Probing the steric requirements of the γ-aminobutyric acid aminotransferase active site with fluorinated analogues of vigabatrin

Article abstract of DOI:10.1016/j.bmc.2012.12.009

We have synthesized three analogues of 4-amino-5-fluorohexanoic acids as potential inactivators of γ-aminobutyric acid aminotransferase (GABA-AT), which were designed to combine the potency of their shorter chain analogue, 4-amino-5-fluoropentanoic acid (

Full text of DOI:10.1016/j.bmc.2012.12.009

Bioorganic & Medicinal Chemistry xxx (2013) xxx–xxx
Contents lists available at SciVerse ScienceDirect
Bioorganic & Medicinal Chemistry
journal homepage: www.elsevier.com/locate/bmc
Probing the steric requirements of the
c-aminobutyric acid
aminotransferase active site with fluorinated analogues of vigabatrin
Jose I. Juncosa Jr. a, Andrew P. Groves a, Guoyao Xia a, Richard B. Silverman a,b,
⇑
a Department of Chemistry, Chemistry of Life Processes Institute, and Center for Molecular Innovation and Drug Discovery, Northwestern University, 2145 Sheridan Road, Evanston,
IL 60208-3113, United States
b Department of Molecular Biosciences, Chemistry of Life Processes Institute, and Center for Molecular Innovation and Drug Discovery, Northwestern University,
2145 Sheridan Road, Evanston, IL 60208-3113, United States
a r t i c l e i n f o
a b s t r a c t
Article history:
We have synthesized three analogues of 4-amino-5-fluorohexanoic acids as potential inactivators of
Received 31 October 2012
Revised 30 November 2012
Accepted 7 December 2012
Available online xxxx
c
-aminobutyric acid aminotransferase (GABA-AT), which were designed to combine the potency of their
shorter chain analogue, 4-amino-5-fluoropentanoic acid (AFPA), with the greater enzyme selectivity of
the antiepileptic vigabatrin (SabrilÒ). Unexpectedly, these compounds failed to inactivate or inhibit the
enzyme, even at high concentrations. On the basis of molecular modeling studies, we propose that the
GABA-AT active site has an accessory binding pocket that accommodates the vinyl group of vigabatrin
and the fluoromethyl group of AFPA, but is too narrow to support the extra width of the distal methyl
group in the synthesized analogues.
Keywords:
c
-Aminobutyric acid aminotransferase
Vigabatrin
4-Amino-5-fluoropentanoic acid
Enzyme inhibition
Ó 2012 Elsevier Ltd. All rights reserved.
Molecular dynamics
1. Introduction
that works by the second mechanism. Vigabatrin irreversibly
inhibits the chief catabolizer of GABA, c-aminobutyric acid amino-
c
-Aminobutyric acid (GABA, 1) is the most important inhibitory
transferase (GABA-AT), which catalyzes the transformation of
GABA into succinic semialdehyde. This metabolite is later oxidized
to succinate and reincorporated into the citric acid cycle.1
Inhibition of GABA-AT by vigabatrin has been shown to involve
two distinct mechanisms (Scheme 1).2 After the formation of a
Schiff base between the drug and the cofactor pyridoxal phosphate
(PLP), Lys329 (the previous anchor point for the aldehyde) deprot-
neurotransmitter in the central nervous system (CNS) of mammals.
Its function as a neurotransmitter involves binding at pre- and
postsynaptic receptors at inhibitory synapses in the brain. This
event triggers the flow of chloride and potassium ions in and out
of the cell, respectively, through specific channels. Low levels of
GABA in the brain have been associated with anxiety, pain, mania,
depression, and seizures, mainly due to overstimulation of other
neurotransmitter receptors by their endogenous agonists.1
An approach that has been successfully used to treat disorders
resulting from low levels of GABA in the brain, involves increasing
its concentration by either stimulating its biosynthesis or inhibit-
ing its metabolism. Vigabatrin (SabrilÒ, 2), which is used in cases
where classic antiepileptics are ineffective, is an example of a drug
onates the c-proton, and tautomerizes the complex in two ways
(pathways a and b). Pathway ‘‘a’’ creates a Michael acceptor moiety
in the substrate (3); Lys329 undergoes 1,4-addition to give 4. Path-
way ‘‘b’’ involves tautomerization through the alkene, leading to
enamine adduct 5, which releases enamine 6 that reattaches to
the PLP to give 7. The observed products of the reaction were
determined to be in the ratio of ꢀ75% for the Michael addition
pathway and ꢀ25% for the enamine pathway.
Aside from its use as an antiepileptic, vigabatrin has more re-
cently emerged as a possible novel treatment for addiction.3–5 How-
ever, its side effect profile is detrimental to its use, especially relating
to the permanent visual field defects (VFDs) induced in 25–40% of
patientsafter chronic dosage. One explanationfor the undesired side
effects of vigabatrin stems from the release of a reactive metabolite.
On the basis of the proposed inactivation pathways of GABA-AT by
vigabatrin, one such metabolite could be the hydrolysis product of
unsaturated imine 3 in the Michael addition pathway, which was
shown to be generated.2 If this is responsible for the side effects,
+H3N
GABA ( )
COO-
+H3N
Vigabatrin ( )
COO-
2
1
Figure 1. Structures of GABA (1) and vigabatrin (2).
⇑
Corresponding author. Tel.: +1 847 491 5653; fax: +1 847 491 7713.
E-mail address: r-silverman@northwestern.edu (R.B. Silverman).
0968-0896/$ - see front matter Ó 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.bmc.2012.12.009
Please cite this article in press as: Juncosa Jr., J. I.; et al. Bioorg. Med. Chem. (2013), http://dx.doi.org/10.1016/j.bmc.2012.12.009

2
J. I. Juncosa Jr. et al. / Bioorg. Med. Chem. xxx (2013) xxx–xxx
Lys329
Lys329
NH2
Lys329
+
NH2
NH
~H+
H2O
-H2O
a
Lys329
H
+
NH2
PMP
NH3
O
COO-
+
PMP
N
COO-
PMP
N
COO-
+
b
H+
H2
4
(~75%)
b
a
3
B
b
a
H
PLP
N
COO-
Lys329
H+
a
Lys329
Lys329
-OOC
NH
NH2
NH+
b
+
H2O
PMP
~PLP
2
PLP
NH2
-OOC
O
+
-H2O
NH4
+
PLP
N
COO-
7
(~25%)
H+
6
5
Scheme 1. Mechanisms of inactivation of GABA-AT by vigabatrin (2).
an approach to mitigate the VFDs caused by vigabatrin could be to
bias the inactivation toward the minor enamine pathway b.
5-Fluoro-4-aminopentanoic acid (8) has long been known as a
very efficient inactivator of GABA-AT that inhibits the enzyme
exclusively through a mechanism (Scheme 2, R = H) that involves
Schiff base formation with PLP followed by elimination of HF to
9 (R = H), leading only to the enamine pathway (10, R = H).6 We
thought we could take advantage of this mechanism by synthesiz-
ing 11 or 12, which are identical to 8 except for the addition of a
gem-methyl group on the carbon with the fluorine atom. However,
elimination of HF from 11 or 12 would lead to the identical enam-
ine intermediate (5, Scheme 1) found by tautomerization of 2. If
this occurs, then 11 or 12 would be a vigabatrin mimic that pro-
ceeds exclusively by vigabatrin’s minor enamine pathway (path-
way b, Scheme 1). This is depicted in Scheme 2, where R = CH3
(compare 9, R = CH3 to 5 in Scheme 1).
Off-target effects of 8 are significant, especially at glutamate
decarboxylase (GAD, required for GABA synthesis) and, to a lesser
extent, at aspartate aminotransferase (Asp-AT).7 Vigabatrin (2),
which is based on a hexanoic acid skeleton, shows no activity at
GAD and weakly affects Asp-AT.8,9 We, therefore, thought that hex-
anoic acids 11 and/or 12 would be excellent inactivators of GABA-
AT, while displaying improved enzyme selectivity when compared
to their shorter chain analogue 8.
2. Results and discussion
2.1. Synthesis
The devised synthetic route was based around the asymmetric
dihydroxylation/lactonization of a hexenoate ester and involved
the use of azide as a nitrogen source, as well as a benzyl ester as
a carboxylate protecting group. These choices were made to sim-
plify the isolation and purification of the zwitterion after the final
hydrogenolysis; under these conditions, only simple recrystalliza-
tion was required instead of ion exchange chromatography
(Scheme 3).
The synthesis of (S,S)-11, (R,R)-11, and 12, started with the
Claisen–Johnson rearrangement of 3-buten-2-ol (13) and triethyl
orthoacetate. The continuous removal of ethanol during this reac-
tion, which was reported to result in good yields,10 led to the for-
mation of product 14 along with its corresponding 3-buten-2-ol
ester; this mixture was difficult to separate, and yields were low
as a result. We recommend against this modification of the classic
conditions.
Asymmetric dihydroxylation/lactonization of unsaturated ester
14 led to hydroxyethyl butenolide (R,R)-15, the key intermediate of
this synthesis.11 Fluorination of this compound using XtalFluor EÒ
((diethylamino)difluorosulfonium tetrafluoroborate) and DBU
(1,8-diazabicycloundec-7-ene) yielded intermediate (R,S)-16,
which was hydrolyzed under basic conditions and esterified to give
acyclic fluorinated alcohol (R,S)-17.
The hydroxyl group in (R,S)-17 was replaced by azide under
Mitsunobu conditions using DPPA (diphenylphosphoryl azide)
and DIAD (diisopropyl azodicarboxylate), and the resulting azide
(S,S)-18 was hydrogenated to give final compound (S,S)-11. The
chiral purity of the product was determined by derivatization as
the Mosher amide, and was found to have 92% ee.
F
F
F
F
+H3N
COO- +H3N
COO-
+H3N
COO- +H3N
11
COO-
(S,R)-12
8
(S,S)-11
(R,R)-
Figure 2. 4-Amino-5-fluorohexanoic acid (8) and the designed hexanoic acid
analogues (11 and 12).
Lys329
Lys329
PLP
Lys329
PLP
Lys329
NH+
PLP
NH
PMP
NH4
NH2
NH2
R
+
R
H2O
~PLP
NH2
R
F
R
H
+
-OOC
10 (R = H)
O
+
-H2O
-OOC
N
COO-
N
COO-
H+
H+
9 (R = H)
8
(R = H)
5
(R = CH3)
11 or 12 (R = CH3)
Scheme 2. Mechanism of inactivation of GABA-AT by fluorinated compounds 8, 11, or 12.
Please cite this article in press as: Juncosa Jr., J. I.; et al. Bioorg. Med. Chem. (2013), http://dx.doi.org/10.1016/j.bmc.2012.12.009

J. I. Juncosa Jr. et al. / Bioorg. Med. Chem. xxx (2013) xxx–xxx
3
O
-
[Et2N=SF2]+BF4
O
BnOOC
HO
BnOOC
N3
-OOC
+H3N
COOEt
1. KOH
MeOH
AD-Mix β
EtCOOH
(cat)
DPPA
DIAD
OH
(XtalFluor E®)
MsNH2
H2, Pd/C
O
O
H
CH3C(OEt)3
45%
tBuOH
H2O, 61%
PPh3
THF
85%
DBU, CH2Cl2
42%
2. BnBr
DMF
89%
MeOH
82%
H
13
F
F
F
F
HO
(R,R)-15
14
(R,S)-16
(R,S)-17
(S,S)-18
(S,S)-11
4-Nitrobenzoic
(R,R)-11 synthesized
using AD-Mix α
acid, DIAD
PPh3, PhCH3, 94%
-OOC
O
O
O
BnOOC
HO
BnOOC
N3
1. KOH
MeOH
1. KOH
DPPA
DIAD
XtalFluor E®
DBU, CH2Cl2
H2O
H2, Pd/C
O
O
O
+H3N
O
2. H2SO4,
THF, H2O
PPh3
THF
78%
2. BnBr
DMF
85%
MeOH
39%
H
H
H
45%
F
F
F
O
HO
F
O2N
72%
12
22
23
20
21
19
Scheme 3. Synthesis of fluorinated GABA analogues 11 and 12.
Compound (R,R)-11 was obtained in a manner similar to that of
Time (min)
50 100
its enantiomer, but using AD-Mix
a instead of its b counterpart,
0
150
and DAST (diethylaminosulfur trifluoride) instead of XtalFluor EÒ
for the fluorination step.
0.2
0
(S,S)-11
(R,R)-11
12
The synthesis of compound 12 was initially attempted through
the double inversion of ent-17 to give 23. However, this strategy
failed, and efforts were shifted toward the inversion of (R,R)-15
via a Mitsunobu protocol. Although the reaction failed with picol-
inic acid,12 the use of 4-nitrobenzoic acid was successful, as re-
ported earlier.11
After subjecting diester 19 to literature conditions designed to
obtain butenolide 20 exclusively (free from the competing tetra-
hydropyrone product),11 we were delighted to obtain only the de-
sired product in a good yield. Fluorination of this compound led to
21, and the synthesis was completed as before (Scheme 3).
The enantiomer of 12 was not synthesized because it is known
that active GABA compounds of this type have S stereochemistry at
the amine carbon. Therefore, the R isomer was not a particularly
attractive synthetic target to warrant the additional work (See
Figs. 1 and 2).
-0.2
-0.4
-0.6
-0.8
-1
Vigabatrin
Linear
(Vigabatrin)
-1.2
-1.4
140
120
100
80
24
(S,S)-11
(R,R)-11
12
60
2.2. Enzymatic assays
40
Compounds (S,S)-11, (R,R)-11, and 12 were tested for time-
dependent inhibition of GABA-AT at 5 mM concentrations; viga-
batrin (2) was used as a positive control. Unfortunately, no inacti-
vation of the enzyme was observed by any of the synthesized
molecules (Fig. 3).
Regression
(24)
20
0
-6
-4
Log [Inhibitor]
-2
Similarly, no reversible inhibition of GABA-AT was observed
with any of the three compounds at 23 mM concentration (Fig. 3,
bottom). In this case, reversible inhibitor 2413 was employed as a
positive control. Consistent with these results, the compounds
did not show significant substrate activity with the enzyme; at a
concentration of 3 mM, (S,S)-11, (R,R)-11, and 12 showed gluta-
mate production relative to GABA of 0.11%, 0.02% and 0.32%,
respectively (n = 2).
Figure 3. Top: Inactivation study of GABA-AT with (S,S)-11, (R,R)-11, 12, and 2.
Bottom: Competitive binding of (S,S)-11, (R,R)-11, 12, and 20, at GABA-AT (n = 3 for
all measurements).
of the compounds in the active site, we found that there is a small,
narrow accessory binding pocket adjacent to the main cavity. This
space is occupied by the terminal vinyl group when vigabatrin is
bound (Fig. 4, middle); in the case of 8, it is mostly empty.
We were surprised to see that the terminal methyl group of
(S,S)-11 did not occupy the same space during about a third of
the simulation time; during a significant interval, it instead pre-
ferred pointing toward the access channel (Fig. 4). This occurred
in spite of the resulting unfavorable syn-pentane interaction with
2.3. Molecular modeling
In an attempt to find an explanation for the low inactivation
properties of the synthesized compounds, we decided to carry
out molecular modeling studies. On the basis of the published X-
ray structure of GABA-AT inactivated by vigabatrin (pdb code
1OHW),14 we performed docking studies of (S,S)-11, 8, and vigabat-
rin (2) in the active site, followed by molecular dynamics (MD)
simulations and energy minimization. Upon analyzing the binding
carbon 2 (a to the carboxylate), which carries with it an energetic
penalty of around 3.6 kcal/mol.
Please cite this article in press as: Juncosa Jr., J. I.; et al. Bioorg. Med. Chem. (2013), http://dx.doi.org/10.1016/j.bmc.2012.12.009

4
J. I. Juncosa Jr. et al. / Bioorg. Med. Chem. xxx (2013) xxx–xxx
intrigued by the conformational shift that occurred during the
MD simulation. We analyzed the evolution of the relevant dihedral
angle during the simulation, and found that there had, in fact, been
a shift from the anti to the gauche conformation and back to anti
toward the end of the simulation (Fig. 4, bottom).
We propose that the reason for these conformational shifts is
the narrow width of the accessory-binding pocket. As a planar sub-
stituent, the vinyl group of vigabatrin is narrower than the distal
methyl group of (S,S)-11 and is accommodated by the available
space. However, the methyl group is apparently too wide to fit into
the pocket without incurring steric penalties. As a crude measure
of the width of the binding pocket, we have calculated the area
of the triangle formed by the C
a of Ile72, Glu270, and Gln301, at
regular time intervals. The results can be seen in Figure 4 (bottom).
In the case of 8, the area has an average of 43.3 0.01 Å2, which can
be taken as the unoccupied width. When vigabatrin is bound, the
pocket is slightly widened at 44.6 0.01 Å2, owing to the alkene
substituent; this substituent appears to be well tolerated. How-
ever, for (S,S)-11, the width goes from 45.3 0.02 Å2 when occu-
pied back to 43.8 0.02 Å2 when empty, and then is again
expanded to the even wider 46.7 0.03 Å2 after returning to the
original conformation.
The observed conformational transitions for the proposed inac-
tivator lead us to believe that adopting the extended anti
conformation, with the terminal methyl group residing in the
accessory-binding pocket, incurs an energetic penalty in the vicin-
ity of the aforementioned 3.6 kcal/mol. We attribute this to the
cost of expanding the accessory-binding pocket. In other words,
both of the two possible conformations of the compound in the ac-
tive site are substantially higher in energy than those for 2 or 8. We
believe this occurrence explains the lack of binding affinity of 11
and 12 for GABA-AT, and thus, their inefficacy as inactivators.
3. Conclusions
We synthesized a series of 4-amino-5-fluorohexanoic acids
(both enantiomers of 11 and 12) as potential inactivators of
GABA-AT. The rationale behind these compounds was that adding
an extra carbon to enamine-pathway inactivator 8, such as the one
in vigabatrin, would improve its selectivity over other enzymes.
Unfortunately, the compounds were shown not to inactivate or
competitively inhibit GABA-AT. After carrying out molecular mod-
eling studies, we found that there exists an accessory-binding
pocket adjacent to the main active site of the enzyme, which
accommodates the flat vinyl group of vigabatrin. For (S,S)-11, the
corresponding methyl group appears to be too voluminous to fit
in the cavity, causing unfavorable steric interactions with the pro-
tein; an alternative conformation was observed, but a steric clash
occurred as well, in this case intramolecularly. Understanding the
topology of this accessory-binding pocket will allow us to refine
our inactivator design process, with the goal of achieving improved
enamine pathway GABA-AT inactivators, with good selectivity for
the enzyme.
4. Experimental section
Figure 4. Top: Observed conformations of (S,S)-11 within the GABA-AT binding
pocket during the MD simulations. Vigabatrin adopts an orientation similar to the
extended (blue) conformer. PLP is shown in gray. Middle: Location of (S,S)-11
relative to the active site surface; yellow: triangle used to track the width of the
accessory binding pocket during MD. Bottom (during MD time frame, averaged over
three values for smoothing): Yellow: C4-C5 dihedral angle for (S,S)-11; Red, green,
and blue: Area of yellow triangle in middle panel during the simulations with
compounds 8, 2, and (S,S)-11, respectively.
4.1. Materials and methods
1H NMR and 13C NMR spectra were recorded on a Bruker Avance
III 500 MHz spectrometer. 19F NMR spectra were obtained on an
Agilent DDR2 400 MHz spectrometer. Chemical shifts for are re-
ported as d values in ppm relative to tetramethylsilane (1H, 13C)
or CFCl3 (
19F), with the CHCl3 signal arbitrarily set as 7.27 (1H) or
77.0 (13C) ppm. Melting points were determined in a Büchi B540
melting point apparatus using open capillary tubes, and are uncor-
rected. Mass spectra were obtained with a Thermo Finnigan LCQ
Because the starting pose of the compound was analogous to
that of vigabatrin in the inactivated enzyme complex, that is, with
the methyl group in the vicinity of the accessory pocket, we were
Please cite this article in press as: Juncosa Jr., J. I.; et al. Bioorg. Med. Chem. (2013), http://dx.doi.org/10.1016/j.bmc.2012.12.009

J. I. Juncosa Jr. et al. / Bioorg. Med. Chem. xxx (2013) xxx–xxx
5
electrospray impact low resolution mass spectrometer, or an Agi-
lent LC-TOF 6210 (accurate mass), using 1:1 dichloromethane/
methanol as an eluent. Optical rotations were measured in an Opti-
cal Activity Limited AA-100 polarimeter, using a 0.5 dm, 1.3 mL
cell. Column chromatography was performed with Sorbent Tech-
nologies silica gel, (60 Å pore size, 230 Â 400 mesh) or on an Agi-
CH2CH2COO), 1.27 (3H, d, J = 6.5 Hz, CH3); 13C NMR (125 MHz,
CDCl3) d (ppm) 177.29, 84.23, 69.82, 28.68, 24.03, 18.45; ESI-MS
(m/z, abundance): 261 (100, [2 M+H]+), 131 (56, [M+H]+).
4.2.3. (S)-5-((S)-1-Hydroxyethyl)dihydrofuran-2(3H)-one ((S,S)-
15)
lent 971-FP machine using pre-packed 50
l
silica columns
This was synthesized similarly to (R,R)-15, but using AD-Mix
a.
(Analogix, Silicycle, or Agilent). Thin layer chromatography was
carried out using Baker-FlexÒ plastic-backed plates coated with sil-
ica gel IB2 and fluorescent indicator. Purity was determined on an
Agilent 1260 reverse phase analytical HPLC, using evaporative light
scattering detection (Agilent 385 ELSD); a C18 column (GeminiÒ
From 5.079 g (35.71 mmol) of ester 10, 3.011 g (23.14 mmol, 65%)
25
of product were obtained. [
a]
+46° (c = 0.49, CHCl3); all other
spectral data were similar to (R,R)-15.
4.2.4. (R)-5-((S)-1-Fluoroethyl)dihydrofuran-2(3H)-one ((R,S)-
5
l
m NX, 110 Å pore size, 50 Â 4.6 mm size) was used with 5% ace-
16)
tonitrile in water (0.05% TFA) as the mobile phase (0.8 mL/min).
All reagents were purchased from Aldrich, and were used as re-
ceived, except when noted. Solvents were dried using cartridge-
filled drying trains.
NMR spectra were analyzed with the help of MNova 7 (Mestr-
eLab Research, Santiago de Compostela, Spain, http://
mestrelab.com).
To a solution of alcohol (R,R)-15 (0.4073 g, 3.130 mmol) in
CH2Cl2 (10 mL) at À78 °C was added DBU (716
lL, 0.729 mg,
4.69 mmol), followed by XtalFluor-EÒ ((diethylamino)dif-
luorosulfonium tetrafluoroborate) (1.075 g, 4.695 mmol). After
30 min, the reaction was allowed to reach room temperature,
and stirring was continued for 24 h. Then, 5% aqueous NaHCO3
(10 mL) was added, and after stirring another 15 min, the aqueous
layer was extracted with CH2Cl2 (2 Â 10 mL). The combined organ-
ic phases were dried (MgSO4), filtered through a pad of silica gel
with abundant CH2Cl2, and the solvent was removed. Column chro-
4.2. Synthetic procedures and characterization
4.2.1. (E)-Ethyl hex-4-enoate (14)
matography (10–30% ethyl acetate in hexanes) gave the product
25
Neat propionic acid (126.2
lL, 125.2 mg, 1.681 mmol) was
(0.1721 g, 1.303 mmol, 42%) as a clear oil. [
a]
+2° (c = 0.44,
added to a mixture of 3-buten-2-ol (25.00 g, 336.3 mmol) and tri-
ethyl orthoacetate (100 mL, 86.73 g, 534.6 mmol) in a 3-necked
flask fitted with a reflux condenser over a liquid addition funnel.
The magnetically stirred solution was heated to 135 °C under
nitrogen. Every 2 h, the funnel was closed and a sample of the con-
densate was analyzed by NMR. When no more starting alcohol was
found in the distillate, the reaction was brought to room tempera-
ture and water (30 mL) was added. Stirring was continued for
30 min, and the low-boiling components were distilled off with
minimal vacuum, using an additional 30 mL ethanol to remove
residual water. Then, a higher vacuum was applied to distill the
product. Column chromatography (20% dichloromethane in hex-
anes) led to the product (21.535 g, 151.44 mmol, 45%) as a clear,
volatile liquid, and significant amounts of 3-buten-2-yl hex-4-eno-
ate as a side product (yield not determined). 1H NMR (500 MHz,
CDCl3) d (ppm) 5.56–5.36 (2H, m, @CH), 4.12 (2H, q, J = 7.1 Hz,
OCH2), 2.38–2.32 (2H, m, C@CCH2), 2.32–2.26 (2H, m, O@CCH2),
1.64 (3H, dq, J = 6.1, 1.2 Hz, C@CCH3), 1.25 (3H, t, J = 7.1 Hz,
OCH2CH3); 13C NMR (125 MHz, CDCl3) d (ppm) 173.25, 129.16,
126.07, 60.19, 34.30, 27.89, 17.85, 14.21.
CHCl3); 1H NMR (500 MHz, CDCl3)
d (ppm) 4.84 (1H, dqd,
J = 48.7, 6.5, 3.4 Hz, CHF), 4.46 (1H, dddd, J = 20.9, 7.9, 6.0, 3.4 Hz,
CHO), 2.69–2.45 (2H, m, CH2COO), 2.40–2.17 (2H, m, CH2CH2COO),
1.38 (3H, dd, J = 23.7, 6.6 Hz, CH3); 13C NMR (126 MHz, CDCl3) d
(ppm) 176.59, 89.75 (d, J = 173.4 Hz), 80.97 (d, J = 23.4 Hz), 27.88,
21.61 (d, J = 4.8 Hz), 16.50 (d, J = 22.1 Hz); 19F NMR (376 MHz,
CDCl3) d (ppm) À191.6 (1F, dp, J = 23.0, 46.9 Hz); ESI-MS (m/z,
abundance): 287 (93, [2 M+Na]+).
4.2.5. (S)-5-((R)-1-Fluoroethyl)dihydrofuran-2(3H)-one ((S,R)-
16)
To a solution of alcohol (S,S)-15 (1.859 g, 14.29 mmol) in dichlo-
romethane (15 mL) at 0 °C, was added DAST (2.37 mL, 2.91 g,
17.1 mmol) dropwise. The reaction was allowed to reach room
temperature over 16 h with magnetic stirring. Again at 0 °C, satu-
rated aqueous NaHCO3 (15 mL) was added slowly, and the mixture
was extracted with dichloromethane (3 Â 15 mL). The combined
organic extracts were dried (MgSO4), the solvent removed, and
the crude subjected to column chromatography (10–30% ethyl ace-
tate in hexanes), giving the product (0.4959 g, 1.888 mmol, 26%) as
25
a clear oil. [
a]
À5° (c = 0.45, CHCl3); all other spectral data were
4.2.2. (R)-5-((R)-1-Hydroxyethyl)dihydrofuran-2(3H)-one ((R,R)-
similar to those of (R,S)-16.
15)
To
a
vigorously stirred mixture of AD-mix15
b
(26.61 g,
4.2.6. (4R,5S)-Benzyl 5-fluoro-4-hydroxyhexanoate ((R,S)-17)
To a solution of lactone (R,S)-16 (0.1557 g, 1.178 mmol in MeOH
(5 mL) was added solid KOH (0.0777 g, 1.25 mmol), and the reac-
tion was stirred at room temperature for 16 h. The solvent was
then evaporated under reduced pressure; then the solid was redis-
solved in DMF (5 mL) and treated with benzyl bromide (0.143 mL,
0.206 g, 1.18 mmol) dropwise. After another 16 h of stirring, the
reaction was diluted with water (10 mL) and extracted with Et2O
(3 x 10 mL). The organic extracts were washed with water
(2 Â 5 mL), dried (Na2SO4), and concentrated under reduced pres-
sure. Column chromatography (10–40% ethyl acetate in hexanes)
Sigma–Aldrich) and methanesulfonamide (1.864 g, 19.01 mmol)
in t-BuOH/H2O (1:1, 100 mL) at 4 °C, was added unsaturated ester
14 (3.00 mL, 2.703 g, 19.01 mmol). Stirring was continued at that
temperature for 4 days. At this point, the reaction was complete
by TLC (1:1 EtOAc/hexanes) and solid NaHSO3 (30 g) was added.
After stirring for another 1 h, the suspension had turned white
and was partitioned between water and ethyl acetate (50 mL each).
The aqueous phase was then extracted with ethyl acetate
(3 Â 50 mL), and the combined organic fractions were dried
(Na2SO4), filtered, and the solvent removed. The crude was sub-
jected to column chromatography (20% EtOAc in hexanes and
yielded the product (0.2522 g, 1.050 mmol, 89%) as a clear oil.
25
0–5% MeOH in CH2Cl2), leading to the product (1.506 g,
[a
]
+24° (c = 0.40, CHCl3); 1H NMR (500 MHz, CDCl3) d (ppm)
25
11.57 mmol, 61%) as a clear oil. [
a]
À46° (c = 0.49, CHCl3); 1H
7.44–7.28 (5H, m, ArH), 5.14 (2H, s, ArCH2), 4.59 (1H, dqd,
J = 47.3, 6.4, 4.1 Hz, CHF), 3.80–3.64 (1H, m, HOCH), 2.69–2.46
(2H, m, CH2COO), 2.26 (1H, s, OH), 1.90 (1H, dtd, J = 14.6, 7.4,
2.8 Hz, CH2CH2COO), 1.72 (1H, ddt, J = 14.0, 9.9, 6.8 Hz,
CH2CH2COO), 1.34 (3H, dd, J = 24.8, 6.3 Hz, CH3); 13C NMR
NMR (500 MHz, CDCl3) d (ppm) 4.35 (1H, td, J = 7.4, 5.4 Hz,
COOCH), 3.79 (1H, p, J = 6.3 Hz, HOCH), 2.66–2.52 (2H, m,
CH2COO), 2.27 (1H, dddd, J = 12.5, 9.5, 7.3, 5.0 Hz, CH2CH2COO),
2.19 (1H, br s, OH), 2.05 (1H, dddd, J = 12.9, 9.9, 9.1, 7.6 Hz,
Please cite this article in press as: Juncosa Jr., J. I.; et al. Bioorg. Med. Chem. (2013), http://dx.doi.org/10.1016/j.bmc.2012.12.009

6
J. I. Juncosa Jr. et al. / Bioorg. Med. Chem. xxx (2013) xxx–xxx
(126 MHz, CDCl3) d (ppm) 173.81, 135.72, 128.58, 128.30, 128.25,
92.54 (d, J = 167.7 Hz), 72.84 (d, J = 22.0 Hz), 66.51, 30.65, 26.45
(d, J = 6.0 Hz), 15.38 (d, J = 22.6 Hz); 19F NMR (376 MHz, CDCl3) d
(ppm) À181.53 (1F, dqd, J = 48.0, 24.1, 17.9 Hz); ESI-MS (m/z,
abundance): 263 (59, [M+Na]+), 480 (43, [2 M]+).
4.2.11. (4S,5S)-5-Fluoro-4-((S)-3,3,3-trifluoro-2-methoxy-2-
phenylpropanamido)hexanoic acid
(R)-(À)-
(7.4 L, 9.9 mg, 39
tion of amine (S,S)-11 (5.8 mg, 39
a
-Methoxy-
mol) was added to a magnetically stirred solu-
mol) and NaHCO3 (98 mg,
a-(trifluoromethyl)phenylacetyl chloride
l
l
l
1.2 mmol) in water and acetone (1.5 mL each). After stirring over-
night at room temperature, the mixture was evaporated under re-
duced pressure, and 3 M aqueous HCl (3 mL) was added. Extraction
with CH2Cl2 (2 Â 3 mL) and CHCl3 (3 mL), followed by drying
4.2.7. (4S,5R)-Benzyl 5-fluoro-4-hydroxyhexanoate ((S,R)-17)
This was synthesized by the same method as that for (R,S)-17
from lactone (S,R)-16 (0.4463 g, 3.378 mmol), obtaining the prod-
25
(Na2SO4), filtration, and evaporation gave the product (14.2 mg,
uct (0.7403 g, 3.081 mmol, 91%) as
a
clear oil.
[
a]
À16°
25
38.9
l
mol, 100%) as a clear oil, [
a
]
À11° (c = 0.25, CHCl3); 1H
(c = 0.64, CHCl3); all other spectral data were similar to (R,S)-17.
NMR (500 MHz, CDCl3) d (ppm) 7.61–7.48 (2H, m, ArH), 7.45–
7.40 (3H, m, ArH), 6.99 (1H, d, J = 9.7 Hz, NH), 4.78 (1H, dqd,
J = 46.9, 6.3, 1.8 Hz, CHF), 4.10 (1H, dtdd, J = 26.7, 9.7, 4.7, 1.7 Hz,
CHNH), 3.57 (1H, br s, COOH), 3.45 (3H, q, J = 1.6 Hz, OCH3),
2.39–2.27 (2H, m, CH2COOH), 2.03–1.88 (2H, m, CH2CH2COOH),
1.36 (3H, dd, J = 24.4, 6.3 Hz, CHFCH3); 13C NMR (126 MHz, CDCl3)
d (ppm) 177.67, 167.04, 132.16, 129.61, 128.71, 127.41, 123.64 (q,
J = 290.2 Hz), 91.34 (d, J = 171.3 Hz), 84.03 (q, J = 26.4 Hz), 55.07,
52.04 (d, J = 18.5 Hz), 30.12, 26.97 (d, J = 3.2 Hz), 17.70 (d,
J = 22.6 Hz); 19F NMR (376 MHz, CDCl3) d (ppm) À69.07 (3F, s,
CF3), À189.97 (1F, ddq, J = 48.5, 27.6, 24.4 Hz, CHF); de = 92% based
on CF3 peak (3.00 F at À69.07 ppm vs 0.13 F at À69.24 ppm); (ESI-
MS (m/z, abundance): 366 (100, [M+H]+), 388 (23, [M+Na]+), 753
(53, [2M+Na]+).
4.2.8. (4S,5S)-Benzyl 4-azido-5-fluorohexanoate ((S,S)-18)
DIAD (0.200 mL, 0.205 g, 0.964 mmol) was slowly added to a
stirred solution of alcohol (R,S)-17 (0.1931 g, 0.8037 mmol), tri-
phenylphosphine (0.2555 g, 0.9644 mmol), and diisopropylethyl-
amine (0.140 mL, 0.104 g, 0.804 mmol) in THF (5 mL) at 10 °C,
and stirring was continued for 15 min. Diphenylphosphoryl azide
(0.214 mL, 0.274 g, 0.964 mmol) was then added slowly at
À15 °C, and the reaction was allowed to reach room temperature
overnight. The solvent was removed under reduced pressure, and
the crude product was directly subjected to column chromatogra-
phy (dichloromethane) to give the product (0.1742 g,
25
0.6747 mmol, 85%) as a clear oil. [
a
]
À20° (c = 0.29, CHCl3); 1H
NMR (500 MHz, CDCl3) d (ppm) d 7.47–7.25 (5H, m, ArH), 5.15
(2H, ABq,
DmAB = 7.28 Hz, JAB = 12.3 Hz, ArCH2), 4.65 (1H, dqd,
4.2.12. (4R,5R)-4-Ammonio-5-fluorohexanoate ((R,R)-11)
Using the same procedure as that for (S,S)-11, azide (R,R)-18
J = 47.4, 6.3, 5.2 Hz, CHF), 3.34 (1H, dddd, J = 18.2, 10.3, 5.2,
3.8 Hz, CHN3), 2.66–2.44 (2H, m, CH2COO), 1.94 (1H, dtd, J = 14.3,
7.8, 3.9 Hz, CH2CH2COO), 1.80 (1H, dddd, J = 14.2, 10.3, 7.4,
6.2 Hz, CH2CH2COO), 1.41 (3H, dd, J = 24.2, 6.3 Hz, CH3); 13C NMR
(126 MHz, CDCl3) d (ppm) 172.46, 135.64, 128.61, 128.37, 128.30,
92.22 (d, J = 173.2 Hz), 66.55, 64.69 (d, J = 19.2 Hz), 30.46, 25.22
(d, J = 5.2 Hz), 17.77 (d, J = 22.7 Hz); 19F NMR (376 MHz, CDCl3) d
(ppm) À181.54 (dqd, J = 48.3, 24.3, 18.1 Hz); ESI-MS (m/z, abun-
dance): 312 (40, [M+H+EtOH]+), 298 (10, [M+H+MeOH]+).
(0.1658 g, 0.6250 mmol) yielded the product (78.3 mg,
25
0.5249 mmol, 84%) as a white powder, mp. 131–132 °C, [a]
À44° (c = 0.22, CD3OD); HRMS (ESI) (m/z): 150.0926 (calc, for
C6H13FNO2+: 150.0925, [M+H]+); ee = 84% (from Mosher amide
de); all other spectral data were similar to (S,S)-11. HPLC purity
(retention time): 94% (0.790 min).
4.2.13. (4R,5R)-5-Fluoro-4-((S)-3,3,3-trifluoro-2-methoxy-2-
phenylpropanamido)hexanoic acid
4.2.9. (4R,5R)-Benzyl 4-azido-5-fluorohexanoate ((R,R)-18)
This compound was obtained through the procedure used for
(S,S)-18, starting from (S,R)-17 (0.2008 g, 0.8357 mmol), and yield-
ing the product (0.1898 g, 0.7155 mmol, 86%) as a clear oil. [a]
+23° (c = 0.34, CHCl3); all other spectral data were similar to
(S,S)-18.
This compound was prepared exactly as the derivative of prod-
uct (S,S)-11, but using amine (R,R)-7 (5.6 mg, 38
the product (13.2 mg, 36.1 mol, 96%) as a clear oil, contaminated
with (S)-Mosher’s acid (3.1 mg, 13.2
CHCl3), Corrected for Mosher’s acid: [
lmol) and giving
l
25
25
l
a
mol); [
]
a]
+8.3° (c = 0.96,
25
+27° (c = 0.78, CHCl3);
1H NMR (500 MHz, CDCl3) d (ppm) 7.62–7.48 (2H, m, ArH), 7.48–
7.36 (3H, m, ArH), 7.08 (1H, d, J = 9.8 Hz, NH), 4.75 (1H, dqd,
J = 46.9, 6.3, 2.2 Hz, CHF), 4.10 (1H, dtdd, J = 26.7, 9.5, 5.3, 2.0 Hz,
CHNH), 3.57 (1H, br s, COOH), 3.39 (3H, d, J = 1.5 Hz, OCH3),
2.54–2.40 (2H, m, CH2COOH), 2.07–1.92 (2H, m, CH2CH2COOH),
1.25 (3H, dd, J = 24.4, 6.3 Hz, CHFCH3); 13C NMR (126 MHz, CDCl3)
d (ppm) 178.05, 167.02, 131.83, 129.60, 128.71, 127.62, 123.73 (q,
J = 290.3 Hz), 91.43 (d, J = 171.3 Hz), 84.06 (q, J = 26.5 Hz), 54.91,
52.14 (d, J = 18.5 Hz), 30.08, 27.02 (d, J = 3.2 Hz), 17.84 (d,
J = 22.6 Hz); 19F NMR (376 MHz, CDCl3) d (ppm) À69.22 (3F, s,
CF3), À189.56 (1F, ddq, J = 48.5, 27.0, 24.5 Hz, CHF); de = 84% based
on CF3 peak (3.00 F at À69.22 ppm vs 0.27 F at À69.04 ppm); (ESI-
MS (m/z, abundance): 729 (100, [2MÀH]À), 364 (12, [MÀH]À).
4.2.10. (4S,5S)-4-Ammonio-5-fluorohexanoate ((S,S)-11)
Azido ester (S,S)-18 (0.1414 g, 0.5330 mmol) was dissolved in
methanol (12 mL) and 10% Pd/C (27 mg, 0.025 mmol) was added.
The flask was flushed under vacuum and filled with hydrogen three
times. A hydrogen-filled balloon was fitted to the sealed flask
through a needle, and stirring was continued for 24 h at room tem-
perature. At that time, the suspension was filtered through a pad of
Celite with additional methanol (50 mL). Solvent and volatiles
were removed from the filtrate under high vacuum. The crude
product was recrystallized from methanol and diethyl ether to give
the product (64.9 mg, 0.435 mmol, 82%) as a white solid, mp. 135–
25
137 °C, [
a
]
+41° (c = 0.36, MeOH); 1H NMR (500 MHz, CD3OD) d
(ppm) 4.76 (1H, dp, J = 48.0, 6.4 Hz, CHF), 3.30–3.17 (1H, m,
CHN), 2.58–2.35 (2H, m, CH2COO), 1.97–1.74 (2H, m, CH2CH2COO),
1.47 (3H, dd, J = 24.7, 6.3 Hz, CH3); 13C NMR (126 MHz, CD3OD) d
(ppm) 181.45, 92.21 (d, J = 170.8 Hz), 58.43 (d, J = 18.4 Hz), 36.44,
27.78 (d, J = 4.9 Hz), 18.75 (d, J = 22.1 Hz); 19F NMR (376 MHz,
CD3OD) d (ppm) À182.91 (1F, dqd, J = 49.0, 24.6, 14.3 Hz); HRMS
(ESI) (m/z): 150.0927 (calc, for C6H13FNO2+: 150.0925, [M+H]+);
ee = 92% (from Mosher amide de); HPLC purity (retention time):
95% (0.784 min).
4.2.14. (S)-1-((R)-5-Oxotetrahydrofuran-2-yl)ethyl 4-
nitrobenzoate (19)
A magnetically-stirred suspension of alcohol (R,R)-15 (0.4981 g,
3.827 mmol), triphenylphosphine (1.318 g, 4.976 mmol), and 4-
nitrobenzoic acid (0.8485 g, 4.976 mmol) in toluene (10 mL) at
0 °C and under nitrogen, was treated dropwise with neat diisopro-
pyl azodicarboxylate (1.1 mL, 1.1 g, 5.0 mmol). The reaction was al-
lowed to reach room temperature overnight, and the solids were
removed using a plug of cotton, washing with an additional
Please cite this article in press as: Juncosa Jr., J. I.; et al. Bioorg. Med. Chem. (2013), http://dx.doi.org/10.1016/j.bmc.2012.12.009

J. I. Juncosa Jr. et al. / Bioorg. Med. Chem. xxx (2013) xxx–xxx
7
10 mL of toluene. The crystalline product was redissolved with
ethyl acetate (10 mL) and the solvent was removed. The toluene
solution was also evaporated under reduced pressure, and the
resulting crude product was subjected to column chromatography
(20–50% ethyl acetate in hexanes), giving the product as a white
4.2.18. (4S,5R)-Benzyl 4-azido-5-fluorohexanoate (23)
This was obtained similarly to (S,S)-18, starting from alcohol 22
(0.1425 g, 0.5931 mmol), giving the product (0.1228 g,
25
0.4629 mmol, 78%) as a clear oil. [
a
]
À34° (c = 0.25, CHCl3); 1H
NMR (500 MHz, CDCl3) d (ppm) d 7.44–7.30 (5H, m, ArH), 5.15
solid (total yield: 1.005 g, 3.599 mmol, 94%), mp: 136–138 °C,
(2H, ABq, DmAB = 4.98 Hz, JAB = 12.3 Hz, ArCH2), 4.69 (1H, dqd,
25
[
a]
+16° (c = 0.66, CHCl3); 1H NMR (500 MHz, CDCl3) d (ppm)
J = 46.5, 6.3, 4.4 Hz, CHF), 3.57 (1H, dddd, J = 13.6, 10.4, 4.4,
3.3 Hz, CHN3), 2.64–2.46 (2H, m, CH2COO), 1.92 (1H, dtd, J = 14.2,
7.9, 3.3 Hz, CH2CH2COO), 1.67 (1H, dddd, J = 14.0, 10.4, 7.8,
5.7 Hz, CH2CH2COO), 1.39 (3H, dd, J = 24.4, 6.3 Hz, CH3); 13C NMR
(126 MHz, CDCl3) d (ppm) 172.45, 135.65, 128.60, 128.36, 128.30,
91.66 (d, J = 172.8 Hz), 66.55, 64.64 (d, J = 21.4 Hz), 30.62, 24.82
(d, J = 6.0 Hz), 16.16 (d, J = 22.7 Hz); 19F NMR (376 MHz, CDCl3) d
(ppm) À177.12 (dqd, J = 45.9, 24.3, 13.5 Hz); ESI-MS (m/z, abun-
dance): 223 (41, [MÀN3]+).
8.30 and 8.18 (4H, AA’XX’, JAX = 8.9 Hz, ArH), 5.39 (1H, qd, J = 6.5,
4.0 Hz, ArCOOCH), 4.69 (1H, td, J = 7.4, 4.0 Hz, lactone COOCH),
2.71–2.55 (2H, m, CH2COO), 2.42 (1H, dddd, J = 13.5, 9.0, 7.6,
6.1 Hz, CH2CH2COO), 2.30–2.14 (1H, m, CH2CH2COO), 1.45 (3H, d,
J = 6.5 Hz, CH3).; 13C NMR (126 MHz, CDCl3) d (ppm) 176.24,
163.79, 150.64, 135.13, 130.71, 123.63, 80.54, 72.16, 28.02, 23.00,
14.99; ESI-MS (m/z, abundance): 581 (100, [2 M+Na]+), 302 (24,
[M+Na]+), 280 (14, [M+H]+).
4.2.15. (R)-5-((S)-1-Hydroxyethyl)dihydrofuran-2(3H)-one (20)
Diester 19 (0.9751 g, 3.492 mmol) was dissolved in EtOH
(15 mL) at room temperature, and KOH (1.350 g, 21.65 mmol)
was added. The reaction was stirred at 60 °C for 2 h, and, after cool-
ing back to room temperature, the solvent was removed under re-
duced pressure. The residue was redissolved in THF and H2O
(25 mL each), and H2SO4 (2.3 mL, 4.2 g, 43 mmol) was added
slowly. After stirring at room temperature for 24 h, the reaction
was extracted with ethyl acetate (3 Â 50 mL), and the combined
extracts were dried (Na2SO4) and the solvent was removed. Col-
4.2.19. (4S,5R)-4-Ammonio-5-fluorohexanoate (12)
This was synthesized by the same route as for compound (S,S)-
11, using azido ester 23 (0.1063 g, 0.4007 mmol), and obtaining the
product (23.2 mg, 0.156 mmol, 39%) as a white solid, mp 158–
25
159 °C, [
a]
+15° (c = 0.24, CD3OD); 1H NMR (500 MHz, CD3OD)
d (ppm) 4.98–4.81 (1H, m, CHF), 3.46–3.28 (1H, m, CHN), 2.57–
2.33 (2H, m, CH2COO), 2.01–1.69 (2H, m, CH2CH2COO), 1.42 (3H,
dd, J = 24.2, 6.5 Hz, CH3); 13C NMR (126 MHz, CD3OD) d (ppm)
180.61, 90.97 (d, J = 171.2 Hz), 56.66 (d, J = 20.8 Hz), 35.81, 24.91
(d, J = 5.0 Hz), 16.00 (d, J = 22.4 Hz); 19F NMR (376 MHz, CD3OD) d
(ppm) À185.92 (1F, dqd, J = 47.8, 24.1, 19.3 Hz); HRMS (ESI) (m/
z): 150.0925 (calcd for C6H13FNO2+: 150.0925, [M+H]+); ee = 93%
(from Mosher amide de); HPLC purity (retention time): 94%
(0.794 min).
umn chromatography (30–50% ethyl acetate in hexanes) gave the
25
product (0.3272 g, 2.514 mmol, 72%) as a clear oil. [
a]
À11°
(c = 0.31, CHCl3); 1H NMR (500 MHz, CDCl3) d (ppm): 4.42 (1H,
td, J = 7.4, 3.3 Hz, COOCH), 4.14 (1H, qd, J = 6.7, 3.4 Hz, CHOH),
2.68–2.48 (2H, m, CH2COO), 2.35–2.12 (2H, m, CH2CH2COO), 2.04
(1H, br s, OH), 1.20 (3H, d, J = 6.5 Hz, CH3); 13C NMR (126 MHz,
CDCl3) d (ppm) 177.36, 83.37, 67.33, 28.62, 20.93, 17.64; ESI-MS
(m/z, abundance): 283 (100, [2M+Na]+), 153 (34, [M+Na]+), 131
(34, [M+H]+).
4.2.20. (4R,5R)-5-Fluoro-4-((S)-3,3,3-trifluoro-2-methoxy-2-
phenylpropanamido)hexanoic acid
This compound was prepared exactly as the derivative of prod-
uct (S,S)-11, but using amine 12 (6.1 mg, 41
product (13.6 mg, 37.3 mol, 91%) as a clear oil, contaminated with
(S)-Mosher’s acid (2.7 mg, 11.3
lmol) and giving the
l
25
4.2.16. (R)-5-((R)-1-Fluoroethyl)dihydrofuran-2(3H)-one (21)
This was prepared as detailed for (R,S)-16, from lactone 20
l
]
mol); [
a]
À17° (c = 0.96, CHCl3),
25
Corrected for Mosher’s acid: [
a
À1.0° (c = 0.80, CHCl3); 1H NMR
(0.3167 g, 2.434 mmol), yielding the product (0.1162 g,
(500 MHz, CDCl3) d (ppm) 7.56–7.46 (2H, m, ArH), 7.46–7.36 (3H,
m, ArH), 7.03 (1H, d, J = 9.6 Hz, NH), 4.73 (1H, dqd, J = 48.4, 6.5,
3.5 Hz, CHF), 4.10 (1H, dddt, J = 24.1, 11.2, 9.7, 3.4 Hz, CHNH),
3.57 (1H, br s, COOH), 3.46 (3H, q, J = 1.6 Hz, OCH3), 2.41–2.22
(2H, m, CH2COOH), 2.08 (1H, dtd, J = 15.1, 7.7, 3.3 Hz,
CH2CH2COOH), 1.75 (1H, dddd, J = 13.9, 11.2, 7.4, 6.0 Hz,
CH2CH2COOH), 1.39 (3H, dd, J = 24.1, 6.4 Hz, CHFCH3); 13C NMR
(126 MHz, CDCl3) d (ppm) 178.30, 166.70, 132.33, 129.63, 128.65,
127.33, 123.58 (q, J = 290.2 Hz), 91.83 (d, J = 171.6 Hz), 83.98 (q,
J = 26.4 Hz), 55.02, 52.28 (d, J = 20.7 Hz), 30.00, 22.51 (d,
J = 4.4 Hz), 17.41 (d, J = 22.2 Hz); 19F NMR (376 MHz, CDCl3) d
(ppm) À69.20 (3F, s, CF3), À188.75 (1F, dp, J = 48.2, 24.0 Hz,
CHF); de = 93% based on CF3 peak (3.00 F at À69.20 ppm vs
0.11 F at À69.30 ppm); (ESI-MS (m/z, abundance): 729 (100,
[2MÀH]À), 364 (12, [MÀH]À).
25
0.8794 mmol, 45%) as a clear oil. [
a]
À31° (c = 0.26, CHCl3); 1H
NMR (500 MHz, CDCl3) d (ppm): 4.68 (1H, dqd, J = 46.9, 6.5,
3.1 Hz, CHF), 4.49 (1H, dddd, J = 24.0, 8.1, 5.9, 3.1 Hz, CH3CHFCHO),
2.64 (1H, dddd, J = 18.3, 10.1, 6.6, 1.7 Hz, CH2COO), 2.52 (1H, dddd,
J = 18.0, 10.0, 7.0, 1.5 Hz, CH2COO), 2.34 (1H, dddd, J = 12.9, 10.0,
8.0, 6.6 Hz, CH2CH2COO), 2.20 (1H, dddd, J = 13.0, 10.1, 7.0,
5.9 Hz, CH2CH2COO), 1.45 (3H, dd, J = 23.9, 6.5 Hz, CH3); 13C NMR
(126 MHz, CDCl3) d (ppm) 176.74, 90.55 (d, J = 174.9 Hz), 80.71
(d, J = 19.7 Hz), 27.95, 23.57 (d, J = 4.7 Hz), 16.63 (d, J = 23.1 Hz);
19F NMR (376 MHz, CDCl3) d (ppm) À190.34 (1F, dp, J = 47.6,
23.9 Hz); ESI-MS (m/z, abundance): 568 (100, [4M+K]+).
4.2.17. (4R,5R)-Benzyl 5-fluoro-4-hydroxyhexanoate (22)
This was obtained similarly to (R,S)-17, from lactone 21
(0.1032 g, 0.7810 mmol), giving the desired ester (0.1597 g,
25
0.6647 mmol, 85%) as a clear oil, [
a]
+14° (c = 0.18, CHCl3); 1H
4.3. Computational methods
NMR (500 MHz, CDCl3) d (ppm): 7.42–7.28 (5H, m, ArH), 5.14
(2H, ABq, AB = 6.37 Hz, JAB = 12.3 Hz, ArCH2), 4.51 (1H, dqd,
D
m
Renderings were performed in PyMOL.16 Docking was per-
formed either with Autodock17,18 or Surflex-Dock.19 The protein
was downloaded from the Protein Data Bank (ID No. 1OHW)14
and was prepared in SYBYL20 by fixing all incomplete sidechains,
and setting histidine, asparagine, glutamine, aspartate and gluta-
mate protonation states and orientations. End caps were set to
acetamides and N-methyl amides for the N- and C-termini, respec-
tively, and hydrogen atoms were added. Only one set of dimers
was used in this work. The structure of vigabatrin with PLP was ex-
tracted, and all extraneous molecules except waters were removed.
J = 48.4, 6.3, 5.3 Hz, CHF), 3.58 (1H, ddtd, J = 16.8, 10.3, 5.3,
3.3 Hz, CHOH), 2.66–2.52 (2H, m, CH2COO), 2.22 (1H, dd, J = 5.3,
2.2 Hz, OH), 1.94–1.71 (2H, m, CH2CH2COO), 1.35 (3H, dd,
J = 24.7, 6.4 Hz, CH3); 13C NMR (126 MHz, CDCl3) d (ppm) 173.53,
135.80, 93.06 (d, J = 167.2 Hz), 73.34 (d, J = 19.9 Hz), 66.41, 30.27,
27.41 (d, J = 5.4 Hz), 16.92 (d, J = 22.6 Hz); 19F NMR (376 MHz,
CDCl3) d (ppm) À186.53 (1F, dqd, J = 49.5, 24.2, 16.6 Hz); ESI-MS
(m/z, abundance): 503 (100, [2M+Na]+), 264 (40, [M+Na]+), 241
(23, [M+H]+).
Please cite this article in press as: Juncosa Jr., J. I.; et al. Bioorg. Med. Chem. (2013), http://dx.doi.org/10.1016/j.bmc.2012.12.009

8
J. I. Juncosa Jr. et al. / Bioorg. Med. Chem. xxx (2013) xxx–xxx
Waters were protonated so as to maximize polar contacts. Finally,
Gasteiger-Marsili charges were calculated for all atoms.
The reversible inhibition studies were carried out following a
different sequence: The enzyme and SSADH were mixed with a
Preparation of the structures for Autodock was further carried
out using AutoDock Tools. All non-polar hydrogens were merged
with their corresponding heavy atoms. During docking, several
binding site sidechains were set as flexible: I72, E270, K329,
Y348, F351 and T353. A binding site of 15 Å around the center of
the vigabatrin–PLP complex was considered.
For Surflex-Dock, the protomol was generated from the prepared
structure, and docking was carried out using the vigabatrin/PLP
complex as a template. The same 15 Å radius was used, and the
structures obtained were the best out of 100 docking poses.
The structures of the ligands in the GABA-AT active site were re-
fined by molecular mechanics, using the GROMACS program.21
Amber99 Force field parameters for the GABA-AT dimer were ob-
tained with the pdb2gmx module, with parameters for the central
FeS cluster derived from an ab initio calculation (NWChem, HF//6–
31G⁄).22 The protein was solvated in a dodecahedron-shaped box
with 1 nm clearance on each side, and some waters were replaced
with sodium and chloride ions, to give an effective concentration of
0.14 M (with adjustments to obtain a neutral system).
Parameters for the ligands and other organic molecules were
obtained using the R.E.D. server23 and transformed into topology
files using the Antechamber module of the AMBER program.24
Because the simulations involved a protein dimer, one of the ac-
tive sites was occupied with the desired ligands, while the other
one included a molecule of PLP covalently bound to Lys329, and
an acetate ion forming an ionic interaction with Arg192. These
structures were obtained from the corresponding ligand-free entry
in the Protein Data Bank (#1OHV).14
solution containing 6.2
centrations) and 133.8
l
L of inhibitor solution (at different con-
lL of assay solution; this mixture was mea-
sured directly for activity, as before. IC50 values were calculated
using GraphPad PRISM version 5.0c for Mac OS X, GraphPad Soft-
ware, San Diego, CA, USA, www.graphpad.com. Ki values were cal-
culated using the Cheng–Prusoff equation. Graphs were replotted
in Microsoft Excel 2010 for publication. Finally, the possibility of
11 and 12 undergoing metabolism and activating SSDH similar to
GABA and succinic semialdehyde was ruled out; no NADPH pro-
duction was observed in the absence of GABA.
Substrate activity assays were performed by measuring gluta-
mate production by GABA-AT in the presence of the test com-
pounds. After turnover of
a
substrate molecule, PLP is
-ketoglutarate to gluta-
regenerated from PMP by transforming
a
mate. The detection of glutamate was carried out using a non-
amplified version of Invitrogen’s AmplexÒ Red glutamate detection
kit. In a 96-well plate, each well contained 0.5
glutamate oxidase (5 units/mL), 0.125 L horseradish peroxidase
(100 units/mL), 0.5
L AmplexÒ Red solution (10 mM), 6
ketoglutarate (100 mM), and of substrate (100 mM) in
89.1 L 100 mM potassium pyrophosphate buffer (pH 8.6), for a to-
tal of 100 L. Fluorescence was measured at 590 nm, with excita-
tion at 530 nm. A glutamate standard curve was created with
concentrations of 0, 2, 4, 6, 8, and 10 M (with 20 min preincuba-
tion), and all the wells in the plate were scaled to the 10 M stan-
lL GABA-AT, 0.8 lL
l
l
lL a-
3 lL
l
l
l
l
dard (at 50,000 RFU, or reference fluorescence units). The highest
rates of glutamate production, which occurred between 15 and
25 min of reaction time, were used for analysis.
After the molecules were in the active site, the system was en-
ergy minimized, and molecular dynamics simulations ensued. The
system was slowly warmed to 300 K (200 ns each at 100, 200, and
300 K), using position restraints (1000 kJ molÀ1 nmÀ2). After this
time, the position restraints were removed, and the simulations
were continued for 4 ns. After a final energy minimization, the
structures were used for evaluation.
Acknowledgments
The authors are grateful to the National Institutes of Health
(GM066132 and DA030604) for financial support of this research
and to Amit Walia for assistance in determining the purity of the
final compounds.
4.4. Enzymatic assays
References and notes
GABA-AT was purified from fresh pig brains following the pre-
viously published procedure,25 and was obtained with a concentra-
tion of 0.294 mg/mL and a specific activity of 1.74 units/mg. The
behavior of the isolated GABA-AT was similar to that observed pre-
viously, with a Km for GABA of 2.2 0.4 mM, and a Vmax of
1. Tolman, J. A.; Faulkner, M. A. Expert Opin. Pharmacother. 2009, 10, 3077.
2. Nanavati, S. M.; Silverman, R. B. J. Am. Chem. Soc. 1991, 113, 9341.
3. Gardner, E. L.; Schiffer, W. K.; Horan, B. A.; Highfield, D.; Dewey, S. L.; Brodie, J.
D., ; Ashby, C. R., Jr. Synapse 2002, 46, 240.
4. DeMarco, A.; Dalal, R. M.; Pai, J.; Aquilina, S. D.; Mullapudi, U.; Hammel, C.;
Kothari, S. K.; Kahanda, M.; Liebling, C. N.; Patel, V.; Schiffer, W. K.; Brodie, J. D.;
Dewey, S. L. Synapse 2009, 63, 87.
(6.1 0.5) Â 10–6 mol LÀ1 minÀ1
.
5. Paul, M.; Dewey, S. L.; Gardner, E. L.; Brodie, J. D.; Ashby, C. R., Jr. Synapse 2001,
41, 219.
Assay solutions consisted of 87 mM potassium pyrophosphate
6. Silverman, R. B.; Invergo, B. J. Biochemistry 1986, 25, 6817.
7. Silverman, R. B.; Muztar, A. J.; Levy, M. A.; Hirsch, J. D. Life Sci. 1983, 32, 2717.
8. Neal, M. J.; Shah, M. A. Br. J. Pharmacol. 1990, 100, 324.
9. Okumura, H.; Omote, M.; Takeshita, S. Arzneimittel-Forschung 1996, 46, 459.
10. Adachi, Y.; Kamei, N.; Yokoshima, S.; Fukuyama, T. Org. Lett. 2011, 13, 4446.
11. Bourque, E.; Kocienski, P. J.; Stocks, M.; Yuen, J. Synthesis 2005, 3219.
12. Sammakia, T.; Jacobs, J. S. Tetrahedron Lett. 1999, 40, 2685.
13. Qiu, J.; Pingsterhaus, J. M.; Silverman, R. B. J. Med. Chem. 1999, 42, 4725.
14. Storici, P.; De Biase, D.; Bossa, F.; Bruno, S.; Mozzarelli, A.; Peneff, C.; Silverman,
R. B.; Schirmer, T. J. Biol. Chem. 2004, 279, 363.
15. Sharpless, K. B.; Amberg, W.; Bennani, Y. L.; Crispino, G. A.; Hartung, J.; Jeong, K.
S.; Kwong, H. L.; Morikawa, K.; Wang, Z. M. J. Org. Chem. 1992, 57, 2768.
16. The PyMOL Molecular Graphics System, Version 1.4.1. Schrödinger, LLC.
17. Huey, R.; Morris, G. M.; Olson, A. J.; Goodsell, D. S. J. Comput. Chem. 2007, 28, 1145.
18. Morris, G. M.; Huey, R.; Lindstrom, W.; Sanner, M. F.; Belew, R. K.; Goodsell, D.
S.; Olson, A. J. J. Comput. Chem. 2009, 30, 2785.
19. Jain, A. N. J. Comput. Aided Mol. Des. 2007, 21, 281.
20. SYBYL 8.0, Tripos International, 1699 South Hanley Rd., St. Louis, MO 63144,
USA.
21. Hess, B.; Kutzner, C.; van der Spoel, D.; Lindahl, E. J. Chem. Theory Comput. 2008,
4, 435.
22. Valiev, M.; Bylaska, E. J.; Govind, N.; Kowalski, K.; Straatsma, T. P.; Van Dam, H.
J. J.; Wang, D.; Nieplocha, J.; Apra, E.; Windus, T. L.; de Jong, W. Comput. Phys.
Commun. 2010, 181, 1477.
buffer (pH 8.6), with 6 mM GABA, 5 mM
a-ketoglutaric acid,
1.25 mM NADP+, and 3.3 mM b-mercaptoethanol. Enzyme activity
was measured through the change in absorbance at 340 nM, corre-
sponding to the transformation of NADP+ to NADPH, based on the
coupled assay of Scott and Jakoby.26 The reactions were carried out
in the presence of excess succinic semialdehyde dehydrogenase
(SSADH), obtained from commercially available GABase (Sigma–
Aldrich) by published methods.27
The inactivation (time-dependent inhibition) studies were per-
formed as follows: GABA-AT (20
phate buffer, pH 7.0, with 1 mM b-mercaptoethanol) was
incubated at 25 °C with b-mercaptoethanol (1.25 L of 100 mM
solution), -ketoglutaric acid (1.25 L of 100 mM solution), and
the tested compounds (2.5 L of 50 mM solution for a final concen-
tration of 5 mM). At different time points, 4 L of this incubation
mixture were transferred to a 96-well plate with 150 L assay
lL in 100 mM potassium phos-
l
a
l
l
l
l
solution and excess SSADH, and activity (reaction rate) was mea-
sured in a Biotek Synergy H1 plate reader.
Please cite this article in press as: Juncosa Jr., J. I.; et al. Bioorg. Med. Chem. (2013), http://dx.doi.org/10.1016/j.bmc.2012.12.009

J. I. Juncosa Jr. et al. / Bioorg. Med. Chem. xxx (2013) xxx–xxx
9
23. Vanquelef, E.; Simon, S.; Marquant, G.; Garcia, E.; Klimerak, G.; Delepine, J. C.;
Cieplak, P.; Dupradeau, F. Y. Nucleic Acids Res. 2011, 39, W511.
24. Wang, J. M.; Wang, W.; Kollman, P. A.; Case, D. A. J. Mol. Graph. Model. 2006, 25,
25. Churchich, J. E.; Moses, U. J. Biol. Chem. 1981, 256, 1101.
26. Scott, E. M.; Jakoby, W. B. J. Biol. Chem. 1959, 234, 932.
27. Jeffery, D.; Weitzman, P. D. J.; Lunt, G. G. Insect Biochem. 1988, 18, 347.
247.
Please cite this article in press as: Juncosa Jr., J. I.; et al. Bioorg. Med. Chem. (2013), http://dx.doi.org/10.1016/j.bmc.2012.12.009