A study on the chiral inversion of mandelic acid in humans

Article abstract of DOI:10.1039/c3ob42515k

Mandelic acid is a chiral metabolite of the industrial pollutant styrene and is used in chemical skin peels, as a urinary antiseptic and as a component of other medicines. In humans, S-mandelic acid undergoes rapid chiral inversion to R-mandelic acid by a

Full text of DOI:10.1039/c3ob42515k

Organic &
Biomolecular Chemistry
View Article Online
View Journal | View Issue
PAPER
A study on the chiral inversion of mandelic
acid in humans†
Cite this: Org. Biomol. Chem., 2014,
12, 6737
Maksims Yevglevskis, Catherine R. Bowskill, Chloe C. Y. Chan, Justin H.-J. Heng,
Michael D. Threadgill, Timothy J. Woodman and Matthew D. Lloyd*
Mandelic acid is a chiral metabolite of the industrial pollutant styrene and is used in chemical skin peels,
as a urinary antiseptic and as a component of other medicines. In humans, S-mandelic acid undergoes
rapid chiral inversion to R-mandelic acid by an undefined pathway but it has been proposed to proceed
via the acyl-CoA esters, S- and R-2-hydroxy-2-phenylacetyl-CoA, in an analogous pathway to that for
Ibuprofen. This study investigates chiral inversion of mandelic acid using purified human recombinant
enzymes known to be involved in the Ibuprofen chiral inversion pathway. Both S- and R-2-hydroxy-2-
phenylacetyl-CoA were hydrolysed to mandelic acid by human acyl-CoA thioesterase-1 and -2 (ACOT1
and ACOT2), consistent with a possible role in the chiral inversion pathway. However, human α-methyl-
acyl-CoA racemase (AMACR; P504S) was not able to catalyse exchange of the α-proton of S- and R-2-
hydroxy-2-phenylacetyl-CoA, a requirement for chiral inversion. Both S- and R-2-phenylpropanoyl-CoA
were epimerised by AMACR, showing that it is the presence of the hydroxy group that prevents epimeri-
sation of R- and S-2-hydroxy-2-phenylacetyl-CoAs. The results show that it is unlikely that 2-hydroxy-2-
phenylacetyl-CoA is an intermediate in the chiral inversion of mandelic acid, and that the chiral inversion
of mandelic acid is via a different pathway to that of Ibuprofen and related drugs.
Received 16th December 2013,
Accepted 10th July 2014
DOI: 10.1039/c3ob42515k
www.rsc.org/obc
of mandelic acid have been investigated as inhibitors of glyoxy-
lase 1 for the treatment of cancer15 and glycosides containing
Introduction
Mandelic acid (2-hydroxy-2-phenylacetic acid 1) is a common mandelic acid have also been shown to possess anti-tumour
chiral α-hydroxyacid produced from styrene, a hazardous activity.16
monomer used in the plastics industry.1,2 During its meta-
Mandelic acid 1 has long been known to undergo chiral
bolism, styrene is converted to phenyloxirane (styrene epoxide) inversion during its metabolism. In bacteria, bi-directional
by CYP2E1, the majority of which is ring opened to 1-phenyl- chiral inversion is catalysed by the Mg2+-dependent mandelate
ethane-1,2-diol (styrene glycol) by epoxide hydrolase. The racemase to produce
a
racemic mixture from either
latter is oxidised to mandelic acid and 2-oxo-2-phenylacetic enantiomer.17–19 In mammals, mandelic acid 1 undergoes uni-
acid (phenylglyoxylic acid).1,3–5 Both 1-phenylethane-1,2-diol directional chiral inversion of the S-enantiomer to form the
and mandelic acid 1 are excreted in urine and monitoring of R-enantiomer.3,4,6 Gao et al.6 proposed that chiral inversion of
their levels is used to assess environmental exposure to S-mandelic acid 1S occurred by
a three step pathway
styrene,6,7 whilst 2-oxo-2-phenylacetic acid has been shown to (Scheme 1), analogous to that for Ibuprofen and other 2-aryl-
induce striatal-motor toxicity in rats.8 Mandelic acid 1 is also propanoic acid (2-APA) drugs,20–22 consisting of: formation of
used pharmaceutically in α-hydroxyacid chemical skin peels S-2-hydroxy-2-phenylacetyl-CoA 2S by an acyl-CoA synthetase
for the treatment of acne, scarring and hyperpigmentation9 (Step 1); epimerisation to R-2-hydroxy-2-phenylacetyl-CoA 2R
and in urinary antiseptics and bladder irrigants.10 It is also a by α-methylacyl-CoA racemase (AMACR) via a deprotonated
component of or used in the manufacture of anti-viral (enolate) intermediate (Step 2); and hydrolysis of R-2-hydroxy-
agents,10–12 contraceptives13 and cyclic peptides.14 Thioesters 2-phenylacetyl-CoA 2R to R-mandelic acid 1R by an acyl-CoA
thioesterase (ACOT) (Step 3). This pathway6 was proposed
based on the structural similarities between mandelic acid 1
Medicinal Chemistry, Department of Pharmacy & Pharmacology, University of Bath,
and Ibuprofen 3, in that both compounds possess an aromatic
side-chain and a chiral centre adjacent to the carboxylic acid.
Claverton Down, Bath BA2 7AY, United Kingdom. E-mail: M.D.Lloyd@bath.ac.uk;
Fax: +44 (0) 1225 386114
S-2-Hydroxy-2-phenylacetyl-CoA 2S was rapidly hydrolysed
†Electronic supplementary information (ESI) available: Substrate NMR and
and R-mandelic acid 1R formed in rat liver cell extracts,6 but
mass spectra data, NMR spectra of substrates incubated with AMACR and
conversion of 2S to 2R was not proven.
kinetic plots of substrates of ACOT enzymes. See DOI: 10.1039/c3ob42515k
This journal is © The Royal Society of Chemistry 2014
Org. Biomol. Chem., 2014, 12, 6737–6744 | 6737

View Article Online
Paper
Organic & Biomolecular Chemistry
implies a multi-step chiral inversion pathway in which either
activation of S-mandelic acid 1S or production of R-mandelic
acid 1R is stereospecific. That of Ibuprofen 3 and related drugs
is from the R-enantiomer to the S-enantiomer (Scheme 2),20
i.e. the stereochemical direction of chiral inversion is opposite
for the two pathways. The direction of chiral inversion for
Ibuprofen 3 is known to be determined by the requirement
of long-chain fatty acyl-CoA synthetase for substrates with
R-configuration.23–25
A number of potential acyl-CoA substrates were required in
order to test whether mandelic acid 1 chiral inversion was
mediated by the Ibuprofen 3 pathway. Gao et al.6 synthesised
racemic 2-hydroxy-2-phenylacetyl-CoA esters 2 by reaction of
phenylglyoxal with CoA-SH followed by treatment with 2,6-
dimethylpyridine.15 In this study a stereoselective synthesis of
2 was developed (Scheme 3A) as the efficiency of conversion of
ACOT substrates is known to be influenced by the epimeric
configuration at the 2-position.22 Reaction of mandelic acid
1R or 1S with carbonyldiimidazole did not yield the expected
product, and therefore a direct synthesis of 2R and 2S
was not possible. The hydroxy groups of mandelic acid 1R
or 1S were therefore protected using TBDMS-Cl in a known
procedure to give the chiral TBDMS ethers 5R and 5S.14 These
were treated with carbonyldiimidazole to give the activated
acids 6R and 6S. Treatment of 6R and 6S with CoA-SH followed
by deprotection with KF gave the desired epimeric acyl-CoA
esters 2R and 2S. Acidification of the TBDMS-protected acyl-
CoA esters 7R and 7S was necessary to prevent base-catalysed
hydrolysis of the acyl-CoA ester upon treatment with KF.
Scheme 1 The proposed chiral inversion pathway for mandelic acid 1.
ACOT, acyl-CoA thioesterase; AMACR, α-methylacyl-CoA racemase.
Scheme 2 The chiral inversion pathway of Ibuprofen 3 and other
2-APA drugs. ACOT, acyl-thioesterase; AMACR, α-methylacyl-CoA
racemase.
The chiral inversion pathway of Ibuprofen 3R and other
2-APA drugs (Scheme 2) is well-established.20–22 Only R-Ibupro-
fen 3R is converted into R-ibuprofenoyl-CoA 4R, by long-chain
fatty acyl-CoA synthetase (Step 1),23–25 which undergoes R- to
S-chiral inversion.20 Chiral inversion of R-ibuprofenoyl-CoA
4R is catalysed by α-methylacyl-CoA racemase (AMACR)
(Step 2),20,21,26,27 whilst hydrolysis of the S-ibuprofenoyl-CoA
product 4S is probably catalysed by acyl-CoA thioesterase-1
(ACOT1) (Step 3).22 The pathway necessitates transport of 4R
from the cytosol into mitochondria or peroxisomes where
AMACR is localised,28–30 followed by export of 4S to the cytosol
for hydrolysis.20,22
This paper reports an in vitro study on the chiral inversion
of mandelic acid 1 using recombinant human AMACR and
ACOT enzymes to determine if chiral inversion can occur by
the same pathway as for Ibuprofen 3. The results show that
2-hydroxy-2-phenylacetyl-CoA 2 is not a substrate for AMACR,
and hence chiral inversion of mandelic acid 1 occurs by a
different pathway to Ibuprofen 3.
Scheme 3 Synthesis of substrates. (A) S-2-Hydroxy-2-phenylacetyl-
CoA 2S; (B) S-2-phenylpropanoyl-CoA 8S. Reagents and conditions:
(i) TBDMS-Cl, imidazole, THF, 4 °C, 1 hour, then 1.0 M NaOHaq., 105 min;
(ii) carbonyldiimidazole, dichloromethane, room temp., 45 min; (iii)
CoA-SH, tri-lithium salt, THF–aq. 0.1 M NaHCO3 (1 : 1), room temp,
>16 h; (iv) HCl to pH ∼3.5, KF, 18 h. R-2-Hydroxy-2-phenylacetyl-CoA
2R and R-2-phenylpropanoyl-CoA 8R were synthesised by identical
routes from R-mandelic acid 1R and R-2-phenylpropanoic acid 9R,
Results and discussion
Chiral inversion of mandelic acid 1 is from the S-enantiomer
to the R-enantiomer (Scheme 1).4,6 This uni-directional chiral
inversion of mandelic acid from 1S to 1R in mammalian cells respectively.
6738 | Org. Biomol. Chem., 2014, 12, 6737–6744
This journal is © The Royal Society of Chemistry 2014

View Article Online
Organic & Biomolecular Chemistry
Paper
R- and S-2-phenylpropanoyl-CoAs 8R and 8S were also syn-
thesised (Scheme 3B) in order to compare the influence of
2-hydroxy vs. 2-methyl groups. Direct reaction of 2-phenylpro-
panoic acid 9R and 9S with carbonyldiimidazole21 gave the
activated acids 10R and 10S, which were treated with CoA-SH
to give 8R and 8S. Myristoyl-CoA 11 was synthesised as a
known substrate of ACOT1 and ACOT2,22 and -fenoprofenoyl-
CoA as a known substrate of AMACR.21
Table 2 Kinetic parameters SE for ACOT2. Ki values are reported for
those substrates showing uncompetitive substrate inhibition
Vmax (nmol
kcat
kcat/Km
Substrate Km (µM)
min−1 mg−1
)
Ki (µM)
(s−1
)
(s−1 M−1
)
11
2S
2R
8S
8R
29 11
172 199 103 104
131 37
≫10 000
5.3 0.3
15 10
152 205 0.027 1703
27 32
None
0.11
0.094
0.018
—
549
142
—
20
≫10 000
27
3
5
None
0.005
178
Conversion of these acyl-CoA substrates by selected
enzymes involved in the Ibuprofen chiral inversion pathway
was then investigated. AMACR is known to catalyse epimerisa-
tion of 2-APA-CoA substrates, and is localised in mitochondria
and peroxisomes28–30 and was selected for study. ACOT1 is
localised in the cytosol31 and is thought to be the primary
enzyme responsible for hydrolysis of 2-APA-CoA esters to their
corresponding acids.22 ACOT2 is located in mitochondria31
and has been implicated in toxicities of 2-APA drugs, so was
also selected for study. Long-chain fatty acyl-CoA synthetase
(ibuprofenoyl-CoA synthetase24) was not selected for study as
its stereochemical requirements23–25 are inconsistent with the
direction of mandelic acid 1 chiral inversion.3,4,6
able kinetic parameters.
Selected acyl-CoA esters were tested as substrates with
ACOT1 and ACOT2. Both enzymes were active with their known
substrate, myristoyl-CoA 11.22,32 The 2-hydroxy-2-phenylacetyl-
CoA (2S and 2R) and 2-phenylpropanoyl-CoA (8S and 8R)
esters were then tested as substrates. All of these acyl-CoA
2-Hydroxy-2-phenylacetyl-CoA (2S and 2R) and 2-phenylpro-
esters were hydrolysed to their corresponding acids and panoyl-CoA (8S and 8R) esters were then tested as substrates
CoA-SH, as judged by the increasing absorbance at 412 nm for purified human recombinant AMACR by assaying for
due to reaction of CoA-SH with DTNB. No hydrolysis above exchange of the α-proton with deuterium.21,26 Under the reac-
background was observed in negative controls containing heat- tion conditions the best known AMACR substrate, -fenopro-
inactivated enzyme, thus showing that the reaction was fenoyl-CoA,21 showed >95% conversion (as measured by
enzyme-catalysed. These results are consistent with the obser- conversion of the 2-methyl group signal from a doublet into
vation that both 2S and 2R were hydrolysed in rat liver cell a single peak, in reality a 1 : 1 : 1 triplet arising from 1H–2H
1
extracts.6
coupling with J = < 1 Hz by H NMR spectroscopy). Assays for
Kinetic parameters for each of the substrates were then 8S and 8R, conducted in parallel under identical conditions,
determined (Tables 1 and 2). Substrates 11, 2S and 8S showed resulted in ∼60% conversion of S-2-phenylpropanoyl-CoA 8S,
substrate inhibition, whilst 2R and 8R showed standard and ∼40% conversion of R-2-phenylpropanoyl-CoA 8R.
Michaelis–Menten behaviour. The 2-hydroxy-2-phenylacetyl- Exchange was not observed in negative controls containing
CoA (2S and 2R) and R-2-phenylpropanoyl-CoA (8R) esters were heat-inactivated enzyme. Exchange of the α-proton is an
modest substrates for ACOT1 and ACOT2 compared to myris- obligatory step in the epimerisation reaction catalysed by
toyl-CoA 11, as judged by kcat/Km values. 2-Hydroxy-2-phenyl- AMACR,20,21,26 and hence it is highly likely that chiral inver-
acetyl-CoA 2S was converted somewhat more efficiently than sion of 8S and 8R also occurred.
2R by both ACOT1 and ACOT2, as judged by kcat/Km values.
Assays of 2S and 2R were also carried out with AMACR
Kinetic plots show that data for 8S fitted the substrate inhi- under identical conditions. As expected, control experiments
bition model (ESI†) for both enzymes, but yielded unreason- showed >95% exchange of the α-proton of -fenoprofenoyl-CoA
occurred with live enzyme but not heat-inactivated enzyme.
The α-proton singlets of 2S and 2R at ca. 5.3 p.p.m. in the
1H NMR spectra were unchanged when assays containing
Table 1 Kinetic parameters SE for ACOT1. Ki values are reported for
active enzyme and heat-inactivated controls were compared
(Fig. 1). Integration of the adenosine CH proton at 6.0 p.p.m.
and the mandelic acid α-proton at ca. 5.3 p.p.m. showed they
had a constant near 1 : 1 ratio, with no reduction of signal
upon incubation with active AMACR. A reduction in the inten-
sity of the α-proton signal of 2R and 2S would be expected if
significant exchange for deuterium had occurred. This result
demonstrates that R- and S-2-hydroxy-2-phenylacetyl-CoA 2R
those substrates showing uncompetitive substrate inhibition
Vmax (nmol
kcat
kcat/Km
Substrate
Km (µM)
min−1 mg−1
)
Ki (µM)
(s−1
)
(s−1 M−1
)
11
2S
2R
8S
8R
84 170
156 138
62 18
≫10 000
22 5.7
68 119
75 56
12 1.5
≫10 000
5.0 0.4
28 69
40 38
None
0.11
0.054
0.060
0.010
—
646
386
159
—
None
0.004
179
This journal is © The Royal Society of Chemistry 2014
Org. Biomol. Chem., 2014, 12, 6737–6744 | 6739

View Article Online
Paper
Organic & Biomolecular Chemistry
acyl-CoA racemase/epimerase. The only other enzyme of this
type in humans is 2-methylmalonyl-CoA epimerase, which cata-
lyses a chiral inversion step during the metabolism of propio-
nyl-CoA to succinyl-CoA. 2-Methylmalonyl-CoA epimerase
utilises a catalytic metal ion17,38,39 in a similar fashion to bac-
terial mandelate racemase.17–19 However, the relative positions
of the carboxylate ligand in methylmalonyl-CoA and the
hydroxy group in 2-hydroxy-2-phenylacetyl-CoA 2 are different.
Citrate, an analogue of the 2-methylmalonyl-CoA substrate,
ligates to the active site metal ion of methylmalonyl-CoA epi-
merase using two carboxylate ligands and not its carboxylate
and hydroxy ligands,38 showing the importance of the relative
positions of these ligating groups. Thus, it is unlikely that
Fig. 1 1H NMR spectra of 2R incubated with AMACR, showing α-proton
at ca. 5.3 p.p.m. Left panel, heat-inactivated enzyme; Right panel, live
enzyme, showing no reduction in signal.
2-hydroxy-2-phenylacetyl-CoA
2
will be able to bind to
manner which is
2-methylmalonyl-CoA epimerase in
competent for catalysis.
a
A second possibility is that 2S is metabolised by derivatisa-
tion of the hydroxy group. These derivatives are unlikely to be
accommodated within the active site of AMACR, based on
structural models of the related MCR enzyme.33–35 The
2-methyl binding pocket in MCR from M. tuberculosis can only
accommodate small groups,34 and it is likely that any modifi-
cation of 2S (e.g. by acetylation) would exclude substrates by
steric hindrance. Similarly, these derivatives of 2S are unlikely
to be able to bind to 2-methylmalonyl-CoA epimerase in a way
which is competent for catalysis since chelation to the active
site metal will be further compromised.
A further aspect to be considered is the requirement for
acyl-CoA formation in order for the proposed pathway
(Scheme 1) to occur. Long-chain fatty acyl-CoA synthetase is
known to be specific for R-APAs,23–25 whilst activation of
S-mandelic acid is required. Furthermore, there is no evidence
that long-chain fatty acyl-CoA synthetase is able to activate
2-hydroxy-fatty acids. Thus, the involvement of this specific
enzyme in the mandelic acid 1 chiral inversion pathway can be
excluded based on these stereochemical and other consider-
ations. However, at least 26 acyl-CoA synthetase enzymes are
present in the human genome,40 so potentially an alternative
enzyme may be able to catalyse the required conversion of
S-mandelic acid 1S to S-2-hydroxy-2-phenylacetyl-CoA 2S. Con-
version of racemic long-chain 2-hydroxyfatty acids to their
corresponding acyl-CoA esters by an undefined acyl-CoA
synthetase has been demonstrated in rat liver and mouse
brain extracts.41,42 Activation of mandelic acid 1 has not been
specifically studied, but naturally occurring long-chain
2-hydroxyfatty acids substrates are the D-isomer, possessing
R-stereochemistry. Activation of racemic 2-hydroxyfatty acids has
also been observed, suggesting that the enzyme may be non-
stereoselective.41 These observations are inconsistent with the
known direction of mandelic acid chiral inversion (S to R).3,4,6
and 2S do not undergo significant α-proton exchange. Since
removal of the α-proton is an obligatory first step in the
AMACR epimerisation reaction,20,21,26 it can be concluded that
2R and 2S are not epimerised.
The failure of 2R and 2S to be epimerised by AMACR is not
due to them having insufficiently hydrophobic phenyl side-
chains, as evidenced by the significant levels of α-proton
exchange with the 2-phenylpropanoyl-CoA substrates, 8S and
8R. The acyl-CoA esters 2R and 2S differ from 8R and 8S only
in having 2-hydroxy or 2-methyl substituents, respectively. The
hydroxy and methyl substituents are of similar size but are
hydrophilic (hydroxy group) and hydrophobic (methyl group),
respectively. No X-ray crystal structure for human AMACR has
been reported, but structures for the highly similar M. tubercu-
losis homologue have been reported.33–35 In this enzyme the
substrate methyl side-chain is accommodated in a hydro-
phobic pocket, comprising of methylene groups from the side-
chains of His-126, Asp-127, Asn-152, Asp-156, and Leu-217,
Tyr-224 and Ile-240 of the second subunit35 (residue numbers
refer to MCR). These residues are identical in human AMACR
1A, corresponding to His-122, Asp-123, Asn-148, Asp-152,
Leu-213, Tyr-220, and Ile-236 (ESI Fig. S1†). Thus, it appears
that 2R and 2S will be substantially excluded from the enzyme
as a result of the hydrophilic 2-hydroxy group been unable to
occupy the hydrophobic methyl group binding pocket.
An alternative possibility is that the presence of the
2-hydroxy group inhibits formation of the enolate intermedi-
ate20,21,26 in the reaction catalysed by AMACR. The pKa of man-
delic acid 1 has been measured as 22 (for formation of the
enol),36 compared to a pKa value of ∼21 for the α-proton of
acyl-CoA esters.37 The α-proton of 2-hydroxy-2-phenylacetyl-
CoAs 2R and 2S are likely to be more acidic than in mandelic
acid 1, suggesting that an enolate could be formed upon
binding. This again supports the idea that lack of metabolism of
2R and 2S is due to lack of binding within the AMACR active site.
There are a number of alternative metabolic pathways
where 2S and 2R could undergo chiral inversion. One possi-
bility is that chiral inversion of 2-hydroxy-2-phenylacetyl-CoA
substrates 2R and 2S is mediated by an alternative 2-methyl-
Conclusions
Mandelic acid 1 is a component of a number of drugs9,10 and
drugs in development10–14 and is a marker of environmental
exposure to styrene.1–5 Its chiral inversion pathway is of inte-
6740 | Org. Biomol. Chem., 2014, 12, 6737–6744
This journal is © The Royal Society of Chemistry 2014

View Article Online
Organic & Biomolecular Chemistry
Paper
rest in order to understand the metabolism of these xeno- fication, unless otherwise noted. Reagents were of analytical
biotics. Although mandelic acid 1 and Ibuprofen 3 have super- grade or equivalent (synthesis) or biochemical grade. Oasis
ficially similar structures, the results in this paper HLB cartridges were obtained from Waters Corporation. Myris-
demonstrate that they are not metabolised by the same toyl-CoA 11 and -fenoprofenoyl-CoA were synthesised as pre-
pathway. This conclusion is reached based on AMACR being viously described.21,22 Reactions were performed at ambient
unable to catalyse α-proton exchange of 2S or 2R and therefore temperature, unless otherwise stated. Solvents were evaporated
it cannot catalyse the required epimerisation reaction. As a under reduced pressure. NMR spectra were recorded on Bruker
consequence, different enzymes must be involved in the chiral Avance III 400.04 MHz or 500.13 MHz spectrometers. Chemical
inversion step of the two pathways. It is also unlikely that shifts are reported to the nearest 0.01 p.p.m. and coupling con-
chiral inversion of 2S or 2R can be performed by 2-methyl- stants (J values) are reported to 0.1 Hz. Multiplicities of peaks
malonyl-CoA epimerase or that epimerisation of a derivative are described as follows: s, singlet; d, doublet; t, triplet; m,
with a modified 2-hydroxy group could be performed by either multiplet; br, broad. Mass spectra were obtained using VG7070E
enzyme, and this implies that chiral inversion of mandelic and Bruker microTOF spectrometers in the ES+ mode at the
acid 1 probably does not proceed via an acyl-CoA intermediate University of Bath Mass Spectrometry Service. Solutions in
at all (as shown in Scheme 1).
organic solvents were dried over anhydrous magnesium sulfate
Moreover, this study demonstrates that ACOT1 and ACOT2 and evaporated under reduced pressure. Aqueous solutions
can hydrolyse both epimers of 2-hydroxy-2-phenylacetyl-CoA 2, were prepared in 18.2 Mega-Ω cm−1 Nanopure water and pH
and hence cannot determine the direction of chiral inversion adjusted with HCl or NaOH solutions as appropriate. Con-
for the proposed pathway as a whole. Metabolism via an acyl- struction of expression plasmids for human AMACR26 and
CoA intermediate6 therefore requires stereospecific conju- ACOT1 and ACOT222 have been previously described. E. coli
gation of S-mandelic acid with CoA-SH to form S-2-hydroxy-2- BL21 (DE3) pLysS and Rosetta2 (DE3) expression strains were
phenylacetyl-CoA 2S. The acyl-CoA synthetases which activate obtained from Novagen.
2-hydroxyfatty acids appear to be either specific for substrates
Synthesis of S-mandelic acid O-TBDMS ether 5S.14
with R-configuration or are non-stereospecific.41 These stereo- TBDMS-Cl (331 mg, 2.2 mmol, 2.2 eq.) and imidazole (163 mg,
chemical observations are inconsistent with the observed di- 2.4 mmol, 2.4 eq.) were added to a solution of S-mandelic acid
rection of mandelic acid 1 chiral inversion for the pathway as a 1S (152 mg, 1.0 mmol) in anhydrous THF (3 mL) cooled to
whole (2S to 2R),3,4,6 and further argues for a pathway not 4 °C and the reaction mixture was stirred for 1 h. The reaction
involving an acyl-CoA intermediate.
was allowed to warm to room temperature and stirred for a
Chiral inversion of mandelic acid 1 by stereoselective oxi- further 3 h before filtration and concentration in vacuo. NaOH
dation and reduction is an obvious alternative pathway which (1.0 M, 3 mL) was added and the mixture stirred for
does not involve acyl-CoA intermediates. This possibility has 105 minutes before dilution with 3 mL water. The reaction was
been investigated, and stereoselective oxidation of S-mandelic extracted with diethyl ether (2 × 5 mL), and the reaction
acid 1S by NAD+-dependent dehydrogenases has been mixture acidified to pH ∼3.5 with 1.0 M citric acid buffer, pH
observed.43 S-Mandelic acid 1S has also been shown to be 3.5 and extracted with diethyl ether (3 × 5 mL). The combined
stereoselectively oxidised by the kidney isoform, but not the organic layers were dried over MgSO4, filtered and concen-
1
liver isoform, of rat L-2-hydroxyacid oxidase.44 Chiral inversion trated to give 5S as a colourless solid (230 mg, 86%): H NMR
of mandelic acid 1 has been observed in rat liver homogenates,6 (400.04 MHz, CDCl3)14 δ 9.30 (br s, 1H), 7.49–7.28 (m, 5H),
implying that L-2-hydroxyacid oxidase cannot be involved. The 5.22 (s, 1H), 0.92 (s, 9H), 0.12 (s, 3H), 0.00 (s, 3H).
oxidation product in this alternative pathway is 2-oxo-2-phenyl-
Synthesis of R-mandelic acid O-TBDMS ether 5R. The title
acetate (phenylglyoxylic acid). Previous studies4 have shown compound was synthesised from 1R (152 mg, 1.0 mmol) by
that the majority of 2-oxo-2-phenylacetate is not reduced the same method to give 5R (232 mg, 87%). NMR data was
in vivo to mandelic acid 1 (in rats), with only around 1% of the identical to 5S.
total substrate been converted. Moreover, the reduction reac-
Synthesis of S-2-hydroxy-2-phenylacetyl-CoA 2S. 5S (58 mg,
tion appears not to be completely stereoselective, with a 0.22 mmol) was dissolved in anhydrous dichloromethane
reported ratio of products of ca. 9 : 1 (1R : 1S). Other studies (2 mL) and stirred with carbonyldiimidazole (71 mg,
have reported that metabolism of S-mandelic acid 1S is stimu- 0.44 mmol, 2.0 eq.) for 45 minutes. The reaction was extracted
lated by NADPH,45 implying that reduction could take place with water (5 × 5 mL), dried over MgSO4, filtered and concen-
under some circumstances. The exact pathway for mandelic trated to dryness. The residue was dissolved in THF (2 mL)
acid chiral inversion therefore remains unclear.
and aq. NaHCO3 (0.1 M, 2 mL) and stirred with reduced co-
enzyme A, tri-lithium salt (20 mg, 0.025 mmol) for >16 h. The
reaction mixture was diluted with water (1 mL), acidified to pH
∼3.5 with 1 M HCl and stirred with anhydrous KF (27 mg,
0.46 mmol) for 18 h. The mixture was extracted with diethyl
ether (5 × 5 mL) and freeze-dried. The product 2S was purified
Experimental
General experimental
All chemicals were obtained from the Sigma-Aldrich Chemical by solid-phase extraction using an Oasis HLB cartridge. After
Co. or Fisher Scientific Ltd and were used without further puri- loading the cartridge was washed with water (3 mL) and eluted
This journal is © The Royal Society of Chemistry 2014
Org. Biomol. Chem., 2014, 12, 6737–6744 | 6741

View Article Online
Paper
Organic & Biomolecular Chemistry
with water–acetonitrile (4 : 1, 7 mL). The elution fraction was enzymes22,32); S-2-hydroxy-2-phenylacetyl-CoA 2S; R-2-hydroxy-
concentrated in vacuo and freeze-dried to give 2S as a colour- 2-phenylacetyl-CoA 2R; S-2-phenylpropanoyl-CoA 8S; and
1
less solid (4.3 mg): H NMR (500.13 MHz, D2O) δ 8.45 (s, 1H), R-2-phenylpropanoyl-CoA 8R. Stock concentrations of acyl-CoA
1
8.15 (s, 1H), 7.41–7.24 (m, 5H), 6.07 (d, 1H, J = 7.0 Hz), 5.24 esters were determined using H NMR.21 Solutions of acyl-CoA
(s, 1H), 4.20–4.08 (m, 2H), 3.94–3.87 (m, 2H), 3.82 (dd, 1H, J = esters were diluted in HEPES-NaOH, pH 7.27 except for myris-
9.5, 4.2 Hz), 3.44 (dd, 1H, J = 9.5, 4.3 Hz), 3.30–3.10 (m, 4H), toyl-CoA 11 where buffer was supplemented with BSA as pre-
3.06–2.85 (m, 2H), 2.20–2.07 (m, 2H), 0.75 (s, 3H), 0.63 (s, 3H); viously described.22 This was required to reduce substrate
HRMS (ES) Calcd for C29H40N7O18P3SNa: 922.1267. Found: inhibition due to the formation of micelles. Kinetic analyses of
922.1240.
11 with ACOT1 and ACOT2 used substrate concentrations of
Synthesis of R-2-hydroxy-2-phenylacetyl-CoA 2R. The title 2–60 μM. All other kinetic analyses used concentrations of
compound was synthesised from 5R (40 mg, 0.15 mmol) 5–200 μM. Assays were initiated by addition of enzyme (50 µL)
by the same method to give 2R (3.0 mg). NMR and other data to substrate: DTNB (2 × stock solution) and the reaction was
were identical to 2S.
monitored for up to 15 minutes. Rates at each substrate
Synthesis of S-2-phenylpropanoyl-CoA 8S. 9S (30 mg, concentration were measured using three dependent repeats.
0.20 mmol) was dissolved in dichloromethane (2 mL) and Reaction rates were obtained by plotting changes in absorb-
stirred with carbonyldiimidazole (65 mg, 0.40 mmol) for ance for the linear progress curve with Excel. Activities in
45 minutes. The reaction was extracted with water (5 × 5 mL), nmol min−1 mg−1 were calculated assuming ε412 = 14.15 mM−1
dried over MgSO4, filtered and concentrated to dryness. The cm−1 at 25 °C.46,47
residue was dissolved in THF (2 mL) and aq. NaHCO3 (0.1 M,
Data was analysed using SigmaPlot 11 and enzyme kinetics
2 mL) and stirred with coenzyme A, tri-lithium salt (16 mg, module 1.3, fitting to the Michaelis–Menten equation with
0.02 mmol) for >16 h. The reaction mixture was diluted with and without uncompetitive substrate inhibition. The correct
water (1 mL) and acidified to pH ∼3.5 with 1 M HCl. The model was chosen based on convergence of fitting and visual
mixture was extracted with diethyl ether (5 × 5 mL) and freeze- inspection of plots. Kinetic plots for all substrates are available
dried. The product was purified by solid-phase extraction in ESI.† Km, Ki and Vmax values are reported in Tables 1 and 2
using an Oasis HLB cartridge. After loading, the cartridge was as mean values SE. Ki values are reported for those substrates
washed with water (3 mL) and eluted with water–acetonitrile showing substrate inhibition.
(1 : 1, 7 mL). The elution fraction was concentrated in vacuo
and freeze-dried to give 8S as a colourless solid (2.0 mg):
AMACR assays
1H NMR (500.13 MHz, D2O) δ 8.58 (s, 1H), 8.31 (s, 1H), Human AMACR was expressed in E. coli Rosetta2 (DE3) grown
7.30–7.16 (m, 5H), 6.11 (d, 1H, J = 5.5 Hz), 4.22–4.10 (m, 1H), at 37 °C and 220 r.p.m. until an O.D.600 = ∼1.5 was reached.
4.00–3.89 (m, 2H), 3.77 (dd, 1H, J = 9.5 4.5 Hz), 3.49 (dd, 1H, Cultures were cooled to 22 °C, induced with 0.25 mM IPTG
J = 9.5 3.9 Hz), 3.30–3.13 (m, 4H), 2.99–2.79 (m, 2H), 2.20–2.07 and incubated overnight under the same conditions.21 Cells
(m, 2H), 1.38 (d, 3H, J = 7.2 Hz), 0.84 (s, 3H), 0.69 (s, 3H); (∼2 g) were lysed using the ‘one shot’ (Constant Systems) in
HRMS (ES) Calcd for C30H43N7O17P3S: 898.1654. Found: ∼30 mL NaH2PO4–NaOH, 300 mM NaCl and 10 mM imidazole
898.1640.
pH 7.2, supplemented with 1 mM PMSF, 1 mM benzamidine-
Synthesis of R-2-phenylpropanoyl-CoA 8R. In similar HCl, and 250 U Benzonase (Novagen) and stirred with
fashion to the preparation of 8S, 8R was obtained from the N-lauroyl-sarcosine [1.5% (w/v)] at 4 °C for 1 hour. The sample
reaction of 9R (30 mg, 0.20 mmol) and coenzyme A, tri-lithium was centrifuged (Beckmann JA-14 rotor, 10 000 r.p.m., 15 300g,
salt (16 mg, 0.02 mmol) in THF (2 mL) and aq. NaHCO3 10 minutes), filtered through a 0.45 µ filter, and purified by
(2 mL, 0.1 M) to give 8R as a colourless solid (2.7 mg). NMR metal-chelate chromatography as previously described.26 Puri-
and other data were identical to 8S.
fied enzyme was dialysed into 20 mM NaH2PO4–NaOH, pH 7.4
and stored at −80 °C. Protein concentration was determined
by UV-visible spectroscopy.21 Protein purity of pooled fractions
ACOT assays
Human ACOT1 and ACOT2 were expressed as recombinant was ca. 95–98% by SDS-PAGE analyses.
His-tag proteins in E. coli BL21 (DE3) pLysS.22 Cells (∼2 g) were
Assays were conducted in 50 mM NaH2PO4–NaOH buffer,
lysed using Bugbuster (Novagen) in the presence of Benzonase pH 7.4 containing ca. 85% (v/v) 2H2O, 100 µM acyl-CoA sub-
and the enzyme was purified by metal-chelate chromatography. strate and 3.5 µM enzyme as previously described,21,26,27 with
Purified enzyme was exchanged into 20 mM HEPES-NaOH, negative controls containing heat-inactivated enzyme. -Feno-
pH 7.27 and protein concentrations determined by UV-visible profenoyl-CoA, an ibuprofenoyl-CoA analogue and the best
spectrometry. Protein purity was ca. 95–98% by SDS-PAGE AMACR substrate reported to date,21 was used as a positive
analyses.
control in all assays. Stock concentrations of acyl-CoA esters
Assays contained 0.109 mg (2.27 nmol) of ACOT1 or were determined using 1H NMR.21 Assays were initiated by
0.099 mg (1.80 nmol) of ACOT2. Reactions were carried out at addition of enzyme to the substrate in buffer, which were incub-
pH 7.27,22 as this minimises spontaneous hydrolysis of ated at 30 °C for 1 hour. After this time samples were heated
DTNB.46 The following substrates (Tables 1 and 2) were used at 50 °C for 10 minutes to inactivate the enzyme and analysed
in the assays: myristoyl-CoA 11 (known substrate for ACOT by 1H NMR (500.13 MHz). Conversion of 2-phenylpropanoyl-
6742 | Org. Biomol. Chem., 2014, 12, 6737–6744
This journal is © The Royal Society of Chemistry 2014

View Article Online
Organic & Biomolecular Chemistry
Paper
CoA substrates 8S and 8R was quantified by conversion of the
2-Me doublet at ca. 1.0 p.p.m. into a single peak, in reality a
2 S. F. Helal and W. S. Elshafy, Toxicol. Ind. Health, 2013, 29,
812–819.
3 L. Drummond, J. Caldwell and H. K. Wilson, Xenobiotica,
1989, 19, 199–207.
4 L. Drummond, J. Caldwell and H. K. Wilson, Xenobiotica,
1990, 20, 159–168.
5 I. Linhart, I. Gut, J. Smejkal and J. Novak, Chem. Res.
Toxicol., 2000, 13, 36–44.
1
1 : 1 : 1 triplet arising from H–2H coupling with J = < 1 Hz, as
previously described for other substrates.21,26 Conversion of
2-hydroxy-2-phenylacetyl-CoA 2S and 2R was monitored by
reduction of the α-proton singlet at ca. 5.3 p.p.m. due to deu-
terium incorporation.
6 L.-B. Gao, J.-Z. Wang, T.-W. Yao and S. Zeng, Chirality,
2012, 24, 86–95.
7 M. Polakova, Z. Krajcovicova, V. Melus, M. Stefkovicova and
M. Sulcova, Cent. Eur. J. Public Health, 2012, 20, 226–
232.
8 G. Terre’Blanche, N. Heyer, J. J. Bergh, L. J. Mienie,
C. J. van der Schyf and B. H. Harvey, Neurotoxic. Res., 2011,
20, 97–101.
Abbreviations
ACOT
AMACR
2-APAs
BSA
Acyl-CoA thioesterase
α-Methylacyl-CoA racemase (a.k.a. P504S)
2-Arylpropanoic acids (‘profens’)
Bovine serum albumin
tert-Butyl
tBu
CoA-SH
CYP2E1
DTNB
E. coli
ES
Coenzyme A (reduced form)
Cytochrome P450 2E1
5,5′-Dithiobis(2-nitrobenzoic acid)
Escherichia coli
9 V. K. Garg, S. Sinha and R. Sarkar, Dermatol. Surg., 2009,
35, 59–65.
10 B. C. Herold, I. Scordi-Bello, N. Cheshenko, D. Marcellino,
M. Dzuzelewski, F. Francois, R. Morin, V. M. Casullo,
R. A. Anderson, C. Chany, D. P. Waller, L. J. D. Zaneveld
and M. E. Klotman, J. Virol., 2002, 76, 11236–11244.
11 M. J. Keller, M. E. Klotman and B. C. Herold, J. Antimicrob.
Chemother., 2003, 51, 1099–1102.
12 P. M. M. Mesquita, S. S. Wilson, P. Manlow, L. Fischetti,
M. J. Keller, B. C. Herold and R. J. Shattock, J. Virol., 2008,
82, 6576–6584.
Electrospray
HEPES
4-(2-Hydroxyethyl)piperazine-1-ethanesulfonic
acid
kcat
1st order rate constant for conversion of substrate
to product
kcat/Km
Ki
Selectivity constant
Inhibitor constant
Km
Mandelic
acid
O.D.600
NAD+
Michaelis constant
2-Hydroxy-2-phenylacetic acid
13 L. J. D. Zaneveld, R. A. Anderson, X. H. Diao, D. P. Waller,
C. Chany, K. Feathergill, G. Doncel, M. D. Cooper and
B. Herold, Fertil. Steril., 2002, 78, 1107–1115.
14 O. Kuisle, E. Quinoa and R. Riguera, J. Org. Chem., 1999,
64, 8063–8075.
15 S. S. Hall, L. M. Doweyko, A. M. Doweyko and
J. S. R. Zilenovski, J. Med. Chem., 1977, 20, 1239–
1242.
Optical density at 600 nm
Nicotinamide adenine dinucleotide (oxidised
form)
Nuclear magnetic resonance
Phenylmethylsulfonyl fluoride
Parts per million
NMR
PMSF
p.p.m.
r.p.m.
16 T. Fukuda, H. Ito, T. Mukainaka, H. Tokuda, H. Nishino
and T. Yoshida, Biol. Pharm. Bull., 2003, 26, 271–273.
Revolutions per minute
SDS-PAGE Sodium dodecyl sulfate polyacrylamide gel 17 R. N. Armstrong, Biochemistry, 2000, 39, 13625–13632.
electrophoresis
Standard error
TBDMS-Cl tert-Butyldimethylsilyl chloride
18 B. Mitra, A. T. Kallarakal, J. W. Kozarich, J. A. Gerlt,
J. G. Clifton, G. A. Petsko and G. L. Kenyon, Biochemistry,
1995, 34, 2777–2787.
SE
THF
Vmax
Tetrahydrofuran
Maximum rate of enzyme reaction
19 M. St. Maurice and S. L. Bearne, Biochemistry, 2000, 39,
13324–13335.
20 M. D. Lloyd, M. Yevglevskis, G. L. Lee, P. J. Wood,
M. D. Threadgill and T. J. Woodman, Prog. Lipid Res., 2013,
52, 220–230.
Acknowledgements
This work was funded by Prostate Cancer UK. Part of this work 21 T. J. Woodman, P. J. Wood, A. S. Thompson,
was performed as a Masters in Pharmacy final year project at
the University of Bath, U.K. MJ, MDT, TJW and MDL are
members of the Cancer Research @ Bath network.
T. J. Hutchings, G. R. Steel, P. Jiao, M. D. Threadgill and
M. D. Lloyd, Chem. Commun., 2011, 47, 7332–7334.
22 X. Qu, A. Allan, G. Chui, T. J. Hutchings, P. Jiao,
L. Johnson, W. Y. Leung, P. K. Li, G. R. Steel,
A. S. Thompson, M. D. Threadgill and M. D. Lloyd,
Biochem. Pharmacol., 2013, 86, 1621–1625.
23 R. Brugger, B. G. Alia, C. Reichel, R. Waibel, S. Menzel,
K. Brune and G. Geisslinger, Biochem. Pharmacol., 1996, 52,
1007–1013.
Notes and references
1 J.-K. Chung, W. Yuan, G. Liu and J. Zheng, Toxicology,
2006, 226, 99–106.
This journal is © The Royal Society of Chemistry 2014
Org. Biomol. Chem., 2014, 12, 6737–6744 | 6743

View Article Online
Paper
Organic & Biomolecular Chemistry
24 R. Brugger, C. Reichel, B. G. Alia, K. Brune, T. Yamamoto, 35 S. Sharma, P. Bhaumik, W. Schmitz, R. Venkatesan,
I. Tegeder and G. Geissinger, Biochem. Pharmacol., 2001,
61, 651–656.
25 C. Sevoz, E. Benoît and T. Buronfosse, Drug Metab. Dispos.,
2000, 28, 398–402.
J. K. Hiltunen, E. Conzelmann, A. H. Juffer and
R. K. Wierenga, J. Phys. Chem. B, 2012, 116, 3619–
3629.
36 Y. Chiang, A. J. Kresse, P. Pruszynski, N. P. Schepp and
J. Wirz, Angew. Chem., Int. Ed., 1990, 29, 792–794.
26 D. J. Darley, D. S. Butler, S. J. Prideaux, T. W. Thornton,
A. D. Wilson, T. J. Woodman, M. D. Threadgill and 37 J. P. Richard and T. L. Amyes, Curr. Opin. Chem. Biol., 2001,
M. D. Lloyd, Org. Biomol. Chem., 2009, 7, 543–552. 5, 626–633.
27 F. A. Sattar, D. J. Darley, F. Politano, T. J. Woodman, 38 L. Shi, P. Gao, X.-X. Yan and D.-C. Liang, Proteins: Struct.,
M. D. Threadgill and M. D. Lloyd, Chem. Commun., 2010,
46, 3348–3350.
28 L. Amery, M. Fransen, K. De Nys, G. P. Mannaerts and
P. P. Van Veldhoven, J. Lipid Res., 2000, 41, 1752–1759.
29 S. Ferdinandusse, S. Denis, L. Ijlst, G. Dacremont,
Funct., Bioinf., 2009, 77, 994–999.
39 A. A. McCarthy, H. M. Baker, S. C. Shewry, M. L. Patchett
and E. N. Baker, Structure, 2001, 9, 637–646.
40 P. A. Watkins, D. Maiguel, Z. Jia and J. Pevsner, J. Lipid
Res., 2007, 48, 2736–2750.
H. R. Waterham and R. J. A. Wanders, J. Lipid Res., 2000, 41 V. Foulon, M. Sniekers, E. Huysmans, S. Asselberghs,
41, 1890–1896.
V. Mahieu, G. P. Mannaerts, P. P. Van Veldhoven and
30 T. J. Kotti, K. Savolainen, H. M. Helander, A. Yagi,
M. Casteels, J. Biol. Chem., 2005, 280, 9802–9812.
D. K. Novikov, N. Kalkkinen, E. Conzelmann, J. K. Hiltunen 42 L. A. Paige, G. Q. Zheng, S. A. Defrees, J. M. Cassady and
and W. Schmitz, J. Biol. Chem., 2000, 275, 20887–20895. R. L. Geahlen, Biochemistry, 1990, 29, 10566–10573.
31 M. C. Hunt, J. Yamada, L. J. Maltais, M. W. Wright, 43 L.-B. Gao, J.-Z. Wang and S. Zeng, Acta Pharm. Sin, 2009,
E. J. Podesta and S. E. H. Alexson, J. Lipid Res., 2005, 46,
2029–2032.
32 M. C. Hunt, A. Rautanen, M. A. K. Westin, L. T. Svensson
and S. E. H. Alexson, FASEB J., 2006, 20, 1855–1864.
33 K. Savolainen, P. Bhaumik, W. Schmitz, T. J. Kotti,
44, 778–784.
44 S. Angermüller, C. Leupold, K. Zaar and H. D. Fahimi,
Histochemistry, 1986, 85, 411–418.
45 L.-B. Gao, J.-Z. Wang, T.-W. Yao and S. Zeng, Zhongguo
Yaolixue Yu Dulixue Zazhi, 2009, 23, 351–356.
E. Conzelmann, R. K. Wierenga and J. K. Hiltunen, J. Biol. 46 P. W. Riddles, R. L. Blakeley and B. Zerner, Anal. Biochem.,
Chem., 2005, 280, 12611–12620. 1979, 94, 75–81.
34 P. Bhaumik, W. Schmitz, A. Hassinen, J. K. Hiltunen, 47 P. Eyer, F. Worek, D. Kiderlen, G. Sinko, A. Stuglin,
E. Conzelmann and R. K. Wierenga, J. Mol. Biol., 2007, 367,
1145–1161.
V. Simeon-Rudolf and E. Reiner, Anal. Biochem., 2003, 312,
224–227.
6744 | Org. Biomol. Chem., 2014, 12, 6737–6744
This journal is © The Royal Society of Chemistry 2014