ARTICLES
© 2008 Nature Publishing Group http://www.nature.com/naturechemicalbiology
13C
NMR snapshots of the complex reaction coordinate
of pyridoxal phosphate synthase
Jeremiah W Hanes, Ivan Keresztes & Tadhg P Begley
The predominant biosynthetic route to vitamin B6 is catalyzed by a single enzyme. The synthase subunit of this enzyme, Pdx1,
operates in concert with the glutaminase subunit, Pdx2, to catalyze the complex condensation of ribose 5-phosphate, glutamine
and glyceraldehyde 3-phosphate to form pyridoxal 5¢-phosphate, the active form of vitamin B6. In previous studies it became
clear that many if not all of the reaction intermediates were covalently bound to the synthase subunit, thus making them difficult
to isolate and characterize. Here we show that it is possible to follow a single turnover reaction by heteronuclear NMR using
13C- and 15N-labeled substrates as well as 15N-labeled synthase. By denaturing the enzyme at points along the reaction
coordinate, we solved the structures of three covalently bound intermediates. This analysis revealed a new 1,5 migration of the
lysine amine linking the intermediate to the enzyme during the conversion of ribose 5-phosphate to pyridoxal 5¢-phosphate.
Vitamin B6 is a composite term for six different vitamers recognized
as pyridoxine (1), pyridoxal (2), pyridoxamine (3) and their corresponding 5¢-phosphorylated derivatives (4, 5 and 6, respectively). The
vitamer pyridoxal 5¢-phosphate (PLP; 5) is known for its catalytic
versatility1. In most cases PLP acts as an enzyme-bound cofactor that
participates in diverse biochemical reactions and pathways, including
amino acid biosynthesis, carbohydrate metabolism and the modification of many amine-containing compounds. It has been estimated that
at least 140 different PLP-dependent enzymes exist, and approximately
1.5% of the genes in a typical prokaryote encode PLP-using enzymes2.
A number of these enzymes are already targets for therapeutic agents,
and many more are thought to be good candidates3. In addition to
these well-documented roles in enzyme catalysis, PLP has recently
been implicated in singlet oxygen resistance4,5.
As vitamin B6 is required for various processes in all organisms, it is
either biosynthesized de novo, as is the case for most microorganisms
and plants, or acquired externally, as is necessary for animals6. Two
independent de novo pathways for the biosynthesis of PLP are
currently known. The best understood pathway, found in Escherichia
coli (1-deoxyxylulose 5-phosphate (7)-dependent), is rarely used
compared with the route present in most other species (ribose
5-phosphate–dependent)7. Pdx1-Pdx2, the biosynthetic enzyme
found in Bacillus subtilis, catalyzes the condensation reaction shown
in Scheme 1 using either D-ribose 5-phosphate (R5P; 8) or D-ribulose
5-phosphate (Ru5P; 9), glutamine (10) and D-glyceraldehyde
3-phosphate (G3P; 11)8,9. Most of the chemistry of this conversion
occurs in the PLP synthase domain, Pdx1, whereas Pdx2 is a
glutaminase responsible for delivering ammonia via a hydrophobic
tunnel to the synthase active site. The structural and mechanistic
enzymology of this pathway has become a topic of intense interest due
to both the unique chemistry involved and the fact that it may
represent an attractive target for antimicrobial agents10–20. However,
given the complexity of this reaction there is still a great deal that is
not well understood.
Previous high-resolution mass spectrometric analysis of Pdx1
revealed that the first substrate used, R5P or Ru5P, becomes covalently
attached to the enzyme through an active site lysine concomitant with
the loss of water (Pdx1 + 212 Da)8,21. At the time, this species
(referred to here as Pdx1-Z1) was proposed to be an imine formed
with the C2 carbonyl of Ru5P, based on the observation that Pdx1 has
R5P-Ru5P isomerase activity and is structurally similar to imidazole
glycerol phosphate synthase (HisF), which catalyzes the formation of
an imine at the C2 carbonyl of a 1-aminoribulose 5-phosphate
derivative (12)8,20. However, our recent biochemical studies on the
early steps of this reaction, as well as the trapping and structural
characterization of a more advanced chromophoric intermediate
(I320), suggested that this interpretation was not correct22,23.
The first hint that the Pdx1-Z1 intermediate was not bound via an
imine was provided by the observation of its relatively high stability. It
was shown that ESI-FTMS spectra of Pdx1-Z1 could be collected even
without NaBH4 reduction23. This was surprising because the preparation of the sample for analysis included a reversed-phase desalting step
that was performed under mildly acidic conditions that should result
in imine hydrolysis23. Furthermore, the addition of only NH4Cl (or
glutamine if Pdx2 was present) to Pdx1-Z1 resulted in the accumulation of the chromophoric species, I320 , which appeared to be bound to
the enzyme via C5 (ref. 22). Based on the fact that only one molecule
of water is lost during the reaction of Pdx1 with R5P to form Pdx1-Z1,
it was difficult to imagine that the substrate was initially covalently
bound to C5. The data suggested the possibility of a C-N bond shift in
going from Pdx1-Z1 to I320, but there was no direct structural data
supporting this claim. In addition to these two intermediates, a third
Department of Chemistry and Chemical Biology, Cornell University, 120 Baker Laboratory, Ithaca, New York 14853, USA. Correspondence should be addressed to
T.P.B. (tpb2@cornell.edu).
Received 30 January; accepted 8 May; published online 30 May 2008; doi:10.1038/nchembio.93
NATURE CHEMICAL BIOLOGY
VOLUME 4
NUMBER 7
JULY 2008
425
© 2008 Nature Publishing Group http://www.nature.com/naturechemicalbiology
ARTICLES
It was shown previously that the enzyme
purifies with 40–60% of the active sites
already occupied with the initial adduct8.
Therefore, the unlabeled compound was
exchanged with the 13C-labeled sugar by incubating Pdx1 overnight at 15 1C in a solution
containing labeled D-ribose, ribokinase and
Mg2+/ATP. Size exclusion chromatography
was used to free the protein from labeled
small molecules to ensure that the NMR
spectroscopy would only detect relevant
tightly bound species.
Scheme 1 The reaction catalyzed by B. subtilis PLP synthase (Pdx1-Pdx2).
intermediate accumulates upon addition of G3P to I320 that, as
described here, appeared to be a PLP-like species (Pdx1-Z3) perhaps
also bound to C5.
In an effort to obtain structural data regarding the PLP synthase
reaction intermediates, we undertook a more thorough and direct
investigation of the reaction with 13C NMR using labeled R5P. If the
13C chemical shifts of one or more of these intermediates could be
determined, then, in addition to gaining structural information, the
C-N bond in the intermediate could be localized using a selective 15N
decoupling analysis to inspect for single-bond 13C-15N coupling
between 15N-labeled protein and the intermediate. 13C chemical shifts
of small molecules bound to enzyme active sites have been previously
determined using solution NMR24. However, it was unknown whether
the resolution, with the large PLP synthase (4400 kDa), would be
sufficiently high to observe 13C-15N coupling because of the line
broadening that occurs with large proteins.
RESULTS
Preparation of 13C-containing Pdx1-Z1
To enhance the 13C NMR signal over that seen using natural abundance
R5P, singly ((1-13C), (2-13C) or (5-13C); 13, 14 or 15, respectively) or
universally labeled ((U-13C5); 16) D-ribose isotopomers were phosphorylated to the corresponding labeled R5P
((1-13C)R5P, 17; (2-13C)R5P, 18; (5-13C)R5P,
19; or (U-13C5)R5P, 20) in situ using E. coli
a
ribokinase23. Figure 1 illustrates the general
method that resulted in the successful pre13
paration of samples, which contained covalently bound species at stoichiometric ratios
close to 1:1 with the enzyme.
Kinetic competence of Pdx1-Z1
To establish the chemical and kinetic competence of Pdx1-Z1, we
measured its rate of conversion to I320. This was accomplished
by the in situ preparation of [32P]R5P using the conditions described
above in order to obtain [32P]Pdx1-Z1, which was purified by
size exclusion chromatography. Following the addition of NH4Cl,
the rate of loss of radioactive phosphate from the protein was
monitored by SDS-PAGE and shown to be identical to the rate of
formation of the chromophoric species (B0.06 min–1) (Supplementary Fig. 1a–c online). Even though this experiment was performed
at a subsaturating concentration of NH4Cl (500 mM), the rate
of formation of I320 was fast enough to account for steady state
turnover (kcat B 0.02 min–1)9. However, given that the two rates are
comparable, it is important to note that, when substituting NH4Cl
for glutamine and Pdx2, the formation of the chromophoric
species is slower and requires very high concentrations of
NH4Cl compared with the conditions under which the kcat was
previously measured (saturating glutamine and Pdx2). Nonetheless, the rate of conversion of Pdx1-Z1 to I320 under our
conditions supports the interpretation that the Pdx1-Z1 intermediate
is kinetically competent.
1
3
4
4
1
2
3
b
Figure 1 13C analysis of covalently bound Pdx1
intermediates. (a) Overview of the method
used for the preparation of each intermediate.
(b) 13C spectra of Pdx1-Z1. The top spectrum
was obtained using (2-13C)R5P, whereas
the bottom spectrum was obtained using
(U-13C5)R5P. (c) 13C spectra of Pdx1-Z2. The
top spectrum was obtained using (5-13C)R5P,
whereas the bottom spectrum was obtained using
(U-13C5)R5P. (d) 13C spectra of Pdx1-Z3. The top
spectrum was obtained by using R5P, whereas
the bottom spectrum was obtained by using
(U-13C5)R5P. Line broadening of 7 Hz was
applied to all data before Fourier transformation.
The compounds to the left of the spectra are
shown for reference and were derived from
reaction intermediates proposed here and
elsewhere22,23. The numbers 1–5 within the
spectra correspond to the numbered carbon
atoms in the compounds to the left.
426
1
c
2
2
d
3
VOLUME 4
NUMBER 7
JULY 2008
NATURE CHEMICAL BIOLOGY
ARTICLES
© 2008 Nature Publishing Group http://www.nature.com/naturechemicalbiology
a
b
c
Figure 2 Summary of 13C chemical shift assignments. (a) Pdx1-Z1.
(b) Pdx1-Z2. (c) Reduced Pdx1-Z3. Specific carbon atoms are colored
similarly to illustrate their identity throughout the reaction.
Structural characterization of Pdx1-Z1
It was clear from our initial attempts at acquiring NMR data that the
enzyme had to be denatured to get adequate resolution. The addition
of 7 M urea, just before NMR data acquisition, was used for this
purpose. Denaturation of the protein greatly alleviates the line broadening typically associated with the slow tumbling of large macromolecules, thus resulting in a large increase in the signal to noise of the
13C spectra. The addition of urea to these samples was of particular
importance because Pdx1 forms a minimum of a dimer of hexamers
(4400 kDa total) in solution at the high concentrations required for
these NMR measurements (B2–3 mM).
The 13C spectra of both (2-13C)- and (U-13C5)-labeled Pdx1-Z1 are
shown (Fig. 1b). The five signals associated with the universally
labeled small molecule (color bars) are easily distinguished from the
protein background. The chemical shift of C2 was 205.5 p.p.m., which
is characteristic of a ketone, not an imine (B160–170 p.p.m.). The
splitting patterns and coupling constants allow for the assignment of
terminal versus interior carbon atoms (doublet versus doublet of
doublets, respectively), but in this case they do not unequivocally
establish the C-C connectivity or the location of the C-N bond
connecting Z1 to the protein.
To determine this connectivity, a 13C double quantum-filtered
correlation spectroscopy (dqfCOSY) experiment was carried out
(Supplementary Fig. 2a,b online). A dqfCOSY spectrum produces
the additional benefit of essentially eliminating the undesirable background associated with protein, which turned out to be an advantage
in the assignment of another intermediate (Pdx1-Z3), where one of
the signals was in an unusually crowded region of the spectrum. This
data and the 1H chemical shift information obtained from a 1H-13C
heteronuclear single quantum coherence (HSQC) experiment (Supplementary Fig. 3 online) were consistent with an open chain species
probably bound to the protein via an amine at C1 (structure 21). The
interim conclusion that C1 was bound to the lysine nitrogen was
based largely on the upfield chemical shift of B53.8 p.p.m. for C1
relative to a similar alcohol (460 p.p.m.). A summary of the carbon
chemical shift information obtained for Pdx1-Z1 is shown (Fig. 2).
Figure 3 13C spectra of intermediates bound to 15N-labeled Pdx1. The data
and arrows are color coded and correlate with the following carbon atoms:
green is C1, red is C2 and blue is C5. (a) C1 region of Pdx1-Z1. Two
different samples are shown, one using (15N)Pdx1 (top) and one using
(14N)Pdx1 (bottom). (b) From left to right: C5, C2 and C1 regions of
Pdx1-Z2. A single sample was prepared using (15N)NH4Cl and (15N)Pdx1.
On the right hand side is an indication of the status of 15N decoupling
(interleaved). (c) C5 region of reduced Pdx1-Z3. Reduced Pdx1-Z3 was
prepared using (14N)NH4Cl, so no C2-N coupling was observed. A shifted
sine bell function was applied to all the data to enhance the resolution
over that seen in Figure 1b–d.
NATURE CHEMICAL BIOLOGY
VOLUME 4
NUMBER 7
JULY 2008
To obtain more convincing data regarding this putative C1-N
linkage, (15N)Pdx1 was prepared by performing the overexpression
in minimal medium containing (15N)NH4Cl as the sole nitrogen
source. In this system, the carbon bound directly to the active site
lysine should reveal an additional coupling to the 15N amine. The C1
region of the 13C spectrum of (15N)Pdx1-Z1 is compared to that of
(14N)Pdx1-Z1 in Figure 3. The expected additional coupling constant
to C1 (7.4 Hz) was observed. Importantly, the splitting patterns of the
other four carbon signals associated with this species were not
significantly altered by the introduction of 15N (Supplementary
Fig. 4a online). An interleaved broadband 15N decoupling experiment
was also performed on Pdx1-Z1 prepared using 15N-labeled Pdx1 and
(1-13C)R5P (Supplementary Fig. 4b). This type of analysis provided
further confirmation that the observed coupling was due to a C1-N
bond rather than differences in data acquisition or sample preparation
or composition.
Structural characterization of Pdx1-Z2
In an effort to obtain direct structural information, such as
the location of the C-N bond connecting I320 to the protein,
NMR experiments analogous to those described for Pdx1-Z1
were performed by denaturing the enzyme with 7 M urea upon
completion of chromophore formation (quenched at pH B 7.6).
Unfortunately, this did not result in a homogenous species bound
to the protein. Strategies such as quenching with weak base,
weak acid or NaBH4 reduction before the addition of the urea
were also explored. Quenching the reaction with HCl to a final
concentration of 50 mM was the only method that produced
a single species. The 13C spectra for Pdx1-Z2 (Pdx1-Z2 refers
to the species bound to the protein after quenching I320 with acid
and denaturing with urea) derived from (5-13C)R5P or (U-13C5)R5P
are shown in Figure 1c. In this case, knowledge of the signal
corresponding to C5 and the C-C coupling constants were used to
a
b
c
427
ARTICLES
© 2008 Nature Publishing Group http://www.nature.com/naturechemicalbiology
Scheme 2 Proposed cyclization reaction of I320 upon quenching with
50 mM HCl and denaturing with 7 M urea.
establish the connectivity of Pdx1-Z2. The connectivity and the carbon
chemical shifts suggest structure 7 (Fig. 2b).
To confirm the proposed heteroatom arrangement in Pdx1-Z2, an
interleaved broadband 15N decoupling analysis was performed; this
revealed that C1, C2 and C5 are bound to nitrogen atoms with
coupling constants of 9.8, 6.7 and 13.7 Hz, respectively (Fig. 3b). By
repeating the same experiment using (14N)NH4Cl, it was then confirmed that, as expected, the C2-N coupling observed resulted from
the incorporation of nitrogen from NH4Cl (Supplementary Fig. 5
online). C3 and C4 were not affected by 15N decoupling (Supplementary Fig. 6 online). However, this left both C1 and C5 bound to
the protein, which was not expected from our proposed structure for
I320. This suggests that I320 may have cyclized upon quenching the
reaction with weak acid such that C1 and C5 became bound to the
same nitrogen atom. Alternatively, it could also be argued that either
C1 or C5 reacted nonspecifically with a nitrogen-containing residue of
the denatured protein. In order to differentiate between these two
possibilities, an arrayed narrowband 15N decoupling experiment was
performed (Supplementary Fig. 7 online). This analysis revealed that
it is likely that the same nitrogen was bound to C1 and C5, because the
relative intensities of the peaks corresponding to these two carbons
varied in concert as a function of the 15N
decoupling frequency, thus supporting 22 as
our proposal for the structure of Pdx1-Z2. We
propose that Pdx1-Z2 is formed from I320 23
under the acidic denaturing conditions of the
NMR experiment as outlined in Scheme 2,
and that it is not an intermediate on the
enzymatic reaction pathway.
Structural characterization of reduced
Pdx1-Z3
Following the formation and purification of
I320, the addition of G3P resulted in an
absorbance that is characteristic of a PLPlike species (Supplementary Fig. 8a online).
Under single turnover conditions (no other
substrates present), we noticed that the species that accumulated was difficult to separate
from the protein. We were unable to isolate
free protein via extensive dialysis, gel filtration
and His6 tag–based affinity chromatography.
However, the addition of weak acid, base or
7 M urea did liberate PLP 1 from the enzyme.
This observation suggested that the final
intermediate was tightly bound to the enzyme
via an imine that was susceptible to hydrolysis
upon denaturation of the protein. Another
piece of evidence that was in support of this
hypothesis was the observation that under
428
these conditions, the lmax of PLP free in solution is 388.5 nm, as
compared with a lmax of 408 nm for Pdx1-Z3.
In the presence of a saturating concentration of G3P, the single
turnover reaction for the conversion of I320 to the putative imine
(Pdx1-Z3) occurred at B0.03 min–1 (Supplementary Fig. 8b) and
exhibited an isosbestic point at 344.5 nm. A comparison of the rate
of the single turnover measured here to the kcat for the reaction
(B0.02 min–1) suggests that the overall rate-limiting step occurs after
G3P binding9. The observation that this species is difficult to separate
from the enzyme merits further investigation and may indicate a
complex mechanism of product release. To determine whether
the final intermediate is an imine, it was formed, reduced with
NaBH4 and analyzed by 13C NMR as described above for Pdx1-Z1
and Pdx1-Z2 (Fig. 1d). The C-C backbone was established by
obtaining a dqfCOSY spectrum (Supplementary Fig. 9a,b online),
and the 13C chemical shifts are given in Figure 2c. The chemical
shifts are in good agreement with the proposed structure 24 and also
suggest that the pyridine ring is bound via the C5 position. Upon
broadband 15N decoupling, the apparent doublet at 45.6 p.p.m.
sharpened and increased in intensity (Fig. 3c). We estimate that
the C5-N coupling constant is approximately 4 ± 2 Hz based on the
change in the linewidth upon decoupling (B16 Hz and B11 Hz,
respectively). Taken together the data support that 24 is the structure
of reduced Pdx1-Z3 and that the last step in the reaction sequence
for PLP formation is the hydrolysis of the imine linking PLP to
the synthase.
DISCUSSION
Our experiments provide three molecular snapshots of the complex
reaction catalyzed by PLP synthase. A mechanistic proposal that is
consistent with these findings is shown in Scheme 3. Ring opening of
R5P to aldehyde 25, followed by imine formation, results in 26.
Scheme 3 Mechanistic proposal for PLP synthase.
VOLUME 4
NUMBER 7
JULY 2008
NATURE CHEMICAL BIOLOGY
© 2008 Nature Publishing Group http://www.nature.com/naturechemicalbiology
ARTICLES
Isomerization to 21 followed by imine formation with ammonia gives
27. Elimination of water gives 28, which then tautomerizes to 29. A
lysine shift followed by loss of phosphate gives I320 23. Imine
formation with G3P gives 30. Two tautomerization reactions followed
by an electrocyclic ring closure and an aromatization gives 31. Imine
hydrolysis gives PLP 5 in the final step of the reaction.
There is now a substantial body of evidence supporting this
mechanism. Ammonia triggers the conversion from 21 to 23, and
both intermediates have been detected by mass spectrometry8,21,23.
Under single turnover conditions, there is a primary C5 deuterium
kinetic isotope effect on the formation of I320 23, which supports the
proposal of a kinetically significant tautomerization of 28 to 29.
Phosphate is released by an elimination rather than a hydrolysis
reaction (32 to 24), and this reaction follows C5 deprotonation23.
Finally, G3P addition to I320 results in the formation of imine 31.
The PLP synthase–catalyzed reaction proceeds via a cascade of
imine chemistry. The unprecedented C1 to C5 lysine migration (29 to
32) discovered in this study is a new variation on known active site
imine chemistry. This migration is likely to greatly enhance the
catalytic versatility of the active site by shifting the intermediate into
a new environment. A full understanding of the significance of this
migration will require the structural characterization of PLP synthase
intermediate complexes.
METHODS
Overexpression and purification of Pdx1. B. subtilis Pdx1 was overexpressed
and purified according to previously published methods except for the
following changes23. Unless otherwise noted all chemicals were obtained from
Sigma Aldrich and were of the highest purity offered. The medium recipe used
for overexpression of the enzyme is given in the Supplementary Methods
online. The (15N)NH4Cl was obtained from Cambridge Isotope Laboratories
and was 499% pure. Following elution of the protein from the Ni2+-based
affinity chromatography column, the protein was buffer exchanged (10DG gel
filtration column, Bio-Rad) into the following buffer: 50 mM phosphate (Na+)
pH 7.6, 300 mM NaCl, 2 mM tris(2-carboxyethyl)phosphine hydrochloride
(TCEP, Soltec Ventures) and 25% glycerol. The protein was aliquoted, flash
frozen using liquid N2 and stored at –80 1C.
Pdx1-Z1 preparation. To prepare Pdx1-Z1, a 2 ml solution of the reaction
mixture was prepared according to details given in the Supplementary
Methods and incubated for 12 h at 15 1C. 13C-labeled D-ribose derivatives
were obtained from Cambridge Isotope Laboratories and were in all cases
greater than or equal to 98% pure. If Pdx1-Z2 and Pdx1-Z3 were desired, a final
concentration of 1 M NH4Cl was added and allowed to react for approximately
1 h at room temperature (22–23 1C). At this time the covalently modified
protein was purified from the excess salt and small molecules using a 10DG gel
filtration column according to the product instructions. The column was preequilibrated with the following buffer: 10 mM phosphate (Na+) pH 7.45,
200 mM NaCl. If Pdx1-Z3 was desired, G3P was added at a final concentration
of 1.5 mM (racemic mixture) and allowed to react for 2 h at room temperature.
Following purification of these species, either (i) solid urea was added to the
protein to a final concentration of 7 M (in the case of Pdx1-Z1), (ii) 50 mM
HCl then urea was added (in the case of Pdx1-Z2) or (iii) 50 mM NaBH4
(15 min reaction time at room temperature) then urea was added (in the case
of Pdx1-Z3). For each of these samples the protein was concentrated by
ultrafiltration using a 10,000 Da cutoff Amicon Ultra-4 centrifugal filter unit
at 4 1C (Millipore) to a volume of 500 ml, then 100 ml of D2O was added.
32P-labeled Pdx1-Z preparation. To form Pdx1-Z , a 450 ml solution was
1
1
prepared according to the details given in the Supplementary Methods for the
preparation of Pdx1-Z1 (volumes reduced to B23% of that listed) and
incubated for 12 h at 15 1C. Pdx1-Z1 was purified using a 10DG gel filtration
column equilibrated with 40 mM phosphate (Na+) pH 7.45 buffer containing
200 mM NaCl. To start the reaction to form I320, 1 M NH4Cl in the same buffer
was mixed 1:1 with Pdx1-Z1. Two samples were created, one radioactive and
NATURE CHEMICAL BIOLOGY
VOLUME 4
NUMBER 7
JULY 2008
one nonradioactive. For the radioactive reaction 2 ml of [g-32P]ATP (PerkinElmer; 5 mCi ml–1 (3,000 Ci mmol–1)) was added in place of 2 ml of H2O.
Samples were removed from the radioactive reaction mixture at various time
points. The time points for the radioactive reaction were taken by quenching,
with two volumes of 160 mM HCl. The phosphate released (32 to 23, Scheme
3) was separated from Pdx1-Z1 using 15% SDS-PAGE, and the gel was dried
and then quantified by phosphorescence using a Storm 860 (GE Healthcare)
imaging system. The nonradioactive sample was analyzed for chromophore
production under identical conditions by measuring the absorbance at 320 nm
using a Hitachi U-2010 UV-visible spectrophotometer.
Characterization of Pdx1-Z3. To obtain I320, the following reaction mixture
was prepared in a total volume of 500 ml: 200 mM Pdx1, 2 mM TCEP, 500 mM
R5P, 1 M NH4Cl, 100 mM NaCl in HEPES buffer at pH 7.6. This mixture was
allowed to react for 1 h, at which time the protein was purified using a 10DG
gel filtration column equilibrated with 10 mM phosphate (Na+) pH 7.45
containing 200 mM NaCl. The reaction to form Pdx1-Z3 was initiated by the
addition of 1.5 mM G3P (racemic mixture) to 60 mM I320. The reaction was
monitored by UV-visible absorbance.
Data fitting and analysis. Kinetic data were analyzed by nonlinear regression
using the following single exponential equation:
Y ¼ Aelt + C
The parameters A and l correspond to the amplitude and observed rate,
respectively. The term C is the offset. Data analysis and plotting were performed
using GraFit 5 (Erithacus Software).
NMR analysis. All of the NMR experiments were run at 22 1C in B12–18%
D2O/H2O using Varian Inova 500 (one-dimensional 13C and dqfCOSY) or
600 (one-dimensional 13C with 15N decoupling and 1H-13C HSQC experiments) MHz instruments. For experiments in which 15N decoupling was
needed, a 5 mm carbon nitrogen direct observe proton decouple probehead
was used. For the 1H-13C HSQC, a 5 mm proton nitrogen observe carbon
decouple probehead was used. One-dimensional 13C spectra in the 500 MHz
instrument (operating at 125.7 MHz for 13C) were acquired observing a
chemical shift range from –5 to 230 p.p.m. with an acquisition time of 1.3 s
and a relaxation delay of 1.5 s using 601 pulses and broadband 1H decoupling.
Approximately 4,000 to 15,000 scans were averaged for total acquisition times
of B3 h to B12 h. Spectra were zero filled to 256 K. Specific processing of data
before Fourier transformation is described in each figure legend.
Note: Supplementary information and chemical compound information is available on
the Nature Chemical Biology website.
ACKNOWLEDGMENTS
This research was supported by grants from the US National Institutes of Health
to T.P.B. (GM069618).
Published online at http://www.nature.com/naturechemicalbiology/
Reprints and permissions information is available online at http://npg.nature.com/
reprintsandpermissions/
1. Eliot, A.C. & Kirsch, J.F. Pyridoxal phosphate enzymes: mechanistic, structural, and
evolutionary considerations. Annu. Rev. Biochem. 73, 383–415 (2004).
2. Percudani, R. & Peracchi, A. A genomic overview of pyridoxal-phosphate-dependent
enzymes. EMBO Rep. 4, 850–854 (2003).
3. Amadasi, A. et al. Pyridoxal 5¢-phosphate enzymes as targets for therapeutic agents.
Curr. Med. Chem. 14, 1291–1324 (2007).
4. Bilski, P., Li, M.Y., Ehrenshaft, M., Daub, M.E. & Chignell, C.F. Vitamin B6 (pyridoxine)
and its derivatives are efficient singlet oxygen quenchers and potential fungal antioxidants. Photochem. Photobiol. 71, 129–134 (2000).
5. Daub, M.E. & Ehrenshaft, M. The photoactivated cercospora toxin cercosporin:
contributions to plant disease and fundamental biology. Annu. Rev. Phytopathol. 38,
461–490 (2000).
6. Garrido-Franco, M. Pyridoxine 5¢-phosphate synthase: de novo synthesis of vitamin B6
and beyond. Biochim. Biophys. Acta 1647, 92–97 (2003).
7. Fitzpatrick, T.B. et al. Two independent routes of de novo vitamin B6 biosynthesis: not
that different after all. Biochem. J. 407, 1–13 (2007).
8. Burns, K.E., Xiang, Y., Kinsland, C.L., McLafferty, F.W. & Begley, T.P. Reconstitution
and biochemical characterization of a new pyridoxal-5¢-phosphate biosynthetic pathway. J. Am. Chem. Soc. 127, 3682–3683 (2005).
429
© 2008 Nature Publishing Group http://www.nature.com/naturechemicalbiology
ARTICLES
9. Raschle, T., Amrhein, N. & Fitzpatrick, T.B. On the two components of pyridoxal 5¢phosphate synthase from Bacillus subtilis. J. Biol. Chem. 280, 32291–32300 (2005).
10. Flicker, K. et al. Structural and thermodynamic insights into the assembly of the
heteromeric pyridoxal phosphate synthase from Plasmodium falciparum. J. Mol. Biol.
374, 732–748 (2007).
11. Tambasco-Studart, M., Tews, I., Amrhein, N. & Fitzpatrick, T.B. Functional analysis of
PDX2 from Arabidopsis, a glutaminase involved in vitamin B6 biosynthesis. Plant
Physiol. 144, 915–925 (2007).
12. Neuwirth, M., Flicker, K., Strohmeier, M., Tews, I. & Macheroux, P. Thermodynamic
characterization of the protein-protein interaction in the heteromeric Bacillus subtilis
pyridoxalphosphate synthase. Biochemistry 46, 5131–5139 (2007).
13. Denslow, S.A., Rueschhoff, E.E. & Daub, M.E. Regulation of the Arabidopsis thaliana
vitamin B6 biosynthesis genes by abiotic stress. Plant Physiol. Biochem. 45, 152–161
(2007).
14. Tambasco-Studart, M. et al. Vitamin B6 biosynthesis in higher plants. Proc. Natl.
Acad. Sci. USA 102, 13687–13692 (2005).
15. Wrenger, C., Eschbach, M.L., Muller, I.B., Warnecke, D. & Walter, R.D. Analysis of
the vitamin B6 biosynthesis pathway in the human malaria parasite Plasmodium
falciparum. J. Biol. Chem. 280, 5242–5248 (2005).
16. Wetzel, D.K., Ehrenshaft, M., Denslow, S.A. & Daub, M.E. Functional complementation between the PDX1 vitamin B6 biosynthetic gene of Cercospora nicotianae and
pdxJ of Escherichia coli. FEBS Lett. 564, 143–146 (2004).
430
17. Dong, Y.X., Sueda, S., Nikawa, J. & Kondo, H. Characterization of the products of the
genes SNO1 and SNZ1 involved in pyridoxine synthesis in Saccharomyces cerevisiae.
Eur. J. Biochem. 271, 745–752 (2004).
18. Ehrenshaft, M. & Daub, M.E. Isolation of PDX2, a second novel gene in the pyridoxine
biosynthesis pathway of eukaryotes, archaebacteria, and a subset of eubacteria.
J. Bacteriol. 183, 3383–3390 (2001).
19. Strohmeier, M. et al. Structure of a bacterial pyridoxal 5¢-phosphate synthase complex.
Proc. Natl. Acad. Sci. USA 103, 19284–19289 (2006).
20. Zein, F. et al. Structural insights into the mechanism of the PLP synthase holoenzyme
from Thermotoga maritima. Biochemistry 45, 14609–14620 (2006).
21. Raschle, T. et al. Reaction mechanism of pyridoxal 5¢-phosphate synthase. Detection
of an enzyme-bound chromophoric intermediate. J. Biol. Chem. 282, 6098–6105
(2007).
22. Hanes, J.W., Keresztes, I. & Begley, T.P. Trapping of a chromophoric intermediate in
the Pdx1-catalyzed biosynthesis of pyridoxal 5¢-phosphate. Angew. Chem. Int. Edn
Engl. 47, 2102–2105 (2008).
23. Hanes, J.W. et al. Mechanistic studies on pyridoxal phosphate synthase: the reaction
pathway leading to a chromophoric intermediate. J. Am. Chem. Soc. 130, 3043–3052
(2008).
24. Anderson, K.S. et al. Observation by 13C NMR of the EPSP synthase tetrahedral
intermediate bound to the enzyme active site. Biochemistry 29, 1460–1465
(1990).
VOLUME 4
NUMBER 7
JULY 2008
NATURE CHEMICAL BIOLOGY