6032 J. Med. Chem.

2009, 52, 6032–6041

DOI: 10.1021/jm900775q

Hydroxamic Acids As a Novel Family of Serine Racemase Inhibitors: Mechanistic Analysis Reveals
Different Modes of Interaction with the Pyridoxal-50 -phosphate Cofactor

)


)
Hillary E. Hoffman,†,‡, Jana Jir a,†,‡, Petr Cı́gler,†,^ Miloslav Sanda,
askov †
Jan Schraml,§ and Jan Konvalinka*,†,‡
†
Gilead Sciences and IOCB Research Center, Institute of Organic Chemistry and Biochemistry of the ASCR, v. v. i., Flemingovo n. 2, 166 10
Prague 6, Czech Republic, ‡Department of Biochemistry, Faculty of Science, Charles University, Hlavova 8, Prague 2, Czech Republic, and

)
§
Institute of Chemical Process Fundamentals of the ASCR, v. v. i., Rozvojov a 135, 165 02 Prague 6, Czech Republic. These two authors
contributed equally to this work. ^ Current address: The Scripps Research Institute, 10550 North Torrey Pines Road, La Jolla, California 92037.

Received June 1, 2009

Mammalian serine racemase (SR) is a pyridoxal-50 -phosphate (PLP) dependent enzyme responsible for
the biosynthesis of the neurotransmitter D-serine, which activates N-methyl-D-aspartate (NMDA)
Downloaded by CZECH ACADEMY OF SCIENCES on November 5, 2009 | http://pubs.acs.org

receptors in the CNS. Aberrant regulation of NMDA receptor signaling has been implicated in a variety
of neuropathologies, and inhibitors of SR would therefore be a worthwhile tool for further investigation
or treatment of such conditions. Here, we identify a series of small aliphatic hydroxamic acids (HAs)
that act as potent SR inhibitors. However, specificity studies showed that some of these HAs can act as
Publication Date (Web): September 11, 2009 | doi: 10.1021/jm900775q

nonspecific inhibitors of PLP-dependent enzymes. We employed NMR, MS, and UV/vis spectroscopic
techniques to reveal that the nonspecific effect is likely due to irreversible interaction of the HA moiety
with PLP to form aldoxime species. We also characterize L-aspartic acid β-hydroxamate as a competitive
and selective SR inhibitor that could be used as a scaffold for further inhibitor development.

Introduction of active-site directed inhibitors with high specificity is a

a 0
Mammalian serine racemase (SR ) is a unique pyridoxal-5 -
phosphate (PLP)-dependent enzyme responsible for the bio-
synthesis of D-serine in the central nervous system. D-Serine
acts as a neurotransmitter and coagonist of N-methyl-
1
D-aspartate (NMDA) receptors, ionotropic glutamate recep-
tors that are found throughout the brain but predominantly
within the forebrain.2 NMDA-receptor-mediated glutamate
excitotoxicity has been implicated in a plethora of neuro-
pathological conditions, but treatment of patients with high-
affinity NMDA receptor blockers often results in undesirable
side effects, such as hallucinations. Thus, alternative methods
of decreasing NMDA receptor overactivation are highly
sought after. Excitotoxic D-serine levels have been implicated
in Alzheimer’s disease3 and amyotrophic lateral sclerosis
(Lou Gehrig’s disease),4 while low D-serine levels result in
NMDA receptor hypofunction, which may lead to symptoms
of schizophrenia.5 SR inhibitors therefore offer a novel
and potentially highly specific approach for attenuation of
NMDA-receptor-mediated excitotoxicity and for further
study of the pathway.
Since almost all PLP-dependent enzymes share a
common transition state, the external aldimine, identification
*Corresponding author. Tel: 420-220183218. Fax: 420-220183578.
E-mail: konval@uochb.cas.cz.
a
Abbreviations: AR, alanine racemase; ATP, adenosine triphos-
phate; DHA, dihydroxamic acid; DNPH, 2,4-dinitrophenylhydrazine;
DTT, dithiothreitol; EDTA, ethylenediaminetetraacetic acid; FDAA,
1-fluoro-2,4-dinitrophenyl-5-L-alanine amide; HA, hydroxamic acid;
HEPES, 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid; IPTG,
isopropyl-β-D-1-thiogalactopyranoside; NMDA, N-methyl-D-aspar-
tate; OD, optical density; PLP, pyridoxal-50 -phosphate; SDH, serine
dehydratase; SR, serine racemase (mSR, mouse SR; hSR, human SR);
TA, transaminase.
pubs.acs.org/jmc Published on Web 09/11/2009
particular challenge.6 Nevertheless, several PLP-dependent
enzymes have been recognized as pharmaceutical targets, as
recently reviewed by Amadasi et al.7 In the case of SR,
however, very few potent and specific inhibitors have been
identified to date; the two most potent competitive inhibitors,
malonic acid and L-erythro-3-hydroxyaspartic acid,8 exhibit
inhibitory constants (Ki) in the micromolar range.
With the aim of identifying more effective SR inhibitors, we
screened a panel of structurally diverse small molecules for the
inhibition of recombinant mouse SR (mSR). As a result of the
screening, we identified a series of hydroxamic acids (HAs)
with potent inhibitory activity. Since their discovery in 1869,9
the HAs have remained an intriguing family of organic
bioligands.10 Their bioactivity is usually associated with their
metal-binding properties [e.g., natural or artificial sidero-
phores for Fe(III) sequestration and strong chelators of Zn-
(II)], but recent interest has focused on the use of HAs as
specific enzyme inhibitors and potential pharmacophores.
HAs have been shown to inhibit a number of enzyme targets
including metalloproteases, hydrolases, ureases, lipoxy-
genases, cyclooxygenases, and peptide deformylases (for a
recent review, see refs 11 and 12). Several of the HAs have
been shown to be cell-permeable, and in at least one case, they
have been shown to cross the blood-brain barrier.13
In the present work, we identify several HAs, including a
series of dihydroxamic acids (DHAs), with inhibitory activity
toward SR. In fact, one of them, succinodihydroxamic acid
(3, Chart 1), appears to be the most potent competitive
SR inhibitor identified to date. However, screening the HAs
for inhibition of a small panel of structurally and func-
tionally diverse PLP-dependent enzymes revealed that the
DHAs capable of inhibiting SR also inhibit all the other
r 2009 American Chemical Society
 Article
Chart 1. Structures of the Hydroxamic Acids Studied Split into
Four Structurally Diverse Groupsa
Downloaded by CZECH ACADEMY OF SCIENCES on November 5, 2009 | http://pubs.acs.org
Publication Date (Web): September 11, 2009 | doi: 10.1021/jm900775q
a screening to facilitate comparison of the results with our
The groups are (I) dihydroxamic acids (DHAs) and (II) hydroxamic
derivatives of amino acids (compound 8 is a mixture of two isomers: the
D-isomer with 2R,3S configuration and the L-isomer with 2S,3R config-
uration), (III) three representatives of benzohydroxamic acids, and (IV)
formohydroxamic acid and the clinical compounds acetohydroxamic
acid (Lithostat) and vorinostat (Zolinza).
enzymes tested. Although the DHAs are effective SR inhibi-
tors, their compromised specificity renders them unattractive
candidates for further drug development. We employ NMR,
MS, and UV/vis spectrophotometry to show that the DHA
inhibitors act as covalent modifiers of PLP, forming an
aldoxime species and in effect depleting the available pool of
cofactor.
Notably, not all of the inhibitory HAs identified by our
screening experiments are able to act nonspecifically via PLP-
aldoxime formation. L-Aspartic acid β-hydroxamate (10,
Chart 1) acts as a competitive SR inhibitor and forms an
aldimine species with PLP that may act as a transition state
mimetic. This compound is selective for SR and closely related
enzymes and could be used as a lead compound in further
inhibitor development studies.
Journal of Medicinal Chemistry, 2009, Vol. 52, No. 19 6033
Figure 1. Graph of relative remaining SR activity in the presence of
selected hydroxamic acids compared to the noninhibited reaction.
Latin numerals indicate the structurally diverse groups of hydro-
xamic acids described in Chart 1. Activity was assayed at pH 8.0 in
the presence of 10 μM PLP (see Experimental Section for details)
with both the substrate (L-serine) and the inhibitor at 5 mM
concentration. A known potent inhibitor, L-erythro-3-hydroxyas-
partic acid,8 was used as the positive inhibition control (inhibitor).
Control represents the noninhibited reaction. The data are averages
of duplicate or triplicate measurements; the error bars show the
standard deviation from the mean. For the effective compounds,
IC50 values (μM) are shown. IC50 values were calculated from
two independent data sets. The standard deviation is within 25%,
except in the case of 4 (within 60%). *No remaining activity was
observed.
Results
Selected Hydroxamic Acids Potently Inhibit Mouse Serine
Racemase. A panel of hydroxamic acid analogues of car-
boxylic and amino acids was analyzed for the inhibition of
mouse serine racemase (mSR). We have recently shown that
mSR and its human orthologue (hSR) exhibit comparable
inhibitor sensitivity.14 Therefore, we used mSR for the initial
previously published mSR data.8 Compounds were screened
for inhibition of mSR-catalyzed L-serine racemization.
L-Serine and the test compound were present at equimolar
concentration (5 mM);9 this setup was chosen to reveal
ligands of comparable or higher affinity toward serine
racemase than the substrate L-serine. The inhibition screen-
ing data is summarized in Figure 1.
The first group of test compounds (I) comprised two to six
and eight carbon chain dihydroxamic acids (Chart 1).
Because of the unavailability of its synthetic precursor, the
seven carbon chain dihydroxamic acid was not included. The
three, four, and five carbon chain DHAs inhibit racemiza-
tion significantly and with comparable IC50 values: malono-
dihydroxamic acid (2), 370 ( 30 μM; succinodihydroxamic
acid (3), 90 ( 20 μM; and glutarodihydroxamic acid (4),
170 ( 110 μM. The two, six, and eight carbon chain
dihydroxamic acids (1, 5, and 6, respectively) do not affect
enzyme activity. The second group of compounds (II) in-
cluded hydroxamic acid derivatives of the amino acid sub-
strates serine and threonine and the inhibitors glycine (Ki =
1640 μM8) and L-aspartic acid (Ki = 1900 μM15). Neither the
 6034 Journal of Medicinal Chemistry, 2009, Vol. 52, No. 19 Hoffman et al.
Downloaded by CZECH ACADEMY OF SCIENCES on November 5, 2009 | http://pubs.acs.org
Publication Date (Web): September 11, 2009 | doi: 10.1021/jm900775q

Figure 2. Mechanism of action of 2 (A), 3 (B), and 10 (C,D) depicted as Lineweaver-Burk (A-C) and Eadie-Hofstee (D) plots. Data for the
plots were analyzed using global analysis nonlinear fit followed by the linear transformation provided by GraFit 5 software.19 Kinetic data were
recorded using an HPLC-based end point assay as described in Experimental Section. The substrate L-serine is abbreviated as LS. Inhibitor
concentrations (μM) are indicted on the right y-axes. Reaction velocity is in arbitrary units of product (D-serine) peak area relative to the
internal standard (glycine) peak in time. The data points represent triplicate experiments with an average standard deviation of 9%.

substrate (7 and 8) nor the glycine (9) R-hydroxamic acid
derivatives showed significant inhibition. However, L-aspar-
tic acid β-hydroxamate (10) inhibited serine racemase with
an IC50 of 1300 ( 200 μM.
Additionally, we screened several other structurally di-
verse hydroxamic acids for SR inhibition (III and IV),
including a panel of benzohydroxamic acids, exemplified
by 11, 12, and 13. None of these compounds significantly
affected serine racemization.
Hydroxamic acids often act by chelating functionally
important metal ions, and divalent cations are important
serine racemase activators.16,17 Since the screening reactions
were performed in the presence of 1 mM MgCl2, we were
concerned that the observed SR inhibition might be an effect
of metal chelation. Therefore, we tested SR inhibition by 2
(as a representative inhibitory compound) in the presence of
varying concentrations of MgCl2. The inhibition of SR by 2
remained unaffected even in the presence of a 5-fold molar
excess of Mg2þ over the inhibitor (data not shown).
Mechanism of Serine Racemase Inhibition by HAs. We
selected 2, 3, and 10, hydroxamic acid analogues of known
competitive inhibitors,8,15 as representative compounds for
further analysis. Compound 4 was also an effective inhibitor,
but since our synthetic preparation was less than 95% pure,
we excluded it from the detailed kinetic analyses. We ana-
lyzed the mechanisms of action of 2, 3, and 10 by determina-
tion of mSR racemization kinetics at various substrate/
inhibitor concentrations (see Experimental Section for
details). The data were processed with DynaFit4 software,
which uses least-squares nonlinear regression analysis to fit
the data to defined kinetic models (in this case, models for
competitive, mixed-type, uncompetitive, or noncompetitive
inhibition).18 DynaFit4 accomplishes precise numerical and
graphical tests to choose the best fitting model. The best
fitting model for 2 is noncompetitive inhibition with a Ki
value of 417 ( 52 μM. Data for 3 best match the competitive
model. The Ki value calculated for 3 is 3.6 ( 0.6 μM,
indicating that 3 is the most potent competitive SR inhibitor
characterized to date. Compound 10 is also a competitive
inhibitor with a Ki of 97.5 ( 23.7 μM. The results of the
mechanistic analysis are presented in Figure 2 as Linewea-
ver-Burk or Eadie-Hofstee plots prepared with GraFit5
software.19
Specificity of the HAs. In order to test the specificity of the
HA inhibitors for serine racemase, we screened 2, 3, and 10
for inhibition of a small panel of PLP-dependent enzymes.
The noninhibitory compound 1 was included in the screen as
a control, as was the previously identified SR inhibitor
malonic acid.8 The enzyme panel comprised mouse SR,
human SR, rat liver serine dehydratase (SDH),20 alanine
racemase from Bacillus stearothermophilus (AR), and bac-
terial transaminase (TA).
The results of the specificity screen are shown in Figure 3.
All enzymatic reactions were performed at pH 8.0 in the
presence of 10 μM PLP using 5 mM compound (see Experi-
mental Section for details). Compounds 2 and 3 showed an
inhibitory effect on all enzymes tested. Interestingly, malonic
acid, the dicarboxylic acid analogue of 2, is specific for mouse
and human SR. Compound 1, which did not inhibit mSR,
did not inhibit any of the other enzymes tested. Alone among
 Article Journal of Medicinal Chemistry, 2009, Vol. 52, No. 19 6035
Downloaded by CZECH ACADEMY OF SCIENCES on November 5, 2009 | http://pubs.acs.org

Figure 3. Inhibition of a panel of PLP-dependent enzymes by selected hydroxamic acids. Inhibition was assayed at pH 8.0 in the presence of 10
μM PLP. The inhibitor concentration was kept at 5 mM, while the substrate concentration was chosen to be near the KM of the enzyme (see
Publication Date (Web): September 11, 2009 | doi: 10.1021/jm900775q

Experimental Section for details). *No remaining activity was observed. #The tested compound acts as a substrate.

Figure 4. Absorbance spectra of PLP and selected compounds recorded at pH 8.0. The final concentration of the test compound was 5 mM,
and the PLP concentration was 0.3 mM.

the hydroxamic acids tested, the L-aspartic acid analogue 10
exhibits some selectivity: it inhibits mSR, hSR, and their
homologue SDH (20% identity). It does not show any
significant inhibitory activity toward AR, and it acts as a
substrate of TA (the natural substrate of which is L-aspartic
acid).
The only common feature of the enzymes included in the
panel is the presence of a PLP-Schiff base in the catalytic
center. Therefore, we hypothesized that the DHAs might act
via modification of PLP, and to test this hypothesis, we
analyzed SR inhibition in the presence of various concentra-
tions of PLP. When the concentration of PLP added to the
reaction mixture exceeded the concentration of 2 or 3 in
molar equivalents, we observed quenching of the inhibition
by excess PLP (data not shown), suggesting that certain HAs
may inhibit SR and other PLP-dependent enzymes by inter-
acting with the cofactor.
Some Hydroxamic Acids React with PLP in Buffered
Aqueous Solution. In order to investigate the putative inter-
action of HAs with PLP in aqueous solution, we analyzed
the reaction of PLP and HAs using UV/vis spectrophoto-
metry. Solutions of PLP and various HAs were prepared in
phosphate buffer, pH 8.0, and absorbance spectra were
recorded in the 310-500 nm range. Results of the screening
are shown in Figure 4 (1, 2, 3, 10, and 16) and Figure 2S
(Supporting Information) (4, 5, and 6). The absorbance
spectrum of PLP alone (Figure 4, black curve) exhibits
maxima at 325 and 390 nm. The major chromophore at
390 nm corresponds to PLP with an unsubstituted aldehyde
moiety, and the minor peak at 325 nm corresponds to PLP-
hydrate.21 As a positive control, we added an excess of
hydroxylamine to PLP. Hydroxylamine reacts rapidly with
the aldehyde of PLP to form a PLP-aldoxime (see Scheme 1).
The absorbance spectrum of this PLP-aldoxime is shown in
Figure 4 (red curve); as expected, one major peak is visible at
335 nm, while the peak at 390 nm has disappeared entirely.
Addition of the inhibitory DHAs 2, 3, or 4 to PLP resulted in
similar spectral shifts. Addition of noninhibitory DHAs 1, 5,
or 6 had no significant effect on the absorbance profile of free
PLP, indicating that these compounds do not react with the
aldehyde of PLP. Addition of 10, a selective and competitive
HA inhibitor of SR, to PLP resulted in a unique absorbance
spectrum (Figure 4, yellow curve). In addition to the forma-
tion of a peak at 330 nm (similar to spectra of PLP in the
 6036 Journal of Medicinal Chemistry, 2009, Vol. 52, No. 19
Scheme 1
Downloaded by CZECH ACADEMY OF SCIENCES on November 5, 2009 | http://pubs.acs.org
presence of 2, 3, 4, and hydroxylamine), a second broad peak
centered around 405 nm appears in the spectrum. This
second peak could indicate that the inhibitor interacts with
PLP via its amino group, as PLP-amino acid Schiff bases are
Publication Date (Web): September 11, 2009 | doi: 10.1021/jm900775q
known to exhibit absorbance maxima in this region.22,23
Two pharmacologically active compounds containing
a hydroxamic acid moiety, acetohydroxamic acid (15,
Lithostat), an inhibitor of bacterial urease,24 and suberoyl-
anilide hydroxamic acid (16, vorinostat, see Chart 1), a
histone deacetylase inhibitor marketed under the brand
name Zolinza,25 were added to the series as controls. Neither
5 mM 16 (data not shown) nor 5 mM 15 (see Figure 1)
exhibits inhibitory activity toward SR, and no change in the
spectrum of PLP upon the addition of 15 or 16 was observed.
PLP as a Reactive Target for DHAs: Characterization of
Reaction Products. In order to analyze the product(s) result-
ing from reactions of HAs with PLP, we monitored the
reactions with 1H NMR spectroscopy (see Supporting In-
formation for background and details). All solutions were
prepared in 100 mM phosphate buffer in D2O, and the pD
was kept as close to 8.0 as possible. Upon addition of 2, 3, or
the noninhibitory compound 1 to PLP, the chemical shift of
PLP’s aldehydic proton does not change, but the signal
intensity decreases until the signal disappears completely
(except in the case of 1). The rate of the decrease in intensity
depends on the nature of the DHA and decreases in the order
3 > 2 > 1. The reaction of 3 with PLP went to completion in
less than a minute and resulted in the formation of two
products. A thorough structural analysis based on 1H and
15
N NMR (see Supporting Information Tables 2S and 3S for
a detailed explanation) resulted in the unambiguous identi-
fication of the major and minor products as the syn- and anti-
aldoxime, respectively, as shown in Scheme 1. The reaction
of PLP and 2 led to the same final products but proceeded
more slowly, and an intermediate product could be observed.
PLP and 1 underwent a very slow reaction (after 12 days,
20% of the starting material remained), which eventually
resulted in the formation of syn- and anti-PLP-aldoxime via
an unidentified intermediate (with chemical shifts different
from those of the intermediate observed in the reaction of
PLP and 2). The NMR analysis was also carried out with
DHAs 4-6, with spectra recorded immediately and after 20
h. The noninhibitory compounds 5 and 6 did not react with
PLP during this time period, while the inhibitory DHA 4 was
reactive. After 20 h, the starting material had been comple-
tely consumed, and the PLP-aldoxime was formed.
Hoffman et al.
Formation of PLP-aldoximes in mixtures of PLP and
DHAs was further confirmed by ESI-MS analysis of the
reaction mixtures. A signal of 261.2 m/z, which corresponds
to PLP-aldoxime, was detected in the spectra of the inhibi-
tory DHAs (2, 3, and 4) mixed with PLP. Additionally, an
artificial ionization signal of 243.2 m/z was observed, which
corresponds to the aldoxime with one water molecule re-
moved. These two signals correspond to the major compo-
nents visible in the spectrum of a mixture of hydroxylamine
with PLP, and thus, both signals can be attributed to PLP-
aldoxime. Most of the spectra also contained a signal of
246.1 m/z, which corresponds to unmodified PLP. The
aldoxime signals did not appear in the spectra of noninhibi-
tory compounds 1, 5, 6, or 15 mixed with PLP; these spectra
contained only the signal corresponding to unmodified PLP.
Interaction of 10 with PLP Occurs by a Different Mechan-
ism. 1H NMR analysis showed that, like the DHAs, 10 reacts
with the aldehyde group of PLP, as evidenced by a decrease
in the intensity of the CHO signal over time. However, the
PLP-aldoximes shown in Scheme 1 are not formed in this
case. Rather, the reaction of PLP and 10 results in the
appearance of two new broad signals at 8.73 and 5.75 ppm
(as well as a variety of minor side-products). A slow sub-
sequent reaction affects the CH-CH2 moiety of 10. The
proton on the R-carbon is abstracted, keeping the two CH2
protons nonequivalent. Exchange of the R-proton with D
can be excluded (no line broadening of the CH2 signals was
observed). These data do not allow for a conclusive determi-
nation of the structure of the product(s) formed from the
reaction of PLP and 10. However, on the basis of the known
interactions of PLP with amino acids26-28 and on the avail-
able NMR data, we propose that one of the major products
of the reaction of 10 with PLP is the aldimine shown in
Scheme 1.
MS analysis of the mixture of 10 and PLP offers support
for this proposal. At PLP concentrations of 4-5 mM, the
major feature of the spectrum is a peak at 246.1 m/z,
corresponding to unmodified PLP, and only trace amounts
of the proposed PLP-10-aldimine (as a potassium adduct,
416.3 m/z) can be observed. However, at higher PLP con-
centration (30 mM), the spectra contained a significant
signal corresponding to the PLP-10-aldimine. Under these
conditions, the corresponding PLP-amino acid-aldimines
can be observed in the spectra of mixtures of PLP and 7, 8,
9, L-serine, or glycine. The fact that the R-HAs 7, 8, and 9
have no effect on SR activity (see Figure 1) suggests that
these compounds probably do not form aldimine species in
the active site, perhaps because of steric or electrostatic clash
of the R-hydroxamic acid moiety with active site residues. To
address this hypothesis, it would be interesting to test the
R-hydroxamic acid analogue of 10; however, since this
compound is not commercially available and its synthesis
is rather challenging, we did not include it here. In the case of
the β-HA 10, a competitive and selective SR inhibitor,
aldimine formation is a probable mechanism of inhibition.
Discussion
Specific and potent inhibitors of mammalian serine race-
mase would be advantageous tools for further research of the
enzyme function in vivo and might have therapeutic potential
for the treatment of neuropathologies associated with gluta-
matergic excitotoxicity. Recently published experiments with
SR knockout mice support this hypothesis, showing that SR
deficiency is correlated with a reduction in neurotoxic effects
 Article
caused by overexcitation of NMDA receptors.29 SR inhibitor
development has been addressed by several groups in recent
years.30,31,15,8 The most effective competitive inhibitors iden-
tified by these groups are small amino and dicarboxylic acids,
with the most potent being L-erythro-3-hydroxyaspartic acid
(Ki = 43 μM) and malonic acid (Ki = 71 μM).8 In order to
conduct experiments in animal models or tissue cultures, more
potent compounds are needed, and identification of novel
structures capable of specifically inhibiting SR is therefore
very relevant.
During our screening of various structurally diverse small
molecules for SR inhibition, we identified a series of hydro-
xamic acids as potent SR inhibitors. Hydroxamic acids have
been shown to inhibit a variety of enzymes, often by chelating
functionally important metal ions. We immediately addressed
the possibility that the observed inhibition was due to chela-
tion of Mg2þ, a potent SR activator present in our screening
assay. Gratifyingly, the HAs retained their inhibitory activity
Downloaded by CZECH ACADEMY OF SCIENCES on November 5, 2009 | http://pubs.acs.org
even in the presence of a large molar excess of Mg2þ over the
test compound, indicating that metal chelation does not play a
significant role in SR inhibition.
Kinetic analysis of the mechanism of action of inhibitory
Publication Date (Web): September 11, 2009 | doi: 10.1021/jm900775q
HAs shows that 10 is a classical competitive inhibitor with a Ki
of 97.5 ( 23.7 μM. Compound 10 is therefore a roughly 20-
fold more potent mSR inhibitor than its carboxylic acid
analogue L-aspartic acid (competitive mechanism, Ki =
1900 μM15). There could be several reasons for the increased
potency of 10 compared to that of L-aspartic acid. Hydro-
xamic acids generally have much higher pKa values than their
corresponding carboxylic acids,32 which could lead to differ-
ent electrostatic interactions with active site residues depend-
ing on the local pH in the active site. Furthermore, the
hydroxamic acid moiety has the potential to form more
hydrogen bonding interactions with active site residues than
its carboxylic acid counterpart. However, since the 3-D
structure of SR has not yet been solved, it is difficult to
speculate about which possibility is more likely.
Succinodihydroxamic acid (3), which behaves as a compe-
titive inhibitor with a Ki of 3.6 ( 0.6 μM, is also more potent
than its dicarboxylic acid analogue succinic acid, which only
moderately inhibits mSR.8 In contrast, malonic acid is one of
the most potent competitive SR inhibitors identified to date
(Ki = 71 μM8), while its dihydroxamic acid analogue 2 acts as
a noncompetitive inhibitor with a Ki value of 417 ( 52 μM. It
is surprising that 2, which differs from 3 only in the length of
the carbon chain, behaves as a noncompetitive inhibitor, while
3 behaves as a competitive inhibitor. It should be noted that
the inhibition mechanisms analyzed in the cases of 2 and 3 are
very complex. Both compounds are able to irreversibly inter-
act with the PLP cofactor, as our NMR, MS, and UV/vis
analyses indicate, and therefore, the kinetic models used may
not be entirely suitable to describe the inhibitor-SR interac-
tion. It is noteworthy, however, that in conditions approach-
ing the initial velocity of the enzymatic reaction (for details,
see Experimental Section) 3 and 2 behave as potent competi-
tive and noncompetitive inhibitors, respectively.
In addition to the mechanism of action and potency, we also
addressed the target specificity of the HAs. The development
of specific inhibitors for PLP-dependent enzymes is always a
challenging task since the mechanisms of almost all PLP-
dependent enzymes, regardless of their substrate or reaction
specificity, proceed via condensation of the amino acid sub-
strate with PLP, forming an external aldimine intermediate.
We tested the selectivity of 2, 3, and 10 for mammalian SR by
Journal of Medicinal Chemistry, 2009, Vol. 52, No. 19 6037
repeating the screening assay with a panel of PLP-dependent
enzymes [mouse SR, human SR, rat serine dehydratase
(SDH), bacterial alanine racemase (AR), and bacterial trans-
aminase (TA)]. Mouse and human SR share 89% sequence
identity and are predicted to be members of the fold type II
subfamily of PLP-dependent enzymes.33 SDH, which elim-
inates L-serine to pyruvate in the liver, shares approximately
20% sequence identity with mammalian SR, and its crystal
structure reveals that it is fold type II.34 AR shares no
significant sequence homology with mammalian SR and is a
member of the fold type III family,35 a family that is quite
structurally distinct from other PLP-dependent enzymes. As
its name suggests, AR catalyzes the interconversion of D- and
L-alanine. TA produces oxaloacetate and L-glutamate from L-
aspartate and R-ketoglutarate and is a member of the fold type
I subfamily, also referred to as the asparate aminotransferase
family. The enzymes included in the panel are therefore quite
structurally and functionally diverse and are representative of
the variety of PLP-dependent enzymes. Direct comparison of
the potency of the inhibitors toward the different enzymes is
not possible under this setup, as potency is affected by the kcat/
KM of the enzymes toward their respective substrates and by
the reaction conditions. The results shown in Figure 3 should
therefore be interpreted qualitatively rather than quantita-
tively. However, it can be concluded that the DHA not
capable of inhibiting mouse SR (1) did not inhibit any of the
other enzymes tested, while the most active DHAs (2 and 3)
significantly inhibit all of the enzymes tested. Strikingly,
malonic acid is a specific inhibitor of SR, while its dihydroxa-
mic acid analogue 2 inhibits all of the PLP-containing en-
zymes tested indiscriminately. Unlike 2 and 3, the L-aspartic
acid analogue 10 is rather selective; it inhibits mouse and
human SR as well as their homologue SDH quite potently but
does not inhibit the more distantly related AR. Compound 10
acts as an inefficient substrate for TA and competes with the
natural substrate L-aspartic acid.
The ability of 2 and 3 to inhibit such a diversity of PLP-
dependent enzymes, together with the observed inhibition
quenching in the presence of excess PLP, prompted us to
analyze the chemical mechanism of the observed enzyme
inhibition in more detail. Hydroxylamine, NH2OH, which is
used in the synthesis of hydroxamic acids, is a reagent
commonly used to remove PLP from PLP-dependent holoen-
zymes. The presence of hydroxylamine in the inhibitor pre-
parations could explain their nonspecific interaction with
PLP. However, all of the synthetic compounds described in
this work were purified as described in Experimental Section,
and analysis of both 15N labeled and unlabeled compounds36
by NMR, mass spectrometry (MS), and elemental analysis
revealed that no significant hydroxylamine was present in the
final preparations. Furthermore, hydroxamic acids are gen-
erally believed to be stable in aqueous solution, although they
are known to undergo acid-catalyzed hydrolysis.37
We set out to analyze this putative interaction of HAs and
PLP by UV/vis spectroscopy, NMR, and mass spectrometry.
We introduce a rapid and simple spectroscopic assay that
enables facile identification of compounds with the ability to
cross-react with PLP’s aldehyde moiety. Addition of the
inhibitory DHAs (2, 3, and 4) to PLP caused a spectral shift
consistent with the formation of PLP-aldoxime. As controls,
we used clinically available compounds containing a hydro-
xamic acid moiety, acetohydroxamic acid (15, Lithostat) and
vorinostat (16, Zolinza). These compounds do not inhibit SR,
and we expected the compounds to be unreactive in the
 6038 Journal of Medicinal Chemistry, 2009, Vol. 52, No. 19
presence of PLP since there have been no reports of vitamin
B6 deficiency in patients treated by these drugs. Indeed, the
addition of noninhibitory compounds (1, 5, 6, 15, and 16) had
no significant influence on the UV/vis spectrum, suggesting
that, as expected, these compounds are stable at physiological
pH and do not cross-react with PLP-dependent enzymes.
Again, 10 is an outlier. The major feature of the spectrum of
the mixture of 10 and PLP is a peak at 330 nm, consistent with
PLP-aldoxime or PLP-hydrate formation. A second broad
peak centered around 405 nm indicates that 10 might interact
with PLP via its amino group, forming the PLP-aldimine
species shown in Scheme 1.
In order to analyze the structures of the putative products of
the reaction of PLP and HAs, we used 1H NMR and MS.
NMR analysis of the interaction of DHAs 1, 2, and 3 with
PLP allowed us to unambiguously identify the reaction
products as the syn- and anti-PLP-aldoximes shown in
Scheme 1. It is interesting to note that the experimentally
Downloaded by CZECH ACADEMY OF SCIENCES on November 5, 2009 | http://pubs.acs.org
observed rates of reaction of DHAs with PLP (i.e., 1 , 2 , 3)
agrees with the published order of hydrolytic stability of the
DHAs in acidic solution (1 > 2 > 3).37 One could speculate
that the aldoximes reported here are in fact products of the
Publication Date (Web): September 11, 2009 | doi: 10.1021/jm900775q
reaction of PLP with hydroxylamine liberated upon hydro-
lysis of DHAs. While HAs are generally known to be stable in
aqueous solutions, acid- or base-catalyzed hydrolysis is a
possibility. Hydrolysis experiments in 100 mM phosphate
buffer at pH 8.0 and pH 7.4 (see Supporting Information)
offer support for this proposal, showing that the DHAs that
react rapidly with PLP (2, 3, and 4) are prone to hydrolysis,
while nonreactive DHAs (1, 5, and 6) are hydrolytically stable
under these conditions. However, while the UV/vis, MS, and
NMR results draw a picture of the interaction of DHAs and
PLP in solution, it is important to note that it remains unclear
whether the observed enzymatic inhibition by DHAs results
from this nonspecific depletion of PLP in solution or from
specific interaction of the inhibitor and enzyme followed by
cofactor depletion within the active site.
The information gleaned from the NMR analysis further
highlights the complexity of interpreting the kinetic data
presented in Figure 2A and B. The enzymatic reaction velo-
cities were recorded at a single time point ranging from 8 to
15 min. However, the NMR measurements revealed that
hydroxylamine and 3 react with PLP to form PLP-aldoxime
within seconds, while the chemical reaction of PLP and 2 takes
considerably longer (on the order of minutes or hours). Like 3,
hydroxylamine behaves as a competitive inhibitor in our
assay conditions (data not shown), while 2 appears noncom-
petitive. Adding a time dimension to the enzymatic assay may
help to further unravel the differences between 2 and 3,
but such a detailed kinetic analysis is beyond the scope of
this work.
Another interesting point raised in this study is the observa-
tion that compounds that are closely structurally related
(differing only in the length of the carbon chain) can have
such different reactivities, both in buffered solution and in the
presence of PLP and enzyme. In this case, the intermediate
chain-length DHAs 2, 3, and 4 proved to be inhibitory and
able to react with PLP in solution according to UV/vis, MS,
and NMR analysis, while the compounds with very short (1)
and very long (5 and 6) carbon chains were not. One possibi-
lity is that some DHAs may undergo intramolecular conden-
sation, resulting in the release of hydroxylamine, a notion that
is supported by hydrolysis experiments (see Supporting In-
formation). The ability of analogues of 3 to cyclize to form
Hoffman et al.
5-membered rings at mildly alkaline pH has been demon-
strated.38 Cyclization of 4 would result in a 6-membered ring,
which is sterically favored. Noninhibitory compounds 1, 5,
and 6 would form 3-, 7-, and 9-membered rings, respectively;
these conformations are less likely to occur. The only outlier in
this scheme is the inhibitory compound 2. Intramolecular con-
densation of 2 would result in the formation of a 4-membered
ring, which is not a sterically favorable conformation but is
also not impossible to imagine. Alternatively, 2 may undergo
hydrolysis or interaction with PLP via a different mechanism,
perhaps reflected in its different mode of inhibition.
The findings presented here underscore the notion that
caution must be taken when analyzing the results of large-
scale inhibitor screening involving HAs and their derivatives.
Indeed, the limited stability of some HAs under physiological
conditions has been reported in other biomedicinal applica-
tions. Although HAs have been identified as one of the most
powerful inhibitors of Zn2þ-containing metalloproteases, in-
corporation of this pharmacophore into drugs is not common,
partly due to the rapid metabolism of the hydroxamic acid
moiety.39,25
In this work, we identify a series of DHAs capable of
nonspecific interaction with PLP-dependent enzymes via
modification of the cofactor to form a catalytically inactive
aldoxime species. In contrast to the DHAs, one of the newly
identified HA inhibitors, 10, derived from L-aspartic acid, is a
potent competitive SR inhibitor and exhibits moderate selec-
tivity for mouse and human SR. Although it also slowly reacts
with the cofactor PLP, it does so less efficiently than the
DHAs and yields a different mixture of products, in particular
an aldimine species that may act as a transition state mimetic.
Compound 10 could serve as a lead molecule for the devel-
opment of the next generation of SR inhibitors, though its cell
permeability and ability to cross the blood-brain barrier
must be addressed in the future. One can speculate that its
similarity to L-aspartic acid would allow it to use the same
physiological pathway.
Experimental Section
Materials. Enzymatic reaction mixtures were resolved on
a Zorbax Extend C18 reversed phase HPLC column (4.6 
250 mm, particle size 5 μm, Agilent Technologies, USA)
mounted on an Alliance 2795 HPLC system (Waters Co.,
Milford, MA, USA) coupled to a Waters 2487 dual wavelength
absorbance detector. For NMR experiments, the pH was mea-
sured using a glass electrode (Spintrode, Hamilton) and a
PHH2220 pH-meter (Radiometer) calibrated with Fisher Scien-
tific Buffers; the pH values were converted into pD values by
adding 0.40 as recommended.40 The 1H, 13C, and 15N NMR
spectra were measured on a Varian INOVA 500 spectrometer.
The instrumental setup and experimental parameters have been
described.41 Mass spectral data were recorded using a Q-TOF
microspectrometer from Waters Co. (Milford, MA, USA).
UV/vis absorbance scans were conducted on a Unicam UV500
UV/vis spectrophotometer.
D,L-Serine hydroxamate (7), D,L-threonine hydroxamate (8),
glycine hydroxamate (9), L-aspartic acid β-hydroxamate (10),
acetohydroxamic acid (15), and pyridoxal-50 -phosphate were
purchased from Sigma Aldrich and used without further pur-
ification. Experimental procedures for the synthesis of 11-14
can be found in Supporting Information. The purity and
identity of synthetic compounds were accessed by elemental
analysis and NMR spectroscopy. All compounds were found to
be at least 95% pure unless otherwise indicated.
Synthesis of Dihydroxamic Acids (DHAs; 1-6). Both 15N
labeled and nonlabeled DHAs were prepared from dicarboxylic
 Article
acid ethylesters by treatment with hydroxylamine.42 Detailed
procedures have been described.41 Briefly, the esters were con-
verted to sodium dihydroxamates using an excess of hydroxy-
lamine and sodium methoxide in dry methanol. The free DHAs
were liberated from aqueous solutions of dihydroxamates on a
weakly acidic catex column in Hþ-form. Lyophilization of
eluates and drying over P2O5 provided the final DHAs in good
yields (80-90%). The final preparation of compound 4 was
roughly 85% pure.
1-Hydroxy-2,5-pyrrolidinedione (N-hydroxysuccinimide)
was identified as a minor side-product (about 8 molar %)
formed during the synthesis of 3. It was purchased in pure
form from Sigma Aldrich. A freshly prepared solution of
N-hydroxysuccinimide (5 mM) did not significantly inhibit
mSR.
Detailed NMR spectral data has been published elsewhere
for compounds 1-6.41,43 NMR parameters for the predominant
Z-conformers are given here.
Oxalodihydroxamic acid (1): 1H NMR (300 MHz, DMSO-
d6) δ 11.47 (bs, 2H), 9.16 (bs, 2H); 13C NMR (75 MHz, DMSO-
Downloaded by CZECH ACADEMY OF SCIENCES on November 5, 2009 | http://pubs.acs.org
d6) δ 157.02; 15N NMR (50 MHz, DMSO-d6) δ -212.5.
Malonodihydroxamic acid (2): 1H NMR (300 MHz, DMSO-
d6) δ 10.39 (bs, 2H), 8.93 (bs, 2H), 2.75 (s, 2H); 13C NMR
(75 MHz, DMSO-d6) δ 163.65 (CO), 38.46 (CH2); 15N NMR
Publication Date (Web): September 11, 2009 | doi: 10.1021/jm900775q
(50 MHz, DMSO-d6) δ -213.4.
Succinodihydroxamic acid (3): 1H NMR (300 MHz, DMSO-
d6) δ 10.36 (bs, 2H), 8.69 (bs, 2H), 2.17 (s, 4H); 13C NMR
(75 MHz, DMSO-d6) δ 168.38 (CO), 27.99 (CH2); 15N NMR
(50 MHz, DMSO-d6) δ -216.1.
Glutarodihydroxamic acid (4): 1H NMR (300 MHz, DMSO-
d6) δ 10.36 (bs, 2H), 8.67 (bs, 2H), 1.94 (t, J = 7.2 Hz, 4H), 1.70
(p, J = 7.2 Hz, 2H); 13C NMR (75 MHz, DMSO-d6) δ 168.88
(CO), 31.92 (CH2, 2C), 21.58 (CH2, 1C); 15N NMR (50 MHz,
DMSO-d6) δ -216.1.
Adipodihydroxamic acid (5): 1H NMR (300 MHz, DMSO-
d6) δ 10.34 (bs, 2H), 8.68 (bs, 2H), 1.93 (unresolved multiplet,
4H), 1.4 (unresolved multiplet, 4H); 13C NMR (75 MHz,
DMSO-d6) δ 169.14 (CO), 32.28 (CH2), 25.02 (CH2); 15N
NMR (50 MHz, DMSO-d6) δ -215.0.
Suberodihydroxamic acid (6): 1H NMR (300 MHz, DMSO-
d6) δ 10.37 (bs, 2H), 8.71 (bs, 2H), 1.92 (t, J = 7.4 Hz), 1.45
(unresolved multiplet, 4H), 1.19 (unresolved multiplet, 4H);
13
C NMR (75 MHz, DMSO-d6) δ 169.26 (CO), 32.41, 28.50,
25.21; 15N NMR (50 MHz, DMSO-d6) δ -215.2.
Mouse Serine Racemase Preparation. Recombinant mouse
serine racemase was prepared as previously described.8 Briefly,
calcium chloride competent E. coli MC1061 cells were trans-
formed with pKS-mSR.8 Cell cultures were grown in nutrient-
rich medium at 37 C and 300 rpm. Serine racemase expression
was induced by 1 mM L-arabinose (Sigma) at OD600 = 0.8.
After 5 h of induction, bacteria were harvested by centrifugation
(10000g, 10 min, 4 C) and stored at -70 C.
The purification procedure was carried out in QA buffer
[20 mM triethanolamine hydrochloride-NaOH, pH 7.0,
1 mM MgCl2, 20 μM PLP, 0.1 mM D,L-dithiothreitol, and
0.02% (w/v) NaN3]. Pellets of induced bacteria were suspended
in QA buffer supplemented with 1 mM phenylmethanesulfonyl
fluoride, 0.2 mg/mL chicken egg lysozyme, and 0.05% (w/v)
sodium deoxycholate and homogenized using a glass Dounce
homogenizer. After the addition of 10 mM MgCl2 and DNase I
(Roche Molecular Biochemicals) to 10 μg/mL, the cell suspen-
sion was sonicated and centrifuged at 20000g for 30 min at 4 C.
The supernatant was subjected to the following three chro-
matographic steps: reverse phase (Phenyl-Sepharose FastFlow,
Pharmacia), ion-exchange (Q-Sepharose, FastFlow, Pharma-
cia), and affinity chromatography (ATP-agarose, Sigma).
The resulting 90% pure protein preparation was dialyzed three
times against 200 volumes of QA buffer at 4 C, concentrated to
1 mg/mL (as determined by quantitative amino acid analysis),
and stored at -70 C.
Journal of Medicinal Chemistry, 2009, Vol. 52, No. 19 6039
SR Inhibition Screening. Inhibition of mSR was analyzed as
described.8 Briefly, the racemization reactions were performed
in a pH 8.0 buffer supplemented with 10 μM PLP, 1 mM ATP,
1 mM MgCl2, and 5 mM DTT. For the initial screening, a 5 mM
concentration of both the substrate L-serine and the test com-
pound was used. Reactions proceeded for 30 min at 37 C and
were stopped by the addition of 0.3 M HClO4. After deri-
vatization with 1-fluoro-2,4-dinitrophenyl-5-L-alanine amide
(FDAA, Marfey’s reagent), the substrate L-serine and the
product D-serine were resolved and detected by HPLC.
For the determination of IC50 values, various inhibitor con-
centrations were used while the L-serine concentration was kept
constant at 5 mM, and the data were analyzed with GraFit
5 software.19 For the determination of the mechanism of
inhibition, 4 or 5 substrate concentrations (from 1.25 to
12 mM L-serine) and a minimum of 2 inhibitor concentrations
and a noninhibited reaction set were used. We performed a
preliminary experiment with multiple reaction time points in
order to choose reaction times that approach the initial velocity
of the enzymatic reaction and result in detectable production of
D-serine (not shown). The reaction time of the experiments
presented here ranged from 8 to 15 min, and the conversion of
L-serine to D-serine was 0.5-2%. All reactions were performed
in triplicate, and the averages and corresponding standard
deviations were analyzed with DynaFit4 software, which calcu-
lates the least-squares nonlinear regressions and compares the
possible mechanisms of action.18 The Ki values were calcu-
lated with DynaFit4 from one representative independent
experiment. The following inhibitor concentrations were used
for the determination of inhibitory constants: 2 (0, 150, and
400 μM), 3 (0, 7.5, 10, and 30 μM), and 10 (0, 100, and 800 μM).
Lineweaver-Burk and Eadie-Hofstee transformations were
performed with GraFit5 software.19
Inhibition Screening with Other PLP-Dependent Enzymes.
Alanine racemase from Bacillus stearothermophilus (AR) and
broad-range bacterial transaminase (TA) were purchased from
Sigma Aldrich as lyophilized powders.
The plasmid pCW-SDH, which encodes rat liver serine
dehydratase (SDH), was a generous gift from H. Ogawa.20 To
express SDH, pCW-SDH was transformed into E. coli BL21-
RIL cells, and the cells were cultured in 1 L of Luria-Bertani
broth at 37 C. When the culture reached an OD600 of approxi-
mately 0.5, IPTG was added to a final concentration of 0.5 mM.
The cells were induced for 17 h at 37 C, then harvested at 2000g
for 15 min at 4 C. The cell pellet was resuspended in 20 mL of
20 mM Tris-HCl buffer, pH 8.0, containing 2 mM EDTA and
5 mM DTT. Ten milligrams of lysozyme and one protease
inhibitor tablet (Complete Mini EDTA free, Roche) were
added, and cells were subjected to freeze/thaw and sonication.
The soluble fraction was clarified by centrifugation, and
SDS-PAGE analysis of the supernatant revealed that SDH
was heavily overexpressed. To partially purify SDH, the lysate
was diluted with buffer to a final volume of 25 mL, and 5 g of
ammonium sulfate was added. After incubation at 4 C for 1 h,
the SDH-containing precipitate was separated by centrifugation
and then redissolved in lysis buffer. This resulted in an SDH
preparation of about 50% purity, as judged by SDS-PAGE
stained by Coomassie Brilliant Blue G-250.
The preparation of recombinant human serine racemase has
been described,14 and its activity was assayed according to the
protocol for mouse SR described above.
AR, SDH, and TA were assayed in 50 mM phosphate buffer,
pH 8.0, containing 10 μM PLP. The enzymes were preincubated
with 5 mM inhibitor at room temperature for 15-20 min.
Reactions were started by the addition of substrate at a con-
centration close to the KM of the enzyme (5 mM D-alanine for
AR, 50 mM L-serine for SDH, and 5 mM L-aspartic acid and
5 mM R-ketoglutarate for TA). Reactions were allowed to
proceed for 10-40 min at 50 C (AR) or 37 C (SDH and
TA). Reactions were quenched by the addition of TCA to a final
 6040 Journal of Medicinal Chemistry, 2009, Vol. 52, No. 19
concentration of 5% (w/v), and protein pellets were separated
by centrifugation. The supernatants were extracted twice with
water-saturated diethyl ether prior to derivatization. SDH
reaction mixtures were derivatized with 2,4-dinitrophenylhy-
drazine (DNPH), and pyruvate formation was monitored on
HPLC as previously described.8,44 AR and TA reaction mix-
tures were derivatized with FDAA and separated according to
the protocol for separation of L- and D-serine described above.
AR mixtures were analyzed for the formation of L-alanine, and
TA mixtures were analyzed for the formation of L-glutamic acid.
Pure L-alanine, D-alanine, L-aspartic acid, and L-glutamic acid
were derivatized and used as calibration standards.
NMR Measurements. An approximately 8 mM stock solution
of PLP was prepared in 100 mM phosphate buffer in D2O, pD
8.4. An appropriate amount of hydroxamic acid (chosen to
ensure an approximately 1:1 molar ratio between the acid and
PLP) was weighed out and mixed into 2 mL of PLP stock
solution. The final pD was adjusted to 7.1-8.1 by the addition
of NaOH powder. For 13C NMR measurements (see Supporting
Information), an approximately 10-fold higher PLP concen-
Downloaded by CZECH ACADEMY OF SCIENCES on November 5, 2009 | http://pubs.acs.org
tration was used.
Mass Spectrometry. A 4, 5, or 30 mM solution of PLP (pH
adjusted to 8.5 with NaOH) was mixed with a 10-fold molar
excess of the test compound and incubated at room temperature
Publication Date (Web): September 11, 2009 | doi: 10.1021/jm900775q
for one day prior to measurement. Mixtures were prepared both
with and without a pH control (200 mM ammonium sulfate, pH
8.0). The mixture was diluted in water 40- to 100-fold and
subjected to ESI-MS in a negative ion mode.
Spectrophotometric Assay for the Reaction of PLP with
Hydroxamic Acids. A 0.6 mM solution of pyridoxal-50 -phos-
phate and 10 mM solutions of test compounds were prepared in
100 mM phosphate buffer, pH 8.0. The PLP and inhibitor
solutions were mixed in a 1:1 ratio so that the final concentration
of PLP was 300 μM, and the final concentration of inhibitor was
5 mM. The reaction mixtures were incubated at room tempera-
ture for 1 h prior to measurement. Absorbance spectra were
collected in the 310-500 nm range.
Acknowledgment. This work was financially supported by
the Ministry of Education of the Czech Republic (grant
1M0508) and the Grant Agency of the Academy of Sciences
of the Czech Republic (grant IAA 400720706).We would like
to thank Mr. Roman Vanı́cek for providing us with the very
helpful Sample Reader1 program and Mrs. Ludmila
Soukupova for the synthesis of some hydroxamic acids. We
would also like to thank the anonymous reviewers of this
manuscript for their inspiring comments.
Supporting Information Available: Synthetic procedures for
compounds 11-14 and additional spectroscopic measurements
(NMR and UV/vis).
This material is available free of charge via the Internet at
http://pubs.acs.org.
References
(1) Mothet, J.-P.; Parent, A. T.; Wolosker, H.; Brady, R. O., Jr.;
Linden, D. J.; Ferris, C. D.; Rogawski, M. A.; Snyder, S. H. D-
Serine is an endogenous ligand for the glycine site of the N-methyl-
D-aspartate receptor. Proc. Natl. Acad. Sci. U.S.A. 2000, 97 (9),
4926–4931.
(2) Monaghan, D. T.; Bridges, R. J.; Cotman, C. W. The excitatory
amino acid receptors: their classes, pharmacology, and distinct
properties in the function of the central nervous system. Annu. Rev.
Pharmacol. Toxicol. 1989, 29, 365–402.
(3) Hashimoto, K.; Fukushima, T.; Shimizu, E.; Okada, S.; Komatsu,
N.; Okamura, N.; Koike, K.; Koizumi, H.; Kumakiri, C.; Imai, K.;
Iyo, M. Possible role of D-serine in the pathophysiology of
Alzheimer0 s disease. Prog. Neuropsychopharmacol. Biol. Psychiatry
2004, 28 (2), 385–388.
(4) Sasabe, J.; Chiba, T.; Yamada, M.; Okamoto, K.; Nishimoto, I.;
Matsuoka, M.; Aiso, S. D-Serine is a key determinant of glutamate
Hoffman et al.
toxicity in amyotrophic lateral sclerosis. EMBO J. 2007, 26 (18),
4149–4159.
(5) Krystal, J. H.; D0 Souza, D. C. D-serine and the therapeutic
challenge posed by the N-methyl-D-aspartate antagonist model
of schizophrenia. Biol. Psychiatry 1998, 44 (11), 1075–1076.
(6) Toney, M. D. Reaction specificity in pyridoxal phosphate enzymes.
Arch. Biochem. Biophys. 2005, 433 (1), 279–287.
(7) Amadasi, A.; Bertoldi, M.; Contestabile, R.; Bettati, S.; Cellini, B.;
di Salvo, M. L.; Borri-Voltattorni, C.; Bossa, F.; Mozzarelli, A.
Pyridoxal 50 -phosphate enzymes as targets for therapeutic agents.
Curr. Med. Chem. 2007, 14 (12), 1291–1324.
(8) Strisovsky, K.; Jiraskova, J.; Mikulova, A.; Rulisek, L.; Konva-
linka, J. Dual substrate and reaction specificity in mouse serine
racemase: identification of high-affinity dicarboxylate substrate
and inhibitors and analysis of the beta -eliminase activity. Bio-
chemistry 2005, 44 (39), 13091–13100.
(9) Lossen, H. Uber die oxalohydroxamsaure. Justus Liebigs Ann.
Chem. 1869, 150 (3), 314–322.
(10) Marmion, C. J.; Griffith, D.; Nolan, K. B. Hydroxamic acids: An
intriguing family of enzyme inhibitors and biomedical ligands. Eur.
J. Inorg. Chem. 2004, 15, 3003–3016.
(11) Muri, E. M. F.; Nieto, M. J.; Sindelar, R. D.; Williamson, J. S.
Hydroxamic acids as pharmacological agents. Curr. Med. Chem.
2002, 9 (17), 1631–1653.
(12) Muri, E. M. F.; Nieto, M. J.; Williamson, J. S. Hydroxamic acids as
pharmacological agents: an update. Med. Chem. Rev. 2004, 1 (4),
385–394.
(13) Hockly, E.; Richon, V. M.; Woodman, B.; Smith, D. L.; Zhou, X.;
Rosa, E.; Sathasivam, K.; Ghazi-Noori, S.; Mahal, A.; Lowden, P.
A.; Steffan, J. S.; Marsh, J. L.; Thompson, L. M.; Lewis, C. M.;
Marks, P. A.; Bates, G. P. Suberoylanilide hydroxamic acid, a
histone deacetylase inhibitor, ameliorates motor deficits in a mouse
model of Huntington0 s disease. Proc. Natl. Acad. Sci. U.S.A. 2003,
100 (4), 2041–2046.
(14) Hoffman, H. E.; Jiraskova, J.; Ingr, M.; Zvelebil, M.; Konvalinka,
J. Recombinant human serine racemase: enzymologic character-
ization and comparison with its mouse ortholog. Protein Expres-
sion Purif. 2009, 63 (1), 62–67.
(15) Dunlop, D. S.; Neidle, A. Regulation of serine racemase activity by
amino acids. Brain Res. Mol. Brain Res. 2005, 133 (2), 208–214.
(16) Strisovsky, K.; Jiraskova, J.; Barinka, C.; Majer, P.; Rojas, C.;
Slusher, B. S.; Konvalinka, J. Mouse brain serine racemase cata-
lyzes specific elimination of L-serine to pyruvate. FEBS Lett. 2003,
535 (1-3), 44–48.
(17) Cook, S. P.; Galve-Roperh, I.; Martinez del Pozo, A.; Rodriguez-
Crespo, I. Direct calcium binding results in activation of brain
serine racemase. J. Biol. Chem. 2002, 277 (31), 27782–27792.
(18) Kuzmic, P. Program DYNAFIT for the analysis of enzyme kinetic
data: application to HIV proteinase. Anal. Biochem. 1996, 237 (2),
260–273.
(19) Leatherbarrow, R. J. GraFit Version 5; Erithacus Software Ltd.:
Horley, U.K., 2001.
(20) Ogawa, H.; Takusagawa, F.; Wakaki, K.; Kishi, H.; Eskandarian,
M. R.; Kobayashi, M.; Date, T.; Huh, N.-H.; Pitot, H. C. Rat liver
serine dehydratase. Bacterial expression and two folding domains
as revealed by limited proteolysis. J. Biol. Chem. 1999, 274 (18),
12855–12860.
(21) Harris, C. M.; Johnson, R. J.; Metzler, D. E. Band-shape analysis
and resolution of electronic spectra of pyridoxal phosphate and
other 3-hydroxypyridine-4-aldehydes. Biochim. Biophys. Acta
1976, 421 (2), 181–194.
(22) Leussing, D. L. Vitamin B6 Pyridoxal Phosphate: Chemical, Bio-
chemical and Medical Aspects Part A; John Wiley & Sons: New York,
1986; Vol. 1.
(23) Kallen, R. G.; Korpela, T.; Martell, A. E.; Matsushima, Y.;
Metzler, C. M.; Metzler, D. E.; Morozov, Y. V.; Ralston, I. M.;
Savin, F. A.; Torchinsky, Y. M.; Ueno, H. Transaminases; John
Wiley & Sons: New York, 1985.
(24) Rosenstein, I. J.; Hamilton-Miller, J. M. Inhibitors of urease as
chemotherapeutic agents. Crit. Rev. Microbiol. 1984, 11 (1), 1–12.
(25) Grant, S.; Easley, C.; Kirkpatrick, P. Vorinostat. Nature Rev. Drug
Discovery 2007, 6 (1), 21–22.
(26) Metzler, D. E.; Snell, E. E. Deamination of serine. I. Catalytic
deamination of serine and cysteine by pyridoxal and metal salts.
J. Biol. Chem. 1952, 198 (1), 353–361.
(27) Snell, E. E. Summary of known metabolic functions of nicotinic
acid, riboflavin and vitamin B6. Physiol. Rev. 1953, 33 (4), 509–524.
(28) Olivard, J.; Metzler, D. E.; Snell, E. E. Catalytic racemization of
amino acids by pyridoxal and metal salts. J. Biol. Chem. 1952, 199
(2), 669–674.
(29) Inoue, R.; Hashimoto, K.; Harai, T.; Mori, H. NMDA- and
beta-amyloid1-42-induced neurotoxicity is attenuated in serine
 Article
racemase knock-out mice. J. Neurosci. 2008, 28 (53), 14486–
14491.
(30) Dixon, S. M.; Li, P.; Liu, R.; Wolosker, H.; Lam, K. S.; Kurth, M.
J.; Toney, M. D. Slow-binding human serine racemase inhibitors
from high-throughput screening of combinatorial libraries. J. Med.
Chem. 2006, 49 (8), 2388–2397.
(31) Koval, D.; Jiraskova, J.; Strisovsky, K.; Konvalinka, J.; Kasicka,
V. Capillary electrophoresis method for determination of D-serine
and its application for monitoring of serine racemase activity.
Electrophoresis 2006, 27 (13), 2558–2566.
(32) Simanenko, Y. S.; Prokop0 eva, T. M.; Popov, A. F.; Bunton, C. A.;
Karpichev, E. A.; Savelova, V. A.; Ghosh, K. K. O-Nucleophilicity
of hydroxamate ions in reactions with ethyl 4-nitrophenyl
ethylphosphonate, diethyl 4-nitrophenyl phosphate, and 4-nitro-
phenyl 4-toluenesulfonate. Russ. J. Org. Chem. 2004, 40 (9), 1337–
1350.
(33) Wolosker, H.; Blackshaw, S.; Snyder, S. H. Serine racemase: a glial
enzyme synthesizing D-serine to regulate glutamate-N-methyl-
D-aspartate neurotransmission. Proc. Natl. Acad. Sci. U.S.A.
1999, 96 (23), 13409–13414.
(34) Yamada, T.; Komoto, J.; Takata, Y.; Ogawa, H.; Pitot, H. C.;
Takusagawa, F. Crystal structure of serine dehydratase from rat
liver. Biochemistry 2003, 42 (44), 12854–12865.
Downloaded by CZECH ACADEMY OF SCIENCES on November 5, 2009 | http://pubs.acs.org
(35) Shaw, J. P.; Petsko, G. A.; Ringe, D. Determination of the
structure of alanine racemase from Bacillus stearothermophilus
at 1.9-Å resolution. Biochemistry 1997, 36 (6), 1329–1342.
Publication Date (Web): September 11, 2009 | doi: 10.1021/jm900775q
Journal of Medicinal Chemistry, 2009, Vol. 52, No. 19 6041
(36) The labeled and unlabeled compounds behaved identically in the
inhibition screening assay.
(37) Ghosh, K. K.; Patle, S. K.; Sharma, P.; Rajput, S. K. A comparison
between the acid-catalysed reactions of some dihydroxamic acids,
monohydroxamic acids and desferal. Bull. Chem. Soc. Jpn. 2003, 76
(2), 283–290.
(38) Notari, R. E. Hydroxamic acids. II. Kinetics and mechanisms of
hydroxyaminolysis of succinimide. J. Pharm. Sci. 1969, 58 (9),
1064–1068.
(39) Breuer, E.; Frant, J.; Reich, R. Recent non-hydroxamate matrix
metalloproteinase inhibitors. Expert Opin. Ther. Pat. 2005, 15 (3),
253–269.
(40) Glasoe, P. K.; Long, F. A. Use of glass electrodes to mea-
sure acidities in deuterium oxide. J. Phys. Chem. 1960, 64,
188–190.
(41) Schraml, J.; Cigler, P. 15N-1H and 15N-13C couplings in 15N-
enriched dihydroxamic acids. Magn. Reson. Chem. 2008, 46 (8),
748–755.
(42) Hurd, C. D.; Botteron, D. G. Hydroxamic acids from aliphatic
dicarboxylic acids. J. Org. Chem. 1946, 11, 207–214.
(43) Schraml, J.; Kvı́calova, M.; Blechta, V.; Soukupova, L.; Exner, O.;
Boldhaus, H.-M.; Erdt, F.; Bliefert, C. Structure determination of
dihydroxamic acids and their trimethylsilyl derivatives by NMR
spectroscopy. Magn. Reson. Chem. 2000, 38 (9), 795–801.
(44) Neidle, A.; Dunlop, D. S. Allosteric regulation of mouse brain
serine racemase. Neurochem. Res. 2002, 27 (12), 1719–1724.
 Supporting Information

Hydroxamic acids as a novel family of serine racemase

inhibitors: mechanistic analysis reveals different modes of

interaction with the pyridoxal-5’-phosphate cofactor

Hillary E. Hoffman1,2#, Jana Jirásková1,2#, Petr Cígler1,##, Miloslav Šanda1,

Jan Schraml3, and Jan Konvalinka1,2*

1
Gilead Sciences and IOCB Research Center, Institute of Organic Chemistry and

Biochemistry of the ASCR, v. v. i., Flemingovo n. 2, 166 10 Prague 6, Czech Republic;

2
Department of Biochemistry, Faculty of Science, Charles University, Hlavova 8, Prague

2, Czech Republic; 3Institute of Chemical Process Fundamentals of the ASCR, v. v. i.,

Rozvojová 135, 165 02 Prague 6, Czech Republic

#
These two authors contributed equally to this work.
##
Current affiliation: The Scripps Research Institute, 10550 North Torrey Pines Road, La

Jolla, California 92037, USA

*
corresponding author. Tel: 420-220183218; Fax: 420-220183578; E-mail:

konval@uochb.cas.cz

S1
 Table of Contents

Synthesis of ring-substituted benzohydroxamic acids (11, 12, 13) S3

Synthesis of formohydroxamic acid (14) S3

NMR Spectra of PLP S4

Table 1S S5

Model experiment with NH2OH: NMR assignment of aldoxime signals S5

Table 2S S7

NMR analysis of the reaction of PLP and DHAs S8

Table 3S S9

Concentration-dependence of the reaction: spectrophotometric analysis S9

Figure 1S S10

UV/Vis data for compounds 4 – 6 S10

Figure 2S S11

Hydrolysis of DHAs at pH 8.0 and 7.4 S11

Figure 3S S12

References S13

S2
Synthesis of ring-substituted benzohydroxamic acids (11, 12, 13)

The synthesis and thorough NMR characterization of these compounds has been

published elsewhere.1 Briefly, the compounds were prepared from carboxylic esters in

methanol or from acyl chlorides in chloroform according to standard procedures.2, 3 In

the case of 11, the reaction of the methyl ester in aqueous solution was followed by

extraction of the acid into ethanol. All compounds were recrystallized from water,

aqueous acetone, acetone-pentane, or aqueous ethanol.

4-aminobenzohydroxamic acid (11) 1H NMR (300 MHz, DMSO-d6) δ 10.730 (bs, 1H),
13
8.643 (bs, 1H), 7.48 (d, J = 8.5 Hz, 2H), 6.53 (d, J = 8.5 Hz, 2H); C NMR (75 MHz,

DMSO-d ) δ 112.73, 119.26, 128.46, 151.81, 165.21; 15N NMR (50 MHz, DMSO-d6) δ -
6

218.3 (NHOH), -314.5 (NH2).

4-chlorobenzohydroxamic acid (12) 1H NMR (300 MHz, DMSO-d6) δ 11.301 (bs, 1H),
13
9.107 (bs, 1H), 7.76 (m, 2H), 7.54(m, 2H); C NMR (75 MHz, DMSO-d ) δ 128.62,
6

128.92, 131.70, 136.06, 163.34; 15N NMR (50 MHz, DMSO-d6) δ -215.3 (NHOH).

4-nitrobenzohydroxamic acid (13) 1H NMR (300 MHz, DMSO-d6) δ 11.558 (bs, 1H),

9.315 (bs, 1H), 8.31 (d, J = 8.6 Hz, 2H), 8.00 (d, J = 8.6 Hz, 2H);13C NMR (75 MHz,

DMSO-d ) δ 123.78, 128.57, 138.71, 149.18, 162.54; 15N NMR (50 MHz, DMSO-d6) δ -
6

213.2 (NHOH), -11.2 (NO2).

Synthesis of formohydroxamic acid (14)

Formohydroxamic acid (14) was prepared from formic acid ethyl ester using a similar

protocol as for DHA synthesis. Briefly, 60 mL of sodium methoxide was added to a

solution of 0.145 mol hydroxylamine hydrochloride in a 2°C water bath. The solution

was filtered to remove NaCl, and the filtrate was treated with 0.189 mol ethyl formate.

S3
The mixture was incubated overnight at room temperature, and methanol was removed in

a rotary evaporator (34°C, 15 mbar). The solid phase was extracted several times with

methylacetate, which resulted in the formation of white crystals. The crystals were

filtered, washed with dichloromethane, and dried. Excess formic acid was removed under

high vacuum. The spectroscopic parameters of the final product were consistent with the

published values.4

NMR Spectra of PLP

Literature 1H NMR data on PLP are summarized in Table 1S along with experimental
13
values obtained in this work. C NMR parameters and their pH dependence have also

been reported in literature.5-7 However, due to the sensitivity issue, they were obtained

either after selective 13C enrichment7 or at concentrations far exceeding those relevant to

biochemical studies.5,6

Our preliminary experiments indicated that 1H NMR spectroscopy is well suited to follow

the putative reaction of PLP and DHAs. The spectra consist of well separated singlets and
31
closely spaced doublets (due to CH2 proton coupling with the P nucleus of the

phosphate moiety), and the reaction with DHAs does not lead to any signal overlap in the

most significant region (δ = 10.5 – 7.0 ppm). A small overlap occurs with CH2O proton

signals, which have signals close to the residual solvent signal.

S4
Table 1S. 1H NMR chemical shifts (and 31P-1H coupling constants) of PLP in water

Solution CHO C6-H 5-CH2O 2-CH3 Reference

(arom)
8, 9
Acidic 10.50 8.20 5.05 2.60

(pD=4.1) 6.53(hydrate)
9
Neutral 10.40 7.77 5.03 (7 Hz) 2.43

(pD=7.25)
8
Neutral 10.40 7.72 5.00 (5 Hz) 2.40

(pD=7.8)
8
Alkaline 10.37 7.62 4.95 2.34

(pD=9.4)

100mM 10.28 7.61 4.94 2.31 this work

phosphate (0.5 Hz) (6.0 Hz)

buffer

(pD=8.42)

Model experiment with NH2OH: NMR assignment of aldoxime signals

15
Two products were formed from the reaction of PLP with NH2OH HCl. The products

were formed in an 88:12 ratio with signals at δ = 8.345 and 7.839 split into doublets by

coupling to the 15N nucleus [with 2J(15N-1H) of 2.4 and 14.0 Hz, respectively, (see Table
15
3S) as confirmed by a N decoupling experiment (1H{15N})]. The major product was

identified as PLP-aldoxime with syn geometry (see Scheme 1), as our values match those

previously reported for the syn isomer of PLP-aldoxime [δ = 8.43 and 2J(15N-1H) = 2.7

S5
Hz].10 Though the corresponding parameters for the anti isomer have not been reported,

the 1H chemical shifts of the minor product roughly agree with those reported for anti

isomers of other aromatic oximes (δ = 7.2 – 7.7 ppm)11 as does the 2J(15N-1H) coupling

[2J(15N-1H) = 13.8 – 18.8 Hz].10 The failure of previous authors10 to observe this anti

isomer is paralleled by similar observations on oximes of other aromatic aldehydes,

especially those containing phenolic OH groups (as found in PLP). The OH group is

thought to autocatalyze rearrangement of anti into syn isomer.12 Purification of the

reaction products by crystallization removes the less abundant isomer, while we directly

measured the unpurified reaction mixture, which enabled us to observe the anti isomer.

[An analogous situation was recognized for aliphatic aldoximes which appeared isolable

in a single form (anti) only.13 In a recent study of closely related PLP-aldimines

Limbach’s group observed only E isomers14 analogous to our syn isomer.]

The reaction could be carried out at sufficiently high concentration (81 mM PLP) to
13
allow measurements of C NMR spectra in a reasonable time. The results are

summarized in Table 2S.

S6
 13 15
Table 2S. C NMR chemical shifts (and N-13C/31P-13C coupling constants in Hz in

parentheses/brackets, respectively) of products of reaction of PLP with 15NH2-OH (pD =

8.04, 81mM PLP); the shifts are referenced to 1,4-dioxane at δ(13C) = 66.30.

Product CH=N C2/3/4/5/6 5-CH2O 2-CH3

syn-PLP- 147.37 (5.3) 146.19/154.49/123.44(4.8)/ 61.35 16.39

oxime 131.38 <8.7>/131.91

anti-PLP- 146.29 (<2.5)a 146.86/157.98/126.58a/ 61.24 16.22

oxime 133.44<8.7>/128.32
a
Coupling not resolved

The assignment of the two products as the syn and anti PLP-aldoxime pictured in Scheme

1 is in agreement with the results of APT and gHMBC experiments. Additionally, the

small, unresolved 1J(15N-13C) coupling in the C=N group of the anti isomer as well as the

coupling observed in the syn isomer (5.3 Hz) are in line with the couplings reported for

other oximes.15 The 15N chemical shifts of the syn and anti-PLP-aldoximes are δ = -26.5

and -15.1 ppm, respectively (relative to neat nitromethane in an external capillary). The

shift of syn-PLP-aldoxime agrees well with the shift reported for syn-benzaldoxime [δ = -

26.3, (E)-benzaldehyde oxime];16 in aliphatic aldoximes the anti isomers are also less

shielded than their syn counterparts (though the differences are smaller than observed

here).16

S7
NMR analysis of the reaction of PLP and DHAs

Three products (A, B, and C) are formed from the reaction between PLP and DHA; the

ratios of their abundance depend on the nature of the acid and change during the course

of the reaction (other factors, such as concentrations of components, ionic strength,

temperature, pD, etc. might also affect the ratio, but a detailed study of their influence

was beyond the scope of this work). In the case of 2 the 1H NMR signals (Table 3S) of

products A and B appear first, but during the course of the reaction product A is

converted completely into B and C. The final reaction mixture consists primarily of B

with 11 – 25 % of product C. In the case of 3 the reaction goes to completion in less than

a minute, and only products B and C can be observed (C constitutes 11 % of the total). In

contrast, the reaction of 1 with PLP is very slow; even after 12 days roughly 20% of the

PLP starting material remains. An intermediate product A is formed, but it has different

chemical shifts than intermediate A observed from the reaction of PLP with 2. Each of

the products (A, B, and C) has the same number of signals as the parent PLP, and the

signals are in the vicinity of those of PLP except for the CHO signal. Characteristic

features of the upfield lines of products B and C are their splittings due to 15N couplings
15
observed in experiments with N-enriched DHAs. These observations suggest that the

products retain the carbon skeleton of PLP and changes occur only at the aldehyde group.

From comparison with the spectral properties of syn- and anti-PLP-aldoxime, we

identified product B as the syn aldoxime and product C as the anti aldoxime (see Table

3S). The structure(s) of intermediate A were not further investigated.

S8
Table 3S. 1H NMR chemical shifts of products A, B, and C of the reaction of PLP with
15
N enriched 2 (as disodium salt at pD = 7.99, 8 mM concentration of PLP) and of
15
aldoximes resulting from reaction of PLP with NH2OH (pD = 8.04, 81 mM PLP).
15
Values in parenthesis/brackets are N-1H/31P-1H coupling constants in Hz, respectively.

The chemical shifts are referenced to the line of 1,4-dioxane at δ(1H) = 3.530.

Product CH=N C6-H (arom) 5-CH2O 2-CH3

Aa 8.77 (0b) 7.58 4.78[5.2] 2.53

B 8.397 (2.6) 7.667 4.699 [5.2] 2.231

C 7.767 (14.1) 7.566 –c 2.246

syn-PLP-oxime 8.278 (2.4) 7.584 4.623 [5.3] 2.075

anti-PLP-oxime 7.772 (14.0) 7.502 4.553 [5.3] 2.184

a
The lines shift to some extent in the course of the reaction. bSinglet. cNot visible due to

overlap with the residual solvent signal.

Concentration-dependence of the reaction: spectrophotometric analysis

PLP (0.1 mM) was incubated with various concentrations of hydroxylamine, 2, or 3.

The reactions were performed in 100 mM HEPES buffer, pH 8.0, and were allowed to

proceed for 30 min at ambient temperature. The absorbance at 388 nm, which

corresponds to unmodified PLP, was then recorded. The decrease in absorbance at 388

nm was plotted as a function of inhibitor concentration (mM) divided by PLP

concentration. The points were fitted using the “IC50 full 4 parameter” function in the

GraFit software package 17.

S9
As can be seen from Figure 1S, hydroxylamine is the most reactive with PLP, followed

by 3 and then 2. A 1:1 ratio of hydroxylamine:PLP is sufficient to convert almost all of

the PLP to aldoxime form, and 3 is only slightly less effective than hydroxylamine. In

contrast, an excess of 2 over PLP must be employed to result in substantial conversion to

the aldoxime form.

Figure 1S. Concentration-dependence of the reaction of PLP and DHAs monitored by

UV/vis spectroscopy.

UV/Vis data for compounds 4-6

UV/vis measurements were conducted as in the Experimental Section. Non-inhibitory

compounds 5 and 6 do not interact with PLP in solution, while addition of 4, which is

inhibitory, to PLP results in a spectral shift consistent with aldoxime formation.

S10
Figure 2S. Absorbance spectra of PLP and compounds 4, 5, 6 recorded at pH 8.0.

Hydrolysis of DHAs at pH 8.0 and 7.4

We performed hydrolysis experiments at pH 8.0 and pH 7.4 by detection of

18
hydroxylamine in solutions of hydroxamic acids as previously described. The method

exploits the use of picrylsulfonic acid, which reacts with primary amines at alkaline pH to

produce a species that absorbs light at 494 nm. This method is very sensitive and can

detect low micromolar concentrations of hydroxylamine; the major disadvantage is that

picrylsulfonic acid takes 15 min to react with hydroxylamine, so detailed kinetic studies

were not possible with this method. Briefly, 1 mL of 100 mM phosphate buffer, pH 8.0

or 7.4, containing 0.5 mM (1, 5, 6) or 0.05 mM (2, 3, 4) DHA (diluted from freshly

prepared 10 mM stocks) was prepared. The compounds were incubated at room

temperature for 0, 30, or 60 min, and then 15 µL picrylsulfonic acid (5 % w/v, Sigma

Aldrich) was added. Absorbance at 494 nm was recorded 15 min after addition of

picrylsulfonic acid.

Compounds 1, 5, 6 (non-inhibitory) did not show any significant absorbance even after 1

hour incubation at pH 8.0 or pH 7.4, indicating that no release of hydroxylamine

occurred. Incubation of 3 in buffer shows immediate evolution of hydroxylamine, and

S11
the absorbance does not substantially increase at subsequent time points. In contrast, 2

and 4 show a slow, steady increase in hydroxylamine evolution throughout the hour at

both pH 8.0 (Figure 3S) and 7.4.

Figure 3S. Evolution of hydroxylamine in buffered solutions of hydroxamic acids 2, 3,

and 4. Times are cumulative (0, 30, or 60 min in buffer plus 15 min reaction with

picrylsulfonic acid)

In order to support the conclusions of the picrylsulfonic acid experiments, we prepared

solutions of selected DHAs in pH 8.0 buffer and subjected them to ESI-MS after 1-2

days. In all cases, the DHA was the predominant species observed. However, in the case

of inhibitory compounds 2, 3, and 4, significant signals corresponding to the respective

monohydroxamic acids were visible, while we did not observe the corresponding

dicarboxylic acids. We also observed signals corresponding to the monohydroxamic

acids minus a water molecule, which could correspond to cyclic intermediates resulting

from intramolecular condensation (or could be ionization artifacts).

S12
References

1. Schraml, J.; Tkadlecova, M.; Pataridis, S.; Soukupova, L.; Blechta, V.; Roithova,

J.; Exner, O., Ring-substituted benzohydroxamic acids: 1H, 13C and 15N NMR spectra

and NH-OH proton exchange. Magn. Reson. Chem. 2005, 43 (7), 535-542.

2. Hauser, C.; Renfrow, W. J., Benzohydroxamic acid. Org. Synth., Colle. Vol. 1959,

2, 67.

3. Stolberg, M. A.; Mosher, W. A.; Wagner-Jauregg, T., Synthesis of a Series of

Vicinally Substituted Hydroxamic Acids. J. Am. Chem. Soc. 1957, 79 (10), 2615-2617.

4. Fritz, H. P.; von Stetten, O. E., 1H-NMR und IR-Spektren der

Formhydroxamsaure und ihres Natriumsalzes. Z. Naturforsch., B: Chem. Sci. 1969, 24b,

947-949.

5. Witherup, T. H.; Abbott, E. H., Carbon-13 nuclear magnetic resonance spectra of

the vitamin B-6 group. J. Org. Chem. 1975, 40 (15), 2229-2233.

6. Harruff, R. C.; Jenkins, W. T., A 13C n.m.r. study of the B6 vitamins and of their

aldimine derivatives. Organic Magnetic Resonance 1976, 8 (11), 548-557.

7. O'Leary, M. H.; Payne, J. R., 13C NMR spectroscopy of labeled pyridoxal 5'-

phosphate. Model studies, D-serine dehydratase, and L-glutamate decarboxylase. J. Biol.

Chem. 1976, 251 (8), 2248-2254.

8. Korytnyk, W.; Singh, R. P., Proton Magnetic Resonance Spectra of Compounds

in the Vitamin B6 Group. J. Am. Chem. Soc. 1963, 85 (18), 2813-2817.

9. Korytnyk, W.; Ahrens, H.; Donald, B. M.; Lemuel, D. W., Nuclear magnetic

resonance spectroscopy of vitamin B6. In Methods Enzymol., Academic Press: 1970; Vol.

Volume 18, Part 1, pp 475-483.

S13
10. Crépaux, D.; Lehn, J. M., Application des mesures de couplages 15N-H à la

détermination de configurations d'oximes. Organic Magnetic Resonance 1975, 7 (10),

524-526.

11. Pejkovic-Tadic, I.; Hranisavljevi-Jakovljevic, M.; Nescaroni, S.; Pascual, C.;

Simon, W., Protonenresonanzspektren von Oximen aromatischer Aldehyde. Helv. Chim.

Acta 1965, 48 (5), 1157-1160.

12. Jerslev, B.; Larsen, S., Hydrogen bonding and stereochemistry of ring-

hydroxylated aromatic aldehyde oximes. Crystal structures of three 4-OH-substituted

benzaldehyde oximes. Acta Chem. Scand. 1991, 45 (3), 285-291.

13. Phillips, W. D., Studies of hindered internal rotation in organic molecules by

nuclear magnetic resonance. Ann. N. Y. Acad. Sci. 1958, 70 (Nuclear Magnetic

Resonance), 817-832.

14. Sharif, S.; Huot, M. C.; Tolstoy, P. M.; Toney, M. D.; Jonsson, K. H. M.;

Limbach, H.-H., 15N Nuclear Magnetic Resonance Studies of Acid-Base Properties of

Pyridoxal-5'-Phosphate Aldimines in Aqueous Solution. The Journal of Physical

Chemistry B 2007, 111 (15), 3869-3876.

13
15. Kalinowski, H.-O.; S.Berger; Braun, S., C-NMR-Spektroskopie. Thieme:

Stuttgart, 1984.

16. Berger, S.; Braun, S.; H.-O.Kalinowski, NMR-Spektroskopie von Nichtmetallen 2.

15
N-NMR-Spektroskopie. Thieme: Stuttgart, 1992.

17. Leatherbarrow, R. J. GraFit Version 5, Erithacus Software Ltd.: Horley, UK,

2001.

S14
18. Pietta, P.; Calatroni, A.; Zio, C., Spectrophotometric determination of residual

hydroxylamine in hydroxamic acids. Analyst 1982, 107, 341-343.

S15