Isolation and characterization of the product of inactivation of γ-aminobutyric acid aminotransferase by gabaculine

Article abstract of DOI:10.1016/S0968-0896(99)00081-4

Gabaculine (5-amino-1,3-cyclohexadienylcarboxylic acid, 1Scheme 1), a naturally occurring neurotoxin isolated from Streptomyces toyocaenis, has been shown to be a mechanism-based inactivator of γ-aminobutyric acid aminotransferase (GABA-AT) (Rando, R. R.

Full text of DOI:10.1016/S0968-0896(99)00081-4

Bioorganic & Medicinal Chemistry 7 (1999) 1581±1590
Isolation and Characterization of the Product of Inactivation of
ꢀ-Aminobutyric Acid Aminotransferase by Gabaculine
Mengmeng Fu and Richard B. Silverman*
Department of Chemistry and Department of Biochemistry, Molecular Biology, and Cell Biology, Northwestern University, Evanston,
IL 60208-3113, USA
Received 3 December 1998; accepted 2 February 1999
AbstractÐGabaculine (5-amino-1,3-cyclohexadienylcarboxylic acid, 1), a naturally occurring neurotoxin isolated from Strepto-
myces toyocaenis, has been shown to be a mechanism-based inactivator of g-aminobutyric acid aminotransferase (GABA-AT)
(Rando, R. R. Biochemistry 1977, 16, 4604). Inactivation results from reaction of gabaculine with the pyridoxal 50-phosphate (PLP)
cofactor. Two HPLC systems for isolating this inactivator-PLP adduct are described as well as a detailed characterization of the
adduct, including the ultraviolet±visible spectrum, electrospray mass spectra, and NMR spectrum. The same spectral characteriza-
tion of the chemically synthesized gabaculine-PLP adduct is also reported. # 1999 Elsevier Science Ltd. All rights reserved.
Introduction
Mechanistic studies of the inactivation of bacterial
GABA-AT by gabaculine were carried out in the mid-
1970s by Rando and co-workers, and an aromatization
mechanism was proposed, leading to the formation of a
modi®ed PLP cofactor (3, Scheme 1).14,15 The same
mechanism and product was proposed in the inactiva-
tion of GABA-AT by isogabaculine (4).16
Gabaculine (5-amino-1,3-cyclohexadienylcarboxylic acid,
1) is a naturally occurring neurotoxin isolated from
Streptomyces toyocaenis.1 Soon after its isolation, it was
found that gabaculine is an exceedingly potent irrever-
sible inactivator of pyridoxal 50-phosphate (PLP)-
dependent mouse brain g-aminobutyric acid amino-
transferase (GABA-AT; EC 2.6.1.19) both in vitro and
in vivo.2 When administered intraperitoneally, brain
GABA-AT is rapidly inactivated, and the brain levels of
the inhibitory neurotransmitter GABA rise by 15±20-
fold, suggesting that gabaculine can cross the blood-
brain barrier.3 Unfortunately, it is too toxic to be useful
as a pharmaceutical agent, probably because it inacti-
vates several other enzymes as well.4 But because of its
remarkable properties, gabaculine has since been used
as a powerful pharmacological tool in a variety of
GABAergic neurotransmission studies.5±13
It was suggested that b-proton removal is an enzyme-
catalyzed event for those PLP enzymes that have a basic
group at the active site to catalyze this process.17 Two
related syntheses of the hypothesized modi®ed cofactor
(3) were carried out; one was the reaction of PLP and
gabaculine at elevated temperature, and the other was
the reaction of PLP with m-anthranilic acid followed by
sodium borohydride reduction.14 On chemical grounds,
both of these reactions are expected to give 3, but the
characterizations of these synthetic products consisted
of thin-layer chromatography, its absorption and ¯uo-
rescence spectra, and an elemental analysis. The product
of the enzymatic reaction was then compared to the
synthetic product by thin-layer chromatography and
electrophoresis. We thought that a more complete
characterization of the synthetic and enzymatic products
was warranted and report here the electrospray mass
spectral and NMR spectral analyses of these products,
which support the original hypothesis of Rando.15
Key words: g-Aminobutyric acid aminotransferase; gabaculine; pyr-
idoxal 50-phosphate; electrospray mass spectrometry.
* Corresponding author. Tel.: (847) 491-5653; fax: (847) 491-7713;
e-mail: agman@chem.nwu.edu
0968-0896/99/$ - see front matter # 1999 Elsevier Science Ltd. All rights reserved.
PII: S0968-0896(99)00081-4

1582
M. Fu, R. B. Silverman / Bioorg. Med. Chem. 7 (1999) 1581±1590
Scheme 1.
Results
Time-dependent inactivation of GABA-AT by gabaculine
Inactivation of GABA-AT by gabaculine was con-
centration and time-dependent; KI and kinact values were
determined by the method of Kitz and Wilson18 to be 29
mM and 6.4 min 1, respectively, at room temperature.
The KI and kinact values at 0ꢀC were determined to be
259 mM and 1.4 min 1, respectively.
Inactivation of [3H]PLP-GABA-AT by gabaculine,
product isolation, and analysis by reverse phase HPLC
There is apparently only one major product formed
from the inactivation of [3H]PLP-GABA-AT by gaba-
culine as shown with two di€erent HPLC elution sys-
tems (TR=36 min, Fig. 1A; TR=10.8 min, Fig. 1B).
This product coelutes with synthetic 3. Less than 1% of
the radioactivity was detected in the protein pellet.
UV±visible spectroscopic studies of the puri®ed
gabaculine inactivation product
The UV±visible spectrum of the puri®ed gabaculine
inactivation product had a maximum at 312 nm (in 2 M
NH4OH) and a shoulder at 245 nm (Fig. 2; Ð). This is
consistent with the UV±visible spectrum of synthetic 3
(80 nmol, Fig. 2; - - - -). From the absorption at 312 nm,
it was estimated that, after puri®cation, approximately
62% (50 nmol) of the product was recovered.
Electrospray ionization mass spectral studies of synthetic
3
Figure 1. Reverse phase HPLC analysis of the product isolated from
[3H]PLP-GABA-AT inactivated by gabaculine. Radioactivity (- - - -)
and absorption (Ð) are plotted versus retention time of the standards
and [3H] labeled product. The absorption peaks correspond to the
standards: PMP (8 min), PLP (16 min), 3 (36 min) in Figure 1A and
PMP (3.5 min), PLP (4.9 min), 3 (10.8 min) in Figure 1B. See solvent
system A in the Experimental for details for Figure 1A and solvent
system B for Figure 1B.
Negative mode electrospray ionization mass spectro-
metry (ESIMS ) of 3 showed the expected molecular
ions ([M H] , m/z=367, Fig. 3). The other prominent
peaks in the spectrum could be assigned as in Scheme 2.
Positive mode electrospray ionization mass spectrometry

M. Fu, R. B. Silverman / Bioorg. Med. Chem. 7 (1999) 1581±1590
1583
Figure 4. Electrospray ionization mass spectrum (positive mode) of 3.
Figure 2. UV±visible spectra of the puri®ed gabaculine inactivation
product and synthetic 3.
Electrospray ionization mass spectral studies of the
highly puri®ed gabaculine inactivation product
The ESIMS , ESIMS+, and MS/MS experiments were
performed on the puri®ed gabaculine inactivation pro-
duct under the same sets of conditions as synthetic 3.
Similar fragmentation patterns from the respective
molecular ions were observed (Figs 6±8, respectively).
NMR Studies of 3 and the puri®ed gabaculine
inactivation product
The 1H NMR spectra obtained after 4096 scans of both
synthetic 3 (carried through the same conditions as the
isolation of the gabaculine inactivation product) and the
puri®ed gabaculine inactivation product are compared
in Figure 9. Despite the small amount of sample used,
all of the peaks expected from the 1H NMR spectrum of
3 were observed both in the spectrum of pretreated
synthetic 3 and the puri®ed gabaculine inactivation
product: (D2O, 400 MHz) d 2.51 (3 H, s), 4.58 (2 H, s),
5.08 (2 H, d, J=6.8 Hz), 7.04 (1 H, m), 7.35 (3 H, m),
7.85 (1 H, s). Many of the peaks that are not assigned
from both samples have similar coupling constants and
chemical shifts, suggesting that these impurities may
have been similarly introduced during the puri®cation
process. The concentration of the puri®ed gabaculine
inactivation product is less than 0.1 mM, and the con-
centration of the pretreated 3 is about 0.25 mM.
Figure 3. Electrospray ionization mass spectrum (negative mode) of 3.
(ESIMS+) of 3 also showed the expected molecular ions
([M+H]+, m/z=369, Fig. 4). Attempts to get a MS/MS
spectrum from the ESIMS peak failed because of the
overwhelming HO3PO signal. MS/MS was therefore
carried out in the ESIMS+ mode on the peak m/z 369.
Two daughter ion peaks were detected (Fig. 5), which
could be assigned as in Scheme 3.
Scheme 2.

1584
M. Fu, R. B. Silverman / Bioorg. Med. Chem. 7 (1999) 1581±1590
Figure 6. Electrospray ionization mass spectrum (negative mode) of
the puri®ed gabaculine inactivation product.
Figure 5. MS/MS of peak m/z 369 from the EIMS+ spectrum of 3.
Discussion
Soon after the isolation of gabaculine from Strepto-
myces toyocaenis,1 an aromatization mechanism
(Scheme 1) was proposed by Rando14,15 to account for
the bacterial GABA aminotransferase inactivation by
gabaculine. The results of these studies successfully
demonstrated the mechanism-based nature of the inac-
tivation of GABA-AT by gabaculine. The conclusion
that catalytic conversion of this inhibitor necessarily
precedes the observed inactivation was made because
(1) gabaculine is only an irreversible inhibitor of the
enzyme when the cofactor is in the PLP form; (2) there
is a deuterium isotope e€ect on the rate of the inactiva-
tion when a deuterium is incorporated into the 5 posi-
tion of gabaculine; (3) the pH versus rate of inactivation
pro®le is similar to that of substrate turnover; and (4)
external nucleophilic trapping agents, such as b-mer-
captoethanol, have no e€ect on the rate of inactivation.
Figure 7. Electrospray ionization mass spectrum (positive mode) of
the puri®ed gabaculine inactivation product.
The enzymatic formation of m-carboxyphenylpyridox-
amine phosphate (3, Scheme 1) was supported2 by thin-
layer chromatographic and paper electrophoretic beha-
vior. The enzymatic product isolated from denatured
[2 3H] gabaculine-inactivated GABA-AT comigrated
with the carrier, namely synthetic 3. Dephosphorylation
of the inactivated radiolabelled inactivation product by
alkaline phosphatase produced a compound that behaved
identically by thin-layer chromatography and paper elec-
trophoresis as synthetic m-carboxyphenylpyridoxamine
Inactivation is not the result of modi®cation of an active
site residue.2 When bacterial GABA-AT was inactivated
by [2 3H]gabaculine, radioactivity incorporation
occurred simultaneously with enzyme inactivation.
Denaturation of the radiolabelled enzyme resulted in
complete loss of radioactivity from the protein. Inter-
mediate 2 (Scheme 1) is a Michael acceptor and is, in
principle, capable of alkylating an active-site residue
(Scheme 4), although this does not occur.
Scheme 3.

M. Fu, R. B. Silverman / Bioorg. Med. Chem. 7 (1999) 1581±1590
1585
(3 without the phosphate group). To show the forma-
tion of only one enzymatic product and to characterize
the product of the enzyme-catalyzed reaction more
fully, HPLC was used to isolate the product, and the
mass spectrum and NMR spectrum are now reported.
As shown by reverse phase HPLC analysis in Figure 1,
there is only one radiolabelled product from the inacti-
vation of [3H]PLP-GABA-AT by gabaculine. This radi-
olabelled product comigrates with synthetic m-
carboxyphenylpyridoxamine phosphate (3) and behaves
di€erently from that of PLP and PMP when analyzed
under the same HPLC conditions. This result is con-
sistent with the thin-layer chromatographic and paper
electrophoretic studies done by Rando.2 Further char-
acterizations of both the synthetic and enzymatic pro-
ducts were carried out. After isolation and puri®cation,
the single product obtained had a UV±vis spectrum vir-
tually identical to that of synthetic 3 (Fig. 2). A max-
imum absorption in this range could be an indication of
the formation of a pyridoxamine adduct.19
Figure 8. MS/MS of peak m/z 369 from the EIMS+ spectrum of the
puri®ed gabaculine inactivation product.
Electrospray ionization mass spectrometry (ESIMS) of
the gabaculine inactivation product was also carried out
(Figs. 6 and 7). Both negative and positive ionization
mode spectra were the same as those for the synthetic 3
(Figs. 3 and 4). The most prominent peak in the
ESIMS+ of the modi®ed coenzyme (Fig. 7) is the
molecular ion with m/z 369, corresponding to 3. This
peak was analyzed by electrospray ionization tandem
mass spectrometry (MS/MS) (Fig. 8), and the molecular
ion (m/z 369) and the major daughter ions at m/z 232
and 134 also were found in the MS/MS of synthetic 3
(Fig. 5). The possible fragmentation sites are shown in
Scheme 3; all are consistent with the proposed product
structure 3.
Figure 9. 1H NMR of 3 and the puri®ed gabaculine inactivation
product.
Further support for structure 3 for the inactivation
product was gathered by proton NMR spectroscopy
Scheme 4.

1586
M. Fu, R. B. Silverman / Bioorg. Med. Chem. 7 (1999) 1581±1590
studies. The NMR spectra for the gabaculine inactivation
adduct and for synthetic 3 are identical (Fig. 9).
Beckman 125P pumps and a Beckman 166 detector. All
of the runs were monitored at 254 nm unless otherwise
speci®ed. The HPLC columns used were Alltech C18
analytical Alltima, Hypersil Elite 5 mm, or semi-prep
Econosil 10 mm columns. Enzyme puri®cation was car-
ried out on a Pharmacia Biotech FPLC system (con-
sisting of a conductivity monitor, a UV-MII detector,
and two P-500 pumps). Electrophoresis was carried out
on a Bio-Rad Mini-Protean II electrophoresis cell with
Bio-Rad 12% Tris±HCl 10 well, 30 ml comb ready gel,
using a Bio-Rad Model 1000/500 power supply.
Additional support for the aromatization mechanism
comes from structural studies of ornithine amino-
transferase inactivated by gabaculine.20 These studies
show that aromatic±aromatic interactions occur
between the enzyme-bound gabaculine molecule and
active site residues Tyr-85 and Phe-177. This ornithine
aminotransferase-gabaculine complex provides the ®rst
structural evidence that the potency of the inhibitor
results from energetically favorable aromatic interac-
tions with residues in the active site.
Reagents
All reagents were purchased from Sigma Chemical Co.
except the following: tri¯uoroacetic acid (TFA), 3-ami-
nobenzoic acid, sodium borohydride, amyl alcohol,
diethylamine, and all of the NMR solvents were pur-
chased from Aldrich; Centricon 10 microconcentrators
were purchased from Amicon; HPLC mobile phases
and organic solvents were purchased from Fisher;
unpacked chromatography columns for Penefsky spin
method,21,22 Dowex 50 and Dowex 1 resins, sodium
dodecyl sulfate, were purchased from Bio-Rad; ultra-
Conclusion
All of the evidence for the structure of the product of
inactivation of GABA-AT by gabaculine supports 3.
This strongly supports the aromatization-type mechan-
ism (Scheme 1).
Experimental
pure urea was
a product of ICN Biomedicals;
Analytical methods
[3H]sodium borohydride was obtained from Amersham.
All of the reagents used for gel electrophoresis were
purchased from Bio-Rad. Pig brains were purchased
from American Meat Protein Corporation (Ames, IA).
All of the bu€ers and solvents used for HPLC or FPLC
analyses were ®ltered through Gelman 0.45 m mem-
branes. Bu€er A (100 mM potassium phosphate bu€er,
pH 7.4, containing 0.25 mM b-mercaptoethanol) was
used in most of the enzyme incubations. Water used in
enzyme related experiments is deionized and doubly
distilled.
GABA-AT assays and UV±visible spectra were recor-
ded on a Perkin±Elmer Lambda 10 spectrophotometer.
Measurements of pH were performed on an Orion 701-
A pH meter with a Ross 8301 combination electrode.
All of the dialyses were done with Pierce Slide-A-Lyzer
dialysis cassettes (molecular weight cut-o€ is 10,000)
unless otherwise speci®ed. Radioactivity was deter-
mined either with a Radiomatic FLO-ONE\Beta Series
A-200 liquid ¯ow scintillation counter or a Packard
TRI-CARB 2100TR liquid scintillation analyzer. The
scintillation ¯uids used were Packard ULTIMA-
FLOTMM and ULTIMA GOLD. Electrospray ioniza-
tion mass spectra were acquired on a Micromass
Quattro II mass spectrometer (Fisons Instruments,
Manchester, UK). NMR spectra were recorded on
either a Varian Unity Plus 400 MHz or a Varian Gemini
300 MHz spectrometer. Chemical shifts are reported as
d values in parts per million down ®eld from tetra-
methylsilane (TMS) in CDCl3 or DMSO and from
sodium 3-(trimethylsilyl)propionate in D2O. Coupling
constants are reported in Hertz. Melting points were
determined on a Mel-Temp capillary tube melting point
apparatus and are uncorrected. A Beckman Microfuge
B was used for microcentrifugations and an IEC clinical
centrifuge was used for the Penefsky spin method.
Cavitator ultrasonic cleaner from Mettler Electronics
Corp. was used for ultrasonications. Column chroma-
tography was done with Whatman DE52 (a pre-swollen
microgranular anion exchanger: diethylamino ethyl cel-
lulose). Thin-layer chromatography (TLC) was per-
formed using Whatman PE SIL/UV silica gel plates
with UV indicator. Amines were visualized on TLC
plates by dipping the plate into a solution of ninhydrin
in n-butanol and then heating. Other compounds were
visualized with I2 or phosphomolybdic acid in ethanol
followed by heating. HPLC analysis was done with
Synthesis of N-m-carboxyphenylpyridoxamine phosphate
(3)
N-m-Carboxyphenylpyridoxamine phosphate (3) was
prepared according to a modi®cation of the method of
Iskander et al.18 Thus PLP (4.94 g, 20 mmol) was dis-
solved in dry methanol (80 mL, distilled over magne-
sium with iodine under nitrogen directly before use)
containing KOH (2.24 g, 40 mmol). To the resulting
yellow solution, a mixture of 3-aminobenzoic acid (2.74
g, 20 mmol) in dry methanol (80 mL) containing KOH
(1.12 g, 20 mmol) was added dropwise. The mixture
turned into a bright reddish-orange solution. After the
addition, the reaction mixture was stirred for an addi-
tional 30 min at room temperature, then sodium bor-
ohydride (0.76 g, 20 mmol) was added portionwise at 0±
5ꢀC. The reaction mixture was stirred for a further 20
min, ®ltered, the ®ltrate acidi®ed with glacial acetic acid
to pH 6 and concentrated in vacuum to dryness. The
solid was dissolved in water (5 mL) and loaded onto a
2Â50 cm Whatman DE 52 resin packed column (pre-
washed and equilibrated with the eluting bu€er). The
compound was eluted with an aqueous solution con-
taining 0.3 M acetic acid and 0.3 M pyridine. Fractions
of 10 mL were collected and checked by TLC (acetone/
t-amyl alcohol/water/diethylamine, 3/4/2/1, silica gel).

M. Fu, R. B. Silverman / Bioorg. Med. Chem. 7 (1999) 1581±1590
1587
The fractions with Rf of 0.51 were combined and cooled
to 0ꢀC. The white needle-shaped crystals formed were
®ltered and washed with the recrystallization bu€er.
After being vacuum dried, the product (3) was obtained
in 30% yield (2.4 g). The UV±visible spectrum had a
maximum at 312 nm (in 2 M NH4OH); 1H NMR
(DMSO, 300 MHz) d 2.41 (3 H, s), 4.32 (2 H, s), 4.96 (2
H, d, J 6.8 Hz), 6.87±6.90 (1 H, m), 7.18±7.20 (2 H, m),
7.26 (1 H, d, J 1.16 Hz), 8.00 (1 H, s).
radioactivity and coeluting with authentic PLP were
combined, and the solvent was removed by lyophiliza-
tion. The second column was run with 100 mM potas-
sium phosphate, pH 7 as the mobile phase. Fractions of
1 mL were collected, and those coeluting with unlabeled
PLP and containing radioactivity were combined and
lyophilized. The third puri®cation was the same as the
®rst one except that 1 mL fractions were collected; those
containing radioactivity and coeluting with authentic
PLP were combined, and the solvent was removed by
lyophilization. The tritiated PLP obtained had a speci®c
radioactivity of 8.38 Ci/mmol, a concentration of 0.28
mM, and a radiopurity of 99%.
Synthesis of tritiated PLP ([4 3H]PLP)
Tritiated PLP ([4-3H] PLP) was synthesized using a
variation of the method of Stock et al.23 PLP (140 mg,
0.569 mmol) was dissolved in 0.5 M sodium hydroxide
(2.5 mL). The solution was protected from light and
cooled to 0ꢀC. [3H]Sodium borohydride from Amer-
sham (100 mCi, 15 Ci/mmol) was dissolved in 0.3 M
solution of sodium borohydride in 0.1 M NaOH (450
mL). This solution was then added to the PLP solution
prepared above and stirred for 40 min at 0ꢀC. The
reaction was monitored by following the 390 nm
absorption change of PLP (UV±visible scan of an ali-
quot of the reaction mixture in 500 mL of 2 M NH4OH).
Further sodium borohydride (total about 2.3 mg) was
added portionwise until the 390 nm absorption change
reached its minimum. The solution was very light yellow
at this point, and the reaction was quenched with con-
centrated HCl (120 mL). Activated manganese dioxide
(75 mg), which was prepared as described by Silverman
and Invergo,24 was added to the solution stirred at room
temperature for about 2 h. Additional manganese diox-
ide (total about 160 mg) was added in three portions
until the return of the 390 nm peak reached its max-
imum. The solution was then divided into two cen-
trifuge tubes and brought to pH 8 with 1 M NaOH.
After being centrifuged in an IEC clinical centrifuge at
maximum speed, the pellet was dark brown, and the
supernatant was dark yellow. Both of the supernatants
were combined, and the solvent was removed by bulb-
to-bulb distillation in a closed system. The brownish
yellow [3H] PLP was puri®ed on a Dowex 1 column
(1.5Â7.5 cm) which had been cleaned by washing with 5
M acetic acid, 1 M NaOH, and 5 M acetic acid once
more. Between each of these washes, the column was
washed with water until the eluent was neutral. Finally,
the column was equilibrated in 4 M sodium acetate. The
[3H] PLP solution was applied to the column and was
eluted with water (10 mL), followed by a gradient of
water (250 mL) and 5 M acetic acid (250 mL). A Phar-
macia Biotech peristaltic pump was attached to the col-
umn and was run at 1.5 mL/min while the fraction
collector changed fractions every 10 min. The fraction
with the highest absorption at 390 nm and all of the
other fractions with higher than 10% of that highest
absorption were combined, and the solvent was
removed by lyophilization. The [3H]PLP was further
puri®ed using three HPLC systems sequentially. All of
the systems utilized an Econosil 10 micron C18 semi-
prep HPLC column with solvent ¯owing at 1 mL/min.
Detection was at 254 nm. The ®rst puri®cation was
done using H2O/0.1% TFA as the mobile phase. Five-
milliliter fractions were collected; those containing
Enzymes and assays
Succinic semialdehyde dehydrogenase (SSDH) was
obtained from GABAase, a commercially available
mixture of GABA-AT and SSDH, by the reported pro-
cedure.25 Protein assays were carried out using bovine
serum albumin (BSA) and Pierce Coomassie protein
assay reagent for standard curves. GABA-AT activity
assays were carried out using a modi®cation of the
coupled assay developed by Scott and Jakoby.26 The
assay solution contained 110 mM GABA, 5.3 mM a-
KG, 1.1 mM NADP+, and 5 mM b-mercaptoethanol in
200 mM potassium diphosphate, pH 8.5. For each
assay, excess SSDH was used. The amount of activity
remaining in an enzyme solution was determined by
adding an aliquot of enzyme solution to the assay solu-
tion with SSDH and monitoring the change in absorp-
tion at 340 nm at 25ꢀC, as a result of the conversion of
NADP+ to NADPH by SSDH.
Further puri®cation of GABA-AT isolated from pig
brains
GABA-AT (55 mg, 0.52 mmol) was isolated from pig
brains (2.5 kg) using a modi®cation of the procedure of
Churchich and Moses.27 The purity of the enzyme prior
to the modi®cation is 77%, and major impurities came
from proteins having molecular weights larger and
smaller than that of GABA-AT according to gel elec-
trophoresis analysis. A published size exclusion chro-
matography procedure28 was, therefore, modi®ed to
further purify the isolated GABA-AT (stored in 25 mL
of 350 mM KH2PO4 bu€er, pH 7.0, containing 1 mM b-
mercaptoethanol). This solution was concentrated (to
2.5 mL) in an Amicon ultra®ltration cell (#8050) con-
taining a PM 10 membrane at a N2 pressure of 70 psi.
The concentrated enzyme was ®ltered through a 0.45 m
GHP Acrodisc purchased from Gelman. The combined
®ltered enzyme and rinses were injected onto a 16/60
Pharmacia Hiprep Sephacryl S-200 high resolution col-
umn, prewashed and equilibrated with two column
volumes of bu€er H (50 mM Na2HPO4 bu€er, pH 7,
containing 0.15 M NaCl, 1 mM b-mercaptoethanol).
The enzyme was then eluted with bu€er H at 0.2 mL/
min for 400 min. The elution was monitored at 280 nm.
Fractions (1 mL) were collected and assayed. All of
the enzyme-containing fractions were analyzed by gel
electrophoresis. Active enzyme fractions of more
than 90% purity (fractions from 210 to 250 min) were

1588
M. Fu, R. B. Silverman / Bioorg. Med. Chem. 7 (1999) 1581±1590
collected and combined. The enzyme was then dialyzed
against bu€er A (3Â1L, bu€er changed every 2 h). The
enzyme obtained (40 mg) had a speci®c activity of 8.5 units/
mL, a protein concentration of 2.1 mg/mL, and a purity of
95%. When this puri®ed enzyme (95% pure, 10 mg) was
subjected to the same puri®cation process as discussed
above, a 99% pure enzyme (8 mg) was obtained.
in a total of 100 mL of Bu€er A (100 mM potassium
phosphate bu€er, pH 7.4, containing 0.25 mM b-mer-
captoethanol) was incubated, while protected from
light, at 25ꢀC for 4h, and the reaction mixture was
assayed. By this time, less than 1% of the enzyme
activity remained. Two controls also were run simulta-
neously, one with no gabaculine and one with no a-
ketoglutarate or gabaculine, but containing GABA (40
mM). Excess inactivator was removed by running the
solutions over Sephadex G-50 packed columns pre-
washed with water, using the Penefsky spin method.23
The solutions obtained were again assayed; no return of
enzyme activity was detected. All three solutions
obtained were subjected to the following denaturation
process.
Preparation of [3H]PLP-reconstituted GABA-AT
GABA-AT (3 units, 6.8 nmol) was incubated with 194
mM GABA in bu€er A at room temperature in the dark
for 30 min. After being cooled to 4ꢀC, monobasic potas-
sium phosphate was added slowly to the incubation mix-
ture over a period of 10 min to make a 0.5 M, pH 5.5
KH2PO4 solution. This mixture was dialyzed ®rst against
1L of 0.5 M KH2PO4 bu€er, containing 0.25 mM b-mer-
captoethanol for 1 h, then against 4 L of 100 mM potas-
sium phosphate bu€er, pH 7.0, containing 0.25 mM b-
mercaptoethanol for 2 h at 4ꢀC. An assay showed less
than 1% activity remaining. This apoGABA-AT was
incubated with a 250-fold excess of [3H]PLP (30 mCi, 1.7
mmol, freshly puri®ed by HPLC) at room temperature
until maximum activity returned. Assays were performed
every hour to monitor the return of activity, which gen-
erally took about 4 h. About 60% of the activity returned.
The excess [3H]PLP was then removed from the recon-
stituted enzyme by running portions (200 mL) over
Sephadex G-50 packed columns prewashed with bu€er
A, using the Penefsky spin method.23 All of the enzyme
samples were combined and dialyzed against bu€er A
(2Â4 L, changed every 2 h) at 4ꢀC. The [3H]PLP-GABA-
AT obtained had a speci®c activity of 1.06 units/mL, a
protein concentration of 0.59 mg/mL, and a speci®c
radioactivity of 0.011 mCi/mg. The yield of the prepara-
tion was 39% (1.06 units, 0.59 mg; yield calculated based
on the enzyme activity).
The pH of the samples were adjusted to 12 using 1 M
KOH and were incubated at room temperature for 1 h.
Then enough tri¯uoroacetic acid (TFA) was added to
quench the base and make 10% v/v TFA solutions.
White denatured protein appeared in the solution after
the acid addition. After being allowed to stand at room
temperature for 10 min, the denatured enzyme solutions
were placed into Centricon 10 micro-concentrators and
were centrifuged for 15 min at 5,000 rpm with a Du
Pont Sorvall RC5B Plus centrifuge, using an SA600
rotor to achieve a complete separation of the protein
and the e‚uents. The protein pellets were rinsed with
0.1% aqueous TFA (50 mL), vortexed and then cen-
trifuged for a further 10 min. This process was repeated
three times. The protein pellets obtained were redis-
solved in water (1 mL) and were counted for radio-
activity. The e‚uent and rinses were combined and
lyophilized to dryness. The solid obtained from lyophi-
lization was dissolved in water (500 mL) to make a stock
solution. The ®nal sample was prepared by taking the
stock sample (100 mL) and adding a standard (20 mL,
containing 4 mM each of PLP, PMP, and 3) which had
been subjected to the same basi®cation and acidi®cation
steps. The samples were subjected to two di€erent
reverse phase HPLC systems. The ®nal sample (pH
adjusted to 1 by 10% aqueous TFA) was injected onto
an Alltech Alltima C18 column (4.6Â250 mm, 5mm).
Mobile phase A was 0.1% aqueous TFA ¯owing at 0.5
mL/min for 15 min. Then a 5-min gradient was run to
50% mobile phase B (80% aqueous acetonitrile). The
column was eluted with 50% mobile phase B for 20 min.
Under these conditions, PLP elutes at 16 min, PMP at 8
min, and CPPp at 36 min. The HPLC eluents were
analyzed for radioactivity with a Radiomatic FLO-
ONE\Beta Series A-200 liquid ¯ow scintillation counter.
After being calibrated with the data from the two con-
trols, the UV absorption at 254 nm and detected radio-
activity of the isolated gabaculine inactivation product
were plotted against the retention time. For the other
system, the column was isocratically eluted with 2%
aqueous CH3CN for 5 min then a 10 min gradient from
2 to 80% aqueous CH3CN was applied. The ¯ow rate
was 1 mL/min. Under these conditions, PLP elutes at
4.9 min, PMP at 3.5 min, and CPPp at 10.8 min. The
HPLC eluent was monitored at 214 nm and analyzed
for radioactivity with the FLO-ONE\Beta Series A-200
liquid ¯ow scintillation counter. After being calibrated
Time-dependent inactivation of GABA-AT by gabaculine
GABA-AT (10 mL; ®nal concentration 0.275 mM) was
added to solutions of gabaculine (90 mL; ®nal concentra-
tions of 0, 5, 6, 8, 10, 14, 16, 20 mM, seven pre-incubation
mixtures for each concentration) in bu€er A, containing 100
mM a-ketoglutarate at 25ꢀC. At timed intervals, aliquots (85
mL) were withdrawn and added to the assay solution (515
mL) containing excess succinic semialdehyde dehydrogenase
(5 mL). Rates were measured spectrophotometrically at 340
nm at 25ꢀC, and the logarithm of the remaining activity was
plotted against time for each concentration of inhibitor. A
secondary plot of 1/slope of these lines versus 1/[inacti-
vator]19 was constructed to determine KI and kinact values
for gabaculine. The inactivation also was carried out at 0ꢀC
under the same conditions except for di€erent gabaculine
concentrations (®nal concentrations of 200, 250, 300, 400,
500, 600, 700, 800, 900 mM).
Inactivation of [3H]PLP-GABA-AT by gabaculine,
denaturation, product isolation, and analysis by reverse
phase HPLC.
A solution of [3H]PLP-GABA-AT (0.59 mg/mL, 50 mL;
7400 dpm), a-ketoglutarate (5 mM), gabaculine (2 mM)

M. Fu, R. B. Silverman / Bioorg. Med. Chem. 7 (1999) 1581±1590
1589
with the data from the two controls, the UV absorption
at 214 nm and detected radioactivity of the isolated
gabaculine inactivation product were plotted against the
retention time.
600 in 1.99 s. For MS/MS analysis, argon was used as
the collision gas at a pressure of ca. 2.2Â10 3 mbar. The
collision energy was 16 eV. The mass range was scanned
from m/z 100 to 400.
UV±visible spectroscopic studies of the puri®ed
gabaculine inactivation product
NMR studies of the puri®ed gabaculine inactivation
product
The gabaculine inactivation product was prepared,
denatured, isolated, and puri®ed as described above
from an incubation mixture containing 1.2 mL (80
nmol) of GABA-AT, 10 mM a-ketoglutarate, and 2
mM gabaculine. After denaturation the gabaculine
inactivation product was puri®ed using the 0.1% TFA
to 50% aqueous acetonitrile elution system. The 35 to
37 min HPLC fractions were collected and freeze-dried.
The dried sample was then dissolved in 2 M NH4OH
(500 mL), placed in a Fisher 1 cm cuvette, and scanned
from 450 nm to 214 nm. Synthetic 3 (80 nmol) was also
dissolved in 2 M NH4OH (500 mL) and scanned under
the same conditions.
GABA-AT (2 mL, 160 nmol) was inactivated with
gabaculine (2 mM) in bu€er A containing 10 mM a-KG
at room temperature in the dark for 4 h. The same
inactivation, denaturation, and product isolation pro-
cedures described above for the UV±visible spectral
study were followed. The 35±37 min HPLC fractions
were collected and freeze dried. The same puri®cation
was applied to synthetic 3 (160 nmol) to serve as a con-
trol for analysis. The control and puri®ed product
samples were dissolved in 1 mL of D2O and freeze dried.
The D2O exchanged samples were each dissolved in 350
mL of D2O, placed in Schigemi BMS-005T NMR tubes,
1
and H NMR spectra were taken on a Varian Unity
Plus 400 NMR spectrometer.
Electrospray ionization mass spectroscopic (ESIMS)
studies of the highly puri®ed gabaculine inactivation
product
Acknowledgements
Care was taken in all of the experimental procedures to
eliminate the introduction of glycerol, polyethylene gly-
col, or other detergent related polymers into samples
intended for mass spectroscopic studies. All glassware
used was treated sequentially with 30% v/v nitric acid in
water, water, methanol, chloroform, methanol, and
water. The enzyme isolated from pig brain was further
puri®ed by dialysis against bu€er A (4Â4 L, changed
every 4 h). The dialysis tubing used, Spectra/por 2,
molecular porous dialysis membrane from Baxter, was
soaked in water (3Â2 L, changed every 4 h) before use.
The puri®ed GABA-AT (1.2 mL, 80 nmol) was inacti-
vated with gabaculine (2 mM) in bu€er A containing 10
mM a-ketoglutarate at room temperature in the dark
for 4 h. The same inactivation, denaturation, and pro-
duct isolation procedure described for the UV±visible
spectroscopic study was followed except that the 2±80%
aqueous acetonitrile elution system was used. The 10±
12.5 min HPLC fractions were collected and freeze-
dried. The same puri®cation procedure was applied to
synthetic 3 (80 nmol) to serve as a control for analysis.
The dried samples prepared above were dissolved in
50% v/v aqueous CH3CN (50 mL) giving solutions with
a product concentration of about 1.6 mM and a control
3 concentration of 1.6 mM. The pH of the solution was
brought to about 10 with the addition of 1 M NH4OH
for negative mode electrospray ionization mass spectro-
metry (ESIMS ) and to 4 by 1% formic acid addition
for positive mode electrospray ionization mass spectro-
metry (ESIMS+) analysis. The control synthetic 3 sam-
ple was run ®rst to ®nd the optimum conditions.
Aliquots of 10-mL each (16 nmol) were loop injected
into the mass spectrometer. All data were acquired
in the multichannel analysis (MCA) mode, and a CsI/
NaI mixture was used to calibrate the instrument. The
resolution of the instrument was 15.0 at both low and
high mass. The mass range was scanned from m/z 160 to
We are grateful to the National Institutes of Health
(NS15703) for ®nancial support of this research.
References
1. Mishima, H.; Kurihara, H.; Kobayashi, K.; Miyazawa, S.;
Terahara, A. Tetrahedron Lett. 1976, 537.
2. Rando, R. R.; Bangerter, F. W. J. Am. Chem. Soc. 1976, 98,
6762.
3. Rando, R. R.; Bangerter, F. W. Biochem. Biophys. Res.
Commun. 1977, 76, 1276.
4. Wood, J. D.; Kurylo, E.; Tsui, D. S.-K. Neurosci. Lett.
1979, 14, 327.
5. Behar, K. L.; Boehm, D. Magn. Reson. Med. 1994, 31, 660.
6. Watts, S. D. M.; Atkins, A. M. UCLA Symp. Mol. Cell.
Biol., New Ser. 1987, 60, 525.
7. Gammon, D. W.; Gingrich, H. L.; Sander, G.; Stewart, R.
R.; Van Der Werf, P. A. ACS Symp. Ser. 1987, 356, 122.
8. Prince, D. J.; Djamgoz, M. B. A.; Karten, H. J. Neurochem.
Int. 1987, 11, 23.
9. Palfreyman, M. G.; Schechter, P. J.; Buckett, W. R.; Tell,
G. P.; Koch-Weser, J. Biochem. Pharmacol. 1981, 30, 817.
10. Aurisano, N.; Bertani, A.; Reggiani, R. Phytochemistry
1995, 38, 1147.
11. Faingold, C. L.; Marcinczyk, M. J.; Casebeer, D. J.;
Randall, M. E.; Arneric, S. P.; Browning, R. A. Brain Res.
1994, 640, 40.
12. Michalik, M.; Nelson, J.; Erecinska, M. Diabetes 1993, 42,
1506.
13. Bernasconi, R.; Maitre, L.; Martin, P.; Raschdorf, F. J.
Neurochem. 1982, 38, 57.
14. Rando, R. R.; Bangerter, F. W. J. Am. Chem. Soc. 1977,
99, 5141.
15. Rando, R. R. Biochemistry 1977, 16, 4604.
16. Metcalf, B. W.; Jung, M. J. Mol. Pharmacol. 1979, 16, 539.
17. Soper, T. S.; Manning, J. M. J. Biol. Chem. 1982, 257,
13930.
18. Kitz, R.; Wilson, I. B. J. Biol. Chem. 1962, 237, 3245.

1590
M. Fu, R. B. Silverman / Bioorg. Med. Chem. 7 (1999) 1581±1590
19. Khomutov, R. M.; Karpeiskii, M. Y.; Severin, E. S. Chemi-
cal and Biological Aspect of Pyridoxal Catalysis. I. U. B. Sym-
posium Series; Pergamon Press: New York, 1963; pp 313±321.
20. Shah, S. A.; Shen, B. W.; Brunger, A. T. Structure 1997, 5,
1067.
21. Penefsky, H. J. Biol. Chem. 1977, 252, 2891.
22. Penefsky, H. Methods Enzymol. 1979, 56, 527.
23. Stock, A.; Ortanderl, F.; P¯eiderer, G. Biochem. Z. 1966,
344, 353.
24. Silverman, R. B.; Invergo, B. J. Biochemistry 1986, 25,
6817.
25. Hopkins, M. H.; Bichler, K. A.; Su, T.; Chamberlain, C.
L.; Silverman, R. B. J. Enzyme Inhib. 1992, 6, 195.
26. Scott, E. M.; Jakoby, W. B. J. Biol. Chem. 1958, 234, 932.
27. Churchich, J. E.; Moses, U. J. Biol. Chem. 1981, 256,
1101.
28. De Biase, D.; Barra, D.; Bossa, F.; Pucci, P.; John, R. A. J.
Biol. Chem. 1991, 266, 20056.