Mechanism of the Glutathione Transferase-Catalyzed Conversion of Antitumor 2-Crotonyloxymethyl-2-cycloalkenones to GSH Adducts

Article abstract of DOI:10.1021/ja030396p

Human glutathione (GSH) transferase (hGSTP1-1) processes with similar kinetic efficiencies the antitumor agents 2-crotonyloxymethyl-2-cyclohexenone (COMC-6), 2-crotonyloxymethyl-2-cycloheptenone (COMC-7), and 2-crotonyloxymethyl-2-cyclopentenone (COMC-5)

Full text of DOI:10.1021/ja030396p

Published on Web 11/12/2003
Mechanism of the Glutathione Transferase-Catalyzed
Conversion of Antitumor
2-Crotonyloxymethyl-2-cycloalkenones to GSH Adducts
Diana S. Hamilton,*,† Xiyun Zhang, Zhebo Ding,‡ Ina Hubatsch,§ Bengt Mannervik,§
K. N. Houk, Bruce Ganem,‡ and Donald J. Creighton*,†
Contribution from the Department of Chemistry and Biochemistry, UniVersity of Maryland
Baltimore County, Baltimore, Maryland 21250, Department of Chemistry and Chemical Biology,
Baker Laboratory, Cornell UniVersity, Ithaca, New York 14853-1301, Department of Chemistry
and Biochemistry, UniVersity of California, Los Angeles, California 90095-1569, and
Department of Biochemistry, Uppsala UniVersity, Biomedical Center, Box 576,
SE-751 23 Uppsala, Sweden
Received July 3, 2003; E-mail: hamilton@umbc.edu; creighto@umbc.edu
Abstract: Human glutathione (GSH) transferase (hGSTP1-1) processes with similar kinetic efficiencies
the antitumor agents 2-crotonyloxymethyl-2-cyclohexenone (COMC-6), 2-crotonyloxymethyl-2-cyclohep-
tenone (COMC-7), and 2-crotonyloxymethyl-2-cyclopentenone (COMC-5) to 2-glutathionylmethyl-2-cyclo-
hexenone, 2-glutathionylmethyl-3-glutathionyl-2-cycloheptenone, and 2-glutathionylmethyl-2-cyclopentenone,
respectively. This process likely involves initial enzyme-catalyzed Michael addition of GSH to the COMC
derivative to give a glutathionylated enol(ate), which undergoes nonstereospecific ketonization, either while
bound to the active site or free in solution, to a glutathionylated exocyclic enone. Free in solution, GSH
reacts at the exomethylene carbon of the exocyclic enone, displacing the first GSH to give the final product.
This mechanism is supported by the observation of multiphasic kinetics in the presence of high
concentrations of hGSTP1-1 and the ability to trap kinetically competent exocyclic enones in aqueous acid
using COMC-6 and COMC-7 as substrates. That the exocyclic enone is formed by nonstereospecific
ketonization of an enol(ate) species is indicated by the observation that COMC-6 (chirally labeled with
deuterium at the exomethylene carbon) gives stereorandomly labeled exocyclic enone. The isozymes
hGSTP1-1, hGSTA1-1, hGSTA4-4, and hGSTM2-2 catalyze the conversion of COMC-6 to final product
with similar efficiencies (Km ) 0.08-0.34 mM, kcat ) 1.5-6.1 s-1); no activity was detected with the rat
rGSTT2-2 isozyme. Molecular docking studies indicate that in hGSTP1-1, the hydroxyl group of Tyr108
might serve as a general acid catalyst during substrate turnover. The possible significance of these
observations with respect to the metabolism of COMC derivatives in multidrug resistant tumors is discussed.
Scheme 1
Introduction
Glutathione transferase (hGSTP1-1) has recently been re-
ported to play a potentially important role in the metabolism of
the antitumor agent 2-crotonyloxymethyl-2-cyclohexenone
(COMC-6, 1b).1,2 This compound is a simple synthetic ana-
logue3 of 2-crotonyloxymethyl-(4R,5R,6R)-4,5,6-trihydroxy-2-
cyclohexenone (COTC, 1a), an antitumor metabolite first
identified in Streptomyces in 1975 (Scheme 1).4
† University of Maryland Baltimore County.
University of California, Los Angeles.
The antitumor activity of COTC was originally hypothesized
to result from competitive inhibition of the methylglyoxal-de-
toxifying enzyme glyoxalase I by the GSH adduct 2a.4 However,
authentic samples of both 2a and 2b proved to be poor com-
petitive inhibitors of human glyoxalase I,1,5 with no significant
antitumor activity in vitro.2 Subsequently, the pi isozyme of
human glutathione transferase (hGSTP1-1) was discovered to
‡ Cornell University.
§ Uppsala University.
(1) Hamilton, D. S.; Ding, Z.; Ganem, B.; Creighton, D. J. Org. Lett. 2002, 4,
1209-1212.
(2) Joseph, E.; Eiseman, J. L.; Hamilton, D. S.; Wang, H.; Tak, H.; Ganem,
B.; Creighton, D. J. J. Med. Chem. 2003, 46, 194-196.
(3) Aghil, O.; Bibby, M. C.; Carrington, S. J.; Double, J.; Douglas, K. T.;
Phillips, R. M.; Shing, T. K. M. Anti-Cancer Drug Des. 1992, 7, 67-82.
(4) (a) Takeuchi, T.; Chimura, H.; Hamada, M.; Umezawa, H.; Yoshioka, O.;
Oguchi, N.; Takahashi, Y.; Matsuda, A. J. Antibiot. 1975, 28, 737-742.
(b) Chimura, H.; Nakamura, H.; Takita, T.; Takeuchi, T.; Umezawa, H.;
Kato, K.; Saito, S.; Tomisawa, T.; Iitaka, Y. J. Antibiot. 1975, 28, 743-
748.
(5) Huntley, C. F. M.; Hamilton, D. S.; Creighton, D. J.; Ganem, B. Org. Lett.
2000, 2, 3143-3144.
9
10.1021/ja030396p CCC: $25.00 © 2003 American Chemical Society
J. AM. CHEM. SOC. 2003, 125, 15049-15058
15049

A R T I C L E S
Hamilton et al.
Scheme 2
Detailed descriptions of the preparation of 1b-d have been
published.1,2 The GSH conjugates 2b1, 2c, and 2d were prepared by
separately incubating 1b (26 mM), 1c (117 mM), and 1d (64 mM)
with equimolar concentrations of GSH in phosphate buffer at pH 7.4
(37 °C) for 20 min, 40 and 48 h, respectively. The crude products were
purified by anion-exchange chromatography (Dowex-1).5
2-Glutathionylmethyl-2-cyclohexenone (2b). (Yield 99%). 300
1
MHz H NMR (D2O): δ 7.13 (t, J ) 4.3 Hz, vinyl H), 4.50 (q, J )
Chart 1
4.8, 8.6 Hz, Cys-CRH), 3.95 (s, Gly-CRH2), 3.94 (t, J ) 6.4 Hz, Glu-
CRH), 3.26 (s, CH2), 2.93 (q, J ) 4.8, 14.5 Hz, Cys-CâHa), 2.75 (q, J
) 8.6, 14.5 Hz, Cys- CâHb), 2.51 (m, Glu-CγH2), 2.45-2.38 (m, ring
2CH2), 2.16 (m, Glu-CâH2), 1.98-1.92 (m, ring CH2). 13C NMR
(D2O): δ 203.3, 176.6, 174.4, 172.9, 172.7, 153.0, 134.2, 52.9, 52.7,
41.1, 37.8, 32.5, 30.9, 30.0, 25.7, 25.6, 22.2. ESI-MS m/z 416 (M + 1,
100%): 287 (9%).
2-Glutathionylmethyl-2-cyclopentenone (2c). (Yield 83%). 300
MHz 1H NMR (D2O, HOD ref): δ 6.92 (m, vinyl H), 4.52 (q, J ) 4.8,
8.4 Hz, Cys-CRH), 3.95 (s, Gly-CR H2), 3.80 (t, J ) 6.4 Hz, Glu-CRH),
3.33 (s, CH2), 2.96 (q, J ) 4.8, 13.9 Hz, Cys-CâHa), 2.78 (q, J ) 8.4,
13.9 Hz, Cys-CâHb), 2.66 (m, ring CH2), 2.53-2.45 (m, Glu-CγH2,
ring CH2), 2.14 (m, Glu-CâH2). 13C NMR (D2O): δ 211.5, 173.9, 171.9,
170.4, 169.9, 163.3, 137.8, 50.7, 50.1, 38.5, 32.2, 29.8, 28.4, 24.2, 23.2,
22.1, 17.6. ESI-MS m/z 402 (M + 1, 100%): 273 (18%).
catalyze the conjugation of GSH to COMC-6 (Scheme 1).1
Preliminary kinetic studies and intermediate trapping experi-
ments suggested that the minimum kinetic mechanism involved
enzyme-catalyzed Michael addition of GSH to COMC-6 to give
an electrophilic exocyclic enone (3b) that reacts with free GSH
to give 2b (Scheme 2).
2-Glutathionylmethyl-2-cycloheptenone (2d). (Yield 38%). 300
MHz 1H NMR (D2O, HOD ref): δ 7.80 (t, J ) 6.3 Hz, vinyl H), 4.52
(q, J ) 4.8, 8.4 Hz, Cys-CRH), 3.97 (s, Gly-CRH2), 3.90 (t, J ) 6.2
Hz, Glu-CRH), 3.35 (s, CH2), 2.96 (q, J ) 5.1, 13.9 Hz, Cys-CâHa),
2.77 (q, J ) 8.8, 13.9 Hz, Cys-CâHb), 2.60 (m, ring CH2), 2.56-2.43
(m, Glu-CγH2, ring CH2), 2.17 (m, Glu-CâH2), 1.76-1.68 (m, ring
2CH2). 13C NMR (D2O): δ 206.1, 171.8, 170.2, 170.0, 146.9, 135.1,
113.1, 111.8, 58.8, 50.3, 50.2, 39.7, 38.4, 30.7, 29.6, 28.3, 24.6, 23.0,
21.6, 18.1. ESI-MS m/z 430 (M + 1, 100%): 402 (34%), 102 (67%).
Synthesis of Chiral Deuterium-Labeled COMC Derivatives. The
deuterium-labeled substrates (R)d1-1b and (S)d3-1b were prepared by
using HLADH to stereospecifically introduce different isotopes of
hydrogen at the pro-R position of the exomethylene function of 5b,
followed by crotonylation of the labeled alcohol. Thus, (R)d1-1b was
prepared as outlined in Scheme 4.
The stereoisomer (S)d3-1b, in which the configuration at the deu-
terium labeled exomethylene carbon is S, was synthesized by the same
procedure, using d4-5b and ethanol as starting materials. Compound
d4-5b was prepared by electrophilic addition of dideuterioformaldehyde
(in D2O) to cyclohexenone following a literature procedure.9 Under
the reaction conditions, the C6 methylene undergoes complete deuterium
exchange with solvent.
Progress of the hydrogen-exchange reactions was followed by
monitoring the time-dependent change in the integrated intensities of
the exomethylene protons of substrates. Incubation times of two to four
weeks, in the presence of 80 units of HLADH (pH 7, 20 °C), were
required to effect >25% isotope exchange. No detectable exchange
was observed in the absence of HLADH. This enzyme is known to
catalyze stereospecific hydrogen transfer between the pro-R hydrogen
of primary alcohols and C4 of the nicotinamide ring of NAD, as
reviewed elsewhere.10 Therefore, alcohol 5b was predicted to be
stereospecifically processed by HLADH.
Thus, the antitumor activities of COTC and COMC-6 might
reflect the reaction of the exocyclic enones with proteins and/
or nucleic acids critical to cell viability.6 From a pharmacological
perspective, a substrate that is transformed to a toxic product
by glutathione transferase (GST) is of considerable interest as
a possible means of inhibiting multidrug-resistant tumors, which
often overexpress specific isozymes of GST.7
To better understand the chemical mechanism of this process
and to identify possible active-site residues involved in catalysis,
the kinetic properties, reaction stereochemistry, and isozyme
specificities of COMC-6 and its five- and seven-membered-
ring homologues have been determined. The results of these
studies, together with molecular docking experiments, suggest
that the COMC derivatives are processed by a mechanism
analogous to that for the GST-catalyzed addition of GSH to
the enone ethacrynic acid (Chart 1).
Experimental Section
Materials. Deuterated reagents, R-(-)-R-methoxy-R-trifluorometh-
ylphenyl acid chloride (R-MTPA chloride), human placental GST
(predominantly the hGSTP1-1 isoform), recombinant hGSTP1-1, and
horse liver alcohol dehydrogenase (HLADH) were purchased from
Sigma-Aldrich Chemical Co. Salts and free GSH were removed from
human placental GST by ultrafiltration prior to use. The isolation and
purification of other recombinant isoforms of GST used in this study
have been described elsewhere.8
Synthesis of COMC Derivatives. The synthesis of the 2-crotonyl-
oxymethyl-2-cycloalkenones 1b-d utilized the known Baylis-Hillman
reaction of commercially available 2-cycloalkenones 4b-d with
formaldehyde to prepare the 2-hydroxymethyl-2-cycloalkenones
5b-d,9 which were then treated with crotonic anhydride3 to give 1b-d
(Scheme 3).
Stereochemical Analysis. The stereospecificities of the HLADH-
1
catalyzed exchange reactions were verified by H NMR spectroscopy
of diastereomeric derivatives of the deuterium labeled products. First,
the product mixtures used to prepare (R)d1-5b and (S)d3-5b were
converted to the esters of the (S)-(-) enantiomer of Mosher’s acid (R-
methoxy-R-trifluoromethylphenylacetic acid, MTPA) using the method
of Ward and Rhee.11 In CDCl3, the diastereotopic exomethylene protons
of unlabeled 5b are observed as an AB quartet at δ ) 4.972 and 4.954.
(6) Zhang, Q.; Ding, Z.; Creighton, D. J.; Ganem, B.; Fabris, D. Org. Lett.
2002, 4, 1459-1462.
(7) Hayes, J. D.; Pulford, D. J. Crit. ReV. Biochem. Mol. Biol. 1995, 30, 445-
600.
(8) (a) Stenberg, G.; Bjo¨rnestedt, R.; Mannervik, B. Protein Expression Purif.
1992, 3, 80-84. (b) Hubatsch, I.; Ridderstro¨m, M.; Mannervik, B. Biochem.
J. 1998, 330, 175-179. (c) Kolm, R. H.; Stenberg, G.; Widersten, M.;
Mannervik, B. Protein Expression Purif. 1995, 6, 265-271. (d) Johansson,
A.-S.; Bolton-Grob, R.; Mannervik, B. Protein Expression Purif. 1999, 17,
105-112. (e) Jemth, P.; Stenberg, G.; Chaga, G.; Mannervik, B. Biochem.
J. 1996, 316, 131-136.
(10) (a) Creighton, D. J.; Murthy, N. S. R. K. The Enzymes 1990, 19, 324-
421. (b) Hummel, W. Trends Biotechnol. 1999, 17, 487-492.
(11) Ward, D. E.; Rhee, C. K. Tetrahedron Lett. 1991, 32, 7165-7166.
(9) Rezgui, F.; El Gaied, M. M. Tetrahedron Lett. 1998, 39, 5965-5966.
9
15050 J. AM. CHEM. SOC. VOL. 125, NO. 49, 2003

Processing Cycloalkenones by Glutathione Transferase
A R T I C L E S
Scheme 3. Synthetic Route to COMC Derivatives and Their GSH Adductsa
a
(a) DMAP, CH2O, THF, rt. (b) Crotonic anhydride, pyridine, DMAP, rt. (c) GSH, phosphate buffer (pH 7.4), 37 °C.
Scheme 4. Preparation of Chirally Labeled (R)d1-1b
The upfield and downfield resonances are assumed to correspond to
the pro-S and pro-R protons, respectively, on the basis of chemical
shift comparisons with MTPA esters of known configuration.12 The
1H NMR spectrum of the MTPA ester of the product mixture (composed
of 50% 5b and 50% (R)d1-5b), obtained by the method of Scheme 4,
shows a quartet characteristic of 5b and a singlet (δ ) 4.929)
corresponding to (R)d1-5b, indicating an isotope shift for the Hs proton
1
of 0.025 ppm (Figure 1a). The H NMR spectrum of the MTPA ester
of the product mixture containing (S)d3-5b shows a singlet at δ ) 4.958,
indicating an isotope shift for the Hr proton of 0.014 ppm (Figure 1b).
These spectra are similar to those of the (S)-MTPA esters of pro-R
and pro-S D-labeled benzyl alcohols.12b The NMR spectra of Figure 1
indicate that (S)d3-5b and (R)d1-5b are isomerically pure, and therefore
crotonylation of these compounds will give isomerically pure (S)d3-1b
and (R)d1-1b.
2-(R)-Deuteriohydroxymethyl-2-cyclohexenone ((R)d1-5b): pre-
pared by incubating 5b (3.6 mmol, 50 mM), HLADH (80 U), NAD
(10 mM) and d5-ethanol (250 mM) for 12 days in 0.1 M HEPES, pH
7.2, 20 °C. The reaction mixture was diluted with brine and extracted
with 3 portions of dichloromethane in the presence of a small amount
of sodium dodecyl sulfate to break the emulsion. The combined organic
layers were washed with water and dried over sodium sulfate, and the
solvent was removed in vacuo to give a colorless liquid which yellowed
slightly on standing. (Yield 75%). 1H NMR analysis indicated that the
Figure 1. Exomethylene region in the 1H NMR spectra (800 MHz, CDCl3/
TMS) of the Mosher’s esters of the product mixture (a) containing 50% 5b
and 50% (R)d1-5b and (b) (S)d3-5b.
product was composed of an approximate 1:1 mixture of 5b and
d1-5b, on the basis of the 25% decrease in the integrated intensity of
the hydroxymethylene proton resonance (δ 4.24). For the product
1
mixture: 300 MHz H NMR (CDCl3, TMS): δ 6.94 (t, J ) 4.0 Hz,
same procedure used to prepare (R)d1-5d to give a colorless liquid.
(Yield 80%). 1H NMR analysis indicated that the product was composed
of an approximate 1:1 mixture of d4-5b and d3-5b, on the basis of the
integrated intensity of the hydroxymethylene proton resonance (δ 4.24).
vinyl H), 4.24 (s, 1.5 H, exocyclic CH2 and CHD), 2.53-2.38 (m, ring
2CH2), 2.06-1.98 (m, ring CH2), 1.61 (s, OH).
2-Dideuteriohydroxymethyl-6,6-dideuterio-2-cyclohexenone
(d4-5b): prepared by reacting 4b with a commercial preparation of
dideuterioformaldehyde (supplied in D2O) under the conditions used
to prepare 5b.9 Under these conditions, complete deuterium exchange
occurs with the methylene protons at C6, as these resonances are absent
from the 1H NMR spectrum. (Yield 18%). 300 MHz 1H NMR (CDCl3,
TMS): δ 6.95 (t, J ) 4.2 Hz, vinyl H), 2.44-2.38 (m, ring CH2), 2.01
(t, J ) 5.7 Hz, ring CH2), 1.64 (s, OH).
(S)-2-Deuteriohydroxymethyl-6,6-dideuterio-2-cyclohexenone ((S)-
d3-5b): prepared by incubating d4-5b (3.6 mmol, 50 mM), HLADH
(80 U), NAD (10 mM), and ethanol (250 mM) for 35 days in 0.1 M
HEPES pH 7.2, 20 °C. The reaction mixture was worked up by the
1
300 MHz H NMR (CDCl3, TMS): δ 6.94 (t, J ) 4.0 Hz, vinyl H),
4.22 (s, 0.5 H, exocyclic CD2 and CHD), 2.44-2.38 (m, ring CH2),
2.01 (t, J ) 5.7 Hz, ring CH2), 1.62 (s, OH).
(R)-2-Crotonyloxydeuteriomethyl-2-cyclohexenone ((R)d1-1b): pre-
pared by treating the (R)d1-5b/5b mixture, obtained as described above,
with crotonic anhydride3 to give a colorless liquid. (Yield 69%). 300
1
MHz H NMR (CDCl3, TMS): δ 7.04-6.94 (m, 2 vinyl H), 5.86 (d,
J ) 15.4 Hz, vinyl H), 4.81 (s, ∼1.5 H, exocyclic CH2 and CHD),
2.49-2.39 (m, ring 2CH2), 2.06-1.98 (m, ring CH2), 1.90 (d, J ) 7.0
Hz, CH3).
(S)-2-Crotonyloxydeuteriomethyl-6,6-dideuterio-2-cyclohex-
enone ((S)d3-1b): prepared by treating the d4-5b/(S)d3-5b mixture,
obtained as described above, with crotonic anhydride3 to give a colorless
liquid. (Yield 59%). 300 MHz 1H NMR (CDCl3, TMS): δ 7.04-6.94
(12) (a) Dale, J. A.; Mosher, H. S. J. Am. Chem. Soc. 1973, 95, 512. (b) Wolfe,
S.; Yang, K.; Weinberg, N.; Shi, Z.; Hsieh, Y.-H.; Sharma, R. D.; Ro, S.;
Kim, C.-K. Chem.-Eur. J. 1998, 4, 886-902.
9
J. AM. CHEM. SOC. VOL. 125, NO. 49, 2003 15051

A R T I C L E S
Hamilton et al.
(m, 2 vinyl H), 5.86 (dq, J ) 2.6, 15.8 Hz, vinyl H), 4.79 (s, ∼0.5 H,
exocyclic CD2 and CHD), 2.49-2.39 (m, ring CH2), 2.05-1.99 (m,
ring CH2), 1.86 (dd, J ) 1.8, 7.0 Hz, CH3).
Kinetic Measurements. For each conjugation reaction, kinetic
measurements were carried out near a wavelength of maximum
difference absorbency of products minus reactants: COMC-6 (1b),
∆ꢀ235 -2400 cm-1 M-1; COMC-5 (1c), ∆ꢀ225 -6600 cm-1 M-1 or
∆ꢀ243 +1600 cm-1 M-1; and COMC-7 (1d), ∆ꢀ235 -3500 cm-1 M-1
.
The kinetic constants associated with the GST-catalyzed conjugation
of GSH with 1b-d were obtained from reciprocal plots of initial rates
(∆OD/min) versus [substrate], pH 6.5, [GSH] ) 1 mM (25 °C), under
conditions where the enzyme-catalyzed step is rate limiting (<0.08 units
GST in the assay cuvette). Units of GST activity were defined using
1-chloro-2,4-dinitrobenzene (CDNB) as substrate. Kinetic constants for
different recombinant GSTs were obtained by curve-fitting the Michae-
lis-Menten equation to the initial rate data.
Figure 2. Rate profile for the reaction of COMC-6 (0.05 mM) with GSH
(1.03 mM). Conditions: Potassium phosphate buffer (0.1 M, pH 6.5), EDTA
(0.05 mM), 25 °C.
Molecular Docking. The structures of ethacrynic acid (EA), the
GS-EA adduct, COMC-6 (1b), and the enolates formed from the
addition of GSH to COMC-6 were optimized using MM3 or AMBER
force fields in MacroModel 7.1.13 HF/6-31G* charges were assigned
to atoms of all of the structures, and the program AutoDock14 was used
to predict preferred binding modes. The X-ray crystal structures of
hGSTP1-1 in complex with EA or with GS-EA (PDB codes 2GSS
and 3GSS, respectively) were used in docking simulations after
removing all crystallographic water molecules and EA or GS-EA from
the active sites. The Lamarckian genetic algorithm was used in certain
AutoDock calculations.14 One hundred twenty simulations were per-
formed for each system. Each simulation was composed of a maximum
of 500 000 energy evaluations and a maximum of 50 000 generations.
Both rigid docking (deleting all rotatable torsion angles) and flexible
docking (allowing all rotatable torsion angles) were applied.
Figure 3. Rate profile for the reaction of COMC-6 (0.05 mM) with GSH
(1.03 mM) in the presence of placental hGSTP1-1 (2.4 CDNB units). Rate
constants were obtained as best-fit values to the equation OD235 ) OD235(t)∞)
+ a(exp(-k1′t)) + b(exp(k2′t)). The inset shows the exponential rise in OD235
due to the reaction of the isolated exocyclic enone with GSH (1.04 mM).
Conditions: Potassium phosphate buffer (0.1 M, pH 6.5), EDTA (0.05 mM),
25 °C.
Results
Kinetic Studies. The antitumor activity of COMC-6 (1b) has
been proposed to arise from a reactive exocyclic enone (3b)
formed during the conjugation reaction between intracellular
GSH and COMC-6 (Scheme 2).1 Most recently, the five- and
seven-membered-ring homologues (COMC-5 (1c) and COMC-7
(1d)) were also found to be potent inhibitors of B-16 melanotic
melanoma in vitro with IC50 values (144 and 29 nM, respec-
tively) similar to that of COMC-6 (46 nM).2 To test whether
the antitumor activities of the homologues could be explained
on the same basis as that of COMC-6, the kinetic properties of
all three compounds, either in the presence or in the absence of
human placental GST (predominantly the hGSTP1-1 isoform),
were compared. The placental and recombinant hGSTP1-1
enzyme preparations gave identical kinetic constants with the
COMC derivatives.
rate constant associated with the second, upward phase of the
reaction is independent of enzyme concentration.
These observations are consistent with enzyme-catalyzed
formation of a reactive species that subsequently dissociates
from the enzyme and reacts with GSH in bulk solvent to give
GSMC-6. Brief incubation of COMC-6, GSH, and high
concentrations of recombinant hGSTP1-1 gave rise to a transient
species that was isolated by reverse-phase HPLC and confirmed
to be an exocyclic enone by NMR spectroscopy.1 Incubation
of the exocyclic enone with GSH gave a species that co-migrated
with authentic GSMC-6 during HPLC. Moreover, the spectro-
photometrically determined pseudo-first-order rate constant for
reaction of the exocyclic enone with GSH under the same buffer
conditions was similar in magnitude to the second phase of the
reaction of COMC-6 with GSH in the presence of hGSTP1-1
(Figure 3, inset). The absorptivity of the exocyclic enone (ꢀ )
4300 cm-1 M-1) was significantly less than that of GSMC-6 (ꢀ
) 7500 cm-1 M-1), which accounts for the overall shape of
the reaction rate profile in the presence of hGSTP1-1.
In the absence of hGSTP1-1, the reaction of COMC-6 with
excess GSH follows a first-order decay, showing no evidence
of an intermediate species (Figure 2).
However, in the presence of placental hGSTP1-1 the reaction
rate profile conforms to a double exponential decay (Figure 3).
The magnitude of the rate constant associated with the initial
downward phase, after subtracting the rate constant for the
nonenzymatic addition of GSH with COMC-6, is linearly
dependent on the concentration of hGSTP1-1 in the range 0 to
2.4 (CDNB) units of enzyme activity. The magnitude of the
The five-membered-ring compound COMC-5 (1c) displayed
biphasic kinetic properties similar to those of COMC-6 (data
not shown). In the absence of hGSTP1-1, the reaction of excess
GSH with COMC-5 followed a simple first-order decay, but in
the presence of high concentrations of placental hGSTP1-1 the
reaction rate profile conformed to a double exponential function.
(13) Mohamadi, F.; Richard, N. G. J.; Liskamp, W. C.; Lipton, M.; Canfield,
C.; Chang, G.; Hendrickson, T.; Still, W. C. J. Comput. Chem. 1990, 11,
440.
(14) Morris, G. M.; Goodsell, D. S.; Halliday, R. S.; Huey, R.; Hart, W. E.;
Belew, R. K.; Olson, A. J. J. Comput. Chem. 1998, 19, 1639-1662.
9
15052 J. AM. CHEM. SOC. VOL. 125, NO. 49, 2003

Processing Cycloalkenones by Glutathione Transferase
A R T I C L E S
As with COMC-6, the magnitude of the rate constant associated
with the initial downward phase, after subtracting the rate
constant associated for the nonenzymatic addition of GSH to
COMC-5, was linearly dependent on the concentration of
hGSTP1-1 in the range 0 to 1.0 CDNB units of enzyme activity.
As for COMC-6, the magnitude of the rate constant associated
with the second, upward phase of the reaction was independent
of enzyme concentration. These observations are consistent with
the formation of an exocyclic enone that dissociates from the
enzyme and subsequently reacts with GSH in bulk solvent to
give GSMC-5 (2c). Brief incubation of COMC-5, GSH, and
recombinant hGSTP1-1 gave rise to a transient species that was
partially purified by reverse-phase HPLC (8% aqueous methanol
with 0.25% acetic acid). Incubation of this species with GSH
in buffered solution gave a species that co-migrated with
authentic GSMC-5 by HPLC. However, the instability of the
transient species precluded the isolation of enough material for
reliable kinetic and NMR characterization. Therefore, this
intermediate could not be confirmed to be an exocyclic enone.
The reactions of the seven-membered-ring compound COMC-7
(1d) with GSH are more complex. In the absence of hGSTP1-
1, the reaction of COMC-7 with excess GSH in aqueous buffer
was best fit using a two-phase exponential decay, implying at
least two sequential reactions (Figure 4).
Figure 4. Rate profile for the reaction of COMC-7 (1d, 0.05 mM) with
GSH (0.99 mM). The rate constants were obtained as best-fit values to the
equation OD235 ) OD235(t)∞) + a(exp(-k1t)) + b(exp(-k3t)). Conditions:
Potassium phosphate buffer (0.1 M, pH 6.5), EDTA (0.05 mM), 1% ethanol,
25 °C.
As a function of time, aliquots from the reaction mixture were
analyzed by reverse-phase HPLC (Waters µBondapak C18, 7.8
mm i.d. × 30 cm, 10 µm) using aqueous 20% methanol
containing 0.25% acetic acid, as an eluting solvent. The initial
phase of the reaction was characterized by the progressive
increase in the intensity of a new peak that co-migrated with
authentic GSMC-7 (2d, 21 min). At long incubation times this
peak progressively decreased in intensity with a corresponding
increase in the intensities of at least three different species with
retention times of 12, 13, and 14 min. ESI FAB MS showed
Figure 5. Rate profile for the reaction of COMC-7 (1d, 0.05 mM) with
GSH (1.05 mM), in the presence of human placental hGSTP1-1 (1 CDNB
unit). The rate constants given in the figure are best-fit values to the equation
OD235 ) OD235(t)∞) + a(exp(-k1′t)) + b(exp(k2′t)) + c(exp (-k3′t)). The
inset shows the exponential rise in OD235 due to the reaction of the isolated
exocyclic enone with GSH (0.99 mM). Conditions: Potassium phosphate
buffer (0.1 M, pH 6.5), EDTA (0.05 mM), 5% ethanol, 25 °C.
that each species gave identical molecular ions (M + Na+
)
760), consistent with the following structure:
NMR spectroscopy confirmed the identity of an exocyclic enone.
Incubation of this species with GSH in buffered solution gave
a species that co-migrates with authentic GSMC-7 by HPLC.
Moreover, the spectrophotometrically determined pseudo-first-
order rate constant for the reaction of the exocyclic enone (ꢀ )
3200 cm-1 M-1) with GSH to give GSMC-7 (ꢀ ) 5000 cm-1
M-1) was similar in magnitude to the second phase of the
reaction of COMC-7 with GSH in the presence of hGSTP1-1
(Figure 5, inset). The slow decrease in absorbance with a rate
constant k3′[GSH] corresponds closely to k3[GSH] for conver-
sion of GSMC-7 to the diglutathione adduct (GS)2MC-7 in the
absence of hGSTP1-1 (Figure 4).
Therefore, the product species are probably different diastere-
omers of the diglutathione adduct (GS)2MC-7, which differ with
respect to the configurations at C2 and C3.
The progress curve for the reaction of COMC-7 with GSH
in the presence of high concentrations of placental hGSTP1-1
could be fit with a triple exponential function (Figure 5).
The magnitude of the observed rate constant for the initial,
rapid phase (after subtracting the rate constant for the nonen-
zymatic reaction) is linearly dependent on enzyme concentration
from 0.1 to 1.0 CDNB units of placental hGSTP1-1. The rate
constant k2′ appears to be associated with the conversion of the
exocyclic enone to GSMC-7. Indeed, brief incubation of COMC-
7, GSH, and recombinant hGSTP1-1 gave rise to a transient
species that was isolated by reverse-phase HPLC, with a
A summary of the nonenzymatic kinetic constants is given
in Table 1.
Under conditions where the enzyme-catalyzed step is rate-
limiting (<0.08 CDNB units GST), the rate constants for
enzyme-catalyzed conjugation of GSH with the three different
COMC derivatives were determined and compared with those
for the uncatalyzed conjugation reaction obtained under the same
buffer conditions (Table 2).
Stereochemistry Studies. To determine the stereochemistry
of the conversion of COMC-6 to the exocyclic enone, COMC-6
1
retention time (26 min) close to that of GSMC-7 (21 min). H
9
J. AM. CHEM. SOC. VOL. 125, NO. 49, 2003 15053

A R T I C L E S
Hamilton et al.
Table 1. Comparison of the Second-Order Rate Constants for
Reaction of GSH with Different COMC Derivatives (k1) and with
the Corresponding Exocyclic Enones (k2)a
The deuterium-labeled substrates(R)d1-1b and (S)d3-1b were
prepared by using HLADH to stereospecifically introduce
different isotopes of hydrogen at the pro-R position of the
exomethylene function of 5b, followed by crotonylation of the
labeled alcohol (Experimental Section). Using either (S)d3-1b
or (R)d1-1b as substrates for hGSTP1-1 under conditions where
>80% of the substrate is processed by the enzyme, we found
that the ratio of the integrated intensities of the upfield versus
downfield resonances of the geminal vinylic hydrogens of the
isolated exocyclic enone was 0.89 ( 0.09 and 1.01 ( 0.1,
respectively (e.g., Figure 6).
derivative
k1, mM-1 min-1 (n)
k2, mM-1 min-1 (n)
COMC-5
COMC-6
COMC-7
0.102 ( 0.006 (5)
0.067 ( 0.003 (5)
0.015 ( 0.002 (5)
ndb
0.861 ( 0.042 (4)
0.063 ( 0.006 (7)
a
Conditions: COMC-X (∼0.05 mM), GSH (∼1.0 mM), potassium
phosphate buffer (0.1 M), pH 6.5, EDTA (0.05 mM), ethanol (0-5%), 25
°C. nd, not determined.
b
Table 2. Kinetic Constants for the Human Placental HGSTP1-1
Catalyzed Conjugation of GSH with Different COMC Derivativesa
A similar ratio was obtained (1.05 ( 0.11) under conditions
where the reaction mixture contained a 5-fold excess of GSH
over (R)d1-1b. The fact that stereorandomly D-labeled exocyclic
enone is observed independent of the concentration of GSH in
the reaction mixture argues against equilibration of the Z and
E stereoisomers after their formation by the enzyme. Thus, the
results indicate the absence of strict stereochemical control over
the conversion of the glutathionylated enol(ate) to the exocyclic
enone.
compound
Km, mM
kcat, min-1
(kcat/Km) × 10-3, min-1 mM-1
COMC-5
COMC-6
COMC-7
0.07 ( 0.02
0.05 ( 0.01
0.18 ( 0.02
70 ( 20
70 ( 20
180 ( 50
1.0 ( 0.4
1.4 ( 0.4
1.0 ( 0.3
a
Conditions: COMC-X (∼0.05 mM), GSH (∼1.0 mM), potassium
phosphate buffer (0.1 M), pH 6.5, EDTA (0.05 mM), ethanol (0-5%),
25 °C.
Isozyme Specificity for COMC-6. To evaluate isozyme
specificity, the steady-state kinetic constants of COMC-6 with
different enzyme variants of the alpha, mu, pi, and theta class
GSTs were determined (Table 3).15
Enzyme assays were carried out under conditions where the
enzyme-catalyzed step is completely rate-limiting (see Experi-
mental Section). The kcat and kcat/Km values for COMC-6 with
the alpha, pi, and mu class enzymes are within a factor of 4 of
one another. The kcat values for COMC-6 and ethacrynic acid
differ by a factor of about 4 using the recombinant hGSTP1-1
isoform. This is the smallest difference found for these two
substrates among all of the GSTs tested. In addition, human
placental hGSTP1-1, composed of the V105 and I105 variants,16
was also found to catalyze the conversion of COMC-6 to
GSMC-6 with kinetic constants (kcat ) 1.2 ( 0.2 s-1, Km
)
Figure 6. Vinyl region in the 1H NMR spectrum (500 MHz, D2O, ref
HOD; solvent suppression applied) of a reaction mixture composed of
hGSTP1-1 (1.7 units), (S)d1-1b (0.83 mM), GSH (0.17 mM) and potassium
phosphate buffer (33 mM) in D2O, pD 6.5, 21 °C. The reaction mixture
(0.6 mL) was stopped after 10 min incubation by the addition of
trichloroacetic acid to 2% final concentration and extracted with CDCl3 to
remove excess COMC-6 before analysis. The exomethylene protons Ha and
Hb occur at δ 5.79 and δ 5.32, respectively. The triplet at δ 7.22 corresponds
to the endocyclic vinyl proton resonance of 2b; the multiplets at δ 7.10
and 5.93 are due to the vinyl protons of crotonic acid.
0.052 ( 0.010 mM) similar in magnitude to those obtained with
ethacrynic acid (kcat ) 1.0 ( 0.03 s-1, Km ) 0.034 ( 0.007
mM) under the same assay conditions (1 mM GSH, pH 6.5, 25
°C). In contrast, there are significant differences in the efficien-
cies with which COMC-6 and ethacrynic acid are processed by
the alpha and mu class enzymes. The kcat values for COMC-6
are about 35-fold and about 70-fold larger than those for
ethacrynic acid with hGSTA1-1 and hGSTM2-2, respectively.
The theta class rGSTT2-2 shows no activity with either
COMC-6 or ethacrynic acid.
Molecular Docking Studies. Docking studies were under-
taken in an attempt to identify possible catalytic residues in the
active site that could promote the Michael addition reaction
between GSH and COMC-6. The AutoDock 3.0 program was
used14 because preliminary experiments showed that this
program could very precisely reproduce the protein-ligand
interactions observed in the X-ray crystal structure of the GS-
(stereospecifically labeled with deuterium at the exomethylene
carbon) was incubated in the presence of high concentrations
of hGSTP1-1. The resultant exocyclic enone was rapidly trapped
in acid solution, and the ratio of the Z and E isomers was
determined by 1H NMR spectroscopy, as detailed in the legend
to Figure 6. We reasoned that if conversion of the putative
glutathionylated enol(ate) to the exocyclic enone takes place in
the asymmetric environment of the active site, either the Z or
E isomer of product would predominate, depending on whether
the preferred conformation of the enzyme-bound enol(ate)
involves a syn or anti relationship between the crotonate-leaving
group and the GS moiety (e.g., Scheme 5).
(15) (a) Mannervik, B.; Awasthi, Y. C.; Board, P. G.; Hayes, J. D.; Dillio, C.;
Ketterer, B.; Listowsky, L.; Morgenstern, R.; Muramatsu, M.; Pearson, W.
R.; Pickett, C. B.; Sato, K.; Widersten, M. Biochem. J. 1992, 282, 305-
306. (b) Montali, J. A.; Wheatley, J. B.; Schmidt, D. E., Jr. Cell. Pharmacol.
1995, 2, 241. (c) Kauvar, L. M. In Structure and Function of Glutathionyl
Transferases; Tew, K., Picket, C., Mantle, T., Mannervik, B., Hayes, J.,
Eds.; CRC Press: Boca Raton, FL, 1993; p 251.
(16) (a) Zimniak, P.; Nanduri, B.; Pikula, S.; Bandorowicz-Pikula, J.; Singhal,
S.; Srivastava, S. K.; Awasthi, S.; Awasthi, Y. C. Eur. J. Biochem. 1994,
224, 893-899. (b) Johansson, A. S.; Stenberg, G.; Widersten, M.;
Mannervik, B. J. Mol. Biol. 1998, 278, 687-698.
Alternatively, if the active site does not exert strict stereo-
chemical control over the elimination reaction or if elimination
takes place after dissociation of the enol(ate) from the active
site (Scheme 6), similar (although not necessarily equal) amounts
of the Z and E isomers would be expected.
9
15054 J. AM. CHEM. SOC. VOL. 125, NO. 49, 2003

Processing Cycloalkenones by Glutathione Transferase
A R T I C L E S
Scheme 5. Possible Stereochemical Routes from (R)d1-1b to the Z or E Isomers of d1-3b in the Active Site of HGSTP1-1a
a The chirality of the glutathionylated carbon is assumed to be R, consistent with the results of molecular docking studies described below.
Scheme 6. Possible Stereochemical Route from (R)d1-1b to a Mixture of Z and E Isomers of d1-3b
Table 3. Steady-State Kinetic Constants for COMC-6 with
Chart 2
Different Classes of Recombinant Glutathione Transferase (GST)a
(kcat/Km) × 10-3
mM-1 min-1
,
[COMC-6]
range, mM
class (variant)
k
cat, min-1
Km, mM
alpha (hGSTA1-1) 90 ( 12 0.13 ( 0.03 0.72 ( 0.12 0.025-0.280
(2.4)b
alpha (hGSTA4-4)
pi (hGSTP1-1)
c
c
2.76 ( 0.36 0.013-0.280
(144)b
138 ( 12 0.08 ( 0.01 1.80 ( 0.18 0.013-0.150
(30)b
mu (hGSTM2-2)
366 ( 48 0.34 ( 0.07 1.08 ( 0.08 0.013-0.280
(5.4)b
theta (rGSTT2-2) ND
ND
ND
0.1
(ND)b
a
Conditions: GSH (1 mM), sodium phosphate (0.1 M, pH 6.5), 30 °C,
∆ꢀ235 ) -2400 M-1 cm-1
.
kcat values for ethacrynic acid (EA) estimated
b
rotate (Chart 2). The binary complex with the 3-(R) enolate was
found to be about 1 kcal/mol more stable than that with the
3-(S) diastereomer (Figure 7).
from published specific activities and subunit molecular weights.8
c
Saturation not achieved. ND, no detectable activity (<0.05 µmol min-1
d
mg-1 at 0.1 mM COMC-6).
Both complexes suggest catalytic roles for the active site
Tyr108 and Tyr7 residues, as discussed below.
EA conjugate in the active site of hGSTP1-1 (PDB code
3GSS).17 Thus, the program should give useful information
about the most stable conformations of similar GSH adducts in
the active site of hGSTP1-1.
Docking experiments were carried out using the two hypo-
thetical diastereomeric enolates, which would form from the
addition of GSH to the 2si-3re face or to the 2re-3si face of
COMC-6 (Chart 2).
Discussion
This work has clarified several aspects of the basic enzyme
chemistry associated with the conjugation of GSH with the
COMC derivatives, which might provide a basis for understand-
ing their tumoricidal activities.
Minimum Kinetic Mechanism. The kinetic measurements
and stereochemical studies suggest that the processing of the
COMC derivatives by GST involves initial enzyme-catalyzed
Michael addition of GSH to the COMC derivative to give a
glutathionylated enol(ate), which undergoes stereorandom loss
of crotonate, while either bound to the active site or free in
solution, to give a glutathionylated exocyclic enone (Schemes
5 and 6). Once free in solution, GSH reacts at the exomethylene
carbon of the exocyclic enone, displacing the first GSH to give
The diastereomers were first constructed and optimized in
Macromodel 7.1 except for the glutathionyl group, which was
taken from the crystal structure and kept rigid during docking.
The optimized structures were then docked into the active site
of 3GSS, allowing only the three bonds highlighted in red to
(17) (a) Oakley, A. J.; Rossjohn, J.; Lo Bello, M.; Caccuri, A. M.; Federici, G.;
Parker, M. W. Biochemistry 1997, 36, 576-585. (b) Lo Bello, M.; Oakley,
A. J.; Battistoni, A.; Mazzetti, A. P.; Muccetelli, M.; Mazzarese, G.;
Rossjohn, J.; Parker, M. W.; Ricci, G. Biochemistry 1997, 36, 6207-6217.
9
J. AM. CHEM. SOC. VOL. 125, NO. 49, 2003 15055

A R T I C L E S
Hamilton et al.
Figure 7. Lowest-energy binding modes of the 3-(R) (left panel) and 3-(S) (right panel) diastereomers of the enolate shown in Chart 2 bound to the active
site of 3GSS.
Chart 3. Possible Catalytic Mechanism for the Glutathione
the final product (Scheme 2). Apparently, all three COMC
derivatives are processed by the same reaction mechanism, as
they all exhibit similar kinetic profiles and kinetically competent
Transferase-Catalyzed Michael Addition of GSH to (A) Ethacrynic
Acid (EA) and to (B) COMC-6
exocyclic enones were successfully isolated from reaction
mixtures containing either COMC-6 or COMC-7 as substrates.
The above mechanism is in sharp contrast to that proposed by
previous workers who believed that the conjugation of GSH
with COTC and COMC-6 involved a simple in-line displace-
ment mechanism leading directly to the GSH conjugate.4
Certainly, in the absence of GST there is a smooth, first-order
conversion of the COMC derivatives to the final GSH adduct,
showing no evidence of an intermediate species (Figures 2 and
4). Since the initial conjugation reaction is rate-determining,
the formation of the exocyclic enone could not be detected
kinetically (Table 1). However, in the presence of high con-
centrations of GST, the multiphase nature of the reaction is
apparent, indicating the formation of a reactive intermediate
prior to formation of the final product. Conceivably, the en-
zymatic reaction might go by a multistep mechanism while the
proposed for the catalyzed addition of GSH to EA.17 The X-ray
nonenzymatic reaction involves an in-line displacement mech-
anism. However, this seems unlikely, as glutathionylated enols
crystal structure of hGSTP1-1 in complex with the GS-EA
adduct suggests that the hydroxyl group of Tyr108 functions
have been detected by mass spectrometry in nonenzymatic
as a general acid catalyst by interacting with the carbonyl group
reaction mixtures containing excess COMC-6 over GSH.6
of the enone function of EA during the addition reaction with
GSH (Chart 3).
Moreover, the hydroxyl group of Tyr7 is in close proximity
The catalytic rate enhancements ((kcat/Km)/k1) for the enzy-
matic processing of COMC-5, COMC-6, and COMC-7 are
modest but significant, corresponding to values of 1.0 × 104,
to the sulfur atom of bound GS-EA and, by inference, probably
2.1 × 104, and 6.7 × 104, respectively (Tables 1 and 2). En-
stabilizes (via H-bonding) the nucleophilic mercaptide anion
zymatic catalysis is probably not limited by chemistry alone.
form of bound GSH prior to reaction.18 The molecular docking
The nonenzymatic rate constants for formation of the exocyclic
studies reported here suggest that the glutathionylated enolate
enones from COMC-5, COMC-6, and COMC-7 progressively
derived from COMC-6 binds to the active site in a fashion
decrease in magnitude with increasing ring size. This trend could
similar to that of GS-EA observed in the X-ray crystal structure
(Figure 7). By inference, the OH group of Tyr108 is well-
positioned to function as a general acid catalyst by interacting
reflect progressively poorer orbital overlap between the carbon-
carbon double bond and the conjugated carbonyl group in the
transition state due to increasing “ring pucker” with increasing
with the carbonyl group of the enone function of COMC-6 in
ring size. However, this trend is not reflected in a significant
the transition state, and the hydroxyl group of Tyr7 is positioned
change in kcat/Km for the COMCs with hGSTP1-1, possibly
to stabilize the mercaptide ion form of GSH. Thus, a common
catalytic mechanism can be envisioned for EA and COMC-6.
This mechanism also comports with the observation that
because catalysis is limited by a conformational transition or
product release from the active site.
Catalytic Mechanism. The mechanism by which hGSTP1-1
COMC-6 and EA are both substrates for the alpha, mu, and pi
catalyzes the initial Michael addition reaction between GSH and
COMC-6 is probably fundamentally similar to that previously
(18) Armstrong, R. N. Chem. Res. Toxicol. 1997, 10, 2-18.
9
15056 J. AM. CHEM. SOC. VOL. 125, NO. 49, 2003

Processing Cycloalkenones by Glutathione Transferase
A R T I C L E S
class variants of GST (Table 3). All three classes of GST are
capable of catalyzing the same general type of chemistry, as
their active sites contain functional groups analogous in position
to those observed in the crystal structure of hGSTP1-1. This
conclusion is based on a large number of sequence alignments,
structural comparisons, and site-directed mutagenesis studies
of the major GST isoforms, extensively reviewed elsewhere.18,19
The explanation for why the recombinant theta class enzyme
rGSTT2-2 does not use EA or COMC-6 as a substrate is
unclear (Table 3). The theta class GSTs show only about 7%
sequence homology with the alpha, mu, and pi class enzymes,
and the functional groups in the active site have not been
completely identified.8e,20 The preponderance of the evidence,
therefore, suggests that general acid catalysis plays a key role
in the Michael addition reactions between GSH and the COMC
derivatives catalyzed by all three major classes of GST.
What happens immediately after formation of the glutathio-
nylated enol(ate) in the active site is less clear. Stereorandom
loss of crotonate from the intermediate implies that elimination
takes place in bulk solvent after dissociation of the intermediate
from the active site (Scheme 6). This type of argument has often
been used to rationalize nonstereospecific steps in metabolic
processes. Nevertheless, there is no absolute assurance that the
structural features of the active site of GST will constrain the
elimination step to follow only one of the stereochemical routes
shown in Scheme 5, particularly in view of reports that some
isozymes of GST can nonstereospecifically process substrates.
For example, the rGSTM2-2 isozyme reportedly catalyzes the
addition of GSH to 4-phenyl-3-buten-2-one with a high degree
of stereoselectivity (90%).21 In contrast, the M1-1 isozyme
catalyzes the very same reaction with little stereoselectivity. In
view of these observations, the stereorandom elimination of
crotonate from the glutathionylated enol(ate) might still be an
active-site process subject to enzymatic catalysis.
What Is the Chemical Basis of Antitumor Activity? The
studies described here were originally motivated by an interest
in the mechanism by which COTC and COMC-6 inhibit tumor
growth (Scheme 1). Tumoricidal activity was once thought to
arise from inhibition of the methylglyoxal-detoxifying enzyme
glyoxalase I by the GSH adducts formed with COTC and
COMC-6.4 However, recent studies show that the adducts are
poor competitive inhibitors of human glyoxalase I.1,5 Even more
importantly, the diethyl ester prodrug form of the GSH conjugate
GSMC-6 was found not to have significant antitumor activity
with B16 melanotic melanoma in vitro (IC50 > 460 µM) versus
COMC-6 (IC50 0.046 µM) and the five- and seven-membered-
ring homologues COMC-5 (IC50 0.144 µM) and COMC-7 (IC50
0.029 µM).2 Therefore, tumoricidal activity cannot be due to
the final GSH conjugate GSMC-6, but must result directly from
COMC-6 and/or from an intermediate species formed during
its conjugation with GSH. Preliminary studies indicating that
the conjugation reaction is indeed a multistep process involving
the intermediate formation of an electrophilic exocyclic enone
led to the current hypothesis that tumor toxicity might reflect,
in part, the ability of exocyclic enone to alkylate biomacromol-
ecules vital to cell function.1
The detailed kinetic work described here strongly indicates
that all three cytotoxic COMC derivatives react with GSH via
mechanisms involving the formation of electrophilic exocyclic
enones, implying that the cytotoxicity of the COMC derivatives
has a common mechanistic explanation. The fact that the
exocyclic enones are better Michael acceptors than the unmodi-
fied COMCs (Table 1) suggests that the exocyclic enones might
be the actual toxic species in tumor cells, at least from a simple
chemical perspective. Nevertheless, the unmodified COMCs are
also good Michael acceptors and more work is needed before
any final conclusions can be reached with respect to the relative
cytotoxicities of the COMCs and their exocyclic enones in
highly compartmentalized tumor cells. Whatever the specific
chemical identity of the cytotoxic species inside tumor cells,
cell death could result from genotoxicity. This follows from
the observation that COMC-6 forms covalent adducts with the
exocyclic amino groups of the nucleotide bases composing
single-stranded oligonucleotides in cell-free systems, either in
the presence or in the absence of GSH.6
What Is the Role of GST in Antitumor Activity? All
human tissues contain GST activity, which is usually differen-
tially distributed among different isozymes of GST based on
tissue type.22 The pi class of GST is of particular interest from
a chemotherapeutic perspective, as it is often overexpressed in
tumor tissue versus normal tissue. GSTs are thought to detoxify
antitumor drugs by catalyzing their conversion to GSH-drug
conjugates, which are subsequently removed from cells by a
membrane-associated protein pump. Repeated exposure to
antitumor agents often leads to even greater expression of GST
activity in tumor cells and the development of multidrug
resistance (MDR).7 For example, MCF7 human breast cancer
cell lines, which have been selected for resistance to different
types of antitumor agents, can overexpress the pi GST isozyme
at levels 10-fold greater than that observed in the wild-type
tumor cell line. Overcoming MDR is one of the major unmet
challenges facing medicinal enzymology.
GST undoubtedly plays an important role in the metabolism
of the COMC derivatives in some types of cells. The fraction
(F) of COMC derivative undergoing enzymatic versus nonen-
zymatic conversion to GSH conjugate is formally given by
(Vmax/KCmOMC
)
∑
F )
(1)
k [GSH] + (Vmax/KmCOMC
)
∑
1
where ∑(Vmax/KCOMC) equals the sum of the Vmax/KCOMC
m
m
activities of the different GST isozymes and k1[GSH] is the
apparent rate constant for the nonenzymatic formation of GSH
conjugate. This equation applies under the most likely physi-
ological conditions, where [GSH] equals about 5 mM and
therefore is .KGmSH (∼0.2 mM),7 and [COMC] , KmCOMC
.
From published enzyme activities,23 wild-type MCF7 cells are
estimated to contain 0.21 CDNB units of hGSTP1-1 activity/
mL cell cytosol. This translates into about 0.011 COMC-6 units/
mL cytosol, using the published activities hGSTP1-1 with
CDNB versus EA7 and the activities of hGSTP1-1 with EA
(19) (a) Mannervik, B.; Danielson, U. H. Crit. ReV. Biochem. Mol. Biol. 1988,
23, 283-337. (b) Daniel, V. Crit. ReV. Biochem. Mol. Biol. 1993, 28, 173-
207. (c) Wilce, M. C. Biochim. Biophys. Acta 1994, 1205, 1-18. (d) Dirr,
H.; Reinemer, P.; Huber, R. Eur. J. Biochem. 1994, 220, 645-661.
(20) Pemble, S. E.; Taylor, J. B. Biochem. J. 1992, 287, 957-963.
(21) Chen, J.; Armstrong, R. N. Chem. Res. Toxicol. 1995, 8, 580-585.
(22) Johansson, A.-S.; Mannervik, B. In InterindiVidual Variability in Human
Drug Metabolism; Pacifici, G. M., Pelkonen, O., Eds; Taylor and Francis:
London, 2001; pp 460-519.
(23) Morrow, C. S.; Smitherman, P. K.; Diah, S. K.; Schneider, E.; Townsend,
A. J. J. Biol. Chem. 1998, 273, 20114-20120.
9
J. AM. CHEM. SOC. VOL. 125, NO. 49, 2003 15057

A R T I C L E S
Hamilton et al.
versus COMC-6 (Table 3). This activity and the k1 and KCOMC
m
[exo]MDR
k [GSH] + (Vmax/KmCOMC
)
∑
1
MDR
)
(2)
values given in Tables 1 and 2 with eq 1 indicates that about
40% of the COMC-6 is processed by hGSTP1-1 in wild-type
MCF7 cells. For MDR cells that overexpress hGSTP1-1 by
greater than 10-fold, the fraction processed by the enzyme is
greater than 87%. A similar analysis for COMC-7 indicates that
in wild-type MCF7 versus MDR MCF7 cells, the fractions
processed by the enzyme are 70% and >96%, respectively.
Therefore, hGSTP1-1 is probably responsible for processing a
significant fraction of COMC-6 and COMC-7 in both wild-
type and MDR MCF7 cells.
k [GSH] + (Vmax/KmCOMC
)
normal
[exo]normal
∑
1
This ratio will be maximal (and differential toxicity greatest)
when the enzymatic rate of formation of the exocyclic enone
greatly exceeds the nonenzymatic rate (k1[GSH] , ∑(Vmax
/
KCmOMC). Under these conditions, [exo]MDR/[exo]normal ap-
proaches the ratio of total GST activity in MDR tumor cells
versus normal cells.
Summary
From the standpoint of tumor toxicity, overexpression of GST
activity could have the following consequences, assuming that
a steady-state kinetic model applies in which the rate of influx
of COMC derivative into tumor cells is equal to the rate of
conversion to GSH conjugate (Scheme 2). First, if the unmodi-
fied COMC derivative is directly responsible for tumoricidal
activity, overexpression of GST activity could protect tumor
cells from cytotoxicity by lowering the steady-state concentra-
tion of intracellular COMC derivative and, therefore, decreasing
the rate of covalent modification of vital biomacromolecules.
Second, if the COMC derivative and the exocyclic enone are
equally toxic to tumor cells, overexpression of GST would have
little effect on cytotoxicity. In this case, the COMC derivatives
should be exceptionally resistant to the development of MDR.
Third, if the exocyclic enone is primarily responsible for anti-
tumor activity, overexpression of GST activity would actually
increase the sensitivity of tumor cells to the COMC derivative
by increasing the steady-state concentration of the exocyclic
enone. Thus, tumor-selective toxicity would be a function of
the ratio of the rates of formation of the exocyclic enone in
MDR cells that overexpress GST versus normal cells that
express lower levels of GST (eq 2).
The work reported here demonstrates that the GST-catalyzed
conversion of COMC derivatives to the corresponding GSH
conjugates is a multistep process involving the formation of a
highly reactive glutathionylated exocyclic enone. Thus, the
antitumor activities of the COMC derivatives might be ex-
plained on the basis of the ability of the exocyclic enone and/
or underivatized COMC to function as Michael acceptors,
resulting in the alkylation of biomacromolecules critical to cell
viability. Finally, the differential expression of GST activity
between normal cells and tumor cells could have a major
influence on the differential toxicity of COMC derivatives to
tumor cells.
Acknowledgment. We thank the University of Maryland,
College of Life Sciences for MS and the UMBC Center for
Structural Biochemistry for NMR spectra. This work was
supported by grants from the NIH (CA 59612 to D.J.C.; GM
24054 to B.G.; GM 61402 to K.N.H.), the U.S. Army Medical
Research and Material Command (DAMD17-99-1-9275 to
D.J.C.), and the Swedish Cancer Society (to B.M.).
JA030396P
9
15058 J. AM. CHEM. SOC. VOL. 125, NO. 49, 2003