Succinylphosphonate esters are competitive inhibitors of MenD that show active-site discrimination between homologous α-ketoglutarate- decarboxylating enzymes

Article abstract of DOI:10.1021/bi901432d

MenD is a thiamin diphosphate-dependent enzyme catalyzing the first unique step in menaquinone biosynthesis in bacteria. We have synthesized acylphosphonate ester analogues of a-ketoglutarate, a substrate of MenD. These compounds are competitive inhibitors of MenD, with K_i values as low as 700 nM. Observed structure-activity relationships are in notable contrast to those reported previously for succinylphosphonate inhibition of the a-ketoglutarate dehydrogenase complex, despite the apparent homology of these enzymes, and the identical decarboxylation reactions catalyzed. Inhibiting menaquinone biosynthesis is a proposed approach to inhibiting Mycobacterium tuberculosis growth. These inhibitors showed no significant inhibition of M. tuberculosis growth in vitro under aerobic and hypoxic conditions but give new information about the binding characteristics of the MenD active site. Site-directed mutation of Ser391 to alanine had only a minor effect on catalysis, but even the conservative mutation of Arg395 to lysine had a significant effect on the catalytic processing of isochorismate.

Full text of DOI:10.1021/bi901432d

2672 Biochemistry 2010, 49, 2672–2679
DOI: 10.1021/bi901432d
Succinylphosphonate Esters Are Competitive Inhibitors of MenD That Show Active-Site
Discrimination between Homologous R-Ketoglutarate-Decarboxylating Enzymes†
Maohai Fang,‡ R. Daniel Toogood,‡ Andrea Macova,‡ Karen Ho,‡ Scott G. Franzblau,§ Michael R. McNeil,
David A. R. Sanders,‡ and David R. J. Palmer*,‡
‡Department of Chemistry, University of Saskatchewan, 110 Science Place, Saskatoon, SK S7N 5C9, Canada,
§Institute for Tuberculosis Research, College of Pharmacy, University of Illinois, 833 South Wood Street, Chicago,
Illinois 60612, and Department of Microbiology, Immunology and Pathology, Colorado State University,
Fort Collins, Colorado 80523
Received August 14, 2009; Revised Manuscript Received February 18, 2010
ABSTRACT: MenD is a thiamin diphosphate-dependent enzyme catalyzing the first unique step in menaquinone
biosynthesis in bacteria. We have synthesized acylphosphonate ester analogues of R-ketoglutarate, a
substrate of MenD. These compounds are competitive inhibitors of MenD, with Ki values as low as 700 nM.
Observed structure-activity relationships are in notable contrast to those reported previously for succinylpho-
sphonate inhibition of the R-ketoglutarate dehydrogenase complex, despite the apparent homology of these
enzymes, and the identical decarboxylation reactions catalyzed. Inhibiting menaquinone biosynthesis is a
proposed approach to inhibiting Mycobacterium tuberculosis growth. These inhibitors showed no significant
inhibition of M. tuberculosis growth in vitro under aerobic and hypoxic conditions but give new information about
the binding characteristics of the MenD active site. Site-directed mutation of Ser391 to alanine had only a minor
effect on catalysis, but even the conservative mutation of Arg395 to lysine had a significant effect on the catalytic
processing of isochorismate.
MenD is an enzyme required for menaquinone (vitamin K2)
biosynthesis in many bacteria (1, 2). The biosynthetic pathway
for menaquinone begins with the reversible conversion of chorismate
to isochorismate by the gene product MenF, followed by a
thiamin diphosphate-dependent decarboxylation and carboligation
catalyzed by MenD. The MenD-catalyzed reaction is shown in
Scheme 1.
another enzyme, MenH (10). MenD is presumed to follow a ping-
pong kinetic mechanism; i.e., the ThDP-dependent decarboxyla-
tion of R-ketoglutarate occurs prior to the binding and reaction of
isochorismate, as shown in Scheme 2, although experiments
demonstrating the kinetic mechanism have not been published.
Recently, the medium-resolution structure of the Escherichia coli
MenD ThDP Mn2þ complex was published (11).
3
3
MenD is a member of a family of homologous thiamin di-
phosphate-dependent decarboxylases that includes pyruvate
decarboxylase, pyruvate dehydrogenase (PDH),1 benzoylformate
decarboxylase (BFDC), the R-ketoglutarate dehydrogenase com-
plex (KGDHC), and others (3, 4). MenD shares a common
chemical step with these enzymes, the ThDP-dependent decarboxy-
lation of a 2-oxo acid, in this case R-ketoglutarate (also called
2-oxoglutarate). MenD is unique, however, in that it catalyzes
formation of a new carbon-carbon bond between the decarboxy-
lated substrate and the β-carbon of a second substrate, isochor-
Phosphonate analogues of 2-oxo acids can act as inhibitors
of ThDP-dependent decarboxylases. For example, methyl
acetylphosphonate and methyl benzoylphosphonate are reversi-
ble inhibitors of PDH (12, 13) and BFDC (14) respectively, while
benzoylphosphonate is an inactivator of BFDC, phosphorylating
an active-site serine residue (15). Bunik et al. showed that the
brain R-ketoglutarate dehydrogenase complex (KGDHC) is
inhibited by some derivatives of succinylphosphonate (4-phos-
phono-4-oxobutanoate, 1) (16). These authors found 1 and its
monoethyl phosphonate ester to be reversible inhibitors of
KGDHC. In contrast, acetylphosphonate shows little or no
inhibition of PDH, suggesting the second negative charge is
deleterious to inhibition for this enzyme. Benzoylphosphonate
inactivation of BFDC indicates that this dianionic inhibitor
binds to the active site. Thus, despite a similar chemical reaction
catalyzed in a conserved ThDP-binding site, the behavior of
acylphosphonates cannot be predicted with certainty.
ismate; the closest nonenzymatic equivalent is
a Stetter
reaction (5). MenD has also been called 2-succinyl-6-hydroxy-
2,4-cyclohexadiene-1-carboxylate (SHCHC) synthase for many
years (6-8), but Jiang et al. have shown that MenD releases
2-succinyl-5-enolpyruvyl-6-hydroxy-3-cyclohexene-1-carboxylate
(SEPHCHC) (9), which is subsequently converted to SHCHC by
†This work was supported by an NSERC Discovery Grant to
D.R.J.P. and a Saskatchewan Health Research Fund Team Grant to
the Molecular Design Research Group.
Scheme 1
*To whom correspondence should be addressed. Telephone: (306)
966-4662. Fax: (306) 966-4730. E-mail: dave.palmer@usask.ca.
1Abbreviations: BFDC, benzoylformate decarboxylase; KGDHC,
brain R-ketoglutarate dehydrogenase complex; KGDC E1, E. coli
R-ketoglutarate dehydrogenase E1 subunit; PDH, pyruvate dehydro-
genase; SHCHC, 2-succinyl-6-hydroxy-2,4-cyclohexadiene-1-carboxy-
late; SEPHCHC, 2-succinyl-5-enolpyruvyl-6-hydroxy-3-cyclohexene-
1-carboxylate; ThDP, thiamine diphosphate.
r
pubs.acs.org/Biochemistry
Published on Web 03/03/2010
2010 American Chemical Society

Article
Biochemistry, Vol. 49, No. 12, 2010 2673
antibiotic properties. Site-directed mutagenesis was then used
to probe the roles of two key residues from the active site of
MenD.
MATERIALS AND METHODS
Synthesis of Inhibitors. (i) General. Experiments that
required anhydrous conditions were performed under an inert
atmosphere of either dry argon or nitrogen gas. Glassware was
dried overnight in an oven set at 120 °C and assembled under a
stream of inert gas. All reagents were obtained from commercial
suppliers and unless indicated otherwise used without further
purification. Dichloromethane was freshly distilled from calcium
hydride. Thin layer chromatography (TLC) was performed on
precoated silica gel plates (Merck Kieselgel 60F254, 0.25 mm
thickness) and visualized with phosphomolybdic acid reagent,
iodine vapors, or ultraviolet light at 254 nm. Flash chromato-
graphy was performed using Merck silica gel 60 (230-400 mesh).
NMR spectra were recorded using a Bruker 500 MHz spectro-
meter with samples dissolved in the appropriate deuterated
solvents. Chemical shifts were reported in parts per million
downfield from tetramethylsilane. Mass spectrometry was per-
formed on an API Qstar XL pulsar hybrid lc/ms/ms apparatus.
Melting points were measured on a Gallencamp melting point
apparatus and were not corrected. NMR, mass spectrometry,
and elemental analysis facilities were a part of the Saskatchewan
Structural Sciences Centre. Chemical reagents, including buffers,
salts, and fine chemicals, were obtained from Sigma-Aldrich
Canada, Ltd. (Oakville, ON) or VWR CanLab (Mississaugua,
ON) and were categorized as molecular biology grade or were the
highest grade available.
(ii) General Procedure for the Preparation of Succinyl-
phosphonates. Succinic anhydride was treated with the appro-
priate alcohol to form the corresponding monoester (23, 24). The
succinate monoester was converted to the acid chloride at 0 °C,
using 1.1 equiv of oxalyl chloride dissolved in CH2Cl2 containing
one drop of DMF. The acid chloride was reacted at 0 °C with
1 equiv of the appropriate trialkyl phosphite. Succinylphos-
phonates 1-4 were made from the monobenzyl ester, and the
benzyl group was subsequently removed by hydrogenolysis over
10% Pd/C in a solution of EtOAc and formic acid at atmospheric
pressure (using a H2-filled balloon). Alkyl phosphonate esters
were cleaved using stoichiometric amounts of LiBr or NaI in
acetonitrile. Details and characterization are supplied in the
Supporting Information.
MenD and KGDHC catalyze the identical decarboxylation
reaction, although there is considerable variation of the two
active sites, with the exception of a small group of residues that
interact directly with the cofactors (11, 17). As described above,
substrate analogues of R-ketoglutarate were synthesized to inhibit
KGDHC, based on the chemical reaction, i.e., without explicit
consideration of the active-site structure. These inhibitors would be
predicted to inhibit MenD also, but the differences in the active-site
topology may result in different structure-activity relationships.
Inhibition of MenD may be of considerable interest because
many bacteria, including Mycobacterium tuberculosis, rely on
menaquinone (vitamin K2) as an electron acceptor, allowing
respiration under low-oxygen conditions (1, 2). Tuberculosis
infections can therefore persist under hypoxic conditions for
long periods of time (18). Controlling the intracellular concen-
tration of menaquinone can inhibit the growth of M. tuber-
culosis (19, 20), and the menaquinone biosynthetic pathway has
been proposed as a target for antibiotic, especially antituber-
culosis, drugs (21, 22). If MenD is to serve as a target for anti-
biotics, then inhibitor selectivity will be a primary concern.
We have therefore explored the ability of succinylphosphonate
derivatives to inhibit MenD. We hypothesized that an R-keto-
glutarate mimic such as methyl succinylphosphonate (2) would
bind to the active site and react with ThDP to give a dead-end
complex, as observed for PDH and BFDC, but we could not
rule out the possibility that 1 would act as an inhibitor, as seen
with KGDHC. Ideally, inhibitors distinct from those that inhibit
mammalian enzymes would be lead compounds for antibacterial
drug design. To test our hypotheses, we synthesized 11 acylphos-
phonates, including four novel compounds, and measured
their inhibitory properties with respect to MenD, and their
Scheme 2
Enzymology. (i) General. Cultures were grown in an
Innova 4230 incubator shaker and were lysed using a Virsonic
600 ultrasonic cell disrupter. A BioCAD Sprint Perfusion Chro-
matography system was routinely used for large-scale protein
purifications and the Bio-Rad Mini-protean 3 used to assess
protein purity. Chelating Sepharose FF columns were purchased
from GE Healthcare (Baie D’Urfe, QC). Centrifugation was
performed using either a Beckman Coulter microfuge 18 and 22R
centrifuge or a Beckman J2-HS refrigerated centrifuge with a
JLA-10.5 or JA-25.5 rotor.
(ii) Protein Purification. The menD gene was subcloned
into the BamHI and KpnI sites of vector pQE80-L (QIAgen,
Mississauga, ON) by PCR amplification of the gene using oligo-
nucleotide primers incorporating these restriction sites, followed
by standard ligation procedures and transformation of E. coli
Nova Blue (Novagen, VWR-Canlab). The insert of the resulting
plasmid was sequenced (National Research Council Plant

2674 Biochemistry, Vol. 49, No. 12, 2010
Fang et al.
Biotechnology Institute DNA Technologies Unit, Saskatoon,
SK) to ensure no mutations had been introduced.
where [I] is the inhibitor concentration, Ki is the inhibition constant,
and [S] is the concentration of the substrate R-ketoglutarate.
Reversibility of inhibition was tested by dilution. MenD was
incubated for 3 min in 10 mM 2 and then diluted 150-fold under
normal assay buffer conditions. The MenD-catalyzed reaction
rate resulting from this treatment was identical to the blank.
To test for the time dependence of inhibition, all reagents,
including 100 μM R-ketoglutarate, 20 μM isochorismate, and
100 or 400 μM 2, were equilibrated at 25 °C, before initiation
of the reaction via addition of MenD (final concentration of
60 nM). The resulting progress curve was linear, with no lag or
apparent increase in the level of inhibition over 30 min. The same
was observed when the reaction was instead initiated by addition
of isochorismate.
(v) Site-Directed Mutagenesis. Point mutations were gene-
rated using the QuikChange site-directed mutagenesis kit from
Stratagene (La Jolla, CA) according to the manufacturer’s
directions, except that an extension time of 2 min per kilobase
gave better results than the recommended time of 1 min per
kilobase. For each mutant, a mutagenic primer was designed to
incorporate a change in the nucleotide sequence encoding the
residue of interest. For other mutants, an additional, silent
mutation was introduced to generate a new restriction site for
the purpose of screening, after transformation of E. coli XL1-
Blue with PCR-amplified DNA. A complete list of primer
sequences is included in the Supporting Information.
After transformants had been screened, selected clones were
tested for expression by transformation of E. coli BL-21Gold. All
selected clones exhibited expression similar to that of the wild-
type enzyme. A clone was selected, and the complete gene was
sequenced (National Research Council Plant Biotechnology
Institute DNA Technologies Unit) to ensure no other mutations
had been introduced. The protein was purified as previously
described for the wild-type enzyme, but using columns which had
never been in contact with wild-type MenD.
Susceptibility Studies. M. tuberculosis H37Rv grown to
midlog in 7H9-OADC medium containing 0.05% Tween 80
was diluted to 2 ꢁ 105 colony forming units/mL for use in these
assays. Each well in a 96-well round-bottomed microtiter plate
received 4 μL of compound in DMSO at its respective concen-
tration, 96 μL of 7H9-OADC Tween 80 medium, and 100 μL of
the diluted H37Rv cells. The starting concentration for each
compound was 200 μg/mL (4 μL of compound in DMSO at
10 mg/mL in a total volume of 200 μL) and serially titrated down
2-fold to as low as 0.4 μg/mL (final concentration). The plates
were incubated for 10 days at 37 °C and scanned for visual
density reads. Wells with no visible settling of cells (clear well)
were considered null, and any settling seen was considered
positive. MIC100 is the first completely clear well.
Gene expression was induced in a Nova Blue cell culture,
shaken at 225 rpm and 37 °C in LB broth containing 50 mg/L
ampicillin, by induction with 1 mM IPTG after the culture had
reached an optical density (OD600) of 0.4, followed by continued
growth for 4 h at 30 °C. Cells were lysed by sonication in buffer
containing 5 mM imidazole, 0.5 M sodium chloride, 12.5%
glycerol, and 20 mM Tris-HCl (pH 7.9). The protein was purified
by being bound to a Ni2þ-charged NTA column, washed with the
same buffer containing 50 mM imidazole, and then eluted with
buffer containing 50 mM EDTA. The purified protein was
dialyzed into 5 mM Tris-HCl (pH 7.4) and 50% glycerol. Protein
was stored in 0.4 mL aliquots at -80 °C after being snap-frozen
in liquid nitrogen.
(iii) Wild-Type Enzyme Assays. Isochorismate was pre-
pared as described previously (8). The isochorismate concentra-
tion was determined enzymatically, by measuring the drop in
absorbance at 278 nm in the presence of MenD and excess
R-ketoglutarate, using an extinction coefficient ε of 8300 M-1
cm-1 (25). Isochorismate consistently accounted for ∼80% of the
absorbance at 278 nm. All assays were conducted at 25 °C using
a Beckman Coulter DU 640 spectrophotometer and were per-
formed in a total volume of 1 mL with a final concentration of
100 mM Tris-HCl (pH 7.4). Apparent rate constants for iso-
chorismate, R-ketoglutarate, ThDP, and MgCl2 were determined
via variation of each substrate or cofactor in saturating concen-
trations of the others, namely, 5 mM MgCl2, 50 μM ThDP,
100 μM R-ketoglutarate, and 20 μM isochorismate. The MenD
concentration was 10 nM. The isochorismate solution was freshly
prepared before each experiment.
The reaction rate was measured by continuously monitoring
the absorbance change at 278 nm due to the disappearance of
isochorismate. All rates measured were corrected with a blank
reaction in the absence of isochorismate. All data points represent
the average of at least two experiments. Kinetic constants were
determined using Leonora (26) by nonlinear least-squares fitting
of the data to the Michaelis-Menten equation (eq 1)
v ¼ Vmax½Sꢀ=ðKm þ ½SꢀÞ
ð1Þ
where v is the measured rate, Vmax is the apparent maximal rate,
Km is the apparent Michaelis constant, and [S] is the concentra-
tion of the varied substrate. The apparent turnover number, kcat
,
is calculated from Vmax by dividing by the enzyme concentration.
(iv) Inhibition Study of R-Ketoglutarate Phosphonate
Analogues. The reaction mixture contained all components and
appropriate inhibitors but R-ketoglutarate. MenD at a final
concentration of 60 nM was used, and the isochorismate
concentration was fixed at 20 μM for the assay. After preincuba-
tion at 25 °C for 3 min, the reaction was initiated by addition of R-
ketoglutarate with various concentrations (10, 15, 20, and 40 μM)
at five fixed concentrations of each inhibitor. The inhibition type
and the inhibition constants were determined graphically with a
Dixon plot (27) (1/v vs inhibitor concentration at four different R-
ketoglutarate concentrations) and a Cornish-Bowden plot (28)
(R-ketoglutarate concentration/v vs inhibitor concentration at
four different R-ketoglutarate concentrations), based on eq 2
The low-oxygen recovery assay (LORA) and microplate
Alamar Blue assay (MABA) were performed following published
procedures (29, 30).
RESULTS AND DISCUSSION
Measuring Rate Constants with MenD. Jiang et al.
effectively showed that monitoring the disappearance of iso-
chorismate (by UV spectroscopy at 278 nm) is the only reliable
way to measure the rate of the MenD-catalyzed reaction, since
the product (SEPHCHC) is essentially UV-silent (9). Although
isochorismate is prone to spontaneous decomposition (31), care-
ful handling of this compound and storage at -80 °C gives
Vmax½Sꢀ
ꢀ
ꢁ
v ¼
ð2Þ
½Iꢀ
½Sꢀ þ Km 1 þ
Ki

Article
Biochemistry, Vol. 49, No. 12, 2010 2675
Table 1: Kinetic Constants Determined for MenD and Its Mutants
enzyme
Km(R-ketoglutarate) (μM)
Km(ThDP) (μM)
Km(Mg2þ) (μM)
Km(isochorismate) (μM)
kcat (min-1
)
wild type
S391A
9.9 ( 0.6
13 ( 1
8 ( 1
17 ( 1
44 ( 1
13 ( 1
trace activity
ND
NDa
ND
17 ( 3
18 ( 1
S391Y
R395K
R395A
45 ( 3
2.2 ( 0.2
ND
5.3 ( 0.9
8.7 ( 0.5
NDb
ND
36c
2.5c
aND, not determined. The apparent Km for isochorismate is below the limit of accurate measurement (see the text). bThe apparent Km values for
R-ketoglutarate and cofactors could not be determined because of the high UV absorbance resulting from concentrations of isochorismate necessary for
saturation. cValue estimated from the Hanes-Woolf plot.
consistent results, but the isochorismate concentration must be
determined enzymatically before kinetic experiments are per-
formed. HPLC analysis of a MenD reaction mixture shows that
in the presence of excess R-ketoglutarate, all isochorismate is
consumed (data not shown).
An inherent challenge is presented by the very low Km value
exhibited by isochorismate. The Km value was previously mea-
sured (9) to be well below 1 μM, and we concur with this finding.
However, this precludes accurate rate measurements at iso-
chorismate concentrations below Km, since even a 100 nM
solution, approximately twice the Km estimated previously, has
an absorbance of only 0.00083 AU. Thus, an accurate apparent
Km value for isochorismate cannot be measured. This also makes
experiments in which both substrates are varied, such as those
that would normally be used to diagnose a ping-pong mechan-
ism, impossible, since values near the Michaelis constant are
appropriate for these experiments. Apparent values for R-keto-
glutarate, as well as cofactors ThDP and Mg2þ, can be measured
in a straightforward manner by saturation with isochorismate.
MenD was expressed from a plasmid derived from pQE80-L,
resulting in a homogeneous protein bearing an N-terminal
hexahistidine tag. Kinetic constants, listed in Table 1, differ
slightly from those determined by Jiang et al. (9). These authors
reported a lower apparent kcat (4.5 min-1) and Km (1.5 μM) for
R-ketoglutarate, a lower Km for ThDP (2.4 μM), and a higher
Km for Mg2þ (80 μM). Some of the variation may be attributable
to differences in reaction temperatures and conditions.
Synthesis of Inhibitors. Acylphosphonates were synthesized
on the basis of the Arbuzov reaction as described by Karaman
et al. (32). Formation of a succinate monoester, followed by
conversion of the acid to the acid chloride, allowed formation of
the carbon-phosphorus bond using a trialkyl phosphite as a
nucleophile. The resulting phosphonate esters were typically
cleaved using a stoichiometric quantity of LiBr. This approach
allowed preparation of succinylphosphonate (1) in >70% yield
and methyl succinylphosphonate (2) in >60% yield starting
from succinic anhydride. Carboxylic esters (see below), requiring
one fewer synthetic step, were isolated in higher yields. This is an
F
IGURE 1: Representative Dixon plots showing competitive inhibi-
improvement on other reported methods for preparing such
compounds; in particular, harsh conditions were avoided by
making use of the benzyl ester protecting group. Bunik’s reported
method (16) for the hydrolysis of phosphonate esters with
aqueous sodium hydroxide did not give satisfactory results in
our hands. Our approach is versatile in that an array of acylphos-
phonates could be prepared in this manner.
MenD Inhibition. The first question to be addressed was
whether succinylphosphonate (1), which is a close steric analogue
of R-ketoglutarate, or methyl succinylphosphonate (2), which is
more similar in charge to the substrate, would act as an inhibitor.
Analogy with KGDHC predicts 1 and 2 would be competitive
tion. Points indicate the average of at least two experiments, and the
lines represent a nonlinear least-squares fit to eq 2 (see Materials and
Methods). The value of Ki is the x-intercept of the intersection of the
lines;the valuesare given inTable 2:(A) inhibitionby1, (B) inhibition
by 2, and (C) inhibition by 7. Conditions are described in Materials
and Methods.
inhibitors, but analogy with PDH suggests that only 2 would
inhibit MenD (vide supra). The inhibitors were analyzed with a
Dixon plot, as shown in Figure 1, and with a Cornish-Bowden
plot (not shown). The analysis is consistent with reversible,
competitive inhibition with respect to R-ketoglutarate. No evi-
dence of time dependence was observed, consistent with previous

2676 Biochemistry, Vol. 49, No. 12, 2010
Fang et al.
by a specific interaction of the phenyl group with the protein,
such as π-stacking or cation-π stabilization; the MenD structure
shows many arginine and lysine residues in the active site. The
increased energy of the more hydrophobic substrate in bulk
aqueous solution may also contribute to the lowered Ki value.
The trend with an increase in the size of the phosphonic ester
alkyl group in the presence of the benzyl ester (compounds 8 and
9) is similar to that seen for carboxylates 2-4: increasing the size
sharply increases the inhibition constant. Note that Bunik et al.
reported a succinylphosphonate bearing a carboxylate ester but
not a phosphonate ester was an effective inhibitor, but a
succinylphosphonate bearing both types of ester was not (16).
Although removal of the negative charge from the “carboxy-
late end” of these inhibitors does not remove the ability to inhibit
the MenD reaction, removal of the carboxyl group itself is
catastrophic to activity. Neither methyl acetylphosphonate (10)
nor methyl butyrylphosphonate (11) exhibited any significant
inhibition at concentrations up to 10 mM.
Table 2: Inhibition of MenD by Acylphosphonates 1-11
Antimicrobial Testing with Acylphosphonates. These
compounds are the first reported inhibitors of MenD. MenD is
a proposed drug target, particularly for antituberculosis drugs;
therefore, we pursued the testing of these compounds as inhibi-
tors of mycobacterial growth, under aerobic conditions and
under hypoxic conditions.
Selectivity was our primary concern with respect to the use of
acylphosphonates as drug candidates, given the similarity to
KGDHC inhibitors, and the possibility that constitutive carboxy-
esterases might produce more potent phosphonate monoesters of
succinylphosphonate in vivo. Bunik et al. did report that
succinylphosphonate and related esters showed no significant
inhibition of R-ketoglutarate-utilizing aminotransferases and
dehydrogenases (11). Compounds were tested for their effects
on the growth of M. tuberculosis. Unfortunately, MICs for all
compounds exceeded 128 μM. Mycobacterial cell walls are well-
known to be highly resistant to permeation, particularly by
hydrophilic compounds (33).
Comparing the Active Sites of MenD and KGDHC.
There is no crystal structure available for human KGDHC, but
the E. coli R-ketoglutarate dehydrogenase E1 subunit (KDHC
E1) has been crystallized and deposited in the Protein Data Bank
(PDB) as entry 2JGD (17). The human and E. coli proteins are
reasonably well-conserved: 38% identical and 56% similar amino
acids. Therefore, comparison of E. coli KGDH E1 and MenD
can be used to rationalize differences in inhibitor recognition.
Unfortunately, no substrate is present in the KGDH E1 struc-
ture, although density identified as oxaloacetate is present, and
some active-site loops are disordered. Similarly, there is no
substrate or substrate analogue in the active site of the MenD
crystal structure (11).
Despite the absence of electron density for some loops in the
active site, Frank et al. (17) were able to identify residues in
KGDH E1 that were essential to efficient catalytic activity,
including His260, which was proposed to interact with a carboxy-
late group of R-ketoglutarate. His260 is in a well-conserved
portion of the protein, as indicated in Figure 2.
MenD and KGDH E1 are distant homologues, with a low
overall level of sequence identity. However, the C-terminal
domain of MenD, sometimes called the PP domain because of
the proximity of residues to the diphosphate region of bound
ThDP, can be structurally aligned with a corresponding domain
of KGDH E1. The alignment of residues 370-540 of MenD with
residues 231-458 of KGDH E1 using DaliLite (34) is shown in
observations with related enzymes; for example, Bunik et al.
reported no time dependence of the inhibition of KGDHC by
similar compounds (16), and Brandt et al. reported no time
dependence of the inhibition of BFDC by methyl benzoylphos-
phonate (14). It was apparent that 2 was a much better inhibitor,
as shown in Table 2. Because 1 is synthesized from a methyl
phosphonate precursor, inhibition observed in the presence of 1
may actually be due to a very small amount of residual 2, since
0.5% contamination would yield the observed inhibition.
A similar phenomenon was observed by Kluger and Pike in their
inhibition study of PDH (12). This result demonstrates that
although MenD and KGDHC have evolved to stabilize a
transition state for the formation of an identical intermediate,
the differences in active-site structure allow selective inhibition: 1
inhibits only KDGHC.
On the basis of the preference of the MenD active site for 2
over 1, we studied other acylphosphonate monoesters. Increasing
the size of the alkyl group of the phosphonate ester from methyl
to ethyl (3) resulted in a 1 order of magnitude increase in the
inhibition constant, and a further increase to isopropyl (4)
resulted in a similar increase (Table 2).
Esterification of the carboxylate was predicted to have a
detrimental effect on inhibition. Indeed, introduction of a methyl
group renders the diester 5 more than 300 times less effective than
2, while the presence of an ethyl group yields an inhibitor 6 that is
2000 times weaker than 2. Unexpectedly, benzyl ester 7 is an
improved inhibitor relative to 5 and 6. This may be rationalized

Article
Biochemistry, Vol. 49, No. 12, 2010 2677
FIGURE 2: Portion of the sequence alignment of the Homo sapiens KGDHC E1 component (HUMAN) and E. coliKGDHC E1. Residues His260
and Arg263, which align structurally with MenD residues Ser391 and Arg395, respectively, are highlighted in bold.
known to be important in KGDH E1 catalysis suggested that this
difference at the active site may contribute to differences in the
recognition of phosphonate inhibitors. Figure 3B shows not only
that KGDH E1 His260 and MenD Ser391 are overlapping but
also that the R-carbon of MenD Arg395 aligns closely with that
of KGDH E1 Arg263. The side chains of the arginine residues are
in different orientations in the two structures, which may or may
not be relevant given the absence of substrate from the structures.
The structure of MenD shows the β-alcohol oxygen of Ser391 to
˚
be ∼2.8 A from the sulfur atom of ThDP and approximately the
same distance from an oxygen atom bonded to the β-phosphorus
atom (11). This seems to contrast with the proposed role of the
structural homologue His280 in KGDHC; for these reasons, the
specific role of Ser391 is ambiguous and begs investigation.
˚
Arg395, ∼6.4
A from C2 of the ThDP, seems more
likely to be involved directly in interactions with isochorismate,
rather than R-ketoglutarate or ThDP. However, the presence of a
structural homologue in KGDH E1 argued against this role,
since that enzyme does not bind isochorismate. We generated
point mutants of Ser391 and Arg395 to probe their roles in
catalysis.
Replacement of Ser391 with an alanine residue had surpris-
ingly little effect on catalysis, as shown in Table 1. The turnover
number is unchanged, and all parameters are within 1 order of
magnitude of those of the wild-type enzyme. The most significant
effect was the decrease in the apparent Km of Mg2þ, followed by
the increased Km of ThDP. The lack of a large effect on catalysis
suggests that Ser391 plays a structural role, or that the role is
primarily served by the backbone, rather than the side chain,
despite the proximity of the Ser391 alcohol group to ThDP (11).
Alternatively, it has been proposed previously that a water
molecule could bind to such a mutant so that the function of
the hydroxyl group was retained, “rescuing” activity (35). Non-
conservative mutant S391Y displayed only trace activity.
Conservative substitution of Arg395 with lysine resulted in a
sharp increase in the Km of isochorismate, to a measurable value
of 5.3 μM, strongly supporting our hypothesis regarding its role
in catalysis. Surprisingly, however, this mutant shows an ex-
tremely low apparent Michaelis constant with respect to Mg2þ
.
Dialyzed enzyme retains virtually all of its Mg2þ-dependent
activity without addition of exogenous metal ion. This activity
can be abolished in the presence of EDTA. This result is very
difficult to rationalize; perhaps crystallization of this mutant will
give some indication of why R395K seems to bind Mg2þ so
tightly. The nonconservative substitution of Arg395 with alanine
results in a mutant with significant activity, but with an even
larger Michaelis constant for isochorismate. This increased value
results in the inverse problem usually encountered with this
enzyme: saturating values of isochorismate have an absorbance
beyond the limit of the spectrophotometer, therefore, the appar-
ent Km values of R-ketoglutarate, ThDP, and Mg2þ could not be
measured, and the apparent Michaelis constant of isochorismate
could only be estimated to be ∼36 μM using a Hanes-Woolf
plot, as shown in Figure 4.
F
IGURE 3: (A) Ribbon diagram of the structural alignment of resi-
dues 370-540 of MenD (cyan) with residues 231-458 of KGDH E1
(orange) using DaliLite (34), showing the similarity of the fold.
Structural coordinates are from PDB entries 2JLA and 2JGD,
respectively. ThDP and Mn2þ bound to MenD are included. (B)
From the same alignment, showing the coincidence of MenD Ser391
and Arg395 (cyan) and KGDH E1 His280 and Arg283 (orange).
Images were generated using VMD, followed by POV-Ray 3.6.
˚
Figure 3A. This alignment results in 2.6 A root-mean-square
deviation, with 144 aligned residues, even though the level of
sequence identity in this region is only 13%. As shown in
Figure 3B, His260 aligns with Ser391 of MenD.
We previously predicted (8) that Ser391 and the nearby
Arg395, which are well-conserved among MenD sequences,
would likely be found at the active site of MenD and be
important for catalysis. The coincidence of Ser391 with a residue
In conclusion, we have reported the first inhibitors of MenD.
They are competitive with R-ketoglutarate and clearly show a

2678 Biochemistry, Vol. 49, No. 12, 2010
Fang et al.
SUPPORTING INFORMATION AVAILABLE
Preparative procedures and spectroscopic data of synthesized
compounds, sequences of mutagenic primers, and an example
time course of a MenD-catalyzed reaction in the absence and
presence of inhibitor. This material is available free of charge via
the Internet at http://pubs.acs.org.
REFERENCES
1. Meganathan, R. (1996) Biosynthesis of the isoprenoid quinones
menaquinone (vitamin K2) and ubiquinone (coenzyme Q). in
Escherichia coli and Salmonella: Cellular and molecular biology
(Neidhart, F. C., Curtiss, R., III, Ingraham, J. L., Lin, E. C. C.,
Low, K. B., Magasanik, B., Reznikoff, W. S., Riley, M., Schaechter,
M., and Umbarger, H. E., Eds.) pp 642-656, American Society for
Microbiology, Washington, DC.
2. Meganathan, R. (2001) Biosynthesis of menaquinone (vitamin K-2)
and ubiquinone (coenzyme Q): A perspective on enzymatic mechan-
isms. Vitam. Horm. (San Diego, CA, U.S.) 61, 173–218.
3. Jordan, F. (2003) Current mechanistic understanding of thiamin
diphosphatedependent enzymatic reactions. Nat. Prod. Rep. 20,
184–201.
4. Duggleby, R. G. (2006) Domain relationships in thiamine diphos-
phate-dependent enzymes. Acc. Chem. Res. 39, 550–557.
5. Myers, M. C., Bharadwaj, A. R., Milgram, B. C., and Scheidt, K. A.
(2005) Catalytic conjugate additions of carbonyl anions under neutral
aqueous conditions. J. Am. Chem. Soc. 127, 14675–14680.
F
IGURE 4: Hanes-Woolf plot used to estimate the apparent kinetic
constants with respect to isochorismate in the presence of saturating
concentrations of R-ketoglutarate and cofactors (see the text). Points
indicate the average of at least two experiments, and the line
represents the fit to the Hanes-Woolf equation (see Materials and
Methods). The slope of the line gives the negative reciprocal of the
apparent Vmax; the x-intercept gives the apparent Km. The values are
listed in Table 1.
6. Palaniappan, C., Sharma, V., Hudspeth, M. E. S., and Meganathan,
R. (1992) Menaquinone (vitamin-K2) biosynthesis: Evidence that the
Escherichia coli MenD gene encodes both 2-succinyl-6-hydroxy-2,4-
cyclohexadiene-1-carboxylic acid synthase and R-ketoglutarate de-
carboxylase activities. J. Bacteriol. 174, 8111–8118.
7. Palaniappan, C., Taber, H., and Meganathan, R. (1994) Biosynthesis
of o-succinylbenzoic acid in Bacillus subtilis: Identification of MenD
mutants and evidence against the involvement of the R-ketoglutarate
dehydrogenase complex. J. Bacteriol. 176, 2648–2653.
8. Bhasin, M., Billinsky, J. L., and Palmer, D. R. J. (2003) Steady-state
kinetics and molecular evolution of Escherichia coli MenD [(1R,6R)-
2-succinyl-6-hydroxy-2,4-cyclohexadiene-1-carboxylate synthase], an
anomalous thiamin diphosphate-dependent decarboxylase-carboli-
gase. Biochemistry 42, 13496–13504.
9. Jiang, M., Cao, Y., Guo, Z. F., Chen, M. J., Chen, X. L., and Guo, Z.
H. (2007) Menaquinone biosynthesis in Escherichia coli: Identifica-
tion of 2-succinyl-5-enolpyruvyl-6-hydroxy-3-cyclohexene-l-carboxy-
late as a novel intermediate and re-evaluation of MenD activity.
Biochemistry 46, 10979–10989.
selectivity for the acylphosphonate monoester, as observed
previously for BFDC, rather than the acylphosphonate, as
observed for KGDHC. Inhibition most likely arises from
MenD-catalyzed formation of a covalent adduct with ThDP in
the active site of the enzyme. The structure-activity study
indicates that a negative charge is preferred but not required in
the position mimicking the δ-carboxylate of R-ketoglutarate,
but the simple four-carbon analogue 11 is insufficient. A
benzyl ester in this position is more potent than smaller alkyl
esters. This reveals binding properties of the active site that
were previously unknown. The active sites of MenD and
KGDH, although apparently products of divergent evolu-
tion, both catalyze the decarboxylation of R-ketoglutarate,
but mechanism-based inhibitors exhibit different structure-
activity relationships.
10. Jiang, M., Chen, X. L., Guo, Z. F., Cao, Y., Chen, M. J., and Guo, Z.
H. (2008) Identification and characterization of (1R,6R)-2-succinyl-
6-hydroxy-2,4-cyclohexadiene-1-carboxylate synthase in the mena-
quinone biosynthesis of Escherichia coli. Biochemistry 47, 3426–3434.
11. Dawson, A., Fyfe, P. K., and Hunter, W. N. (2008) Specificity and
reactivity in menaquinone biosynthesis: The structure of Escherichia
coli MenD (2-succinyl-5-enolpyruvyl-6-hydroxy-3-cyclohexadiene-
1-carboxylate synthase). J. Mol. Biol. 384, 1353–1368.
Although some of the inhibitors, particularly 2, are very
potent, they do not exhibit antituberculosis activity in vitro.
Even those bearing benzyl esters may be too polar to cross the
mycobacterial cell wall. These compounds should be of use in
crystallographic studies of MenD, since existing structural data
do not include substrates.
12. Kluger, R., and Pike, D. C. (1977) Active-site generated analogs of
reactive intermediates in enzymic reactions: Potent inhibition of
pyruvate dehydrogenase by a phosphonate analog of pyruvate.
J. Am. Chem. Soc. 99, 4504–4506.
13. Jordan, F., Nemeria, N. S., Zhang, S., Yan, Y., Arjunan, P., and
Furey, W. (2003) Dual catalytic apparatus of the thiamin diphosphate
coenzyme: Acid-base via the 10,40-iminopyrimidine tautomer along
with its electrophilic role. J. Am. Chem. Soc. 125, 12732–12738.
14. Brandt, G. S., Kneen, M. M., Chakraborty, S., Baykal, A. T., Nemeria,
N., Yep, A., Ruby, D. I., Petsko, G. A., Kenyon, G. L., McLeish, M. J.,
Jordan, F., and Ringe, D. (2009) Snapshot of a reaction intermediate:
Analysis of benzoylformate decarboxylase in complex with a benzoyl-
phosphonate inhibitor. Biochemistry 48, 3247–3257.
15. Bera, A. K., Polovnikova, L. S., Roestamadji, J., Widlanski, T. S.,
Kenyon, G. L., McLeish, M. J., and Hasson, M. S. (2007) Mechan-
ism-based inactivation of benzoylformate decarboxylase, a thiamin
diphosphate-dependent enzyme. J. Am. Chem. Soc. 129, 4120.
16. Bunik, V. I., Denton, T. T., Xu, H., Thompson, C. M., Cooper, A. J. L.,
and Gibson, G. E. (2005) Phosphonate analogues of R-ketoglutarate
inhibit the activity of the R-ketoglutarate dehydrogenase complex iso-
lated from brain and in cultured cells. Biochemistry 44, 10552–10561.
17. Frank, R. A. W., Price, A. J., Northrop, F. D., Perham, R. N., and
Luisi, B. F. (2007) Crystal structure of the E1 component of the
Mutagenesis experiments have provided insight into the roles
of two active-site residues. Ser391 can be replaced with alanine
with very little effect on catalysis, showing only weak influences
on the Michaelis constants of the cofactors, despite the proximity
of the hydroxyl group to the thiazolium and the diphosphate
moieties of ThDP. Arg395 likely interacts directly with the
substrate isochorismate; this is the first experimental evidence
supporting direct interaction of an amino acid of MenD with this
substrate. The result is consistent with the prediction of Dawson
et al. based on modeling of the medium-resolution X-ray
structure of MenD (11).
ACKNOWLEDGMENT
We thank the Saskatchewan Health Research Foundation
for funding the Molecular Design Research Group of the
University of Saskatchewan, Ken Thoms and Jason Maley
(Saskatchewan Structural Sciences Centre), and Kimberly Hanson.

Article
Biochemistry, Vol. 49, No. 12, 2010 2679
Escherichia coli 2-oxoglutarate dehydrogenase multienzyme complex.
J. Mol. Biol. 368, 639–651.
26. Cornish-Bowden, A. (1995) Analysis of Enzyme Kinetic Data, Oxford
University Press, New York.
18. Wayne, L. G. (1994) Dormancy of Mycobacterium tuberculosis
and latency of disease. Eur. J. Clin. Microbiol. Infect. Dis. 13,
908–914.
19. Weinstein, E. A., Yano, T., Li, L. S., Avarbock, D., Avarbock, A.,
Helm, D., McColm, A. A., Duncan, K., Lonsdale, J. T., and Rubin,
H. (2005) Inhibitors of type II NADH:menaquinone oxidoreductase
represent a class of antitubercular drugs. Proc. Natl. Acad. Sci. U.S.A.
102, 4548–4553.
20. Dhiman, R. K., Mahapatra, S., Slayden, R. A., Boyne, M. E.,
Lenaerts, A., Hinshaw, J. C., Angala, S. K., Chatterjee, D., Biswas,
K., Narayanasamy, P., Kurosu, M., and Crick, D. C. (2009) Mena-
quinone synthesis is critical for maintaining mycobacterial viability
during exponential growth and recovery from non-replicating persis-
tence. Mol. Microbiol. 72, 85–97.
21. Truglio, J. J., Theis, K., Feng, Y. G., Gajda, R., Machutta, C., Tonge,
P. J., and Kisker, C. (2003) Crystal structure of Mycobacterium
tuberculosis MenB, a key enzyme in vitamin K-2 biosynthesis. J. Biol.
Chem. 278, 42352–42360.
22. Lu, X. Q., Zhang, H. N., Tonge, P. J., and Tan, D. S. (2008)
Mechanism-based inhibitors of MenE, an acyl-CoA synthetase in-
volved in bacterial menaquinone biosynthesis. Bioorg. Med. Chem.
Lett. 18, 5963–5966.
23. Sharma, S. K., Miller, M. J., and Payne, S. M. (1989) Spermexatin and
spermexatol: New synthetic spermidine-based siderophore analogs.
J. Med. Chem. 32, 357–367.
24. Riegel, B., and Lilienfeld, W. M. (1945) The synthesis of β-keto esters
by the decomposition of acylated malonic esters. J. Am. Chem. Soc.
67, 1273–1275.
27. Dixon, M. (1953) The determination of enzyme inhibitor constants.
Biochem. J. 55, 170–171.
28. Cornish-Bowden, A. (1974) A simple graphical method for deter-
mining the inhibition constants of mixed, uncompetitive and non-
competitive inhibitors. Biochem. J. 137, 143–144.
29. Cho, S. H., Warit, S., Wan, B. J., Hwang, C. H., Pauli, G. F., and
Franzblau, S. G. (2007) Low-oxygen-recovery assay for high-
throughput screening of compounds against nonreplicating Myco-
bacterium tuberculosis. Antimicrob. Agents Chemother. 51, 1380–1385.
30. Franzblau, S. G., Witzig, R. S., McLaughlin, J. C., Torres, P.,
Madico, G., Hernandez, A., Degnan, M. T., Cook, M. B., Quenzer,
V. K., Ferguson, R. M., and Gilman, R. H. (1998) Rapid, low-
technology MIC determination with clinical Mycobacterium tuber-
culosis isolates by using the microplate Alamar Blue assay. J. Clin.
Microbiol. 36, 362–366.
31. DeClue, M. S., Baldridge, K. K., Kast, P., and Hilvert, D. (2006)
Experimental and computational investigation of the uncatalyzed
rearrangement and elimination reactions of isochorismate. J. Am.
Chem. Soc. 128, 2043–2051.
32. Karaman, R., Goldblum, A., Breuer, E., and Leader, H. (1989)
Acylphosphonic acids and methyl hydrogen acylphosphonates: Phy-
sical and chemical properties and theoretical calculations. J. Chem.
Soc., Perkin Trans. 1, 765–774.
33. Jarlier, V., and Nikaido, H. (1994) Mycobacterial cell wall: Structure
and role in natural resistance to antibiotics. FEMS Microbiol. Lett.
123, 11–18.
34. Holm, L., and Park, J. (2000) DaliLite workbench for protein
structure comparison. Bioinformatics 16, 566–567.
25. Rusnak, F., Liu, J., Quinn, N., Berchtold, G. A., and Walsh, C. T.
(1990) Subcloning of the enterobactin biosynthetic gene EntB:
Expression, purification, characterization, and substrate-specificity
of isochorismatase. Biochemistry 29, 1425–1435.
35. Lewendon, A., Murray, I. A., Shaw, W. V., Gibbs, M. R., and Leslie,
A. G. W. (1990) Evidence for transition-state stabilization by serine-
148 in the catalytic mechanism of chloramphenicol acetyltransferase.
Biochemistry 29, 2075–2080.