Phosphonodifluoropyruvate is a mechanism-based inhibitor of phosphonopyruvate decarboxylase from Bacteroides fragilis

Article abstract of DOI:10.1016/j.bmc.2017.06.013

Bacteroides fragilis, a human pathogen, helps in the formation of intra-abdominal abscesses and is involved in 90% of anaerobic peritoneal infections. Phosphonopyruvate decarboxylase (PnPDC), a thiamin diphosphate (ThDP)-dependent enzyme, plays a key role in the formation of 2-aminoethylphosphonate, a component of the cell wall of B. fragilis. As such PnPDC is a possible target for therapeutic intervention in this, and other phosphonate producing organisms. However, the enzyme is of more general interest as it appears to be an evolutionary forerunner to the decarboxylase family of ThDP-dependent enzymes. To date, PnPDC has proved difficult to crystallize and no X-ray structures are available. In the past we have shown that ThDP-dependent enzymes will often crystallize if the cofactor has been irreversibly inactivated. To explore this possibility, and the utility of inhibitors of phosphonate biosynthesis as potential antibiotics, we synthesized phosphonodifluoropyruvate (PnDFP) as a prospective mechanism-based inhibitor of PnPDC. Here we provide evidence that PnDFP indeed inactivates the enzyme, that the inactivation is irreversible, and is accompanied by release of fluoride ion, i.e., PnDFP bears all the hallmarks of a mechanism-based inhibitor. Unfortunately, the enzyme remains refractive to crystallization.

Full text of DOI:10.1016/j.bmc.2017.06.013

Bioorganic & Medicinal Chemistry xxx (2017) xxx–xxx
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry
journal homepage: www.elsevier.com/locate/bmc
Phosphonodifluoropyruvate is a mechanism-based inhibitor of
phosphonopyruvate decarboxylase from Bacteroides fragilis
Katharina Pallitsch a, Megan P. Rogers b, Forest H. Andrews b, Friedrich Hammerschmidt a,
,
⇑
Michael J. McLeish b,
⇑
a Institute of Organic Chemistry, University of Vienna, Vienna, Austria
b Department of Chemistry and Chemical Biology, Indiana University-Purdue University Indianapolis, Indianapolis, USA
a r t i c l e i n f o
a b s t r a c t
Article history:
Bacteroides fragilis, a human pathogen, helps in the formation of intra-abdominal abscesses and is
involved in 90% of anaerobic peritoneal infections. Phosphonopyruvate decarboxylase (PnPDC), a thiamin
diphosphate (ThDP)-dependent enzyme, plays a key role in the formation of 2-aminoethylphosphonate, a
component of the cell wall of B. fragilis. As such PnPDC is a possible target for therapeutic intervention in
this, and other phosphonate producing organisms. However, the enzyme is of more general interest as it
appears to be an evolutionary forerunner to the decarboxylase family of ThDP-dependent enzymes. To
date, PnPDC has proved difficult to crystallize and no X-ray structures are available. In the past we have
shown that ThDP-dependent enzymes will often crystallize if the cofactor has been irreversibly inacti-
vated. To explore this possibility, and the utility of inhibitors of phosphonate biosynthesis as potential
antibiotics, we synthesized phosphonodifluoropyruvate (PnDFP) as a prospective mechanism-based inhi-
bitor of PnPDC. Here we provide evidence that PnDFP indeed inactivates the enzyme, that the inactivation
is irreversible, and is accompanied by release of fluoride ion, i.e., PnDFP bears all the hallmarks of a mech-
anism-based inhibitor. Unfortunately, the enzyme remains refractive to crystallization.
Received 25 April 2017
Revised 11 June 2017
Accepted 12 June 2017
Available online xxxx
Keywords:
Thiamin diphosphate
CD
UV–Vis
Phosphonate
Enzyme inhibition
Ó 2017 Elsevier Ltd. All rights reserved.
1. Introduction
As shown in Scheme 1, there are two enzymes that are common
to the biosynthesis of virtually all biogenic phosphonates.4,5 The
Phosphonates, i.e., compounds having a CAP bond, are found in
archaea, bacteria and eukaryotes and it has been suggested that
they are important intermediates in the global phosphorus cycle.1,2
The relative stability of the CAP bond makes phosphonolipids and
phosphonoglycans less susceptible to enzymatic hydrolysis than
phosphates, thereby providing a selective advantage in phos-
phate-limited environments.2 As phosphonates are structurally
similar to phosphate esters and carboxylates, that same stability
can also contribute to biological activity, with antibacterial, antivi-
ral, antiparasitic and herbicidal activity all having been attributed
to phosphonates.1,3
first, PEP mutase, catalyzes the conversion of phosphoenolpyruvate
(PEP, 1) to phosphonopyruvate (PnP, 2). This reaction is thermody-
namically unfavorable6 so a second enzyme, phosphonopyruvate
decarboxylase (PnPDC), is used to introduce an irreversible step.7
Phosphonoacetaldehyde (PnAA, 3), the product of the decarboxyla-
tion reaction, is subsequently converted to a variety of phosphonic
acid natural products.
Some of these products have found therapeutic use. Fosfomycin,
a broad-spectrum antibiotic, is used in the treatment of cystitis.8
Plumbemycin5 and dehydrophos9 also act as antibiotics while
phosphinothricin, as its tripeptide with two alanine residues (bia-
laphos), is a natural herbicide.10 2-Aminoethylphosphonate (2-
AEP) is used by microorganisms in a variety of ways. For many, it
provides a source of useable carbon, nitrogen and phosphorus.11
For others, 2-AEP provides a counterpart to the lipid polar head
group, phosphoethanolamine, that is resistant to chemical and
enzymatic hydrolysis.12 Still others use it in the synthesis of
Abbreviations: PnDFP, phosphonodifluoropyruvate; Phtase, phosphonatase from
P. putida; PDC, pyruvate decarboxylase; PnPDC, phosphonopyruvate decarboxylase;
ThDP, thiamin diphosphate; YADH, yeast alcohol dehydrogenase; DTT,
dithiothreitol.
⇑
Corresponding authors at: Währinger Straße 38, A-1090 Wien, Austria (F.
bioactive compounds such as argolaphos,
a broad-spectrum
Hammerschmidt), 402 N. Blackford St., Indianapolis, IN 46202, USA (M.J. McLeish).
E-mail addresses: friedrich.hammerschmidt@univie.ac.at (F. Hammerschmidt),
mcleish@iupui.edu (M.J. McLeish).
antibacterial phosphonopeptide.13 It is also notable that several
http://dx.doi.org/10.1016/j.bmc.2017.06.013
0968-0896/Ó 2017 Elsevier Ltd. All rights reserved.
Please cite this article in press as: Pallitsch K., et al. Bioorg. Med. Chem. (2017), http://dx.doi.org/10.1016/j.bmc.2017.06.013

2
K. Pallitsch et al. / Bioorganic & Medicinal Chemistry xxx (2017) xxx–xxx
dehydrophos
plumbemycin
methylphosphonic acid
fosfomycin
O
O
O
PEP Mutase
PnPDC
THDP
H2O3P
OH
H2O3P
H
H2O3PO
OH
O
3
2
1
phosphinothricin
fosfazinomycin
2-AEP
Scheme 1. Common steps in the biosynthesis of phosphonates.
well-known mammalian and plant pathogens produce phospho-
nates. Fosfomycin, for example, although it does have a therapeutic
use, is produced by Pseudomonas syringae, a plant pathogen which
infects fruit-bearing trees.14 Naegleria fowleri, known colloquially
as the ‘‘brain-eating amoeba”, causes meningoencephalitis in
humans, and has been shown to produce phosphonates.15 In Bac-
teroides fragilis, a human pathogen involved in abscess formation,
a 2-AEP-containing capsular polysaccharide has been identified
as the major virulence factor,12 while the genes for 2-AEP biosyn-
thesis have been identified in Trypanosoma cruzi, the causative
agent of Chagas’ disease.16
Since phosphonate metabolism is not present in humans or
higher plants/animals, it offers a promising target for drug design
against these phosphonate producing and using pathogens.
Intriguingly, the antibiotic action of dehydrophos provides a clue
as to how this problem may be approached. Dehydrophos acts as
a ‘‘Trojan horse” in that is taken into the cell by peptide permeases,
whereupon it is activated by intracellular proteases to unmask
methyl acetylphosphonate (MAP),4,17 a substrate analogue of pyru-
vate,18 and a potent mechanism-based inhibitor of both the E1
subunit of the pyruvate dehydrogenase complex19 and 1-deoxy-
elimination of fluoride. The resulting vinyl alcohol tautomerizes
to give acetyl-ThDP which likely reacts with a nucleophile in the
active site thereby inactivating the enzyme.21,22
Given that phosphonopyruvate decarboxylase, which catalyzes
the irreversible step in phosphonate biosynthesis,7 is also a
ThDP-dependent decarboxylase, it would appear to be an ideal tar-
get for similar mechanism-based inhibitors. One such compound,
shown in Scheme 2, is 3-phosphono-3,3-difluoropyruvate (PnDFP,
4) an analogue of the natural substrate, phosphonopyruvate (PnP,
2). It is expected that PnDFP will be accepted by the enzyme and
react with the ThDP ylid to provide intermediate 5. However, fol-
lowing decarboxylation, the resulting carbanion 6 will eliminate
fluoride to produce the inactive enol-ThDP complex 7. Potentially
this could tautomerize to yield a phosphonofluoroacetyl-ThDP
complex 8 but, regardless, the enzyme will remain inactivated.
In this paper we describe the synthesis and testing of 4 with the
PnPDC from Bacteroides fragilis (BfPnPDC). PnDFP proved to be a
potent inhibitor of the enzyme, and showed all the characteristics
of a mechanism-based inhibitor.
2. Results
D
-xylulose-5-phosphate (DXP) synthase.20 The inhibition is attrib-
uted to the reaction of these thiamin diphosphate (ThDP)-depen-
dent enzymes with MAP producing stable phosphonate
analogue of the reaction intermediate, -lactylThDP. The analogue
2.1. Synthesis of 3-phosphono-3,3-difluoropyruvate (PnDFP, 4)
a
a
PnDFP (4) could be obtained in three steps from diethyl difluo-
romethylphosphonate (9) by deprotonation with freshly prepared
lithium diisopropylamide (LDA) at ꢀ78 °C (Scheme 3). The result-
ing difluoromethyllithium could be reacted with diethyl oxalate
((CO2Et)2), followed by acidic work up to give the desired product
in its hydrate form (10a) in 60% yield. The hydrate 10a can be
binds tightly and, in effect, irreversibly, to its targets.19 b-Fluoropy-
ruvate, another substrate analogue, also inactivates the pyruvate
dehydrogenase complex by a ThDP-dependent mechanism.21,22
However, in this instance, addition of ThDP to fluoropyruvate is fol-
lowed by decarboxylation, generating a carbanion and leading to
Scheme 2. Proposed mechanism of inactivation by DFPP (4).
Please cite this article in press as: Pallitsch K., et al. Bioorg. Med. Chem. (2017), http://dx.doi.org/10.1016/j.bmc.2017.06.013

K. Pallitsch et al. / Bioorganic & Medicinal Chemistry xxx (2017) xxx–xxx
3
O
P
EtO
HO OH
OEt
EtO
CF2
O
O
P
O
P
HO OH
1. LDA/-78 °C
10a
1. TMSBr/allylTMS/60 °C
OEt
OEt
OH
F2HC
HO
HO
CF2
O
2. NaOH pH 10
3. Dowex 50/H+
4. NaOH to pH 7
2. (CO2Et)2/-95 °C
3. AcOH
O
P
EtO
O
9
OEt
EtO
4
as its sodium salt
CF2
O
10b
Scheme 3. Synthesis of PnDFP.
partly converted to the ketone (10b) by bulb-to-bulb distillation
(ratio 10a:10b after distillation = 1:1.1).
of 3 mM (ꢂKm) and 30 mM (10 ꢁ Km). Using the Cheng-Prusoff equa-
tion the Ki value for PnDFP was determined to be 0.60 0.05 lM.
The final deprotection was a two-step process. First, the phos-
phonic acid moiety was deprotected using TMSBr/allylTMS for
24 h at 60 °C. The relatively high temperature and long reaction
time were necessary due to the large electron withdrawing effect
of the CF2 group. Next, the ester was saponified at pH >10 in aque-
ous NaOH, and the final product passed through Dowex 50W ꢁ 8,
H+ form, in order to remove excess base. The pH was adjusted to
7 using aqueous NaOH and the sodium salt of 4 was isolated by
lyophilisation as a colourless powder.
2.3.1. PnDFP irreversibly inactivates BfPnPDC
To determine if the inhibition was (i) time-dependent and (ii)
irreversible, a dilution study was performed. A mixture of enzyme
(6 lM) and 4 (50 lM) was incubated at 30 °C before an aliquot was
diluted 1:100 into an assay mixture containing saturating PnP. No
activity was observed at any of the time points tested. Reducing
the inhibitor concentration to 10 lM gave similar results. Further,
no enzyme activity was recovered after overnight dialysis. Taken
together this indicates that the inhibitor binds both rapidly and,
in effect, irreversibly.
2.2. Expression, purification and kinetic characterization of BfPnPDC
The expression and purification of the C-terminal His-tagged
BfPnPDC proceeded smoothly and in good yield. The steady state
kinetics of BfPnPDC were determined using a modified coupled
assay.12,23 In a typical assay for a pyruvate decarboxylase (PDC),
the coupling enzyme is a yeast alcohol dehydrogenase (YADH),
that catalyzes the reduction of acetaldehyde and the concomitant
oxidation of NADH, thus allowing the reaction to be followed spec-
trophotometrically at 340 nm. Phosphonoacetaldehyde (3), the
product of the BfPnPDC reaction, is not a substrate for YADH. How-
ever, the phosphono group can be removed by phosphonoacetalde-
hyde hydrolase (Phtase) which then allows further reaction with
YADH. Accordingly, the Phtase from P. putida was also expressed
and affinity purified. Its utility in the coupled assay was verified
using authentic phosphonoacetaldehyde and YADH.
2.3.2. Titration of BfPnPDC with PnDFP monitored by UV–Vis
spectroscopy
The first step in the reaction of ThDP-dependent enzymes is
generally the formation of a covalent nucleophilic adduct between
the C2 atom of the thiazolium ring of ThDP and the carbonyl group
of the substrate.28 The reaction continues via a series of covalently
bound ThDP-complexes, many of which can be observed by both
UV spectrophotometry and circular dichroism (CD) spectrosopy.29
Assuming PnDFP is a mechanism-based inhibitor of BfPnPDC
(Scheme 2), it may be expected that titration of the inhibitor would
provide a ‘‘signature” UV or CD signal. This proved to be true as the
addition of PnDFP to the enzyme resulted in a yellow color.
Accordingly the titration of BfPnPDC (ꢂ50
lM) with 4 (0–
lM) was monitored via UV–Visible spectrophotometry and it
50
The his-tagged BfPnPDC exhibited Michaelis-Menten type
kinetics with its natural substrate, phosphonopyruvate, with no
evidence of the allosteric activity often observed with the yeast
pyruvate decarboxylases.24,25 The Km value was determined to be
2.9 0.2 mM with a kcat value of 10.1 0.3 sꢀ1. These values are in
excellent agreement with those measured for the untagged
was found that the color change was reflected in an increase in
absorbance at 440 nm (Fig. 1). The intensity increased linearly until
a plateau was reached at around 35 lM inhibitor (see inset Fig. 1),
suggesting that (i) the enzyme was saturated with inhibitor and (ii)
the concentration of enzyme may be lower than expected on the
enzyme.12 The PnPDC from Streptomyces viridochromogenes
23
Tü494 had a similar Km value but a 10-fold lower value for kcat
.
Overall, the data provide a kcat/Km of 3500 mMꢀ1 sꢀ1 for the reac-
tion of BfPnPDC with its natural substrate. By comparison, PDC
from Zymomonas mobilis has a kcat/Km value for pyruvate of
440 mMꢀ1 sꢀ1 26
,
ꢂ8-fold lower, while at 45 mMꢀ1 sꢀ1, Saccha-
romyces cerevisiae PDC has a kcat/Km value almost 80-fold lower
than that of BfPnPDC.27 It has been suggested that much of the dif-
ference in catalytic efficiency results from the presence of the
‘‘phosphono” group which is the major source of enzyme-substrate
binding energy.12
2.3. Reaction of PnDFP with BfPnPDC
In control experiments the inhibitor was tested against both
Phtase and YADH. No inhibition was observed. By contrast, PnDFP
proved to be an excellent inhibitor of BfPnPDC. The low Km value
for phosphonopyruvate precluded standard inhibition assays so
IC50 values were obtained at phosphonopyruvate concentrations
Fig. 1. The titration of BfPnPDC (ꢂ50
lM) with PnDFP (4) monitored by UV–Vis
spectroscopy. The peak at 280 nm is due to ThDP.
Please cite this article in press as: Pallitsch K., et al. Bioorg. Med. Chem. (2017), http://dx.doi.org/10.1016/j.bmc.2017.06.013

4
K. Pallitsch et al. / Bioorganic & Medicinal Chemistry xxx (2017) xxx–xxx
to analyze the products of the titration of the enzyme (ꢂ100 mM)
with PnDFP (100 mM). Fig. 3 shows a linear relationship between
added 4 and release of fluoride ion. This 1:1 relationship is clear over
the range 0–50 lM 4, but tapers off around 75 mM. This value is con-
sistent with that obtained for UV–Vis and CD studies, again suggest-
ing that the concentration of enzyme may be lower than anticipated.
2.5. Screening for crystals of BfPnPDC
Attempts were made to crystallize BfPnPDC in the presence and
absence of 4. All attempts were unsuccessful.
3. Discussion
Biosynthesis of phosphonates is important to many pathogenic
organisms.12,14–16 Given that they are not produced in higher
plants, animals and humans, it seemed as though the phosphonate
biosynthetic pathway would be worth exploring as a target for
novel antibiotics. An important consideration is that two enzymes,
PEP mutase and PnPDC, catalyze the first two steps in the biosyn-
thesis of virtually all known phosphonates (Scheme 1). Accord-
ingly, by focusing on these enzymes it is possible to envisage a
general strategy to interfere with all phosphonate biosyntheses.
PnPDC was chosen as our initial target as it (i) carries out an irre-
versible biosynthetic step and (ii) is of fundamental interest in that
it is thought to closely resemble the fused domain common ances-
tor of all ThDP-dependent enzymes.35,36 To date, only two PnPDCs,
those from Bacteroides fragilis (BfPnPDC)12 and Streptomyces viri-
dochromogenes Tü494 (SvPnPDC)23 have been fully characterized.
The former is from a human pathogen while the latter organism
produces the natural herbicide bialaphos and the antibiotic, strep-
tazolin. Another PnPDC, that from Streptomyces hygroscopicus, has
been purified but not well characterized.37 Confusingly, gel filtra-
tion analysis has suggested that the native enzymes existed as a
homotrimer, a homodimer and a homotetramer, respectively. The
oligomeric status could be clarified by an X-ray structure but, as
yet, no PnPDC has been crystallized. It was hoped that formation
of an inhibitor:ThDP adduct may facilitate crystallization, as has
been observed for other ThDP enzymes such as benzoylformate
decarboxylase33,38 and benzaldehyde lyase.39,40
In the first instance we proposed that 3,3-difluoro-3-phospho-
nopyruvate (PnDFP, 4) would act specifically as mechanism-based
inhibitor for PnPDC. To explore this possibility the sodium salt of
PnDFP was synthesized in three steps from diethyl difluo-
romethylphosphonate (Scheme 3). An alternative strategy employ-
ing dibromodifluoromethane as the starting material41–45 was also
successful, although the yields were much more variable. Con-
comitantly we were able to express and purify BfPnPDC as a C-ter-
minal his-tagged variant, and used a coupled assay to confirm that
its kinetic data were identical to that of the wild type enzyme.
Testing showed that 4 was a potent inhibitor of BfPnPDC, with a
Ki value in the sub-micromolar range.
Fig. 2. The titration of BfPnPDC (ꢂ50 mM) with PnDFP (4) monitored by CD
spectropolarimetry.
basis of Bradford measurements. The latter is not unexpected as
variation in concentrations arising from the Bradford method are
common.30 Unfortunately, interference due to the presence of
ThDP in the storage buffers renders OD280 measurements even less
reliable.
2.3.3. Titration of BfPnPDC with PnDFP monitored by CD
spectropolarimetry
CD was also used to monitor a similar titration of BfPnPDC with
4, with results shown in Fig. 2. Two maxima were observed. The
first, around 315 nm, has been attributed to formation of the
10,40-iminopyrimidine tautomer of enzyme-bound ThDP29,31 or
intermediates related to C2
maximum, around 440 nm, has been attributed to formation of a
charge transfer band, likely arising from an interaction of a bond
a
-adducts of ThDP.32–34 The second
p
on the C2 side-chain of ThDP and the positively charged thiazolium
ring as donor and acceptor, respectively.29 As with the UV data, the
CD data at 440 nm showed a linear increase with inhibitor concen-
tration, again with a plateau around 35 lM inhibitor (Fig. S1).
2.4. Reaction of PnDFP with BfPnPDC releases fluoride ion
The postulated mechanism for the reaction of PnDFP with the
ThDP of BfPnPDC (Scheme 2) requires the elimination of a fluoride
ion. To test this hypothesis, a fluoride ion-selective probe was used
Silverman46 has defined several criteria, most of which should
be met by 4, in order for it to be characterized as a mechanism-
based inhibitor. These include (a) time-dependent inactivation,
(ii) saturation kinetics, (iii) substrate protection, (iv) irreversible
inactivation, (v) 1:1 stoichiometry of binding, (vi) involvement of
a catalytic step and (vii) that binding occurs at the active site.
PnDFP is clearly an analogue of the substrate, PnP, and shows the
qualities necessary to call it a mechanism-based inhibitor. Indeed,
its reaction with BfPnPDC was sufficiently rapid to preclude the
dilution assay normally used to demonstrate that time-dependent
inactivation was accompanied by saturation kinetics. Nonetheless
we could show that inactivation could not be reversed by either
dilution or exhaustive dialysis. Increasing the substrate concentra-
Fig. 3. Fluoride ion analysis of the titration of PnPDC (ꢂ100
lM) with PnDFP (4).
Please cite this article in press as: Pallitsch K., et al. Bioorg. Med. Chem. (2017), http://dx.doi.org/10.1016/j.bmc.2017.06.013

K. Pallitsch et al. / Bioorganic & Medicinal Chemistry xxx (2017) xxx–xxx
5
tion from 3
14 M, thereby satisfying the criterion of substrate protection.
Titration of the enzyme with PnDFP resulted in the formation of
l
M to 10
l
M saw the IC50 value increase from 1.3 to
(BfPnPDC) was a gift of Dr. Debra Dunaway-Mariano (University
of New Mexico, Albuquerque, NM). plasmid containing
recombinant C-terminal His-tagged phosphonoacetaldehyde
hydrolase (Phtase) from Pseudomonas putida (pET22Phtase-
His)23 was a gift of Prof. Georg A. Sprenger (University of Stutt-
gart, Germany).
l
A
a yellow color which could be monitored at 440 nm by both UV–
Vis spectrophotometry (Fig. 1) and CD spectroscopy (Fig. 2). Both
methods provided clear evidence of a 1:1 relationship between
enzyme and inhibitor concentration. A similar color was observed
1H, 13C, 19F and 31P NMR spectra were measured in CDCl3 or D2O
on a Bruker Avance 400 (1H: 400.27 MHz, 13C: 100.65 MHz, 19F:
376.50 MHz, 31P: 162.03 MHz) or DRX 400 (1H: 400.13 MHz, 13C:
100.61 MHz, 31P: 161.98 MHz) spectrometer as indicated. The
chemical shifts were referenced to residual CHCl3 (dH 7.24), HOD
(dH 4.80); CDCl3 (dC 77.23) external H3PO4 (85%) (dP 0.00) and FCCl3
(dF 0.00). Chemical shifts (d) are given in ppm and coupling con-
stants (J) in Hz. IR spectra were recorded on a Bruker VERTEX 70
IR spectrometer in ATR mode. Spectra for compounds 4 and 10a
are provided as Supplementary material.
TLC was usually carried out on 0.25 mm thick Merck plates
precoated with silica gel 60 F254. Spots were visualised by UV
and/or dipping the plate into a solution of (NH4)6Mo7O24ꢃ4H2O
(25.0 g) and Ce(SO4)2ꢃ4H2O (1.0 g) in 10% aqueous H2SO4
(500 mL), followed by heating with a heat gun. Flash (column)
chromatography was performed with Merck silica gel 60 (230–
400 mesh).
Diethyl difluoromethylphosphonate, bromotrimethylsilane,
allyltrimethylsilane and diethyl oxalate were from Sigma Aldrich.
THF was refluxed over potassium and distilled prior to use.
Authentic samples of PnP (2) and PnAA (3) were prepared as their
sodium salts using variations on established methods. Details are
provided in the Supplementary material. All other chemicals and
buffer components were used as purchased from Sigma-Aldrich,
Acros, Fluka or Merck.
by Merski and Townsend47 during the reaction of
D-glyceralde-
hyde-3-phosphate with N2-(2-carboxyethyl)arginine synthase, a
ThDP-dependent enzyme involved in clavulanic acid biosynthesis.
In that instance the yellow chromophore was attributed to the for-
mation of an acryloyl-ThDP adduct, which is not dissimilar to the
adduct predicted for PnPDC in Scheme 2.47 In addition to the max-
imum at 440 nm, the CD spectra showed a second maximum at
315 nm. There are several instance where the Jordan group has
demonstrated that this band arises from formation of the 10,40-
iminopyrimidine tautomer of enzyme-bound ThDP or enzyme
bound C2a
-adducts of ThDP.29,31–34 Thus, observation of the band
at 315 nm provides further evidence for 4 binding at the active site.
Perhaps the most important criterion for a mechanism-based
inhibitor is the involvement of a catalytic step. For 4 we were able
to confirm this by using an ion-selective electrode to establish that
enzyme inactivation was accompanied by the release of a fluoride
ion. Fig. 3 shows that, when BfPnPDC was titrated with 4, there was
a 1:1 relationship between the concentration of free Fꢀ and added
inhibitor. Taken together, all data point to PnDFP being a potent
mechanism-based inhibitor of phosphonopyruvate decarboxylase.
At this time it is not clear which ThDP complex (7 or 8, Scheme 2)
is responsible for the inhibition, and analogues of both keto and
enol complexes have been reported on ThDP-dependent
enzymes.29 It is notable that 8 is an analogue of acetylthiamin
diphosphate which may be expected to be hydrolyzed. Two obser-
vations mitigate against this. First, there is no rapid loss of the
absorbance at 440 nm and second, there is no rapid regeneration
of active enzyme. It appears that, while acylthiamin diphosphate
intermediates can hydrolyze rapidly on some enzymes, on others,
such as pyruvate oxidase48 and benzoylformate decarboxylase,49
it can be quite slow. PnPDC would seem to fit into the latter cate-
gory. Another possibility is that the equilibrium lies in favor of 8,
which is subsequently attacked by an active site nucleophile. Such
a mechanism was proposed for the inactivation of the E1 subunit of
the pyruvate dehydrogenase complex.22 It was hoped that this
issue could be resolved by crystallography but, unfortunately, X-
ray quality crystals are not yet available. Screening for suitable
conditions is continuing.
While inhibition of PnPDC is the prime focus of this paper, it must
be remembered that PnP is also a substrate of PEP mutase
(Scheme 1). There remains the possibility that 4 could be converted
into difluoroPEP and, given that monofluoroPEP inhibits several PEP
utilizing enzymes,50 there is a good chance that difluoroPEP would
behave similarly. Even assuming that difluoroPEP acts as a substrate
for the PEP enzymes, subsequent transformations would result in
the formation of fluorinated pyruvate analogues which are also tox-
ic. It is also important to note that PEP mutase is not found in higher
organisms and, therefore, no fluorinated PEP or subsequentproducts
will be formed in mammaliancells. Given this, and the successof 4 as
a PnPDC inhibitor, future work will focus on the development of 4 as
an in vivo inhibitor of phosphonate biosynthesis.
4.1.1. Ethyl 3-(diethoxyphosphoryl)-3,3-difluoro-2,2-dihydroxy-
propanoate (10a):51
nBuLi (3.1 mmol, 1.24 mL, 2.5 M solution in hexanes) was added
dropwise to a solution of iPr2NH (334 mg, 3.3 mmol, 0.46 mL) in
dry THF (3 mL) at ꢀ40 °C under argon. The mixture was stirred
at this temperature for 15 min and afterwards cooled to ꢀ95 °C.
Diethyl difluoromethylphosphonate (600 mg, 3.1 mmol, 0.50 mL),
dissolved in dry THF (3 mL) was added dropwise and the resulting
solution was stirred for 5 min at the same temperature. Diethyl
oxalate (453 mg, 3.1 mmol, 0.42 mL) dissolved in dry THF (3 mL)
was added and the solution was allowed to warm to ꢀ78 °C in
the cooling bath, whereupon it turned pale yellow. Upon addition
of glacial acetic acid (186 mg, 3.1 mmol, 0.17 mL) the solution
became colourless again. After extraction with NaHCO3 solution
(3 ꢁ 10 mL, sat.) the combined aqueous phases were washed with
EtOAc (2 ꢁ 10 mL), were dried (Na2SO4) and concentrated in vacuo.
The crude product was purified by bulb-to-bulb distillation (105–
115 °C/0.49 mbar) to give a pale yellow, viscous liquid (569 mg,
60%). Subsequent flash chromatography (ethyl acetate (EtOAc)/
hexanes = 2:1, Rf = 0.44) yielded the difluorinated phosphonate in
its hydrate form (10a) as a colourless oil; 1H NMR (CDCl3,
400.13 MHz): d 5.15 (br s, 1.5 H, 2 ꢁ OH), 4.37 (q, J = 7.2 Hz, 2 H,
C(O)OCH2), 4.38–4.24 (m, 4 H, 2 ꢁ P(O)OCH2), 1.37 (td, JHH = 7.1 Hz,
JHP = 0.7 Hz, 6 H, 2 ꢁ CH3), 1.35 (t, J = 7.2 Hz, 3 H, C(O)CH2CH3); 13
C
NMR (CDCl3, 100.61 MHz): d 167.45 (d, JCP = 10.9 Hz, C@O), 115.77
(td, JCP = 198.8 Hz, JCF = 275.4 Hz, CF2), 92.76 (td, JCP = 12.6 Hz,
JCF = 27.4 Hz, C(OH)2), 65.95 (d, JCP = 6.6 Hz, 2 ꢁ OCH2), 64.24 (s,
OCH2), 16.50 (d, JCP = 5.9 Hz, 2 ꢁ CH3), 14.13 (s, CH3); 31P NMR
(CDCl3, 161.98 MHz): d 6.90 (t, JPF = 95.1 Hz); 19F NMR (CDCl3,
4. Experimental section
4.1. General methods
376.50 MHz): d ꢀ120.68 (d, JPF = 95.1 Hz); IR (ATR):
m = 3275,
2989, 1744, 1393, 1370, 1249, 1158, 1085, 1013 cmꢀ1; Anal. calc.
for C9H17F2O7P (306.20): C 35.50%, H 5.60%, F 12.41%; found: C
35.31%, H 5.49%, F 12.16%.
A
plasmid (bf-Pyrdecarb-pET3A)12 containing recombinant
phosphonopyruvate decarboxylase from Bacteroides fragilis
Please cite this article in press as: Pallitsch K., et al. Bioorg. Med. Chem. (2017), http://dx.doi.org/10.1016/j.bmc.2017.06.013

6
K. Pallitsch et al. / Bioorganic & Medicinal Chemistry xxx (2017) xxx–xxx
Ki ¼ IC50=ð1 þ ð½SꢄÞ=Km
Þ
ð1Þ
4.1.2. 3,3-Difluoro-2,2-dihydroxy-3-phosphonopropanoic acid (4 as its
sodium salt)
Ethyl 3-(diethoxyphosphoryl)-3,3-difluoro-2-oxo-propanoate
(10, 0.153 g, 0.5 mmol) was dissolved in dry 1,2-dichloroethane
(DCE, 5 mL) under argon at room temperature. Allyltrimethylsilane
(0.114 g, 1.0 mmol, 0.16 mL) and bromotrimethylsilane (0.459 g,
3.0 mmol, 0.40 mL) were added quickly one after the other and
the reaction mixture was heated at 60 °C for 23 h. Then, all volatile
compounds were removed in vacuo at room temperature. DCE
(5 mL) was added and volatiles were removed a second time; first
at room temperature and then at 40 °C. The residue was dried in
vacuo for 2 h and then dissolved in water (2 mL). The pH of the
solution was adjusted to 10 and it was stirred at room temperature
for 1 h. Excess base was then removed by ion exchange chromatog-
raphy (Dowex 50W ꢁ 8, H+ form, 12 mL, water as eluent). All acidic
fractions were collected, neutralized (pH 7.1) with aqueous NaOH
solution (0.5 M) and finally lyophilized to give a white powder of
3,3-difluoro-2,2-dihydroxy-3-phosphonopropanoic acid (4) as its
sodium salt (0.138 g, about 0.48 mmol); 1H NMR (D2O,
400.13 MHz): no signals detected (despite some minimal impuri-
ties which can be quantified by means of an internal standard);
13C NMR (D2O, 100.61 MHz): d 173.84 (s, C@O), 118.77 (td,
JCP = 166.7 Hz, JCF = 270.3 Hz, CF2), C-2 signal of hydrate was not
detected; 31P NMR (D2O, 161.98 MHz): d 6.30 (t, JPF = 80.1 Hz);
19F NMR (D2O, 376.50 MHz): d ꢀ122.01 (d, JPF = 80.1 Hz).
4.2.3. Time-dependent inactivation of BfPnPDC
BfPnPDC (6 M) and 4 (10 M or 50 M) were mixed in 50 mM
HEPES (pH 8.0) containing 300 mM NaCl, 10 mM MgCl2, and 1 mM
DTT and incubated at 30 °C. At appropriate time intervals, an ali-
quot (10 lL) was diluted 1:100 into a standard assay mixture con-
l
l
l
taining 30 mM (i.e., saturating) phosphonopyruvate.
4.3. Reaction of phosphonodifluoropyruvate with BfPnPDC monitored
by UV–VIS spectrophotometry and circular dichroism (CD)
spectropolarimetry
BfPnPDC (ꢂ50 mM), in a 2 mL volume of BfPnPDC storage buffer,
was titrated with 4 at concentrations ranging from 0 to 50 mM.
PnDFP was introduced in 10 lL aliquots with each addition result-
ing in a 5 mM increase in inhibitor concentration. Spectra were
obtained using a 1 cm path length cell. After the addition of each
of the inhibitor aliquots, the solution was gently mixed and
allowed to sit for a minimum of 3 min at 25 °C. The reaction mix-
tures were monitored on a Cary 50 Bio UV–Vis spectrophotometer
at 440 nm to ensure the reaction was complete, before a scan was
recorded from 550 to 250 nm. The Abs440 titration curves were
constructed using SigmaPlot 12.3.
The procedure was repeated using the Jasco J-810 CD spec-
tropolarimeter. The reaction was initially monitored at 440 nm
on a UV–VIS spectrophotometer to ensure the reaction was com-
plete. Subsequently the mixture was then scanned from 550 to
250 nm on the CD spectropolarimeter. The data obtained for each
inhibitor concentration was the average of five individual scans.
Titration curves for the change in ellipticity at 440 nm were con-
structed using SigmaPlot 12.3.
4.2. Expression and purification of phosphonopyruvate decarboxylase
and phosphonoacetaldehyde hydrolase
The QuikChange site-directed mutagenesis kit (Stratagene) was
used to replace the stop codon of BfPnPDC with a Xho1 restriction
site. Subsequently the NdeI-XhoI fragment containing the modified
BfPnPDC gene was inserted into pET24b (Novagen) to give the
expression vector, pET24bBfPnPDC-His. The fidelity of the amplifi-
cation and the presence of the changed nucleotides was confirmed
by sequencing (University of Michigan DNA Sequencing Core
Facility).
4.4. Fluoride ion analysis of the reaction of phosphonodifluoropyruvate
with BfPnPDC
Release of fluoride ions from the reaction of 4 with BfPnPDC was
monitored using a fluoride ion-selective electrode. Samples con-
taining BfPnPDC (ꢂ100 mM) were mixed with PnDFP (0–75 mM)
in a total volume of 1 mL. In addition, two controls were used.
The first contained BfPnPDC buffer and water, and the second con-
tained BfPnPDC buffer, 75 mM 4 and water in a total volume of
1 mL. After gentle mixing reactions were incubated at room tem-
perature for 30 min. The enzyme was removed from the samples
using 10,000 kDa cut-off microspin columns (Microcon, Millipore).
Total ionic strength adjustment buffer II solution (TISAB II) was
added to the samples to a final volume of 2 mL. The fluoride ion
concentration in each sample was determined.
Following transformation of the plasmids, pET24bBfPnPDC-His
and pET22Phtase-His into E. coli BL21 (DE3) cells, expression and
purification of the enzymes was achieved using routine proce-
dures.52 The enzymes were quantitated with the Bradford
method53 using bovine serum albumin as the standard before
being concentrated to ꢂ15 mg/mL. BfPnPDC was stored at ꢀ80 °C
and Phtase was stored at 4 °C in 50 mM HEPES (pH 8.0) containing
300 mM NaCl, 10 mM MgCl2, and 1 mM DTT.
4.2.1. Steady-state kinetic analysis of phosphonopyruvate
decarboxylase
The activity of purified BfPnPDC was determined using a cou-
pled assay. The assay mixture contained, in a total of 1 mL,
100 mM HEPES (pH 7.0) 5 mM MgCl2, 0.2 mM NADH, 60 U/mL
4.5. Attempted crystallization BfPnPDC
YADH, 100 lg phosphonoacetaldehyde hydrolase, and varying con-
For crystallization trials the enzyme was concentrated to
15 mg/mL in 25 mM HEPES pH 7.0, containing 0.2 mM ThDP and
0.1 mM MgCl2. Screening was initiated by mixing the enzyme (in
the presence and absence of 4) 1:1 with the conditions of Hampton
Crystal Screen II.
centrations of phosphonopyruvate (0.5–15 mM). The reaction was
carried out at 30 °C and initiated by the addition of BfPnPDC. Each
assay was performed in triplicate. The kinetic parameters were
determined by fitting the initial rate data to the Michaelis-Menten
equation using the enzyme kinetics package from SigmaPlot 12.3.
4.2.2. Determination of IC50 and Ki values for
Acknowledgements
phosphonodifluoropyruvate
Here the standard assay was used with the concentration of
phosphonopyruvate held constant at 3 mM (ꢂKm) or 30 mM
(10 ꢁ Km) while the concentration of 4 was varied. The IC50 value
was determined by plotting the concentration of inhibitor (mM)
against residual activity. The Ki was determined from the IC50 value
using the Cheng-Prusoff equation (Eq. (1)).
We thank National Science Foundation (CHE 1306877 to MJM)
and the Austrian Science Fund (P27987-N28 to FWF) for financial
support. We thank Johannes Theiner for performing combustion
analysis, and Susanne Felsinger for recording 19F NMR spectra.
We also acknowledge the help of the Lippert laboratory at the Indi-
ana University School of Dentistry with the fluoride ion analyses,
Please cite this article in press as: Pallitsch K., et al. Bioorg. Med. Chem. (2017), http://dx.doi.org/10.1016/j.bmc.2017.06.013

K. Pallitsch et al. / Bioorganic & Medicinal Chemistry xxx (2017) xxx–xxx
7
22. Frey PA, Flournoy DS, Gruys K, Yang YS. Ann N Y Acad Sci. 1989;573:21–35.
23. Johnen S, Sprenger GA. J Mol Catal B. 2009;61:39–46.
24. Hübner G, Weidhase R, Schellenberger A. Eur J Biochem. 1978;92:175–181.
Malea Kneen for critical reading of the manuscript, and Prof. Ron
Kluger for his advice on ThDP intermediates.
25. Krieger F, Spinka M, Golbik R, Hübner G, König S. Eur
J Biochem.
2002;269:3256–3263.
26. Siegert P, McLeish MJ, Baumann M, et al. Prot Eng Des Sel. 2005;18:345–357.
27. Li H, Jordan F. Biochemistry. 1999;38:10004–10012.
A. Supplementary data
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.bmc.2017.06.013.
28. Kluger R, Tittmann K. Chem Rev. 2008;108:1797–1833.
29. Patel H, Nemeria NS, Andrews FH, McLeish MJ, Jordan F. Biochemistry.
2014;53:2145–2152.
30. Lu TS, Yiao SY, Lim K, Jensen RV, Hsiao LL. N Am J Med Sci.. 2010;2:325–328.
31. Nemeria N, Korotchkina L, McLeish MJ, et al. Biochemistry.
2007;46:10739–10744.
References
32. Nemeria N, Baykal A, Joseph E, et al. Biochemistry. 2004;43:6565–6575.
33. Brandt GS, Kneen MM, Chakraborty S, et al. Biochemistry. 2009;48:3247–3257.
34. Nemeria NS, Chakraborty S, Balakrishnan A, Jordan F. FEBS J.
2009;276:2432–2446.
35. Costelloe SJ, Ward JM, Dalby PA. J Mol Evol. 2008;66:36–49.
36. Duggleby RG. Acc Chem Res. 2006;39:550–557.
1. Kamat SS, Williams HJ, Dangott LJ, Chakrabarti M, Raushel FM. Nature.
2013;497:132–136.
2. Yu X, Doroghazi JR, Janga SC, et al. Proc Natl Acad Sci USA.
2013;110:20759–20764.
3. Seto H, Kuzuyama T. Nat Prod Rep. 1999;16:589–596.
4. Peck SC, van der Donk WA. Curr Opin Chem Biol. 2013;17:580–588.
5. Horsman GP, Zechel DL. Chem Rev. 2017;117:5704–5783.
6. Bowman E, McQueney M, Barry RJ, Dunaway-Mariano D. J Am Chem Soc.
1988;110:5575–5576.
7. Metcalf WW, van der Donk WA. Annu Rev Biochem. 2009;78:65–94.
8. Patel SS, Balfour JA, Bryson HM. Drugs.. 1997;53:637–656.
9. Whitteck JT, Ni W, Griffin BM, et al. Angew Chem Int Ed Engl.
2007;46:9089–9092.
37. Nakashita H, Watanabe K, Hara O, Hidaka T, Seto H.
1997;50:212–219.
J Antibiot (Tokyo).
38. Chakraborty S, Nemeria NS, Balakrishnan A, et al. Biochemistry.
2009;48:981–994.
39. Brandt GS, Nemeria N, Chakraborty S, et al. Biochemistry. 2008;47:7734–7743.
40. Brandt GS, Kneen MM, Petsko GA, Ringe D, McLeish MJ. J Am Chem Soc.
2009;132:438–439.
41. Kawamoto AM. Campbell MM. J Chem Soc Perkin Trans. 1997;1:1249–1253.
42. Burton DJ, Ishihara T, Maruta M. Chem Lett. 1982;11:755–758.
43. Burton DJ, LaSprague LG. J Org Chem. 1988;53:1523–1527.
44. Burton DJ, LaSprague LG. J Org Chem. 1989;54:613–617.
45. Tsai H-J. Phosphorus, Sulfur Silicon Relat Elem. 1997;122:247–259.
46. Silverman RB. Methods Enzymol. 1995;249:240–283.
10. Duke SO, Dayan FE, Romagni JG, Rimando AM. Weed Res. 2000;40:99–111.
11. Morais MC, Zhang W, Baker AS, et al. Biochemistry. 2000;39:10385–10396.
12. Zhang G, Dai J, Lu Z, Dunaway-Mariano D. J Biol Chem. 2003;278:41302–41308.
13. Ju KS, Gao J, Doroghazi JR, et al. Proc Natl Acad Sci USA. 2015;112:12175–12180.
14. Katayama N, Tsubotani S, Nozaki Y, Harada S, Ono H.
1990;43:238–246.
J Antibiot (Tokyo).
47. Merski M, Townsend CA. J Am Chem Soc. 2007;129:15750–15751.
48. Tittmann K, Golbik R, Ghisla S, Hübner G. Biochemistry. 2000;39:10747–10754.
15. Carter RF. J Pathol.. 1970;100:217–244.
16. Berriman M, Ghedin E, Hertz-Fowler C, et al. Science. 2005;309:416–422.
17. Circello BT, Eliot AC, Lee JH, van der Donk WA, Metcalf WW. Chem Biol.
2010;17:402–411.
18. Walsh CT. Tetrahedron. 1982;38:871–909.
19. Kluger R, Pike DC. J Am Chem Soc. 1977;99:4504–4506.
49. Dirmaier LJ, Garcia GA, Kozarich JW, Kenyon GL.
1986;108:3149–3150.
J Am Chem Soc.
50. Garcia-Alles LF, Erni B. Eur J Biochem. 2002;269:3226–3236.
51. Phillion DP, Cleary DG. J Org Chem. 1992;57:2763–2764.
52. Andrews FH, Horton JD, Shin D, et al. Biochim Biophys Acta.
2015;1854:1001–1009.
20. Brammer LA, Smith JM, Wade H, Meyers CF.
2011;286:36522–36531.
J
Biol Chem.
53. Bradford MM. Anal Biochem. 1976;72:248–254.
21. Flournoy DS, Frey PA. Biochemistry. 1989;28:9594–9602.
Please cite this article in press as: Pallitsch K., et al. Bioorg. Med. Chem. (2017), http://dx.doi.org/10.1016/j.bmc.2017.06.013