γ-Glutamyl transpeptidase-dependent mutagenicity and cytotoxicity of γ-Glutamyl derivatives: A model for biochemical targeting of chemotherapeutic agents

Article abstract of DOI:10.1002/(SICI)1098-2280(1998)32:4<377::AID-EM12>3.0.CO;2-1

Many carcinomas in humans are rich in glutamyl transpeptidase (GGT), a plasma membrane enzyme that reacts with extracellular substrates. Thus, biochemical targeting of chemotherapeutic agents may be achieved by converting anticancer drugs into their γ-glutamyl derivatives. Chemical conversion of phenylhydrazine (PH) and biochemical modification of daunomycin (DM) into their γ-glutamyl derivatives γ-glutamyl phenylhydrazine (GGPH) and γ-glutamyl DM (GGDM) resulted in the abolishment of their mutagenicity and cytotoxicity, as judged by decreased viability and increased mutant yields in cultures of several Salmonella Ames strains. Commercial γ- glutamyl-p-nitroanilide (GGPNA) was not toxic or mutagenic. Mutagenicity and/or cytotoxicity of these γ-glutamyl derivatives were restored upon reaction with GGT, with concomitant release of PH, and p-nitroaniline (PNA). The GGT-dependent release of DM from GGDM was demonstrated by thin layer chromatography (TLC), spectral analysis, and specific mutagenicity. Mutagenicity and/or cytotoxicity of γ-glutamyl derivatives increased in the presence of glycylglycine, a GGT activator, and decreased in the presence of serine-borate, a GGT inhibitor. GGDM retained considerable DNA binding capacity. Its inability to kilt and mutagenize was due to altered transport properties. The results are compatible with the notion that γ-glutamylation is a feasible method for biochemical targeting of drugs containing a primary amino group to GGT-rich tumors.

Full text of DOI:10.1002/(SICI)1098-2280(1998)32:4<377::AID-EM12>3.0.CO;2-1

Environmental and Molecular Mutagenesis 32:377–386 (1998)
g
-Glutamyl Transpeptidase-Dependent Mutagenicity and
Cytotoxicity of -Glutamyl Derivatives: A Model for
g
Biochemical Targeting of Chemotherapeutic Agents
Ronit Keren and Avishay-Abraham Stark*
Department of Biochemistry, Tel-Aviv University, Israel
Many carcinomas in humans are rich in g-glutamyl concomitant release of PH, and p-nitroaniline (PNA).
transpeptidase (GGT), a plasma membrane enzyme The GGT-dependent release of DM from GGDM
that reacts with extracellular substrates. Thus, biochem- was demonstrated by thin layer chromatography
ical targeting of chemotherapeutic agents may be (TLC), spectral analysis, and specific mutagenicity.
achieved by converting anticancer drugs into their
g-
Mutagenicity and/or cytotoxicity of g-glutamyl deriv-
glutamyl derivatives. Chemical conversion of phenyl- atives increased in the presence of glycylglycine, a
hydrazine (PH) and biochemical modification of dau- GGT activator, and decreased in the presence of
nomycin (DM) into their
myl phenylhydrazine (GGPH) and
g
-glutamyl derivatives
g-gluta- serine-borate, a GGT inhibitor. GGDM retained
g
-glutamyl DM considerable DNA binding capacity. Its inability to
(GGDM) resulted in the abolishment of their mutage- kill and mutagenize was due to altered transport
nicity and cytotoxicity, as judged by decreased viabil- properties. The results are compatible with the no-
ity and increased mutant yields in cultures of several tion that
Salmonella Ames strains. Commercial -glutamyl-p- chemical targeting of drugs containing a primary
nitroanilide (GGPNA) was not toxic or mutagenic. Mu- amino group to GGT-rich tumors. Environ. Mol. Mu-
tagenicity and/or cytotoxicity of these -glutamyl deriv- tagen. 32:377–386, 1998
atives were restored upon reaction with GGT, with
g-glutamylation is a feasible method for bio-
g
g
᭧ 1998 Wiley-Liss, Inc.
Key words: -glutamyl transpeptidase; phenylhydrazine; daunomycin; mutagenicity; cytotoxic-
g
ity; Salmonella typhimurium; drug targeting
INTRODUCTION
Example 2: Release of a Chemically Reactive Agent
From an Adduct
Biochemical targeting is an interesting approach to in-
crease the specificity of drugs for tumor cells. Biochemi-
cally targeted drugs are substrates for enzymes that exist
at high level/activity in tumor cells as compared to normal
ones, and are specifically converted into active forms
upon reaction with the enzymes.
L-g-glutamyl-4-(hydroxymethyl) phenylhydrazine (agar-
itine) is another metabolite of A. bisporus. Removal of the
-glutamyl moiety forms the cytotoxic 4-(hydroxymethyl)-
phenylhydrazine. The latter is converted to the reactive di-
azonium ion by microsomal enzymes [Hiramoto et al.,
1995a,b; Walton et al., 1997]. The first step of activation
g
is catalyzed by g-glutamyl transpeptidase (GGT), a plasma
membrane enzyme that is induced to high levels at early
stages of experimental hepatocarcinogenesis in rodents.
Example 1: Conversion of a Prodrug Into a
Chemically Reactive Agent
Abbreviations: DM, daunomycin; GGDM, putative
g
-glutamyl-dau-
-glutamyl-p-nitro-
-glutaminyl-p-hyr-
g
-Glutaminyl-p-hydroxybenzene (GHB) is a tyrosine
analog and a natural product of Agaricus bisporus. Tyro-
sinase oxidizes GHB to -glutaminyl-3,4-benzoquinone;
nomycin; GGT,
anilide; GGPH,
g
-glutamyl transpeptidase; GGPNA,
-glutamyl phenylhydrazine; GHB,
g
g
g
g
doxybenzene; GSH, glutathione; PH, phenylhydrazine; PNA, p-nitroani-
the latter converts nonenzymatically to the corresponding
2-hydroxy-4-iminoquinone, which ultimately inhibits
many enzymes involved in normal energy metabolism,
leading to cell death. Most melanoma cells are rich in
tyrosinase, whereas other cells are deficient in tyrosinase:
GHB is a tyrosinase-targeted specific antimelanoma agent
[Vogel et al., 1977, 1979; Burger et al., 1979; Chen et
al., 1979; Boekelheide et al., 1980].
line; TLC, thin layer chromatography.
Contract grant sponsors: the Rekanati Foundation for Medical Research,
the Eizen Foundation for Cancer Research, and the Ella Kodesz Institute
for Research on Cancer Development and Prevention.
*Correspondence to: Dr. Avishay-Abraham Stark, Department of
Biochemistry, Tel-Aviv University, Ramat-Aviv 69978 Israel. E-mail:
aastark@ccsg.tau.ac.il
Received 12 February 1998; Revised and accepted 3 September 1998
᭧
1998 Wiley-Liss, Inc.
/ 8I31$$8011
11-23-98 17:02:02
wlemas
W Liss: EM EM-98011

378
Keren and Stark
Fig. 1. Structures of GGPNA, GGPH, PH, and GGDM.
High GGT levels persist in carcinomas in later stages. In- MATERIALS AND METHODS
creased GGT levels are also found in many human hepatic
carcinomas, and in other human and animal carcinomas
Materials
[Gerber and Thung, 1980; Selvaraj et al., 1981; Hanigan
and Pitot, 1985; Farber, 1986; Bannasch, 1986; Sarma et
al., 1986; Pitot, 1990; Stark, 1991]. Thus, g-glutamyl deriv-
GSH, glycylglycine,
g-glutamyl-p-nitroanilide (GGPNA), amino
acids, phenylhydrazine, pyroglutamic acid, bovine GGT (25 U/mg pro-
tein), GGPNA, NADP, glucose-6-phosphate, G6PD, and calf thymus
DNA were from Sigma (Milwaukee, WI). Kieselgel-60 was from Merck
(Darmstadt, Germany). Sodium glyoxylate was from ICN (Costa-Mesa,
CA). Silica gel-60 TLC plates were from Riedel-De Ha¨en (Hannover,
atives are potentially GGT-targeted prodrugs.
The use of GGT for drug targeting is appealing in that
the active site of membranal GGT is exclusively directed
towards the extracellular space [Tsao and Curthoys, 1980].
ˆ
Germany). Daunomycin was from Rhone-Poulenc (Courbevoie, France).
Salmonella typhimurium strains TA1538, TA98, and TA100 were kindly
provided by B.N. Ames, Division of Biochemistry and Molecular Biol-
ogy, University of California, Berkeley.
The action of GGT on a g-glutamyl derivative of a nonpolar
drug may release the nonpolar drug in the vicinity of the
GGT-rich tumor cell, thus increasing the effective concen-
tration (and uptake) of the drug at the tumor site. In order
to test this possibility, we synthesized and partially charac-
Methods
Preparation of g-Glutamyl Derivatives
terized a g-glutamyl derivative of daunomycin (Fig. 1)
g
-Glutamyl phenylhydrazine (GGPH) (Fig. 1) was synthesized from
(GGDM), and compared its mutagenic and cytotoxic proper-
ties with those of daunomycin (DM). The results indicate
that mutagenicity and cytotoxicity of GGDM depend on
pyroglutamic acid and PH, and was purified as described previously
[Levenberg, 1970]. GGDM was prepared enzymatically in a reaction
mixture containing 250 mM HEPES buffer, pH 8.5, 3 U/ml GGT, 13
mM DM, and 130 mM GSH (adjusted to pH 7) in a final volume of
1.5 ml. Mixtures for small-scale preparations contained 250 mM
GGT activity and that
its uptake properties.
g-glutamylation of DM alters mainly
/ 8I31$$8011
11-23-98 17:02:02
wlemas
W Liss: EM EM-98011

Prodrug Activation by GGT
379
HEPES, pH 8.5, 4.3 mM DM, 50 mM GSH, and 3 U/ml GGT. The
mixture was incubated up to 25 hr at 25 C in the dark and was extracted
Њ
four times with an equal volume of chloroform. Phase separation and
precipitation of denatured proteins was by centrifugation (10,000g,
25
Њ
C, 5 min). The aqueous phase was loaded onto 20
1
20 cm, 5
mm-thick silica gel-60 plates (750
m
l/plate) and chromatographed with
chloroform:methanol:water (13:6:1) [Levin and Sela, 1979] until the
solvent front reached 15 cm. The gel area containing the polar product
(orange to deep red) was scraped, extracted 3–5 times with metha-
nol:water (5:1) and was centrifuged as above. Methanol was evaporated
under reduced pressure at 40
and lyophilized, and the final product (red threads) was stored desiccated
at 20 C. Preparation, handling and experiments with GGDM were
carried out under yellow ( 550 nm) light in order to prevent photodeg-
ЊC, the remaining aqueous phase was frozen
0
Њ
ú
radation. Prior to the experiments, DM or the aglycone resulting from
dissociation or degradation during storage were removed by dissolving
GGDM in 50 mM HEPES, pH 8, and extracting 4–5 times with chloro-
form as above. Purity of GGDM was tested on TLC plates developed
as above. The extinction coefficients of DM in ethanol and in HEPES
buffer pH 8.5 are 11,500 M01 cm01 and 7,419 M01 cm01 at 480 nm,
respectively. Since GGDM is insoluble in ethanol, its concentration was
determined in HEPES buffer using the coefficient of DM in HEPES.
IR spectra were carried out with DM and GGDM in KBr pellets.
Fig. 2. GGT-dependent cytotoxicity of GGPH in strain TA1538. (A)
Reaction mixtures (as described in Materials and Methods) contained 3.5
1
108 cells/ml, 40 mM glycylglycine, and the indicated concentrations of
phenylhydrazine (
GGPH, 3 U/ml GGT and 5mM serine-10 mM borate. Reactions were
shaken at 37 C for 6 hr. Viability was determined as described in Materi-
छ); GGPH (᭺); GGPH and 3 U/ml GGT (᭞) or (ᮀ)
Determination of GGT Activity With Various
Њ
g
-Glutamyl Derivatives
als and Methods. Presented are means of triplicates. (B) Release of PH
from GGPH. Samples of reactions containing GGPH were centrifuged
and the concentration of PH was determined by reaction with glyoxylate
Hydrolysis of L-g-glutamyl-p-nitroanilide was essentially as described
previously [Stark et al., 1993] in reaction mixtures containing 100 mM
Tris HCl, pH 8.2, 20 mM glycylglycine, 0.1–2 mM GGPNA, and 100
mU/ml GGT, and the rate of increase in A412 was followed (e Å 8,800
M01 cm01). p-Nitroaniline was used as a standard. Hydrolysis of GGPH
was determined in 50 mM phosphate buffer pH 8.5 containing 0.3–
5 mM GGPH, 300 mU/ml GGT, and 5 mM glyoxylate. Continuous
as described in Materials and Methods. (
᭺) GGPH; (ᮀ) GGPH and 3
U/ml GGT; ( ) GGPH, 3 U/ml GGT and 5 mM serine-10 mM borate.
᭝
Presented are means of triplicates.
monitoring of A320 (phenylhydrazone-
a
-keto acid adduct, e Å 14,000
M01 cm01) was used to determine reaction rates. Commercial phenylhy-
drazine was used as a standard.
viability was determined as above. The release of biologically active
compounds from their g-glutamyl derivatives was linear with time. Thus,
the cells were exposed to linearly accumulating concentrations, and
actual exposure at a certain time point was half of that of the released
material at that time point. Specific toxicity was defined as the accumu-
lated concentration of a test compound that caused one lethal hit.
Rat kidney and rat liver microsomes were prepared by homogeniza-
tion of kidney cortex or liver tissue in 100 mM Tris buffer, pH 8.0, and
centrifugation at 10,000g for 30 min. The supernatant was centrifuged
at 130,000g for 60 min and the pellet was suspended in the same buffer,
aliquoted, and stored at
0
80ЊC. GGT activity was 10 U/mg protein and
5 mU/mg protein in rat kidney and rat liver microsomes, respectively.
Mutagenicity Assays
Cytotoxicity Assays
Bacterial cultures, growth media, and the procedure for the standard
The toxicity of GGPH, phenylhydrazine, and GGPNA was assayed plate incorporation mutagenicity test were essentially as described by
in Salmonella strain TA1538 [Malca-Mor and Stark, 1982]. Cells were Maron and Ames [1983] except that rat kidney microsomes replaced
grown in minimal medium [Maron and Ames, 1983]. Exponentially S9. The NADPH generation system was not included because neither
growing cells were diluted to 0.8
8.35, 40 mM MgCl2 , 27.5 mM KCl, 27 mM glucose, 5
0
1
1
108/ml in 60 mM HEPES, pH DM nor PH require oxidative metabolism for their activation. The mi-
m
M biotin, 100 crosomes were used solely as a GGT source. Reaction mixtures with
m
M histidine, and, when appropriate, 3 U/ml GGT, 40 mM glycylgly- liver microsomes included the NADPH generation system. Reaction
cine, and the test compound. Cultures were shaken at 37
Њ
C and samples mixtures contained 108 TA98 or TA100 cells, 83 mM HEPES, pH 8.0,
were withdrawn with time for viable counts (triplicates) on LB agar. 12 mM MgCl2 , 27.5 mM KCl, 10 mM glucose, and, when appropriate,
The release of p-nitroaniline from GGPNA was assayed by withdrawing 40 mM glycylglycine and rat kidney microsomes at 7 mg protein/ml
5
m
l samples at various time points (as indicated in the figures), diluting and test compound in a final volume of 0.5 ml. The pH of the top agar
in 150 mM NaCl and determination of p-nitroaniline at A412 . The release was adjusted to 8.0. Triplicate plates were poured for each concentration
of phenylhydrazine was assayed by withdrawing 50 l samples into point. Colonies were counted manually.
1.2 ml of 5 mM Na-glyoxylate in 150 mM NaCl, centrifugation, and Binding of DM and GGDM to DNA was determined by fluorescence
m
determination of A320 of the supernatant. Cytotoxicity of DM and quenching [Levin and Sela, 1979]. The fluorescence of reaction mixtures
GGDM in strain TA98 was assayed on exponential cultures in LB containing 10 mM Tris.HCl, 100 mM NaCl, 0.5 mM EDTA, pH 7, and
medium at 4
in LB, pH 8.5, containing, when appropriate, 40 mM glycylglycine, 5– nm,
20 U/ml GGT, and test compounds. Cultures were shaken at 37 C and incremental addition of calf thymus DNA up to 90
1
108 cells/ml. Cultures were diluted to 4
1
107 cells/ml
5
m
M DM (
em 590 nm) was determined in the absence of DNA, and after
g/ml. Scatchard
lex 485 nm, lem at 580 nm) or 10 mM GGDM (lex 503
l
Њ
m
/ 8I31$$8011
11-23-98 17:02:02
wlemas
W Liss: EM EM-98011

380
Keren and Stark
plots were used to determine frequency and affinity constants of binding
[Levin and Sela, 1979].
RESULTS AND DISCUSSION
GGT catalyzes the cleavage of a
g-glutamyl moiety of
a
g-glutamyl-containing compound (hydrolysis of a do-
nor) and transfers it to an acceptor molecule containing
a primary amino group (transpeptidation) (Equations 1,
2, below). Its specificity is low in that numerous com-
pounds can serve as donors or acceptors, and the kinetics
of the reactions depend on the nature of the donor and
the acceptor. A serine-borate complex is a transition state
analog, and thus a competitive inhibitor of GGT, whereas
the acceptor glycylglycine increases the cleavage of the
g-glutamyl bond by 6–10-fold. We utilized the broad
specificity of GGT to test whether it can release muta-
genic/cytotoxic compounds from their respective
myl analogs.
g-gluta-
The kinetic parameters of the reaction of GGT with
GGPH were determined at 0.312–3.75 mM GGPH, 20
mM glycylglycine, and 1 U/ml GGT. Vmax was 3 nmol/
min/40
mg protein (75 nmol/min/mg protein), and Km
was 0.8 mM. Reaction rate was linear with time at every
GGPH concentration. At 10 mM serine/20 mM boric acid,
Vmax was unchanged and Km was 2.2 mM, indicative
of competitive inhibition (data not shown). Cleavage of
GGPH by GGT is approximately 170-fold slower than
Fig. 3. GGT-dependent mutagenicity of GGPH and GGPNA in strain
TA100, plate incorporation assay. Reaction mixtures (as described in
Materials and Methods) contained 1
amounts of test compounds: ( ) PH; (
) GGPH or GGPH with liver microsomes; (
microsomes; ( ) PNA or PNA with liver or kidney microsomes; (
GGPNA; ( ) GGPNA with kidney or liver microsomes. Kidney and
1
108 cells and the indicated
) PH with kidney microsomes;
the reaction with GGPNA (Vmax
protein (12.5 mol/min/mg protein) at 100 mU GGT/ml,
1 mM GGPNA, 20 mM glycylglycine).
Å
50 nmol/min/4 mg
l
ᮀ
(
᭡
᭞) GGPH with kidney
m
छ
᭺)
᭿
liver microsomes were at 7 mg protein/ml. Presented are means
of triplicate plates.
{
SD
g-Glutamyl-NH-R
/
GGT r
g
-glutamyl-GGT
/
R-NH2 Hydrolysis (1)
the cleavage of GGPH by GGT. The killing obtained at
constant (2 mM) concentration of commercial PH for 6
hr (1.22 logs) was comparable to that obtained at 5 mM
GGPH with GGT (1.2 logs), where free PH accumulated
to 3 mM within the same time. Since PH accumulation
was linear with time, the effective concentration of the
g
-Glutamyl-GGT
/
R
؅
-NH2
r
g-glutamyl-NH-R؅ / GGT Transpeptidation (2)
GGT-Dependent Cytotoxicity of GGPH and GGPNA
Synthetic GGPH and commercial GGPNA (Fig. 1) released PH (integral of zero to 6 hr) was 1.5 mM. The
served as model compounds. The synthesis of GGPH addition of serine-borate at 6 mM GGPH resulted in re-
from pyroglutamic acid yields a racemic mixture of L- lease of 1.6 mM PH (effective concentration 0.8 mM)
and D-GGPH; however, GGT cleaves L- and D-g-gluta- and in 0.5 logs killing, correlated with 0.6 logs killing at
myl moieties equally well. Salmonella cells were exposed 1 mM commercial PH. The toxicity of GGPNA was GGT-
to PH, or to GGPH with and without GGT, for 6 hr. dependent, was partially inhibited by serine borate, and
Figure 2A shows that PH was toxic in a concentration- correlated with the amounts of PNA released from
dependent manner. GGPH without GGT was not toxic, GGPNA: 3.5 and 0.7 lethal hits were obtained at 20 mM
whereas marked toxicity was obtained in the presence of GGPNA, 1U/ml GGT for 5 hr without and with serine-
GGT at 4, 5, and 6 mM GGPH. Addition of serine-borate borate, respectively. Under these conditions, PNA accu-
complex (a competitive inhibitor of GGT) resulted in mulated to 16 and 8 mM with and without serine-borate,
lower toxicity. The amounts of released PH (Fig. 2B) respectively (data not shown). The specific toxicity of the
under each condition indicate that toxicity depended on released PH (0.72 mM) was 2.9-fold higher than that of
/ 8I31$$8011
11-23-98 17:02:02
wlemas
W Liss: EM EM-98011

Prodrug Activation by GGT
381
Fig. 4. Synthesis of GGDM. The procedure was essentially as de- determined at 480 nm. (A) (
scribed in Materials and Methods. Reaction mixtures contained 4.3 mM and serine-borate; ( ) DM and GSH, DM and GGT, or DM and serine-
DM and as indicated, 50 mM GSH, 3 U/ml GGT, 12.5 mM serine, and borate. (B) TLC of 2 l samples of reaction mixtures shown in A, lanes
25 mM boric acid in a final volume of 500 l. At the indicated time 1 and 8, DM standard; lane 2, DM and GSH, 25 hr time point; lanes
points, 30 l samples were withdrawn into 500 l 200 mM HEPES pH 3–7, complete reaction, 0, 4, 8.5, 16, and 25 hr time points, respectively.
8.5 and the mixture was extracted with chloroform. The volume of the Circled are fainter spots. The aglycone of DM appears as a thin red line
aqueous phase was adjusted to 700 l with HEPES buffer (to allow at the top (apolar) position in lanes 1 and 6.
ᮀ) DM, GSH, and GGT; (᭝) DM, GSH,
᭺
m
m
m
m
m
measurement in a microcuvette) and the concentration of GGDM was
the released PNA (2.07 mM), very similar to the ratio No mutagenicity of GGPH and GGPNA was observed
obtained at pH 7.4 [Malca-Mor and Stark, 1982]. Thus, with liver microsomes, which lack GGT. Mutagenicity of
the effective cell death obtained with GGPNA (a precur- PH did not require metabolic activation, and inclusion of
sor for the less toxic PNA) as compared to that of GGPH liver or kidney microsomes slightly decreased its muta-
(a precursor for the more toxic PH) was due to the fact genic activity.
that GGPNA is a good substrate for GGT, whereas GGPH
is a poor one.
The above results indicate that cytotoxicity/mutage-
nicity of a chemically reactive compound is abolished
upon conjugation with a
cleavage of the -glutamyl residue by GGT restores
cytotoxicity and/or mutagenicity. PH and PNA are not
g-glutamyl residue, and that
g
GGT-Dependent Mutagenicity of GGPH and GGPNA
The plate incorporation assay was used to determine used as anticancer drugs. We therefore tested whether
whether GGT can activate GGPH and GGPNA to muta- a known mutagenic/cytotoxic chemotherapeutic agent
gens. Here, rat kidney or rat liver microsomes replaced can be converted into its
soluble GGT in order to test whether metabolism by en- whether the biological activities of the adduct depend
zymes other than GGT converts -glutamyl derivatives on GGT activities.
g-glutamyl derivative and
g
into mutagens. Figure 3 shows that GGPH and GGPNA
were not mutagenic without kidney microsomes. With
kidney microsomes, significant dose-dependent muta-
genicity was obtained only with GGPH. A very weak
response was obtained with GGPNA and kidney micro-
GGT-Dependent Synthesis and Partial
Characterization of a Putative GGDM Adduct
Daunomycin (Fig. 1) seemed a good candidate, in that
somes, similar to the weak mutagenicity of PNA in the it contains a single primary 3
؅-amino group in the dauno-
liquid preincubation assay [Malca-Mor and Stark, 1982]. samine (amino sugar) moiety. Attempts to attach a
g
-
/ 8I31$$8011
11-23-98 17:02:02
wlemas
W Liss: EM EM-98011

382
Keren and Stark
Fig. 5. (A) Activity spectrum of (bottom to top): DM, GGDM, and DM released from the adduct by
GGT. (B) IR spectrum of DM (top line) and GGDM (bottom line).
zero (lane 3). Determination of the concentration of the
extracted polar product (Fig. 4A) revealed that its accu-
mulation was slow, linear with time, and that addition
of serine-borate partially inhibited its accumulation. The
above indicate that the formation of the product depended
on the activity of GGT, and thus it may be the putative
GGDM adduct. Considerable amounts of DM were appar-
ent at early time points, and a small amount was still
visible at 25 hr. Since GGT was present, it was unclear
glutamyl residue to DM by reacting it with pyroglutamic
acid were unsuccessful: no polar products were formed,
and most of the DM remained unchanged save some
cleavage of the daunosamine and appearance of the agly-
cone, as judged by TLC (data not shown). We attempted,
therefore, to use this 3
GSH as a -glutamyl donor in a transpeptidation reaction
؅-amino group as an acceptor and
g
catalyzed by GGT. Such a reaction resulted in the time-
dependent appearance of a polar product in TLC of mix-
tures that contained GGT (Fig. 4B, lanes 4–7), but not whether it was due to incomplete conversion, GGT-de-
without GGT (lane 2) or in a complete mixture at time pendent or nonenzymatic cleavage of GGDM.
/ 8I31$$8011
11-23-98 17:02:02
wlemas
W Liss: EM EM-98011

Prodrug Activation by GGT
383
The activity spectrum of GGDM was very similar to
that of DM (Fig. 5A). Comparison of the IR spectra of
DM and GGDM (Fig. 5B) revealed a broad peak ap-
pearing at 1,630–1,680 cm01 [Williams and Fleming,
1995], consistent with an amide bond. We could not deter-
mine the molecular mass of GGDM due to the instability
of the glycosidic bond, and the amounts of GGDM were
too low to allow NMR and CMR analyses, but other
evidence (see below) indicates that GGDM is a
derivative of DM.
g-glutamyl
Binding of DM and GGDM to DNA
Binding of DM to DNA occurs by intercalation of the
aglycone between the bases, and the daunosamine moiety
lies in the minor groove [Gao et al., 1996] without bond-
ing to DNA [Quigley et al., 1980]. However, modifica-
tions of the 3؅-amino group in the daunosamine moiety
alter the DNA-binding properties of DM more drastically
than modifications of the aglycone [Jolles et al., 1996;
Fig. 6. Scatchard plotting of DNA binding data. (A) DNA was at 5
g/ml in the presence of 0.2–2.4 M DM. (B) DNA was at 10 g/ml
in the presence of 0.2–2.2 M GGDM. DNA-bound DM or GGDM do
not fluoresce. One g DNA contains 3,080 pmols deoxyribonucleotide
phosphate. The bound fraction (B) was calculated from the difference
between the fluorescence values in the absence and the presence of
DNA. The minimal target size (1/n, or B max) and the apparent associa-
tion constant (Kapp) were calculated according to the relationship
m
m
m
m
m
Chaires et al., 1996]. Further, N-alkylations of the 3؅-
amino group by a small moiety (e.g., N,N-dimethyldauno-
mycin) abolish or greatly reduce their mutagenicity in
bacterial and mammalian cells [Westendorf et al., 1984].
We therefore compared the DNA binding properties of
GGDM and DM as judged by fluorescence quenching.
The binding parameters, calculated from Scatchard plots
of DM (Fig. 6A) indicate that at saturating DM concentra-
B
B max
0
B
Å
Free
Kapp
where the slope Å 01/Kapp and intercept
Å
B max.
tions, the minimal target size was 7–9 basepairs (n
0.11–0.146), and the apparent association constant 102 hr (see below); thus, the effective DM concentrations
(Kapp) was 2.1–2.82
106 M01. The affinity for a base- which accumulated within 6 hr at 20, 60, 120, and 200
pair (2n Kapp) was 6.3 M GGDM were 0.4, 1.2, 2.4, and 4 M, respectively.
105 M01. The values calcu-
lated for saturating GGDM (Fig. 6B) were target size of This indicates that the toxicity without GGT was mainly
32–40 basepairs (n 0.025–0.0316), Kapp 2.27–3.14 due to release of DM, which readily penetrates the cells,
106 M01 and 2n 104 M01. DNA rather than due to the penetration of the intact GGDM
Kapp 1.53
Å
1
1
1
m
m
Å
Å
1
1
Å
1
binding of GGDM was not due to nonenzymatic release molecule. GGDM was highly toxic to TA98 cells in the
of DM (see below) in that the latter process is very slow, presence of GGT, where cell death depended on GGDM
whereas its reaction with DNA is instantaneous.
concentration at a constant concentration of GGT (Fig.
8A) or on GGT concentration at constant concentration
of GGDM (Fig. 8B). The cytotoxic agent is suggested
to be DM released from GGDM by GGT, in that TLC
Cytotoxicity of DM and GGDM
In view of the fact that GGDM had a significant capa- of reaction mixtures revealed that the appearance of
bility to bind DNA, we tested whether the molecule was spots with Rf corresponding to DM was GGT- and
cytotoxic as compared to DM. TA98 cells were exposed time-dependent (data not shown), and the activity spec-
to various concentrations of GGDM or DM without GGT, trum of the chloroform-soluble material was identical
and viability was followed with time. Figure 7 shows that to that of DM (Fig. 5A, top line). Prolonged incubation
GGDM at concentrations up to 200
but decreased growth rates in a concentration-dependent release of DM.
manner. DM at 0.9–3.5 M similarly decreased growth
rate and was cytotoxic at 7.2 M. The ability of GGDM
mM was not toxic, of GGDM alone also resulted in limited, nonenzymatic
m
m
GGT-Dependent Mutagenicity of GGDM
to decrease growth rate was approximately 60-fold less
than that of DM. This was unexpected, in that the concen-
Determination of mutagenicity is a more sensitive assay
trations of GGDM exceeded by far those required to satu- than cytotoxicity for the detection of the biological activ-
rate DNA. The half-life time of GGDM without GGT was ity of potent mutagens such as DM. Since the plate incor-
/ 8I31$$8011
11-23-98 17:02:02
wlemas
W Liss: EM EM-98011

384
Keren and Stark
Fig. 8. GGT-dependent cytotoxicity of GGDM in strain TA98. Expo-
nentially growing TA98 cells were diluted into fresh LB medium con-
taining 40 mM glycylglycine and GGDM in one of the following combi-
nations, shaken at 37
as described in Material and Methods. (A) Response to GGDM concen-
tration. Cultures contained ( ) 80 M GGDM; ( ) (top to bottom) 20
U/ml GGT and 10, 20, 40, and 80 M GGDM. (B) Response to GGT
concentrations. Cultures contained ( ) 20 M GGDM without GGT
and ( ) 20 M GGDM in the presence of 5, 10, and 20 U/ml GGT.
ЊC, and samples were withdrawn for viable counts
l
m
᭺
m
l
m
᭺
m
Presented are means of triplicates.
Fig. 7. Cytotoxicity of DM and GGDM in strain TA98 without activa-
tion. Exponentially growing TA98 cells were diluted into fresh LB
medium (adjusted to pH 8.5) containing GGDM or DM. Cultures were
and glycylglycine, whereas GGDM without GGT was not
mutagenic (Table I). The identity of the activity spectra
of authentic DM and the chloroform-soluble material re-
leased from GGDM nonenzymatically or by GGT (Fig.
5A), and the specific mutagenicity of the released material
which was indistinguishable from that of DM (Table I)
indicate that the released material was DM.
shaken at 37
scribed in Material and Methods. GGDM (
20, 60, 120, and 200 M. DM ( ) was at (top to bottom) 3.5 and 7.2
M. Presented are means of triplicates.
ЊC and samples were withdrawn for viable counts, as de-
᭺) was at (top to bottom) 0,
m
l
m
Although transport experiments with labeled GGDM
were not performed, the results presented here suggest
that the low toxicity/mutagenicity of GGDM was mainly
due to its transport. The less polar DM readily enters the
cells and binds DNA, resulting in cell death and mutagen-
esis. The more polar GGDM, although capable of DNA
binding, is not readily transported. The possibility of re-
lease of DM from GGDM by the bacterial, periplasmic
GGT is remote, in that coliforms contain negligible (less
than 0.1 mU/109 cells) GGT activity [Suzuki et al., 1986].
The transport properties of GGDM into mammalian cells
are, however, unknown. The results are also consistent
with the idea that in addition to the conversion of a pro-
poration assay involves prolonged exposure which would
lead to significant nonenzymatic release of DM, the enzy-
matic reactions of GGDM with GGT and the mutagenicity
assay were separated. Samples from the synthesis mixture
were withdrawn with time, the released DM was extracted
with chloroform, dried, dissolved in ethanol, and the con-
centration of DM was determined spectrophotometrically.
The solution was diluted 1:1 in water and mutagenicity
was determined in TA98 cells by the plate incorporation
assay. Table I shows that a slow release of DM from
GGDM occurred without GGT. The rate of release was
higher with GGT and further increased in the presence
of glycylglycine. The half-life of GGDM (calculated from
the remaining GGDM vs. time) was 102, 70, and 32 hr,
respectively. Likewise, mutagenicity of extracts from re-
actions with GGT was higher than from those without
drug into a biologically active compound, g-glutamylation
may increase the specificity of a chemotherapeutic agent
by biochemical targeting of their transport. The slow en-
GGT, and highest in extracts from reactions with GGT zymatic release of DM from its g-glutamyl derivative may
/ 8I31$$8011
11-23-98 17:02:02
wlemas
W Liss: EM EM-98011

Prodrug Activation by GGT
g-Glutamyl-daunomycin by GGT
385
TABLE I. Mutagenicity of Daunomycin Released From
4
m
l/plate
8
m
l/plate
16
ml/plate
Rxn
no.
nmol/
platea
Net Rev/
nmol
nmol/
Net Rev/
nmol
nmol/
platea
Net Rev/
nmol
GGT
Gly-gly
Rev/plate
platea
Rev/plate
Rev/plate
1
2
3
4
5
6
7b
(
(
(
0)
/
/
0)
/
/
0)
(
(
0
0
/
0
0
/
/
)
)
0.19
0.26
0.38
0.28
0.36
0.53
0.32b
129
222
266
190
287
532
50
{
{
{
{
{
{
{
32
12
20
36
41
34
10
426
669
574
507
664
913
0.38
0.53
0.76
0.57
0.72
1.06
0.64b
205
401
523
518
579
931
58
{
{
{
{
{
{
{
29
36
57
30
34
64
17
413
666
625
824
738
833
0.76
1.06
1.53
1.14
1.44
2.12
1.28b
367
750
998
738
952
1458
51
{
{
{
{
{
{
{
26
84
96
41
49
277
15
420
662
621
605
628
665
(
(
)
)
(
0)
(0)
(0)
Reaction mixtures contained 50 mM HEPES buffer pH 8.5, 400
6) 40 mM glycylglycine in a final volume of 1.5 ml. Samples (500
at 37 C. Samples were extracted four times with chloroform; the organic phases were pooled, evaporated, and dissolved in 200
activity spectrum (220–650 nm) of each sample was identical to that shown in Figure 5. Concentrations of DM in extracted samples were 235,
330, 478, 356, 444, and 661 M in reactions 1–6, respectively. Extracts were diluted 5-fold in ethanol:water (1:1) and 4, 8, and 16 l aliquots
were mixed with 1.6
108 TA98 cells in top agar and plated onto minimal plates. Further operations were as described in Materials and Methods.
The spontaneous mutant yield was 48 7 revertants/plate. Authentic DM at 1 and 2 nmol/plate yielded 702 54 and 1.256 131 revertants/
plate, respectively (mean 629 35 revertants/nmol).
aThe amount of DM was calculated from the concentrations of the extracted material.
bAliquots of an aqueous solution of 80
M GGDM were tested for mutagenicity as above.
m
M GGDM (Nos. 1, 4) and 20 U/ml GGT without (Nos. 2, 5) or with (Nos. 3,
l) were withdrawn after 48 hr (Nos. 1–3) and 96 hr (Nos. 4–6) incubation
l ethanol. The
m
Њ
m
m
m
1
{
{
{
{
m
radical generated from 4-(hydroxymethyl)-benzenediazonium
salt, a carcinogen in mushroom. Chem Res Toxicol 8:356–362.
Jolles B, Laigle A, Priebe W, Garnier-Suillerot A (1996): Comparison
of DNA sequence selectivity to anthracycline antibiotics and
be an advantage in that it may serve as a ‘‘time-release’’
capsule in chemotherapy.
their 3
؅
-hydroxylated analogs. Chem Biol Interact 100:165–176.
-glutamyl transferase (mushroom).
REFERENCES
Levenberg B (1970): Agaritine and
g
In: Colowick SP, Kaplan NO, (eds.), Methods in enzymology,
vol. XVII (A). New York: Academic Press, pp 877–879.
Levin Y, Sela B (1979): Studies on amino acid and peptide derivatives
of daunorubicin. FEBS Lett 98:119–122.
Bannasch P (1986): Preneoplastic lesions as end points in carcinogenic-
ity testing. I. Hepatic neoplasia. Carcinogenesis 7:698–695.
Boekelheide K, Graham DG, Mize PD, Koo EH (1980): Melanotoxicity
Malca-Mor L, Stark AA (1982): Mutagenicity and toxicity of carcino-
genic and other hydrazine derivatives: Correlation between toxic
potency in animals and toxic potency in Salmonella typhimurium.
Appl Environ Microbiol 44:801–808.
Maron DM, Ames BN (1983): Revised methods for the Salmonella
mutagenicity test. Mutat Res 113:173–215.
Pitot HC (1990): Altered hepatic foci: Their role in murine hepatocarci-
nogenesis. Annu Rev Toxicol Pharmacol 30:465–500.
Quigley GJ, Wang AH, Ughetto G, van der Marel G, van Boom JH,
Rich A (1980): Molecular structure of an anticancer drug-DNA
complex: Daunomycin plus d(CpGpTpApCpGp). Proc Natl Acad
Sci USA 77:7204–7208.
Sarma DSR, Rao MS, Rajalakshmi S (1986): Liver tumour promotion
by chemicals: Models and mechanisms. Cancer Surv 5:781–798.
Selvaraj P, Balasubramanian KA, Hill PG (1981): Isolation of gamma-
glutamyl transpeptidase from human primary hepatoma and com-
parison of its kinetic and catalytic properties with the enzyme
from normal adult and fetal liver. Enzyme 26:57–63.
and mechanism of activation of g-L-glutaminyl-4-hyrdoxyben-
zene. J Invest Dermatol 75:322–327.
Burger PC, Kemper LAK, Vogel SF (1979): The cytotoxicity of g-L-
glutaminyl-4-hydroxybenzene for cells that contain tyrosinase, a
study of melanocytes in the hair follicle of the mouse. Cancer
Res 39:1494–1503.
Chaires JB, Satyanarayana S, Suh D, Fokt I, Przewloka T, Priebe W
(1996): Parsing the free energy of anthracycline antibiotic bind-
ing to DNA. Biochemistry 35:2047–2053.
Chen YM, Lim BT, Chavin W (1979): Serum tyrosinase in malignant
disease, its activity, and the electrophoretic patterns of the en-
zyme as carried by immunoglobulins. Cancer Res 39:3485–
3490.
Farber E (1986): Some emerging general principles in the pathogenesis
of hepatocellular carcinoma. Cancer Surv 5:695–718.
Gao YG, Priebe W, Wang AH (1996): Substitutions at the C2؅ of
daunosamine in the anticancer drug daunorubicin alter its DNA-
binding specificity. Eur J Biochem 240:331–335.
Stark AA (1991): Oxidative metabolism of glutathione by g-glutamyl
Gerber MA, Thung SN (1980): Enzyme pattern in human hepatocellular
carcinoma. Am J Pathol 98:395–400.
Hanigan MH, Pitot HC (1985): g-Glutamyl transpeptidase: Its role in
transpeptidase and peroxisome proliferation: The relevance to
carcinogenesis. A hypothesis. Mutagenesis 6:241–245.
Stark AA, Zeiger E, Pagano DA (1993): Glutathione metabolism by g-
hepatocarcinogenesis. Carcinogenesis 6:165–172.
Hiramoto K, Kaku M, Kato T, Kikugawa K (1995a): DNA strand break-
ing by the carbon-centered radical generated from 4-(hydroxy-
methyl)-benzenediazonium salt, a carcinogen in mushroom Aga-
ricus bisporus. Chem Biol Interact 94:21–36.
glutamyl transpeptidase leads to lipid peroxidation: Characteriza-
tion of the system and relevance to hepatocarcinogenesis. Carci-
nogenesis 14:183–189.
Suzuki H, Kumagai H, Tochikura T (1986): g-Glutamyltranspeptidase
Hiramoto K, Kaku M, Sueyoshi A, Fujise M, Kikugawa K (1995b):
DNA base and deoxyribose modification by the carbon-centered
from Escherichia coli K-12: Formation and localization. J Bacte-
riol 168:1332–1335.
/ 8I31$$8011
11-23-98 17:02:02
wlemas
W Liss: EM EM-98011

386
Keren and Stark
Tsao B, Curthoys NP (1980): The absolute asymmetry of orientation of Walton K, Coombs MM, Catterall FS, Walker R, Ioannides C (1997):
Bioactivation of the mushroom hydrazine, agaritine, into inter-
mediates that bind covalently to proteins and induce mutations
in the Ames test. Carcinogenesis 18:1603–1608.
g
-glutamyl transpeptidase and aminopeptidase on the surface of
rat renal brush border membrane. J Biol Chem 255:7708–7711.
Vogel SF, Kemper LAK, Jeffs PW, Cass MW, Graham DG (1977):
g
-
Westendorf J, Marquardt H, Marquardt H (1984): Structure-activity
relationship of anthracyclin-induced genotoxicity in vitro. Cancer
Res 44:5599–5604.
L-glutaminyl-4-hydroxybenzene, an inducer of cryptobiosis in
Agaricus bisporus and a source of specific metabolic inhibitors
for melanogenic cells. Cancer Res 37:1133–1136.
Williams DH, Fleming I (1995): Spectroscopic methods in organic
Vogel SF, Kemper LAK, Boekelheide K, Graham DG, Jeffs PW (1979):
chemistry, 5th edition. London: McGraw-Hill.
Intracellular activation of g-L-glutaminyl-4-hydroxybenzene by
Accepted by—
tyrosinase, a mechanism for selective cytotoxicity against mela-
nocarcinoma. Cancer Res 39:1490–1493.
E. Zeiger
/ 8I31$$8011
11-23-98 17:02:02
wlemas
W Liss: EM EM-98011