CHEMMEDCHEM
FULL PAPERS
Results and Discussion
drogen-independent.[14] This may allow treatment of patients
with advanced hormone-resistant prostate cancer to be treat-
ed with AMACR inhibitors and androgen ablation.
Chemistry
Prospects for drug treatment with AMACR inhibitors are
good, because although AMACR is present in many organs,
the major clinical manifestation of congenital racemase defi-
ciency is adult-onset sensory motor neuropathy, which is
caused by the slow and prolonged accumulation of branched-
chain fatty acids.[15] This effect could be prevented during
treatment with AMACR inhibitors with a diet low in phytanic
acid. AMACR is also involved in bile acid biosynthesis; however,
an alternative pathway that does not rely on AMACR has re-
cently been demonstrated in AMACR-deficient mice, which are
healthy and fertile.[16] Bile acid intermediates di- and trihydrox-
ycoprostanoyl CoA (DHT- and THC-CoA 2)[17] and various
NSAID profenoyl CoA esters such as ibuprofenoyl CoA 3 are
also known unnatural substrates for isomerization by
AMACR.[18]
We previously reported a series of AMACR inhibitors, some of
which are substrate analogues containing 2-trifluoromethyl
groups or b-fluorine-a-methyl analogues.[22] We also reported
that both R and S isomers of ibuprofenoyl CoA are inhibitors.
We hypothesized that the presence of the b-fluorine atoms or
indeed the presence of an aryl group lowered the pKa of the
a-proton, making these analogues better substrates. This
would imply a slower release from the active site because pro-
tonation of the enzyme-bound enolate would be slower for
these inhibitors. The competitive assays were carried out with
THC-CoA as the substrate, and the best inhibitor, 2-trifluorome-
thyltetradecanoyl CoA 4, had a Ki value of 0.9 mm. For practical
reasons we have now switched to (S)-ibuprofenoyl CoA as the
substrate, as it is much less expensive and quicker to assay by
HPLC.
In our previous work,[21] when using ibuprofenoyl CoA 3 as
an inhibitor against the substrate THC-CoA 2, R and S isomers
were found to have a ~4.5-fold difference in Ki. This result is
likely to have reflected a difference in inhibition early on in rac-
emization, or that the crude liver enzyme was not effective for
racemization of this substrate, as we now know that ibuprofe-
noyl CoA is a substrate for racemization when using pure
hAMACR.
The bacterial homologue (MCR) from Mycobacterium tuber-
culosis has 43% sequence identity with human AMACR, and
the MCR crystal structure shows the enzyme as an interlocked
dimer with the active site at the interface between the large
and small subunits of each monomer.[8] The active site geome-
try with bound ligands suggests a 1,1-proton transfer mecha-
nism in which the acid/base-pair residues are His126 and
Asp156.[9] The substrate binds such that the a-position sits be-
tween the active site bases and undergoes reversible deproto-
nation to from an enzyme-stabilized enolate.[19] A large methio-
nine-rich hydrophobic pocket accommodates a range of acyl
groups. Minimum substrate requirements for turnover appear
to be a linear chain length of at least eight carbon atoms and
an a-methyl group,[20] although low-level activity has recently
been demonstrated without an a-methyl group.[21]
We synthesized a new series of inhibitors (5–9, Figure 2) de-
signed to test requirements for inhibition, particularly in rela-
tion to the requirement for the a-proton, the a-methyl group,
We previously reported a series of fluorinated substrate ana-
logues as inhibitors of AMACR.[22] We demonstrated that the
free acid corresponding to the best inhibitor, 2-trifluoromethyl-
tetradecanoyl CoA 4, inhibits cancer cell proliferation in vitro
and correlates with AMACR expression levels. More recently,
Wilson et al. developed and applied a high-throughput screen
based on tritium release from [2,3-3H]pristanoyl CoA to discov-
er the first non-substrate-based inhibitors.[23] However, the best
inhibitors to date still have modest Ki values of ~1 mm, and
there is a need to discover more potent inhibitors to further
validate AMACR as a druggable target and to better under-
stand interactions with the active site to optimize inhibitors.
For initial kinetic studies, we previously used rat liver micro-
somes as a source of AMACR. Others have used AMACR puri-
fied from human[20] and rat liver, an E. coli recombinant
AMACR–MBP fusion[23] or recombinant MCR.[24] The use of vari-
ous enzyme sources and various assays and substrates by dif-
ferent groups has made comparison of kinetic data difficult. In
this study we used human AMACR (hAMACR) isolated from
HEK293 kidney cell cultures (Origene), as we have found this
enzyme to possess much higher activity than recombinant
hAMACR from E. coli (see Supporting Information S16 for
a comparison of activities for hAMACR, rAMACR, and MCR).
Figure 2. New inhibitors designed to test structural requirements for the in-
hibition of hAMACR.
and the possibility to include groups for stabilization of the
proposed enzyme-bound enolate. Although we were con-
scious that separate diastereomers of inhibitor (Æ)-5 may pos-
sess different potencies, individual isomers were not tested.
Given the similarities in Ki values obtained for ibuprofenoyl
CoA isomers,[22] inhibitor 6, which is also likely to undergo race-
mization, was tested as a mixture of isomers.
To determine the requirement for an a-proton we synthe-
sized a-fluoroibuprofenoyl CoA 5, which is non-enolizable
(Scheme 1). This was made by treatment of ibuprofen methyl
ester 11 with lithium diisopropylamide followed by fluorination
with N-fluorobenzenesulfonimide (NFSI). The a-fluoro methyl
ester 12 was then hydrolyzed prior to CoA ester formation by
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemMedChem 2013, 8, 1643 – 1647 1644