively small number of bacteria.
Another more prevalent route for PLP biosynthesis (R5P
Pyridoxal-5′-phosphate (PLP; 1) is the active cofactor form
of Vitamin B6 and is essential for primary metabolism in all
known organisms. The cofactor participates in amino acid and
carbohydrate metabolism and has recently been implicated in
singlet oxygen resistance.1 More than 140 different PLP-
dependent enzymes have been identified, and it has been
estimated that 1.5% of the genes in a typical bacterium encode
PLP-dependent enzymes.2 Two de noVo pathways for the
biosynthesis of PLP are currently known to exist. The system
employed in E. coli is the most thoroughly characterized,
requires six enzymes, and uses 1-deoxy-D-xylulose-5-phosphate
and 4-hydroxy-L-threonine as direct precursors (DXP pathway).3-9
pathway) is present in prokaryotes, eukaryotes, and archaea and
involves only two gene products collectively termed Pdx1/Pdx2,
SNZ/SNO, or YaaD/YaaE, depending on the organism of
origin.10-12 The R5P PLP biosynthesis pathway from Bacillus
subtilis has recently been reconstituted in Vitro and shown to
use ribose-5-phosphate (R5P; 2), glyceraldehyde-3-phosphate
(G3P; 3) and glutamine as the substrates (Scheme 1).13,14
The PLP synthase holoenzyme has been structurally charac-
terized and consists of two subunits: a glutaminase (Pdx2 or
YaaE) and a PLP synthase (Pdx1 or YaaD). A highly chro-
mophoric intermediate, with a λmax of approximately 315 nm,
accumulates in the active site of Pdx1 upon addition of
glutamine and R5P. This intermediate is converted to PLP in
† Cornell University.
‡ University of California at San Diego.
(1) Daub, M. E.; Ehrenshaft, M. Annu. ReV. Phytopathol. 2000, 38, 461-490.
(2) Percudani, R.; Peracchi, A. EMBO Rep. 2003, 4, 850-854.
(3) Banks, J.; Cane, D. E. Bioorg. Med. Chem. Lett. 2004, 14, 1633-1636.
(4) Cane, D. E.; Du, S.; Robinson, J. K.; Hsiung, Y.; Spenser, I. D. J. Am.
Chem. Soc. 1999, 121, 7722-7723.
(8) Yeh, J. I.; Du, S.; Pohl, E.; Cane, D. E. Biochemistry 2002, 41, 11649-
11657.
(9) Zhao, G.; Winkler, M. E. J. Bacteriol. 1995, 177, 883-891.
(10) Belitsky, B. R. J. Bacteriol. 2004, 186, 1191-1196.
(11) Ehrenshaft, M.; Bilski, P.; Li, M. Y.; Chignell, C. F.; Daub, M. E. Proc.
Natl. Acad. Sci. U.S.A 1999, 96, 9374-9378.
(5) Cane, D. E.; Du, S.; Spenser, I. D. J. Am. Chem. Soc. 2000, 122, 4213-
4214.
(12) Mittenhuber, G. J. Mol. Microbiol. Biotechnol. 2001, 3, 1-20.
(13) Burns, K. E.; Xiang, Y.; Kinsland, C. L.; McLafferty, F. W.; Begley, T. P.
J. Am. Chem. Soc. 2005, 127, 3682-3683.
(6) Garrido-Franco, M.; Laber, B.; Huber, R.; Clausen, T. J. Mol. Biol. 2002,
321, 601-612.
(7) Sivaraman, J.; Li, Y.; Banks, J.; Cane, D. E.; Matte, A.; Cygler, M. J.
Biol. Chem. 2003, 278, 43682-43690.
(14) Raschle, T.; Amrhein, N.; Fitzpatrick, T. B. J. Biol. Chem. 2005, 280,
32291-32300.
9
10.1021/ja076604l CCC: $40.75 © 2008 American Chemical Society
J. AM. CHEM. SOC. 2008, 130, 3043-3052
3043

A R T I C L E S
Hanes et al.
Scheme 1. Substrates Used by Pdx1 and Pdx2 in the Formation
of PLP
over CaH2 when noted as dry. Proton and carbon NMR were recorded
on a Bruker ARX-300.
Preparation of Deuteroribose Derivative 9 (Scheme 2). (-)-
Isoborneol-1-2H (1.6 g, 10.5 mmol)19 was dissolved in dry ether (4
mL) and added to a previously prepared solution of n-butylmagnesium
bromide [n-butyl bromide (1.2 mL, 11.4 mmol) in dry ether (4 mL)
was added slowly to Mg in ether (4 mL) followed by 1 h of reflux].
After 30 min, dry benzene (13 mL) was added and the ether was
distilled through a vigreux column under argon. Aldehyde 816 (195
mg, 0.70 mmol) was dissolved in dry benzene (2.5 mL) and added to
the prepared reagent. After refluxing for 3 h, the solution was cooled
and treated with 0.1 N HCl (∼30 mL) and dichloromethane (30 mL).
The layers were separated, and the organic phase was washed with
saturated NaHCO3 and brine. After drying over MgSO4, evaporation
of solvent was followed by column chromatography with 25% EtOAc
in hexanes to provide the alcohol 9 (109 mg, 55% yield) as an oil; 1H
NMR (300 MHz, CDCl3) δ 7.40-7.26 (m, 5H), 5.72 (d, J ) 3.7 Hz,
1H), 4.76 (d, J ) 11.9 Hz, 1H), 4.58 (d, J ) 11.9 Hz, 1H), 4.59-4.55
(m, 1H), 4.10 (dd, J ) 3.1, 9.1 Hz, 1H), 3.84 (dd, J ) 4.3, 9.1 Hz,
1H), 3.60 (dd, J ) 2.5, 8.4 Hz, 1H), 1.83 (d, J ) 8.6 Hz, 1H), 1.59 (s,
3H), 1.36 (s, 3H); 13C NMR (75 MHz, CDCl3) δ 137.5, 128.4, 128.0,
127.9, 113.0, 104.0, 78.6, 77.4, 76.5, 72.3, 60.2 (t, J ) 21.5 Hz), 26.8,
26.4.
the presence of G3P15 and is covalently attached to the protein
via Lys 81.16 MS analysis of the labeled peptide indicated a
mass increase of 95 Da which is consistent with loss of water
and phosphate from a Ru5P-Pdx1 adduct. Here we provide
additional information regarding this intermediate that enables
us to further narrow down the mechanistic possibilities for PLP
formation. Specifically, we are able to monitor the production
of the intermediate under mild conditions using ESI-FTMS,
observe a primary deuterium kinetic isotope effect at C5 of R5P
on the formation of the intermediate, and determine that this
isotope effect is due to removal of the pro-R proton. We also
demonstrate that ammonia is covalently incorporated into the
intermediate and that G3P is a much better substrate than
dihydroxyacetone-phosphate (DHAP). Additionally, we show
that phosphate loss occurs by an elimination reaction rather than
by a hydrolysis reaction and that this occurs after C5 deproto-
nation and ammonia addition.
Conversion of Alcohol 9 into (5S)-[5-2H1]-Ribose 4. The alcohol
9 (32 mg, 0.11 mmol) was dissolved in AcOH (2 mL) and treated with
10% Pd/C (10 mg). Stirring was maintained under a balloon of H2 for
3 h, and the reaction mixture was then filtered through Celite.
Evaporation of the solvent followed by column chromatography on
silica gel (3% MeOH in CHCl3) afforded the diol 10 (21.5 mg, 99%)
1
as an oil; H NMR (300 MHz, CDCl3) δ 5.80 (d, J ) 3.8 Hz, 1H),
4.57 (t, J ) 4.7 Hz, 1H), 4.05-3.95 (m, 1H), 3.82 (dd, J ) 3.8, 8.9
Hz, 1H), 3.75-3.68 (m, 1H), 2.60 (d, J ) 10.4 Hz, 1H), 2.24 (br s,
1H), 1.56 (s, 3H), 1.36 (s, 3H); 13C NMR (75 MHz, CDCl3) δ 112.7,
103.9, 80.5, 78.7, 70.8, 60.4 (t, J ) 24.2 Hz), 26.48, 26.45.
The diol 10 (21 mg, 0.11 mmol) was taken up in H2O (3 mL) and
treated with Dowex-50W resin (25 mg previously washed with 10%
HCl, water, EtOH, and diethyl ether) at 50 °C. After 24 h, the resin
was filtered off, NH4OH was added, and the filtrate was lyophilized to
provide (5S)-[5-2H1]-ribose 4 (16 mg, 96%) as an oily solid; 1H NMR
(300 MHz, D2O) (chemical shifts for R- and â-pyranose and furanose)
δ 5.38-5.20 and 4.90-4.80 (m, 1H), 4.20-4.07 (m, 1H), 3.89-3.78
(m, 2H), 3.70-3.47 (m, 1H); 13C NMR (75 MHz, D2O) (chemical shifts
for R- and â-pyranose and furanose) δ 101.2, 96.0, 94.0, 93.8, 83.5,
83.2, 75.3, 71.3, 70.7, 70.3, 69.5, 69.3, 67.6, 67.4, 63.3 (t, J ) 23.4
Hz).
Experimental Section
Protein Overexpression and Purification. Overexpression and
purification were performed as previously published13 except for the
following changes. A 1 mL HisTrap HP column (GE Healthcare Bio-
Sciences Corp. Piscataway, NJ) was used according to the product
literature. Subsequent to elution of Pdx1 and Pdx2 from the column
the protein was buffer exchanged into the following buffer system:
50 mM Tris (Trizma base) pH 8.0 @ RT, 100 mM NaCl, 2 mM TCEP,
and 10% glycerol using an Econo-Pac 10DG desalting column (Bio-
Rad Laboratories, Hercules, CA). A portion of Pdx1 and Pdx2 were
combined at this time and used for experiments where both were
necessary. The two proteins were mixed at a 1:1.5 molar ratio (Pdx1:
Pdx2) according to the Coomassie Plus Assay Reagent (Pierce
Biotechnology, Rockford, IL). To free the enzyme from the bound Ru5P
adduct that copurifies with Pdx1, 10 mM glutamine and 1 mM G3P
were added and allowed to incubate overnight at 4 °C. For samples
containing only Pdx1, 50 mM NH4Cl was added in place of the
glutamine. After the overnight incubation, the samples were again
purified using the HisTrap procedure as detailed above, and subsequent
to elution the proteins were buffer exchanged into 50 mM HEPES pH
7.5 @ RT, 100 mM NaCl, 2 mM TCEP, and 20% glycerol. 30 µL
aliquots of protein were flash frozen using liquid nitrogen and stored
at -80 °C. Control reactions demonstrated that this method of enzyme
storage did not result in any activity loss.
Inversion of Configuration at the 5-Position of 9 (Scheme 3).
Alcohol 9 (47 mg, 0.17 mmol) was dissolved in THF (5 mL) followed
by addition of PPh3 (175 mg, 0.67 mmol) and p-nitrobenzoic acid (112
mg, 0.67 mmol). After cooling to 0 °C, DEAD (116 mg, 0.67 mmol)
was added dropwise and the solution was slowly warmed to room
temperature. After 4 h, the solution was evaporated to an oil and
chromatographed on silica with 10% EtOAc in hexanes to provide the
p-nitrobenzoate ester 11 (72 mg).
The ester (72 mg, 0.17 mmol) was then dissolved in MeOH (3 mL)
and treated with K2CO3 at room temperature. Evaporation of solvent
followed by chromatography on silica gel (30% EtOAc in hexanes)
provided the inverted alcohol 12 (46 mg, 99% over two steps); 1H NMR
(300 MHz, CDCl3) δ 7.40-7.31 (m, 5H), 5.71 (d, J ) 3.6 Hz, 1H),
4.75 (d, J ) 11.9 Hz, 1H), 4.58 (d, J ) 11.9 Hz, 1H), 4.56-4.52 (m,
1H), 4.10 (dd, J ) 2.5, 9.1 Hz, 1H), 3.90-3.88 (m, 1H), 3.83 (dd, J )
4.3, 9.1 Hz, 1H), 1.90 (br s, 1H), 1.59 (s, 3H), 1.36 (s, 3H); 13C NMR
Synthesis of (5S)-[5-2H1]- and (5R)-[5-2H1]-Ribose. The ribose
diastereomers 4 and 5 were prepared from the known alcohol 9 using
a modified procedure.17,18 All solvents were reagent grade and distilled
(17) Kundu, M. K.; Foldesi, A.; Chattopadhyaya, J. HelV. Chim. Acta 2003,
86, 633-642.
(18) Ono, A.; Ono, A.; Kainosho, M. Tetrahedron Lett. 1997, 38, 395-398.
(19) Crich, D.; Neelamkavil, S. PCT/US2002/019274(WO/2003/002526), 41 pp.
2003. Ref Type: Patent.
(15) Burns, K. E. Dissertation, 2006, Cornell University. Ref Type: Thesis/
Dissertation.
(16) Raschle, T.; Arigoni, D.; Brunisholz, R.; Rechsteiner, H.; Amrhein, N.;
Fitzpatrick, T. B. J. Biol. Chem. 2007, 282, 6098-6105.
9
3044 J. AM. CHEM. SOC. VOL. 130, NO. 10, 2008

Studies on the Mechanism of PLP Synthase
A R T I C L E S
Scheme 2. Synthesis of (5S)-[5-2H1]-Ribose Diastereomer 4
Scheme 3. Synthesis of (5R)-[5-2H1]-Ribose Diastereomer 5
(75 MHz, CDCl3) δ 137.5, 128.4, 128.0, 127.9, 113.0, 104.0, 78.6,
77.5, 76.5, 72.3, 60.1 (t, J ) 21.2 Hz), 26.7, 26.4.
glycerol using an Econo-Pac 10DG desalting column and flash frozen
by submersion in liquid nitrogen in 80 µL aliquots. To carry out the
phosphorylation reaction of the various C5-deuterium labeled ribose
preparations, a total reaction volume of 5 mL was used according to
the following conditions: 4 mg of E. coli ribokinase, 35 mM ribose,
45 mM ATP, 50 mM Tris, 100 mM KCl, and 2 mM TCEP. Prior to
the addition of the kinase, the reaction mixture was adjusted to a pH
of 7.2 using NaOH. The reaction was followed by TLC using
polyethyleneimine-cellulose (PEI-cellulose-F) plates (EM Science,
Gibbstown, NJ) run with 0.2 M KH2PO4 and visualized by p-
anisaldehyde staining. The reaction was complete after 1 h at which
time the protein was removed using an Amicon Ultra-15, 10 kDa cutoff
centrifugal concentrator (Millipore, Billerica, MA). The flow through
was diluted to 50 mL using ddH2O and loaded onto a 20 mL HiPrep
16/10 Q XL anion exchange column (GE Healthcare, Uppsala Sweden)
equilibrated with ammonium acetate pH 7. A gradient from 20 mM to
1 M ammonium acetate was run over 10 column volumes at a rate of
4 mL/min. The fractions containing R5P were pooled and lyophilized
to constant weight and resuspended in ddH2O giving a stock concentra-
tion of 50 mM.
Conversion of 5R-Alcohol 12 into (5R)-[5-2H1]-Ribose 5. The
alcohol 12 (46 mg, 0.16 mmol) was converted, as described for 9,
1
providing the diol 13 (30.5 mg, 98%) as an oil; H NMR (300 MHz,
CDCl3) δ 5.80 (d, J ) 3.8 Hz, 1H), 4.58 (t, J ) 4.4 Hz, 1H), 4.09-
3.95 (m, 1H), 3.92 (br s, 1H), 3.82 (dd, J ) 2.7, 8.9 Hz, 1H), 2.55 (d,
J ) 10.4 Hz, 1H), 2.14 (br s, 1H), 1.56 (s, 3H), 1.37 (s, 3H); 13C NMR
(75 MHz, CDCl3) δ 112.8, 104.0, 80.5, 78.7, 70.8, 60.4 (t, J ) 21.4
Hz), 26.5, 26.46.
The diol 13 (25 mg, 0.13 mmol) was treated, as described for 10, to
1
give 19 mg (96%) of (5R)-[5-2H1]-ribose 5 as an oily solid; H NMR
(300 MHz, D2O) (chemical shifts for R- and â-pyranose and furanose)
δ 5.41-5.20 and 5.00-4.80 (m, 1H), 4.24-4.09 (m, 1H), 4.00-3.75
(m, 2H), 3.73-3.45 (m, 1H); 13C NMR (75 MHz, D2O) (chemical shifts
for R- and â-pyranose and furanose) δ 101.2, 94.1, 93.8, 83.5, 83.2,
75.3, 71.3, 70.7, 70.3, 69.6, 69.4, 67.6, 67.4, 63.3 (t, J ) 23.4 Hz).
Enzymatic Synthesis and Purification of C5-Deuterium Labeled
R5P. The E. coli ribokinase overexpression vector20,21 was a gift from
the S.L. Mowbray Laboratory. The enzyme was purified on a 1 mL
HisTrap HP column using the following protocol. The overexpression
strain was E. coli BL21, and overexpression was performed on 2 L of
culture grown in LB media to an OD600 of 0.6 at which time the culture
was induced by the addition of 1 mM isopropyl â-D-1-thiogalactopy-
ranoside. After growth overnight at 15 °C, the cells were pelleted by
centrifugation at 4 °C and then resuspended in the following buffer
for lysis by sonication: 50 mM Na+-phosphate pH 8.0, 300 mM NaCl,
2 mM TCEP, and 10 mM imidazole. The cell debris was removed by
centrifugation at 39 800 × g for 45 min at 4 °C. The lysate was passed
through the HisTrap HP column and washed with 50 mL of 50 mM
Na+-phosphate buffer pH 8.0 containing 300 mM NaCl, 2 mM TCEP,
and 20 mM imidazole and eluted with the same buffer but with 300
mM imidazole. Following elution the protein was buffer exchanged
into 50 mM Tris pH 8.0 @ RT, 100 mM NaCl, 2 mM TCEP, and 20%
Intermediate Formation. Intermediate formation was measured
under the following conditions: 4 µM Pdx1/6 µM Pdx2, 2 mM R5P
(or deuterium labeled R5P), 2 mM TCEP, 50 mM HEPES pH 7.6 @
RT, and 100 mM NaCl. R5P was preincubated with the enzyme for 15
min at room temperature prior to the initiation of the reaction upon
addition of a final concentration of 15 mM glutamine. The absorbance
was scanned or simply monitored at 320 nm using a Hitachi (Berkshire,
UK) U-2010 UV/visible spectrophotometer. The concentrations of both
glutamine and R5P were sufficiently high that the observed rate was
essentially saturated. The reactions were performed in triplicate and
averaged for analysis. Approximately 14 s of the reactions were missed
during mixing and sample manipulation, therefore the data were shifted
accordingly along the x-axis. The data were analyzed by nonlinear
regression using the following double exponential equation:
(20) Andersson, C. E.; Mowbray, S. L. J. Mol. Biol. 2002, 315, 409-419.
(21) Sigrell, J. A.; Cameron, A. D.; Jones, T. A.; Mowbray, S. L. Protein Sci.
1997, 6, 2474-2476.
∆Abs320 ) A1e-λ t + A2e-λ t + C
(eq 1)
1
2
9
J. AM. CHEM. SOC. VOL. 130, NO. 10, 2008 3045

A R T I C L E S
Hanes et al.
The parameters A1 and A2, λ1 and λ2 correspond to the amplitudes
and observed rates of the fast and slow phases of the reaction,
respectively. The term C is the offset. Data analysis and plotting were
performed in the computer program GraFit 5 (Erithacus Software,
Surrey, UK).
and 20% glycerol) was diluted with 270 µL of water, followed by 400
µL of 250 mM HEPES buffer (pH 7.6, containing 500 mM NaCl). 27
µL of 50 mM R5P were then added, and the reaction mixture was
incubated for 15 min at room temperature. 215 µL of 1 M NH4Cl were
then added to initiate chromophore formation. Two parallel control
reactions were set up where the NH4Cl or both the NH4Cl and the R5P
were substituted with an equivalent amount of water and were treated
identically.
After allowing the reaction mixtures to incubate at room temperature
for 80 min, 100 µL aliquots from each of them were quenched with
100 µL of 20% formic acid. Protein from the quenched reaction
mixtures was desalted into 10 µL of 78% acetonitrile containing 0.1%
TFA using C18 Zip Tip following the manufacturer’s protocol. This
was diluted to 50 µL with 50% methanol containing 0.1% formic acid,
and the resulting solution was used for ESI-FTMS analysis (Thermo-
Finnigan LTQ-FT). The resulting data were processed with the computer
program Extract.
PLP Synthase Selectivity for G3P over DHAP. A steady state
assay was performed under the following conditions using Pdx1 and
Pdx2 which were isolated using the improved purification procedure
as detailed in the “protein overexpression and purification section”: 4
µM Pdx1, 6 µM Pdx2, 5 mM R5P, 10 mM glutamine in a buffer
composed of 50 mM HEPES pH 7.6 @ RT, 100 mM NaCl, and 2 mM
TCEP. The reactions were started by the addition of 5 mM DHAP or
1 mM G3P, and the UV-visible absorbance spectrum was monitored
over time. The absorbance was converted to the concentration of PLP
using an extinction coefficient of 6300 M-1 cm-1 at a wavelength of
390 nm.
31P NMR Analysis of I320 Formation. Purified Pdx1 (1.5 mM) was
incubated with R5P (1.5 mM) in a total volume of 1 mL for 2 h at
which time the protein was buffer exchanged into 50 mM HEPES pH
7.5, 200 mM NaCl, 1 mM TCEP, and 2 mM MgCl2. This solution was
then diluted 2× with H218O containing 1.2 M NH4Cl to initiate I320
formation. The reaction was allowed to incubate overnight at room
temperature at which time D2O was added to a final concentration of
10% for NMR analysis. In order to catalyze the incorporation of 18O
into phosphate to generate a reference sample, a final concentration of
10 nM (by Bradford assay) of yeast inorganic pyrophosphatase (Sigma-
Aldrich) was added and the reaction mixture was incubated for an
additional 48 h at room temperature. 31P spectra were recorded on a
Varian INOVA 600 spectrometer operating at 242.785 MHz for 31P
observation using a 10 mm broadband direct observe probe head.
Spectra were acquired observing a 31P chemical shift range from 50 to
-100 ppm with an acquisition time of 1.6 s and relaxation delay of
1
0.2 s using 90° pulses and broadband H decoupling. Approximately
3500 scans were averaged for a total acquisition time of approximately
1.75 h. Spectra were zero filled to 256K complex points, and an
unshifted Gaussian window was applied prior to Fourier transformation.
31P chemical shifts were referenced to 85% phosphoric acid as an
external standard.
Kinetics of Phosphate Elimination and Release. The kinetics of
phosphate release were monitored by two different methods. First was
by using the malachite green assay in a similar fashion to that reported
elsewhere.22 This assay is based on the formation, under acidic
conditions (∼1 M HCl), of a malachite green/molybdate/phosphate
complex that absorbs at 660 nm. The reaction mixture consisted of 20
µM Pdx1, 30 µM Pdx2, 2 mM R5P in a buffer composed of 50 mM
HEPES pH 7.6 @ RT, 100 mM NaCl, and 2 mM TCEP. The enzyme
was preincubated with R5P for 15 min prior to the addition of glutamine
to a final concentration of 15 mM to start the reaction. The second
method for measuring phosphate release was carried out using the
EnzChek Phosphate Assay Kit from Invitrogen (Carlsbad, California).
This assay is based on the formation of the chromophoric 2-amino-6-
mercapto-7-methylpurine (λmax ) 360 nm) from 2-amino-6-mercapto-
7-methylpurine riboside (λmax ) 330 nm) by the purine nucleoside
phosphorylase catalyzed substitution with phosphate. The conditions
for this assay were as follows: 8 µM Pdx1, 12 µM Pdx2, 2 mM R5P
in addition to the assay components according to the manufacturer’s
instructions. This mixture was preincubated for 15 min at room
temperature, and the reaction was initiated by the addition of a final
concentration of 20 mM glutamine.
Synthesis and Assay of 2′-Hydroxypyridoxol-5′-phosphate. 2′-
Hydroxypyridoxol was synthesized as previously described.23-25 2′-
Hydroxypyridoxol-5′-phosphate was synthesized with PdxK and pu-
rified by HPLC (LC-18-T column, 15 cm × 4.6 mm, 3 µm, isocratic
elution with 25 mM NH4OAc, pH 7). To test 2′-hydroxypyridoxol-5′-
phosphate as a substrate for Pdx1, Pdx1 (50 µM) and Pdx2 (75 µM)
were mixed with 330 µM 2′-hydroxypyridoxol-5′-phosphate in 25 mM
Tris, pH 8, 60 µL total. After 16 h at 37 °C, the protein was removed
with a Microcon and the sample was analyzed by HPLC (same column
as above with the following solvent system: A, water; B, 100 mM
KH2PO4 pH 6.6; C, methanol, 0-6 min, 100% B; 7 min, 10% C; 20
min, 30% C, 30% A; 22 min, 100% B).
Results and Discussion
Observation of a Chromophoric Species in the Active Site
of Pdx1. When glutamine is added to a preincubated solution
containing Pdx1 (4 µM), Pdx2 (6 µM), and R5P, a chromophoric
species (I320) accumulates in the active site (Figure 1). While
this work was in progress, the formation and initial characteriza-
tion of this intermediate was reported.16 The intermediate is
covalently bound to Lys 81 and has an extinction coefficient
of approximately 16 200 M-1 cm-1 at 315-320 nm.16 MS
analysis of an I320-containing tryptic peptide suggested that it
is formed by loss of water and phosphate from a protein bound
Ru5P adduct and that the glutamine derived ammonia is not
covalently incorporated. Addition of G3P results in its conver-
sion to PLP.
Formation of the Chromophoric Intermediate Using NH4Cl.
Reactions to form the chromophoric intermediate were carried out in
the absence of the glutaminase subunit (Pdx2). Absorbance scans were
taken under the following conditions: 20 µM Pdx1, 2 mM R5P, 2 mM
TCEP in 50 mM HEPES pH 7.6 @ RT, with 100 mM NaCl. The
reaction was initiated by the addition of solid NH4Cl to a final
concentration of 1 M, and absorbance scans were taken intermittently.
To characterize the concentration dependence of the ammonia induced
production of the enzyme bound intermediate, absorbance measurements
were taken in a 96-well plate format (Greiner plate model 655801)
purchased from Omega Scientific (Tarzana, CA) in a Synergy HT Multi-
Detection Microplate Reader (BioTek Instruments, Winooski, VT).
Isotope Effect on I320 Formation. The formation of I320
requires multiple elimination reactions to form a conjugated
molecule from R5P. This means that there are also multiple
deprotonation events leading up to I320 and suggests the
possibility of a mechanistically informative deuterium isotope
ESI-FTMS Analysis. The chromophoric intermediate was generated
in the following manner. 25 µL of 538 µM Pdx1 stock solution (in 50
mM HEPES buffer pH 7.6 containing 100 mM NaCl, 2 mM TCEP
(23) Korytnyk, W.; Srivastava, S. C.; Angelino, N.; Potti, P. G.; Paul, B. J.
Med. Chem. 1973, 16, 1096-1101.
(24) Pocker, A. J. Org. Chem. 1973, 38, 4295-4299.
(25) Underwood, G.; Paul, B.; Becker, M. J. Heterocycl. Chem. 1976, 13, 1229-
1232.
(22) Sharma, K.; Gupta, M.; Krupa, A.; Srinivasan, N.; Singh, Y. FEBS J. 2006,
273, 2711-2721.
9
3046 J. AM. CHEM. SOC. VOL. 130, NO. 10, 2008

Studies on the Mechanism of PLP Synthase
A R T I C L E S
Figure 1. Formation of I320 in the active site of Pdx1. Pdx1 was
preincubated with Pdx2 and a saturating concentration of R5P. Glutamine
was added to initiate the reaction, and UV-absorbance scans were taken
over a period of 200 s. No G3P was present in solution.
Figure 2. C5-deuterium kinetic isotope effect on I320 formation. The time
dependence of I320 formation was monitored using the natural pentose (R5P,
denoted “H2”) substrate, the C5-dideuterium labeled R5P analogue (denoted
“D2”), the specifically pro-S-D (denoted “pro-S-D”) labeled R5P analogue,
or the specifically pro-R-D (denoted “pro-R-D”) labeled R5P analogue.
effect on the kinetics of its formation. Previous structural and
biochemical data showed that Pdx1 has pentose-5-phosphate
isomerase activity and that ribulose-5-phosphate (Ru5P) forms
an adduct with an active site lysine.13,26 Therefore, given the
information that I320 seemed to be triggered by ammonia addition
to the ribulose-5-phosphate adduct, the isotope effects on I320
formation were measured starting with Pdx1 already loaded with
Ru5P to circumvent any possible rate-limiting steps involved
in isomerization and imine formation that would attenuate the
isotope effect. However, before an accurate deuterium kinetic
isotope effect could be measured it was necessary to free the
active site of the unlabeled Ru5P that is present following the
overexpression and purification. This was accomplished by first
purifying Pdx1 and Pdx2, combining them in a ratio of 1:1.5,
and carrying out a single turnover reaction by adding glutamine
and G3P. This converted the bound Ru5P to PLP. The resulting
protein was subjected to an additional round of purification to
give enzyme with no detectable bound Ru5P.
using the C5-2H2 labeled R5P (labeled “D2”) as a substrate and
the data were again biphasic with fast and slow phase rates of
0.0204 ( 0.0003 s-1 and 0.0147 ( 0.0001 s-1. The isotope
effect was apparent on both phases of the reaction and was
nearly identical in magnitude (2.45 and 2.41). The observation
of a primary deuterium isotope effect of greater than 2.4 for
deprotonation at C5 is strong evidence that the removal of one
of the protons on C5 is partially rate limiting in the formation
of I320 from the Pdx1-Ru5P adduct. The fact that the data show
a clear lag phase is indicative that there are at least two reactions
occurring in sequence at comparable rates prior to I320 produc-
tion. It is not clear, at this time, what the identity of the other
event is that limits I320 formation, but it is important to report
that the lag phase cannot be altered by increasing the concentra-
tion of glutamine or R5P and is therefore likely to be a first-
order reaction following substrate binding. Similar experiments
were carried out using [3-2H1]-ribose and [4-2H1]-ribose. No
isotope effects were observed for these isotopomers.
To obtain substrate to measure the KIE at C5, commercially
available C5 dideuterium labeled ribose ([5-2H2]-ribose) was
enzymatically phosphorylated using recombinant E. coli riboki-
nase and ATP20,21 and the product was purified using anion
exchange chromatography. The rate of formation of I320 from
the Pdx1-[5-2H2]-Ru5P adduct was then compared to the rate
of formation of I320 from the Pdx1-Ru5P adduct under pre-
steady-state conditions. For this measurement, Pdx1 and Pdx2
were present at a molar ratio of 1:1.5, in which Pdx1 was at a
concentration of 4 µM. The natural (C5-H2) and labeled (C5-
2H2) Ru5P substrates were loaded into the active site by
preincubating the corresponding R5P with the enzyme for 15
min prior to the addition of the glutamine. The reactions were
initiated with a final concentration of 15 mM glutamine, and
the results are shown in Figure 2 as an increase in absorbance
at 320 nm plotted over time. The data, which were collected in
triplicate, exhibited biphasic behavior and were therefore
analyzed by nonlinear regression using a double exponential
equation (eq 1) to extract the observed rates for the two phases.
The fast and slow phases using the natural pentose (labeled “H2”
on the plot) occurred at 0.050 ( 0.001 s-1 and 0.0354 ( 0.0002
s-1, respectively. Simply a visual inspection of the plot reveals
that there is a surprisingly large KIE for I320 formation when
Stereochemistry of the C5 Deprotonation. The com-
mercially available C5 dideuterium labeled ribose did not allow
us to assign a specific stereochemistry to the deprotonation
event. To obtain this information, syntheses of the two singly
labeled R5P isomers were carried out.
Synthesis of (5S)-[5-2H1]- and (5R)-[5-2H1]-Ribose. The
syntheses of (5S)-[5-2H1]- and (5R)-[5-2H1]-ribose utilize the
known deuterium-labeled ribose derivative 9 (Scheme 2).18 The
absolute stereochemistry of this structure has been previously
established by conversion to the relevant thymidine derivatives
followed by 2D NMR analysis.18,27
The 5S-diastereotopically enriched alcohol 9 was prepared
in several steps from the commercially available allofuranose
6.17 Straightforward hydrogenolysis followed by acetonide
removal resulted in ribose 4. Based on 1H NMR data, the
configuration at C5 was retained during the final two steps.
Access to the 5R-stereochemistry was realized by inversion of
9 using Mitsunobu conditions followed by ester hydrolysis to
afford the alcohol 11 in quantitative yield (Scheme 3). The same
deprotection sequence resulted in the (5R)-[5-2H1]-ribose 5 in
excellent yield.
(26) Zein, F.; Zhang, Y.; Kang, Y. N.; Burns, K.; Begley, T. P.; Ealick, S. E.
(27) Tate, S.; Kubo, Y.; Ono, A.; Kainosho, M. J. Am. Chem. Soc. 1995, 117,
7277-7278.
Biochemistry 2006, 45, 14609-14620.
9
J. AM. CHEM. SOC. VOL. 130, NO. 10, 2008 3047

A R T I C L E S
Hanes et al.
Figure 4. Time dependence of Pi elimination. Phosphate elimination during
I320 formation was monitored over time by the malachite green method
(b). The solid and dashed lines were data obtained using a continuous assay
based upon an increase in absorbance at 360 nm as described in the main
text. The solid line represents a reaction in which the natural substrate,
R5P, was used (denoted “H2”), and the dashed line represents a reaction in
which the R5P was dideuterium labeled at position 5 (denoted “D2”).
Figure 3. 31P NMR analysis of phosphate release. (A) 31P spectrum of a
reaction mixture in which I320 was prepared stoichiometrically with Pdx1
in buffer containing 50% H218O. The signal denoted “16O4” refers to
inorganic phosphate with natural abundance oxygen bound in all four
positions. (B) To generate a reference sample of 18O-phosphate yeast
inorganic pyrophosphatase was added to the reaction mixture described in
(A) in order to catalyze the incorporation of 18O into the released phosphate.
The new peaks are denoted with the stoichiometric ratio of 16O to 18O
bonded to the phosphorus atom of Pi.
before C5 deprotonation, there should be no large (or primary)
isotope effect on the rate of phosphate formation while elimina-
tion after deprotonation should exhibit a kinetic isotope effect.
Two separate approaches were taken to determine the kinetics
of phosphate elimination. One was to perform a pre-steady-
state reaction under similar conditions to those described above
for measuring the kinetics of I320 formation. However, instead
of monitoring the absorbance at 320 nm, the reaction was
quenched at specific times and analyzed for phosphate using
the malachite green assay22 to determine the time course for
phosphate elimination. Because this assay involves quenching
the reaction mixture with a solution containing a final concen-
tration of approximately 1 M HCl, the data represent the time
course for the C-O bond cleavage reaction. As shown in Figure
4 (filled circles) the elimination of phosphate essentially mirrors
the accumulation of I320 suggesting that loss of phosphate and
chromophore formation are limited by the same microscopic
rate constant or constants. Because of the difficulty in obtaining
accurate measurements for free Pi in reactions performed under
single turnover conditions using the malachite green assay, an
alternative assay was developed. This alternative assay is based
on the phosphorolysis of a purine nucleoside analog by purine
nucleoside phosphorylase. The measurement of an increase in
absorbance at 360 nm is an indicator of the liberation of
inorganic phosphate from Pdx1 as the reaction occurs. The data
denoted “H2” were obtained under similar conditions as those
detailed for the malachite green assay except that Pi liberation
was measured in real time in the spectrophotometer (solid line
in Figure 4). The absorbance increase mirrors the data obtained
using the malachite green assay.
The stereochemistry of the C5 deprotonation was determined
using the deuterium KIE seen on the rate of formation of I320
.
Shown in Figure 2 are the experiments performed with the
stereospecifically labeled R5P derivatives (labeled “pro-R-D”
and “pro-S-D” on the plot). In the experiment in which the pro-R
isomer was present, there was a clear KIE, whereas there was
not when using the pro-S isomer. Therefore, it can be unam-
biguously stated that the pro-R hydrogen is abstracted during
the formation of I320 and that this deprotonation is partially rate
limiting under these reaction conditions.
Phosphate is removed by elimination rather than hydroly-
sis during I320 formation. We know from previous studies that
the Pi of R5P is lost during the course of I320 formation;16
however, it is not known whether it is removed by elimination
or hydrolysis. To differentiate between these two possibilities
the intermediate was formed in the presence of 50% H218O and
31P NMR analysis of the Pi released was performed. If the Pi
group from R5P was removed by hydrolysis then 50% of the
Pi would have an upfield isotopic shift of approximately 0.02
ppm28 compared to that of the all 16O containing Pi. As shown
in Figure 3A, the 31P spectrum of the reaction mixture is
consistent with a single species. To illustrate that the resolution
was adequate to clearly distinguish 16O4 phosphate from 16O318O
phosphate under the conditions used in this experiment,
inorganic pyrophosphatase was added to the sample and 31P
NMR analysis was performed again (Figure 3B). Two new
signals emerged upon addition of pyrophosphatase whose
chemical shifts are consistent with Pi containing one and two
18O atoms. These data demonstrate that Pi is eliminated in route
to I320 formation.
One important new observation can be made from this data.
Because the malachite green assay is measuring the observed
rate of C-O bond cleavage and the optical assay includes the
additional step of Pi dissociation from the enzyme active site,
then it can be concluded that Pi dissociation must be relatively
rapid following the elimination step. Since the release of
phosphate occurs in concert with the formation of I320, it is likely
that Pi elimination follows C5 deprotonation. However, a more
rigorous test of this hypothesis would be to determine whether
there is a C5 deuterium KIE on Pi elimination. Therefore, this
experiment was repeated using (5-2H2)-R5P, and the results are
C5 deprotonation occurs prior to phosphate elimination.
To determine the relative order of phosphate elimination and
C5 deprotonation, we measured the deuterium isotope effect
on phosphate elimination. If phosphate elimination occurred
(28) Cohn, M.; Hu, A. Proc. Natl. Acad. Sci. U.S.A 1978, 75, 200-203.
9
3048 J. AM. CHEM. SOC. VOL. 130, NO. 10, 2008

Studies on the Mechanism of PLP Synthase
A R T I C L E S
Figure 5. ESI-FTMS analysis of Pdx1. (A) Native Pdx1, not treated with R5P or NH4Cl, (B) Pdx1 treated with R5P, and (C) Pdx1 treated with both R5P
and NH4Cl. Masses corresponding to free Pdx1 (34 209 Da), Pdx1 to which Ru5P is bound (34421 Da; ∆m ) +212), and adducts generated from Ru5P
bound Pdx1 after ammonia incorporation via (i) phosphate loss only (34322 Da; ∆m ) +113) and (ii) both phosphate and water loss (34304 Da; ∆m ) +95)
are marked with blue lines. The protocol used in desalting the sample preparation introduced TFA resulting in the observation of TFA adducts of all the
peaks (+114 Da). These adducts are indicated with red arrows. In the deconvoluted spectrum “A”, both the mono (34 323 Da) and the bis (34 437 Da) TFA
adducts of Pdx1 are observed. These adducts are not observed with the complete conversion of free Pdx1 to the Ru5P bound Pdx1, as shown in spectrum
“B”. Addition of NH4Cl results in the appearance of the two new peaks at 34 304 Da (∆m ) +95) and 34 322 Da (∆m ) +113) respectively, as shown in
spectrum “C”. The species with a mass of 34 322 Da overlaps with the mono TFA adduct of free Pdx1. However, the absence of free Pdx1 in spectrum “C”
rules out the emergence of this species due to the mono TFA adduct of free Pdx1. The peak observed at the position overlapping with Pdx1 to which Ru5P
is bound in “C” (observed mass 34 418 Da, slightly shifted from the line depicting the mass 34 421 Da) corresponds to the mono TFA adduct of the species
with a mass of 34 304 Da rather than Pdx1 to which Ru5P is bound.
shown in Figure 4 along with the data for the natural substrate.
There is clearly a KIE effect on phosphate elimination. This
provides strong evidence that Pi elimination follows C5 depro-
tonation in route to the chromophoric intermediate.
regarding the identity of I320. Previous data demonstrated that
a tryptic peptide with an extra 95 Da could be isolated from
the Pdx1-I320 complex.16 This suggested that I320 was formed
by loss of water and phosphate from the Pdx1-Ru5P adduct.
However, since this experiment involved extensive sample
manipulation, the possibility existed that the species detected
High-resolution ESI-FTMS analysis of I320 formation.
ESI-FTMS was performed to gain additional information
9
J. AM. CHEM. SOC. VOL. 130, NO. 10, 2008 3049

A R T I C L E S
Hanes et al.
Figure 7. G3P is greatly preferred over DHAP. Shown in the plot is the
steady-state accumulation of PLP using Pdx1 and Pdx2 at a ratio of 1:1.5.
The substrates G3P and DHAP were present at saturating concentrations.
The horizontal dotted line on the plot is to illustrate the concentration of
Pdx1 used in the assay and is therefore at a point which multiple turnovers
would be observed. When using DHAP as a substrate multiple turnovers
are not observed in the time scale examined.
with a hydration product of I320. Again, these data do not tell
us if ammonia has been incorporated.
Is the nitrogen atom of ammonia covalently incorporated
into I320? I320 can be formed by treating glutamine and R5P
with Pdx1/Pdx2 or by treating R5P and ammonia with Pdx1.15,16
This suggests that the nitrogen atom of ammonia might be
covalently attached to I320. However, MS characterization of
the I320 labeled tryptic peptide did not contain ammonia,16 and
the ESI-FTMS analysis described above did not differentiate
between -OH and -NH2, which differ by 1 Da.
Figure 6. Addition of ammonia triggers I320 formation. (A) A sample of
Pdx1 was preincubated with R5P and then NH4Cl was added as a solid to
a final concentration of 1 M. UV-absorbance scans were taken intermittently
over a period of approximately 10 min. (B) The concentration dependence
of I320 formation was characterized under similar conditions as in (A), except
that varying concentrations of NH4Cl were added to start the reaction (added
as a solution). Starting with the reaction represented by the curve with the
highest final amplitude to the one with the lowest final amplitude, the
concentrations of NH4Cl present were as follows: 320, 160, 80, 40, and
20 mM.
When Pdx2 and glutamine are used as the ammonia source,
the intermediate can be produced stoichiometrically with the
protein in the absence of G3P. The enzyme bound intermediate
can then be purified by gel filtration to remove all small
molecules, and subsequent addition of only G3P results in PLP
formation. At least two possible conclusions can be drawn from
this observation: either ammonia is incorporated into the
intermediate covalently or ammonia is tightly bound to Pdx2.
To differentiate between these possibilities, R5P was preincu-
bated with Pdx1 and NH4Cl was then added to start the reaction.
The results are shown in Figure 6A as UV-absorbance scans
taken over time. From the large increase in absorbance at 320
nm, it is clear that the addition of NH4Cl alone triggered I320
formation. Shown in Figure 6B is the concentration dependence
of this reaction showing that good reconstitution of I320 requires
exceedingly high concentrations of NH4Cl, illustrating that the
enzyme’s affinity for ammonia from solution is weak (not yet
saturated at concentrations up to 1 M NH4Cl). However, once
reconstituted this intermediate could not be separated from the
protein and could be converted to PLP by the addition of only
G3P following purification of the protein/I320 complex by gel
filtration into a solution lacking ammonia. It is highly improb-
able that NH4Cl induces the formation of I320 without covalent
incorporation because the ammonia remains bound to Pdx1
during the gel filtration. This should not occur in the absence
of a covalent bond because our analysis of the ammonia
concentration dependence of I320 formation demonstrates that
ammonia binds very weakly to Pdx1. Therefore, we conclude
that I320 contains covalently bound ammonia and that the
extensive sample preparation used in the reported MS analysis
generated a hydrolysis product.16
by MS was a decomposition product of I320. We therefore carried
out ESI-FTMS analysis on the intact Pdx1-I320 complex,
prepared by reacting Pdx1 with R5P and NH4Cl (see next section
for details). This strategy required only mild desalting of the
complex prior to MS analysis. The results are shown in
Figure 5.
Two informative species are seen in the Pdx1 sample. These
correspond to free Pdx1 (34209 ( 1 Da) and Pdx1 bound to an
adduct of Ru5P (34421 ( 1 Da) in a ratio of approximately
2:3, respectively (Figure 5A). The additional species with mass
values of 34 322 and 34 436 Da correspond to the mono- and
bistrifluoroacetic acid (+114 Da) adducts of free Pdx1. When
Pdx1 is treated with R5P, the Pdx1-Ru5P adduct (34 421 Da)
is the major species observed (Figure 5B). Treatment of the
Pdx1-Ru5P adduct with NH4Cl results in the emergence of
two new species with masses of 34 304 ( 1 Da and 34 322 (
1 Da, respectively, as well as their monotrifluoroacetic acid
adducts (Figure 5C). The species with a mass of 34 304 Da is
consistent with an intermediate generated from the Pdx1-Ru5P
adduct by the elimination of phosphate (-98 Da) and water
(-18 Da). This species is likely to correspond to I320. These
data do not tell us if ammonia has been incorporated into this
species because the precision of the MS analysis ((1 Da) does
not differentiate between -OH and -NH2, which differ by 1
Da. The species with a mass of 34 322 Da corresponds to one
generated by phosphate elimination. This species is consistent
9
3050 J. AM. CHEM. SOC. VOL. 130, NO. 10, 2008

Studies on the Mechanism of PLP Synthase
A R T I C L E S
Scheme 4. Mechanistic Proposal for PLP Synthase
G3P is the preferred substrate. Pdx1 can use either G3P
or DHAP as the three carbon sugar for the synthesis of PLP.13,14
This implied that Pdx1 has triose phosphate isomerase activity.
Whether this activity is intrinsic to PLP synthase or due to a
trace contaminating triose phosphate isomerase activity was
initially unclear. During the course of this study, it became
apparent that the enzyme prepared by a more thorough purifica-
tion procedure had a much reduced ability to make PLP using
DHAP. In order to quantify this observation, an experiment was
performed under saturating conditions using either G3P or
DHAP as the triose phosphate, and the results are shown in
Figure 7.
production of an advanced intermediate under mild conditions
using ESI-FTMS. We observe a primary deuterium kinetic
isotope effect at C5 of R5P on the formation of the intermediate
and determine that this isotope effect is due to removal of the
pro-R proton. We also demonstrate that ammonia is likely to
be covalently attached to the intermediate and that G3P is a
much better substrate than dihydroxyacetone-phosphate (DHAP).
Additionally, we show that phosphate loss occurs by an
elimination reaction rather than by a hydrolysis reaction and
that this occurs after C5 deprotonation and ammonia addition.
A mechanistic proposal for PLP synthase that is consistent with
these new observations is outlined in Scheme 4.
Clearly DHAP was not utilized as efficiently as G3P, and its
consumption could not account for even a single catalytic
turnover (the enzyme concentration used is denoted by the
horizontal dashed line on the plot). This suggests that the
previously observed triose phosphate isomerase activity was due
to a low concentration of the E. coli isomerase present in the
preparation. This is not surprising considering that E. coli triose
phosphate isomerase operates with a kcat of nearly 10 000 s-1
(kcat/Km ≈ 1 µM-1 s-1), whereas Pdx1 catalyzes the formation
of PLP with a kcat of approximately 0.000 87 s-1, a difference
of over ten million-fold.
2′-Hydroxypyridoxol-5′-phosphate is not converted to PLP
by Pdx1. To test a previous suggestion that 2′-hydroxypyri-
doxol-5′-phosphate was an intermediate in the formation of PLP
in Saccharomyces cereVisiae,29 this compound was synthesized
and incubated with Pdx1 and Pdx2 (close orthologues to SNZ
and SNO in S. cereVisiae) or with Pdx1 alone. HPLC analysis
of the reaction mixtures revealed no conversion to PLP.
Proposed Mechanism of PLP Formation. In this paper we
have described additional properties of the pyridoxal phosphate
synthase catalyzed reaction. Specifically, we can follow the
In this mechanism, ring opening of ribose-5-phosphate 2 gives
14. Imine formation with Lys 81 results in 15. Isomerization to
16 sets up imine formation using ammonia generated by the
Pdx2-catalyzed glutamine hydrolysis to give 17. Elimination
of water yields 18, which tautomerizes to give 19. Elimination
of K81 from C1 results in 20, which then reacts at C5 with the
same residue to give 21. Elimination of phosphate provides 22
which is our proposed structure for I320. Imine formation
between 22 and G3P results in 23. Two tautomerization
reactions followed by an electrocyclic ring closure and aroma-
tization give 27. Imine hydrolysis gives PLP 1 in the final step
of the reaction.
This mechanistic proposal now has a substantial level of
experimental support. Treatment of the enzyme with R5P
followed by proteolysis with trypsin gave a peptide with mass
and sequence consistent with imine 15 attached to Lys 81.16
However, such an imine is likely to be hydrolyzed during sample
preparation. It is therefore more likely that the isolated tryptic
peptide was labeled with aminoketone 16 which should be
stable. The formation of imine 17 occurs using high concentra-
tions of ammonia or using ammonia generated by the Pdx2-
catalyzed hydrolysis of glutamine. The deprotonation of 18 to
give 19 involves removal of the pro-R proton at C5. This
(29) Zeidler, J.; Ullah, N.; Gupta, R. N.; Pauloski, R. M.; Sayer, B. G.; Spenser,
I. D. J. Am. Chem. Soc. 2002, 124, 4542-4543.
9
J. AM. CHEM. SOC. VOL. 130, NO. 10, 2008 3051

A R T I C L E S
Hanes et al.
deprotonation reaction shows a primary deuterium isotope effect
of 2.4 indicating that it is a partially rate-limiting step in the
conversion of 16 to 22. Phosphate elimination occurs after C5
deprotonation because we observe a primary deuterium isotope
effect on phosphate formation. The addition elimination mech-
anism for phosphate removal is consistent with the absence of
18O incorporation into the phosphate as determined by 31P NMR
analysis. A similar mechanism of phosphate removal has been
previously reported for the E. coli PLP synthase which uses
1-deoxy-D-xylulose-5-phosphate and 1-amino-acetone-3-phos-
phate and makes chemical sense in that it provides an easy way
to control the point at which phosphate is lost during the reaction
sequence. We propose structure 22 for I320 based on the
following: Formation of I320 requires high concentrations of
ammonia suggesting that noncovalent interactions of ammonia
with Pdx1 are weak. In contrast Pdx1-I320 can be purified by
gel filtration and reacts with G3P to give PLP in the absence
of ammonia. This suggests that ammonia is covalently bound
to I320. I320 must be a highly conjugated system to account for
its long wavelength absorption. The observation of a primary
deuterium isotope effect on the formation of I320 from 16
from 16 by ESI-FTMS is consistent with intermediate 22 and
its hydration product. This technique requires minimal protein
manipulation and therefore minimizes the possibility of I320
decomposition. The mechanistic details for the conversion of
I320 to PLP have not yet been experimentally verified. G3P is
the preferred substrate suggesting the formation of imine 23.
In addition 2′-hydroxypyridoxol-5′-phosphate, a previously
proposed intermediate, is not a substrate for Pdx1. This supports
our proposal for early loss of water from the C1 of ribose (14
to 15).
The mechanism described in Scheme 4 is the fourth genera-
tion PLP synthase mechanism. As additional experimental facts
on the mechanism of this fascinating enzyme emerge, the
mechanism will no doubt continue to evolve. In particular,
experimental evidence for the proposed linkage shift from C1
to C5 needs to be established as well as the steps involved in
the conversion of I320 to PLP and the role of K149. As 16 and
22 are stable intermediates they should be amenable to NMR
and crystallographic characterization.
Acknowledgment. We thank Dr. Brian Lawhorn for the
synthesis of 2-hydroxypyridoxol and the Mowbray lab at
Uppsala University in Sweden for the ribokinase and ribospho-
sphate isomerase plasmids. We also thank Dr. Ivan Keresztes
for his expertise in the 31P NMR data collection. This research
was supported by a grant from the NIH (GM069618).
suggests that the C5 pro-R proton of R5P is absent in I320
.
Phosphate elimination is stoichiometric with the production I320
demonstrating that I320 does not contain phosphate. MS analysis
of a tryptic peptide modified with I320 is consistent with this
structure. However this peptide did not contain the amino group
suggesting that it had been hydrolytically removed during
sample preparation. Finally, analysis of the formation of I320
JA076604L
9
3052 J. AM. CHEM. SOC. VOL. 130, NO. 10, 2008