Glutathione Conjugation of the Cytostatic Drug Ifosfamide and the

Georgia Anderer , Martin Schrappe , Annette Brechlin , Melchior Lauten , Paola Muti , Karl Welte , Martin Stanulla. Pharmacogenetics 2000 10 (8), 715-...
0 downloads 0 Views 925KB Size
Download PDF
Chem. Res. Toxicol. 1995,8, 979-986

979

Glutathione Conjugation of the Cytostatic Drug Ifosfamide and the Role of Human Glutathione S-Transferases Hubert A. A. M. Dirven,? Luc Megens, Martin J. Oudshoorn, Maria A. Dingemanse, Ben van Ommen, and Peter J. van Bladeren" Division of Toxicology, TNO Nutrition and Food Research Institute, P.O. Box 360, 3700 AJ Zeist, The Netherlands Received November 15, 1994@

Development of drug resistance against alkylating cytostatic drugs has been associated with higher intracellular concentrations of glutathione (GSH) and increased expression of glutathione S-transferase (GST) enzymes. Therefore, enhanced detoxification by the glutathione/glutathione S-transferase pathway has been proposed a s a major factor in the development of drug resistance toward alkylating agents. In this paper we describe 31PNMR and HPLC studies on the spontaneous and glutathione S-transferase catalyzed formation of glutathionyl conjugates of two metabolites of ifosfamide, i.e., 4-hydroxyifosfamide and ifosfamide mustard. At 25 "C activated ifosfamide (=4-hydroxyifosfamide aldoifosfamide) disappeared faster in the presence of a 10-fold excess of GSH (t112= 107 min) compared to incubations without GSH (t112 = 266 min). No evidence for the formation of 4-glutathionyl ifosfamide was found. The ultimate alkylating species of ifosfamide is ifosfamide mustard (IM). In the absence of glutathione, the rate constant for the disappearance of the ifosfamide mustard signal a t 25 "C (pH 7) was 1.98 x min-l (tv2 = 350 min). In the presence of a 10-fold molar excess of glutathione, this rate constant was 1.95 x min-l (tv2 = 355 min), indicating that the spontaneous formation of a n aziridinium ion is the rate-limiting event in the reaction with glutathione. The aziridinium ion formed from IM can deprotonate upon formation, leading to the formation of a (noncharged) aziridine species. This intermediate (N-(2-chloroethyl)-N'phosphoric acid diamide) was characterized by 31P,lH, and 13C NMR spectra. When 2 mM ifosfamide mustard was incubated with 1 mM GSH in the presence of 40 ,uM GST Pl-1, the formation of monoglutathionyl ifosfamide mustard was 2.3-fold increased above the spontaneous level. The other major human isoenzymes tested (Al-1, A2-2, and M l a - l a ) did not influence the formation of monoglutathionyl ifosfamide mustard. The results of these studies demonstrate that increased levels of GST P1-1 can contribute to an enhanced detoxification of ifosfamide.

+

Introduction Cyclophosphamide (CP)l and ifosfamide (IF) are both alkylating, cytostatic drugs, with the same molecular weight and sum formula. The key difference between these two compounds is the position of one of the two chloroethyl groups. The two 2-chloroethyl groups of CP are both linked to the extracyclic nitrogen, while in IF one 2-chloroethyl group is linked to the extracyclic nitrogen and the other is located at the endocyclic nitrogen. This minimal difference results in markable changes in the pharmacokinetics and pharmacodynamics (review: ref 1). IF is used extensively in the treatment of solid and hematological malignancies. Preclinical tests have shown that IF has a broader antitumoral spectrum * Address correspondence to this author, at The Division of Toxicology, TNO Nutrition and Food Research Institute, P.O. Box 360, 3700 AJ Zeist, The Netherlands. FAX 31 30 695 7224. Present address: Nycomed Imaging, Department of Toxicology, Oslo, Norway. @Abstractpublished in Advance ACS Abstracts, August 15, 1995. Abbreviations: CP, cyclophosphamide; IF, ifosfamide; 4-OOHIF, 4-hydroperoxyifosfamide; 4-OHCP, 4-hydroxycyclophosphamide; aldoIF, aldoifosfamide; PM, phosphoramide mustard; 4-OHIF, 4-hydroxyifosfamide; IM, ifosfamide mustard; FAB-MS, fast atom bombardment mass spectrometry; CGSCP, 4-glutathionyl cyclophosphamide; IMGS, monoglutathionyl ifosfamide mustard; GST, glutathione Stransferases; HPLC-LS, high-performance liquid chromatographyliquid scintillation; CDNB, l-chloro-2,4-dinitrobenzene. +

than cyclophosphamide. IF was also proven to be effective in some cyclophosphamide-resistant experimental tumors (review: ref 1). The essential features of the metabolism of IF parallel those of CP. Cytochrome P450 3A4 ( 2 )dependent oxidation provides 4-hydroxyifosfamide(4-OHIF),which then undergoes a spontaneous and reversible ring-opening reaction to form aldoifosfamide (aldoIF) (Figure 1). AldoIF can fragment to acrolein and the ultimate alkylating agent ifosfamide mustard (IM). In analogy to phosphoramide mustard it is assumed that IM alkylates DNA through the formation of an intermediate, i.e., an aziridinium ion. IF can also undergo cytochrome P450 mediated N-dechloroethylation reactions, leading to the formation of chloroacetaldehyde and 2- and 3-dechloroethyl-IF (2).These N-dechloroethylation reactions play a more important role in the metabolism of IF compared to CP (review: ref 1). Development of drug resistance in tumor cells is a major problem in chemotherapy. Inactivation by both nonenzymatic and glutathione S-transferase (GST) catalyzed glutathione (GSH) conjugation has been postulated as a mechanism in the development of drug resistance of tumor cells toward alkylating agents. The evidence for this relation, however, is largely qualitative. In many resistant cell lines or clinical samples intracellular levels

0893-228x/95/2708-0979$09.00/0 0 1995 American Chemical Society

Dirven et al.

980 Chem. Res. Toxicol., Vol. 8, No. 7, 1995

4-hydroxyifosfamide

ifosfa m i d e (IF)

(4-OH IF)

aldoifosfamide (aldo-IF)

hemiglutathionyi acetal ifosfamide

U

NH i-P' 8'OH

/ GS

-

dig Iu tath i o n y l i f o s f a m i d e m u s t a rd IM(GS) 2

GS

CI GSH

H

#

monoglutathionyl ifosfa m ide m u s t a r d (IMGS)

N (2-chioroethyi) "(aziridinyl) i f o s f a m ide m u s t a r d

aziridinium intermediate

Figure 1., Metabolism of ifosfamide (simplified) and possible conjugation reactions with glutathione.

of GSH and total cytosolic GST activity appear to be higher (review: ref 3). GSH is a n intracellular cysteinecontaining tripeptide present a t a concentration of 5-10 mM in most mammalian cells (review: ref 3). A concentration of 4-40 pg of GSTlmg of cytosolic protein has been reported in tumors or tumor cell lines (4-6). This corresponds to an intracellular GST concentration of 30240 pM. Two of the metabolites of ifosfamide might be expected to react with the cell-protective nucleophile GSH, i.e., 4-OHIF and IM. The hydroxy group of 4-OHIF might be substituted by GSH like it is in 4-glutathionyl cyclophosphamide (7, 8). The aziridinium ion derived from IM can react with GSH. Reaction of this intermediate species with GSH would yield the monoglutathionyl conjugate (IMGS). This monoconjugate could react further to yield the diconjugate (IM(GS)z). Acrolein and chloroacetaldehyde can react with GSH as well, but are thought not to be important for cytostatic action. In our studies, we describe the formation of glutathionyl conjugates of alkylating drugs, the reaction kinetics, and the relative contribution of purified GST isoenzymes to these reactions. The ultimate goal of these experiments is to elucidate if the measured levels of GSH and GST in resistant tumor cells are sufficient to detoxify the active alkylating species. In previous studies on cyclophosfamide we found two types of glutathione conjugates, i.e., 4-glutathionyl cyclophosphamide and monoglutathionyl and diglutathionyl conjugates of phosphoramide mustard (7,8). In studies with purified GST's, all major human GST's were able to catalyze the formation of 4-glutathionyl cyclophosphamide, but only GST A l - l ( a ) was able to enhance the formation of monoglutathionyl phosphoramide mustard (8). In the present report, we describe the results of our studies on the formation of glutathione conjugates of IF metabolites. No evidence for the formation of 4-glu-

tathionyl ifosfamide was found. However, monoglutathionyl and diglutathionyl ifosfamide mustard conjugates were found, and the formation was described kinetically using 31PNMR. We found a catalytic effect of GST P l - l ( n ) on the rate of formation of monoglutathionyl ifosfamide mustard. These results are discussed in relation with findings of studies on the glutathionelglutathione S-transferase dependent biotransformation of other alkylating cytostatic drugs.

Materials and Methods Chemicals. 4-Hydroperoxyifosfamide (4-OOHIF, cis-isomer, 3-(2-chloroethyl)-2-((2-chloroethyl)amino)tetrahydro-2-0~0-~1,3,2-oxazaphosphorin-4-yl hydroperoxide) and ifosfamide mustard (N,N'-bis(2-chloroethyl)phosphorodiamidicacid) were kindly provided by Dr. J. Pohl from ASTA Medica, Frankfurt am Main, Germany. 4-OOHIF is used as a precursor of 4-OHIF, because of its higher stability in crystalline state and easier synthesis. Glutathione was from Boehringer Mannheim GmbH (Germany), glutathione reductase and NADPH were obtained from Sigma Chemical Co. (St. Louis, MO). [35S]GSHwas obtained from Du Pont de Nemours, NEN division (Den Bosch, The Netherlands). All other chemicals used were of the highest purity obtainable. GST Purification and Assay. GST isoenzymes were purified from human liver and human placenta using affinity chromatography and chromatofocusing a s described previously (8). GST activity was assayed using l-chloro-2,4-dinitrobenzene (CDNB) as a substrate (9). Specific activities (pmoli(min.mg of protein)) were 59, 14, 84, and 109, respectively, for GST Al-1, GST A2-2, GST Mla-la(pu), and GST P1-1. All enzyme concentrations were expressed as the concentration of the subunit (M, 25 900, 25 900, 26 700, and 24,800, respectively, for subunits Al, A2, M l a , and P1). NMR Experiments. lH, 31P,and I3C NMR spectra were collected on a Varian Unity 400 spectrometer (resonance frequency 400, 161.9, and 100.6 MHz, respectively). A switchable 5 mm probe was used, which was held at constant temperature.

Chem. Res. Toxicol., Vol. 8, No. 7, 1995 981

GSH Dependent Biotransformation of Ifosfamide In 31P NMR experiments, data accumulation was initiated using a 25 kHz spectral window, 64K data points, 45" pulse angle, and low power 'H waltz decoupling corresponding to 0.5 W decoupling power. This power gave complete decoupling with minimal dielectric heating of the sample. Pulse recycle time was 3.28 s. Number of transients was 128. FID's were exponentially multiplied, resulting in an additional 1 Hz line broadening in the frequency-domain spectra. All chemical shifts are reported with 85% H3P04 in DzO as external reference. The I3C NMR spectra were recorded with a 25 kHz spectral window, 'H waltz decoupling power, and a pulse repetition time of 3.3 s. A total of 21 000 scans were collected and Fourier transformed with a 1 Hz line broadening. In 'H NMR experiments data accumulation was initiated using a 8 kHz spectral window solvent suppression and a 45" pulse angle. Pulse recycle time was 6.1 s, and the number of transients was 1024. FID's were exponentially multiplied, resulting in a n additional 1Hz line broadening in the frequencydomain spectra. All chemical shifts are reported with the water signal set at 4.7 ppm. Analysis of spectra in kinetic experiments was performed using the deconvolution program of t h e Varian NMR software, assuming Lorentzian lineshapes. Incubations were carried out in 0.07 M phosphate buffer (pH 7.0) with 5 mM EDTA, 3 mM NADPH, 0.8 unit/mL glutathione reductase, and 10% D20, 6 mM IM or 4-OOHIF, with and without 60 mM GSH. Samples were placed in the spectrometer, which was kept at constant temperature. The 31Pmeasurements were started when the lock signal indicated that the sample temperature was constant. Ten measurements of 6 min were made, followed by measurements of 24 min. Peaks i n the spectra were quantified using deconvolution. From these data, rate constants were determined by fitting an appropriate kinetic formula to these data, using the curve fit option of Slidewrite plus version 5.0 (Advanced Graphics Software, Carlsbad, CAI. For IM rate constants were derived from a least-squares fitting of the observed concentrations of IM (A), IMGS ( B ) ,and IM(GS)2 (C) according to:

A(t) = A, exp( -kdt)

(1)

is the rate constant of the disappearance of IM, k1 is the observed rate constant of formation of IMGS, k d , 2 is the rate constant of the disappearance of IMGS, and k~ is the observed rate constant of the formation of IM(GS)2. Water and GSH react competitively with the aziridinium intermediate of ifosfamide mustard according to a second-order rate law. The rates of formation however are expressed as if formation was a firstorder reaction. Rate constants for the reactions of 4-OHIF were derived from a least-squares fitting of the data points to the following equations:

kd

A(t) = A, exp( -kdt)

(4)

+

[trans-4-OHIFl + [aldoIFl; B = [IMI; k d is the rate constant for the disappearance of A, k1 is the rate constant for the p-elimination of acrolein and kz is the rate constant for the disappearance of IM. Half-lives were derived from the rate constants ( t l l ~= [-ln( W Y k 1.

A = [cis-4-OHIFl

Experiments with Purified Glutathione S-Transferases. Incubations were carried out i n 0.07 M phosphate buffer (pH 7.0) with 5 mM EDTA, 3 mM NADPH, 0.8 unit"L glutathione reductase, and 1 mM GSW[35Sl GSH, with or without 40 ,uM GST. Reactions were started with the addition

of IM, which was dissolved immediately before use. Incubation temperature was 37 "C. Incubations were terminated by the addition of N-ethylmaleimide until a final concentration of 10 mM and were stored at -20 "C until analysis by HPLC with a n on-line radiometric detector. The HPLC system used consisted of a Pharmacia HPLC pump 2248 (Uppsala, Sweden) equipped with a flow-through radioactivity detector (Radiomatic Flo-one A500, Meriden, USA). A 2 mL flow cell was used. As scintillation cocktail, Flo-Scint A (Packard Instrument, Groningen, The Netherlands) was used with a flow of 2 m u m i n . The formation of mono- and diglutathione conjugates of ifosfamide mustard was assayed on a Mono Q HR 515 anion exchange column (Pharmacia LKB, Uppsala, Sweden). Eluent A was 0.02 M Tris buffer (pH 6.8) and eluent B was 0.02 M Tris buffer (pH 6.81, with 0.15 M NaC1. Flow was 1 m u m i n . The solvent program started isocratically with 100% A for 5 min, followed by a linear gradient to 15% B in 10 min.

Fast Atom Bombardment Mass Spectrometry (FABMS). Peaks used for quantification of IMGS were isolated, lyophilized, and characterized with FAB mass spectrometry. Mass spectra were acquired on a Finnigan-MAT 900 spectrometer with a magnetic scan range of 50-1000 amu. The primary energy of the cesium ionizing beam was approximately 70 eV. Accelerating voltage was 5 kV, dynode voltage was 16 kV, and the electron multiplier voltage was 1.9 kV. All measurements were performed at a resolution of 1100 (10%) with a source temperature of 50 "C. Glycerol was used a s matrix.

Results Experiments with 4-Hydroxyifosfamide. cis-4Hydroperoxyifosfamide was used as a precursor for 4-OHIF. After treatment with dimethylsulfide (used to reduce the hydroperoxy function) the reactions of 4-OHIF and its products in time were monitored with 31PNMR. The signal assigned to cis-4-OHIF (6 13.58) disappeared, while signals assigned to the trans-isomer (6 13.181, aldoIF (6 18.63), and IM (6 13.5) increased (for chemical structures, see Figure 1). After 120 min the relative proportions of cis-4-OHIF, trans-4-OHIF, aldoIF, and IM were 36:30:17:17. The total peak area of cis-4-OHIF, trans-kOHIF, and aldoIF decreased according to firstorder kinetics (k = 2.6 x min-', t l l z = 266 min). In incubations of 6 mM 4-OHIF with 60 mM GSH the rate of disappearance was 2.5-fold higher than without GSH (k = 6.5 x min-', tl/z = 107 min). No signals were found which could be assigned to the GSH conjugate a t the C-4 position of IF, analogous to 4-glutathionyl cyclophosphamide. Using FAB-MS, samples were studied for the presence of GSH conjugates on the C-4 position. No evidence for their formation was found. After 120 min the relative proportions of cis-4-OHIF, trans-kOHIF, aldoIF, and IM were 21:16:20:43. AldoIF undergoes p-elimination of acrolein to yield IM. The rate constant for the formation of IM in incubations without GSH was 2.2 x min-l. In incubations with GSH the formation of IM was approximately 2.5 times as fast (rate constant = 5.9 x min-'1. In the presence of GSH the signal assigned to aldo-IF (18.63 ppm) was approximately 50% higher a t its maximum height than in incubations without GSH. Experiments with Ifosfamide Mustard. The spontaneous reactions of IM with GSH were monitored using 31PNMR spectroscopy. An initial concentration of 6 mM IM made it possible to detect even minor reaction products within a measuring time that allowed obtaining enough time points for kinetic analysis. In the 31PNMR spectra a signal a t 6 13.5 was found upon dissolving of IM. Based on their formation in time,

982 Chem. Res. Toxicol., Vol. 8, No. 7, 1995

Dirven et al.

B

A

I

i

r

80

0

240

160

320

400

0

time (min)

600

1200

1800

2400

3000

time (min)

Figure 2. (A) 31PNMR derived time course for the breakdown of ifosfamide mustard in 0.07 M phosphate buffer (pH 7.0, 25 "C). (B) 31PNMR derived time course for the breakdown of ifosfamide mustard in the presence of a 10-fold molar excess of glutathione in 0.07 M phosphate buffer (pH 7.0; 25 "C). Legend: ( 0 )ifosfamide mustard; (+I monohydroxy ifosfamide mustard (6 14.3); (v) dihydroxy ifosfamide mustard (6 13.8); (0) monoglutathionyl ifosfamide mustard; ( 0 ) diglutathionyl ifosfamide mustard; ( 0 ) monohydroxy,monoglutathionyl ifosfamide mustard (6 14.36); (A)product a t 6 17.5 (N-(2-chloroethyl)-N'-ethanophosphoric acid diamide). 1

-/

-\------

I "

17 5

17 0

16 5

16 0

15 5

' I

15 0

" " "

14

5

L

' I " ' ' , ' ' '

14 0

234

'

13 5 opm

Figure 3. Stack plot of 161.9 MHz 31PNMR spectra recorded during the reaction of 6 mM IM with a 10-fold molar excess of glutathione. The time interval between each time-averaged spectrum was 48 min. Peak 1 is the product at 6 17.5 N-(2-chloroethyl)N'-ethanophosphoric acid diamide, peak 2 is diglutathionyl ifosfamide mustard, peak 3 is monoglutathionyl ifosfamide mustard, and peak 4 is ifosfamide mustard. Table 1. 31PNMR Derived Rate Constants (Expressed as 10-smin-l) for the Reactions of Ifosfamide Mustard under Various Conditions in 0.07 M Phosphate Buffer (pH 7.0)" 6 mM 6 mM 6 mM 6 mM 6 mM

IM, 25 "C IM 60 mM GSH, 25 "C IM 60 mM GSH, 37 "C IM 6 mM GSH, 37 "C IM 60 mM GSH 10 pM GST A l - l , 3 7 "C

+ + + +

+

1.98 f 0.09 1.95 f 0.01 14.1 f 0.6 14.1 f 0.6 15.9 f 0.6

1.75 f 0.06 13.4 f 0.5 3.0 f 0.2 14.1 f 0.6

1.05 f 0.02 6.4 f 0.5 0.5 f 0.7 6.6 f 0.3

0.79 f 0.02 5.3 f 0 . 2 ndb 5.7 f 0.05

kd,l is the rate constant for the disappearance of IM; k1 is the rate constant for the formation of monoglutathionyl IM; k d , 2 is the rate constant for the disappearance of IMGS; kp is the rate constant for formation of diglutathionyl IM. nd, not determined.

signals of the hydrolysis products with water were identified. The signal a t 6 14.3 was assigned to monohydroxy ifosfamide mustard and the signal a t 6 13.8 to dihydroxy ifosfamide mustard. A relatively large signal a t 6 17.5 was also found. This signal was assigned to a (noncharged) aziridine intermediate (Figure 1). The identification of this intermediate (N-(2-chloroethyl)-N'ethanophosphoric acid diamide) is described below. All signals were quantitated by peak intensity, and the results are shown in Figure 2A. The rate constant (kd) for disappearance of the IM signal was 1.98 x min-' (25 "C), and the calculated half-lifetime was 350 min. When IM was incubated with a 10-fold molar excess of GSH, two signals were found in the 31PNMR spectra which were not present in spectra of incubations without GSH (Figure 3). The signal a t 6 13.60 was assigned to monoglutathionyl ifosfamide mustard (IMGS), and the

signal a t 6 13.76 ppm was assigned to diglutathionyl ifosfamide mustard (IM(GS)Z). The formation of these two products was confirmed by FAELMS ( m l z 491 for monoglutathionyl IM and m l z 762 for diglutathionyl IM). The reaction of 6 mM IM and 60 mM GSH is shown in Figure 2B. The intensity of the peak a t 6 17.5 was smaller in incubations with GSH than in incubations without GSH. The rate constant (kd) for disappearance of the IM signal was 1.95 x min-', and the calculated half-lifetime was 355 min. Also, rate constants for the disappearance and formation of IMGS and IM(GS)zwere calculated at both 25 and 37 "C (Table 1).The disappearance of IM is approximately twice as fast as the disappearance of IMGS. Formation of IMGS is dependent on the concentration GSH: with 60 mM GSH and 6 mM IM the rate of formation of IMGS is more than 4 times as high as with 6 mM IM and 6 mM GSH (Table

Chem. Res. Toxicol., Vol. 8, No. 7, 1995 983

GSH Dependent Biotransformation of Ifosfamide 50

I

* I

50

-

40

-

v)

0

2 5,

20

30-

20 10

10

n

0

0

- GST

A 1 -1

A2-2

hila-la

10

20

30

40

50

70

60

80

P1-1

time (min)

Figure 4. Formation of monoglutathionyl ifosfamide mustard in incubations of 2 mM IM and 1 mM GSH without or with 40 pM glutathione S-transferases. Incubations were performed for 1 h a t 37 "C. The formation of the conjugate with enzyme was compared to the spontaneous formation by means of the paired t-test. P values found were 0.75, 0.24, 0.34, and 0.004, respectively, for GST Al-1, A2-2, M l a - l a , and P1-1.

" 30

O

I

v)

1). The rate of disappearance of IM was not influenced. IMGS reacts with water to a mixed hydroxy, glutathionyl conjugate (6 14.36) or with GSH to yield the diglutathionyl conjugate of IM. This diglutathionyl conjugate of IM is a rather stable compound. After 48 h a t room temperature there was no significant degradation. The addition of 10 pM GST Al-1 did not influence the rate of disappearance of IM or the rate of formation of IMGS (Table 1). To further study the influence of GST's on the formation of IMGS, a HPLC procedure for the quantification of IMGS was developed using lower concentrations of both IM and GSH. Using anion-exchange chromatography in combination with on-line radiometric detection, a peak was found with a retention time of 11.0 min. With FAB mass spectrometry this peak was identified as monoglutathionyl ifosfamide mustard ( ml z 491). When 2 mM IM was incubated with varying concentrations of GSH (range 0.2-5 mM), a GSH concentration-dependent increase in the formation of IMGS was observed (results not shown). The role of GST transferases on the formation of IMGS was studied by incubation of 2 mM IM and 1 mM GSH for 60 min in the absence or presence of 40 pM purified GST. In Figure 4 it is shown that in the presence of GST P1-1 the formation of IMGS is 2.3-fold enhanced above the spontaneous levels. GST Al-1, A2-2, and Mla-la did not influence the rate of formation of IMGS. The spontaneous and GST P1-1 catalyzed formation of IMGS was studied in time (Figure 5A). At all time points studied, the presence of 40 pM GST P1-1 increased the formation of IMGS 1.8-2.9-fold above the nonenzymatic levels. In addition, the influence of different concentrations of GST P1-1 on the formation of IMGS was studied (Figure 5B). With all concentrations of GST P1-1 tested, a n increased formation of IMGS was found. Since tumors might have a poorly organized vasculature, necessitating anaerobic glycolysis with the production of lactic acid, the pH in some parts of a tumor might be lower than in normal tissues (10).So we studied the formation of IMGS in the pH range 5.5-8 (Figure 6). At pH 6, the presence of 40 pM GST P1-1 increased the formation of IMGS 4.3-fold above the spontaneous levels. At pH 8 the formation of monoglutathionyl ifosfamide

d

z f

20

10

0 10

0

20

40

30

50

pM GST P i - 1 Figure 5. (A) Time course of formation of monoglutathionyl ifosfamide mustard in an incubation mixture of 2 mM IM and 1 mM GSH a t 37 "C with and without 40 pM GST P1-1. Legend: (m) without enzyme; ( 0 )with GST P1-1. (B) Formation of monoglutathionyl ifosfamide mustard in the presence of varying Concentrations of GST P1-1. The nonenzymatic formation of IMGS is subtracted from all values. 2 mM IM and l mM GSH were incubated for 1h a t 37 "C. 40

I

0' 5.00

T

5.50

6.00

8.50

7.00

7.50

8.00

8.50

PH Figure 6. The nonenzymatic and glutathione S-transferase P1-1 catalyzed formation of monoglutathionyl ifosfamide mustard at pH 5.5, 6 , 6.5, 7, and 8. 2 mM IM and 1mM GSH was incubated for 1 h at 37 "C with and without 40 pM GST P1-1. Legend: (m) without enzyme; ( 0 )with GST P1-1.

mustard was 2-fold increased above spontaneous levels by GST P1-1. Identification of N-(2-Chloroethyl)-N'-ethanophosphoric Acid Diamide. Upon dissolving of 6 mM IM in 0.07 M phosphate buffer (pH 7), a signal was found a t 17.5 ppm (Figure 7A). The relative intensities of the signals a t 17.5, 14.3, 13.8, and 13.54 ppm were 22:7:7:64. The lH NMR spectrum of this mixture is

Dirven et al.

984 Chem. Res. Toxicol., Vol. 8, No. 7, 1995

the spontaneous and GST catalyzed reactions of GSH with 4-OHIF and IM, two metabolites of IF. In contrast to CP, no evidence for the formation of a glutathione conjugate on the C4 position of IF was found. Kwon et al. (12) described two mechanisms for the formation of 4-thioCP conjugates. One mechanism ( a base-catalyzed elimination of a proton from the endocyclic nitrogen, leading to the formation of iminocyclophosphamide) is not possible with IF, since no proton is present on the endocyclic nitrogen (Figure 1). The other mechanism involves the formation of a hemithioacetal from aldoifosfamide that is subsequently cyclized to 4-thioI conjugates (Figure 1). -R.i*ri L .& L.+,".In our experiments the sum of the signals assigned to ' ~ ~ ' " ' , " ' 1 ' " ' , 4-hydroxyifosfamide and aldoifosfamide disappeared faster 17 5 17 0 16 5 16 0 15 5 15 0 14 5 14 0 13 5 opm in the presence of GSH than without glutathione. This might indicate the formation of the glutathionyl aldoiB fosfamide conjugate. The signal a t 6 18.63 was higher N-CH -CH -CI N-CH -CH -CI N/cHz in incubations with GSH compared to incubations with4 H , out GSH. This signal might be assigned to aldoIF, aldoIF hydrate, or hemithio(=glutathionyl)acetal aldoIF (13). Presumably because of sterical hindrance by the chloroethyl group at the endocyclic nitrogen atom, ring closure to 4-glutathionyl ifosfamide is prevented (Figure 1). An alternative explanation for our observations is a nucleophilic attack of GSH at position 6 of aldoIF with the direct formation of an acrolein-GSH conjugate and IM. This might explain the increased concentration of IM as found in our experiments. It has been proposed that 4-OHIF can enter cells, while IM cannot enter cells due to its polar nature (14). So 4-OHIF might thus be considered as a transport form of IM across cell membranes. IM is formed in the cytoplasm 3 8 3 6 3 4 3 2 3 0 2 8 2 6 2 4 2 2 2 0 pom of normal as well as malignant cells. The disappearance Figure 7. The 161-MHz 31PNMR spectrum (A) and the 400of IM in 0.07 M phosphate buffer (pH 7.0) was much MHz IH NMR spectrum (B) of a mixture of 6 mM ifosfamide slower than that of PM. Half-lifetimes were 350 min for mustard in 0.07 M phosphate buffer (pH 7.0). Based on relative IM and 128 min for PM (7). IM should therefore survive intensities and the presence of characteristic aziridine signals in the IH NMR spectrum, the signal at 17.5 ppm in the 31PNMR longer in the cytoplasm of cells and might have a greater spectrum was assigned to N-(2-chloroethyl)-N'-ethanophosphoric efficacy than PM. The presence of GSH did not acceleracid diamide. ate the rate of disappearance of IM. Similar findings have been reported for the reaction of PM with glushown in Figure 7B. A characteristic doublet for an tathione (7),indicating that the rate-limiting event in the aziridine group was found a t 1.82 ppm with a 3 J p of ~ 13.7 glutathione conjugation reactions of both IM and PM is Hz. In a lH NMR spectrum of thiotepa, a compound with the formation of the aziridinium intermediate. Yuan et 3 aziridine groups linked to a phosphorus atom, a al. (15) demonstrated that the reaction of PM with comparable doublet was found a t 2.2 ppm with a 3 J p of ~ glutathione proceeds not by a direct displacement of 16.5 Hz. In the 13CNMR spectrum of IM, a signal was chloride by the thiol group, but rather by a reaction of found a t 24.3 ppm, also characteristic for an aziridine glutathione with a reactive aziridinium ion formed from group (11). PM. In relatively low chloride concentrations this reacThe relative intensities of the signals in the 'H NMR tion is irreversible (16). The chemical structure of the spectrum a t 1.82, 3.0 and 3.5 ppm were 24~38~38. This aziridine species formed from IM is different from the indicates that the content of N-(2-chloroethyl)-N'-ethaaziridine species formed from PM (17). PM gives rise to nophosphoric acid diamide in the mixture is 24%. Based an aziridinium ion where, in addition to phosphorus, on the relative intensities in the 31PNMR spectrum, the three carbon atoms are attached t o nitrogen. This content of N-(2-chloroethyl)-N'-ethanophosphoric acid charged ion is highly reactive and will react immediately diamide in the mixture is 22%. with nucleophiles. This species has a very short halfBased on the presence of characteristic aziridine lifetime, and it is unlikely to be detectable by 31PNMR signals in both 'H NMR and 13C"VIR spectra, these data spectroscopy. In contrast, IM gives rise to an aziridinium provide evidence for the formation, in solution, of N42ion where, in addition to phosphorus, two carbon atoms chloroethy1)-N'-ethanophosphoric acid diamide. and one proton are attached to the ring nitrogen. Upon formation this aziridinium ion will deprotonate, leading to the formation of a aziridine species, i.e., N-(2-chloroDiscussion ethyl)-N'-ethanophosphoric acid diamide (Figure 1)(17). This type of intermediate is relatively stable as judged The molecular basis 'of acquired resistance toward from our experiments with thiotepa, a compound with alkylating cytostatic drugs is not known. We studied the three noncharged aziridinyl groups linked to a phosphopossible role of glutathione and glutathione S-transrus atom (18). Based on relative intensities and the ferases in the development of resistance by examining

A

LA;

I

I

.

-

,

I '

" I

'

I

"

'

,

'

'

,

'

1

GSH Dependent Biotransformation of Ifosfamide presence of characteristic aziridine signals in lH and 13C NMR spectra, the signal a t 6 17.5 was assigned to this species. This signal was found in experiments with IM as well as in experiments with 4-OHIF. This signal has not been reported by Boal et al. (17) in their study of alkylation kinetics of IM, probably because it was outside their spectral window of 5 kHz. The experiments of Boal et al. (17 )were performed a t pH 7.4 and 9.4. Of course, the pH of the reaction medium will determine the extent of deprotonation. Once this aziridine species is formed, it can react nonenzymatically with nucleophiles like water or glutathione. The reaction with glutathione is favored since the pKa value of glutathione is approximately 1000-fold lower than the pKa value of water. If higher concentrations of glutathione are present, more monoglutathionyl ifosfamide mustard conjugates will be formed. The formation of IMGS is thus dependent on the concentration of glutathione, as demonstrated in our 31PNMR spectroscopy experiments as well as in our HPLC experiments. In our studies we observed that the rate of disappearance of IM is approximately twice a s fast as the disappearance of IMGS. The observed differences are expected since in IM there are two independently reacting centers (-NHCHzCHzCl)for intramolecular displacement of chloride and in IMGS only one (17). Knowledge of the specificity of GST’s for the conjugation reactions of alkylating agents is necessary to provide insight in the role of GST’s in the development of tumor cell resistance. Elevated levels of GST n and a have been associated with cellular resistance to alkylating agents (reviews: refs 3 and 19). The critical event of tumor cells in the development of drug resistance to alkylating cytostatic drugs is the ability of tumor cells to lower the concentrations of aziridinium ions present in the cytoplasm. This can be achieved by effective scavenging of the aziridinium ions formed. The concentration of alkylating cytostatic drugs in tumor cells is likely to be very small; the concentration of the aziridinium ions derived from these drugs is even smaller. It can be assumed that the intracellular concentration of both GSH and GST is much larger than the concentration of aziridinium ions, favoring both enzymatic and nonenzymatic glutathione-dependent detoxification of these drugs. The results presented in this study provide evidence that GST n enzymes can enhance the formation of monoglutathionyl IM. In studies with other alkylating agents like melphalan, chlorambucil, and phosphoramide mustard (8, 20-23) an increase in the formation of monoglutathione conjugates was found with GST a enzymes. The formation of monoglutathionyl conjugates of thiotepa was increased by both GST n and GST a (18). The conjugation of melphalan, chlorambucil, thiotepa, IM, and PM with glutathione are all believed to proceed through the aziridinium intermediate. The aziridiniudaziridine intermediate of IM is a substrate for GST Pl-1, while the aziridinium intermediates of chlorambucil, melphalan, and phosphoramide mustard are substrates for GST Al-1. The aziridiniudaziridine intermediate of thiotepa is a substrate for both GST P1-1 and GST Al-1. Since in ifosfamide mustard the two chloroethyl groups are not linked to the same nitrogen atom as in melphalan, chlorambucil, and phosphoramide mustard, steric differences might explain the observed differences in isoenzyme specificity. In addition, it is possible that GST

Chem. Res. Toxicol., Vol. 8, No. 7, 1995 985

a enzymes have a higher affinity for conjugation of the charged aziridinium ion formed from melphalan, chlorambucil, and phosphoramide mustard, than for the noncharged aziridine formed from IM. The finding that GST n enzymes can enhance the conjugation of IM with GSH is of clinical interest since many tumor cells have increased GST n levels (19). The GST P1-1 catalyzed formation of IMGS in the pH range 6-7 is independent of the pH. Even a t lower pH values expression of GST n might result in an enhanced detoxification of IM and thus in the development of drug resistance. If GST’s do indeed play a role in the development of drug resistance toward alkylating drugs, the findings that phosphoramide mustard is detoxified by GST a enzymes and ifosfamide mustard is detoxified by GST n enzymes might indicate that tumor cells which are resistant to phosphoramide mustard still respond on a treatment with ifosfamide mustard. This might be one of the determinants in the observed lack of crossresistance between CP and IF in experimental tumors (1). Monoglutathionyl IM cannot form DNA crosslinks, while diglutathionyl IM cannot interact with DNA a t all. The proposed model indicates that both glutathione levels and the presence of GST P1-1 can influence the number of DNA alkylations and hence the antitumor activity of ifosfamide. Therefore, i t is likely that a relationship exists between increased levels of GSH and GST P1-1 (z)in tumor cells and the development of drug resistance toward IF.

Acknowledgment. The authors wish to thank Dr. J. C. Venekamp (TNO-SIA) for her assistance with the NMR measurements and Mr. L. G. Gramberg (TNO-SIA) for his assistance with the FAB measurements. This study was supported by Grant TNOV-92-93of the Dutch Cancer Society. References (1) Nowrousian, M. R., Burkert, H., Herdrich, K. and Pohl, J. (1993)

Ifosfamide in cancer therapy: a comparison with cyclophosphamide, ISBN 3-86007-072)3,Universitats Verlag Jena GmbH, Jena, Germany. (2) Walker, D., Flinois, J.-O., Monkman, S. C., Beloc, C., Boddy, A. V., Cholerton, S., Daly, A. K., Lind, M. J., Pearson, A. D. J., Beaune, P. H., and Idle, J. R. (1994) Identification of the major human hepatic cytochrome P450 involved in activation and N-dechloroethylation of ifosfamide. Biochem. Pharmacol. 47, 1157-1163. (3) Tsuchida, S., and Sato, K. (1992) Glutathione transferases and cancer. CRC Crit. Reu. Biochem. Mol. Biol. 27, 337-394. (4) Peters, W. H. M., and Roelofs, H. M. J . (1992) Biochemical characterization of resistance to mitoxantrone and adriamycin in Caco-2 human colon adenocarcinoma cells: a possible role of glutathione S-transferases. Cancer Res. 52, 1886-1890. (5) Peters, W. H. M., Wobbels, T., Roelofs, H. M. J., and Jansen, J . B. M. J . (1993) Glutathione S-transferases in esophageal cancer. Carcinogenesis 14,1377-1380. ( 6 ) van der Zee, A. G. J., van Ommen, B., Meijer, C., Hollema, H., van Bladeren, P. J., and de Vries, E. G. E. (1992) Glutathione S-transferase activity and isoenzyme composition in benign ovarian tumours, untreated malignant ovarian tumours, and malignant ovarian tumours after platinudcyclophosphamide chemotherapy. Br. J. Cancer 66, 930-936. (7) Dirven, H. A. A. M., Venekamp, J . C., van Ommen, B., and van Bladeren, P. J. (1994) The interaction of glutathione with 4-hydroxycyclophosphamide and phosphoramide mustard, studied by 31Pnuclear magnetic resonance spectroscopy. Chem-Biol. Interact. 93, 185-196. ( 8 ) Dirven, H. A. A. M., van Ommen, B., and van Bladeren, P. J. (1994) Involvement of human glutathione S-transferase isoenzymes in the conjugation of cyclophosphamide metabolites with glutathione. Cancer Res. 54, 6215-6220.

986 Chem. Res. Toxicol., Vol. 8, No. 7, 1995 (9) Habig, W. H., Pabst, M. J.,and Jakoby, W. B. (1974) Glutathione S-transferases, The first step in mercapturic acid formation. J . Biol. Chem. 249, 7130-7139. (10) Wike-Hooley, J. L., Haveman, J., and Reinhold, H. S. (1984) The relevance of tumour pH to the treatment of malignant disease. Radiother. Oncol. 2, 343-366. (11) Kalinowski, H. O., Bergen, S., and Braun, C. (1984) 23CNMR spectroskopie, Georg Thieme Verlag, New York. (12) Kwon, C.-H., Borch, R. F., Engel, J., and Niemeyer, U. (1987) Activation mechanisms of mafosfamide and the role of thiols in cyclophosphamide metabolism. J . Med. Chem. 30, 395-399. (13) Martino, R., Crasnier, D., Chouini-Lalanne, N., Filard, V., Niemeyer, U., de Forni, M., and Malet-Martino, M.-C. (1992)A new approach to the study of ifosfamide metabolism by the analysis of human body fluids with 31P nuclear magnetic resonance spectroscopy. J . Pharmacol. Exp. Ther. 260, 1133-1144. (14) Lind, M. J., McGown, A. T., Hadfield, J. A., Thatcher, N., Crowther, D., and Fox, B. W. (1989) The effect of ifosfamide and its metabolites on intracellular glutathione levels in vitro and in vivo. Biochem. Pharmacol. 38, 1835-1840. (15) Yuan, 2.-M., Smith, P. B., Brundrett, R. B., Colvin, M., and Fenselau, C. (1991)Glutathione conjugation with phosphoramide mustard and cyclophosphamide-A mechanistic study using tandem mass spectrometry. Drug Metab. Dispos. 19, 625-629. (16) Colvin, 0. M., Friedman, H. S., Gamcsik, M. P., Fenselau, C., and Hilton, J. (1993) Role of glutathione in cellular resistance to alkylating agents. Adu. Enzyme Regul. 33, 19-26. (17) Boal, J. H., Williamson, M., Boyd, V. L., Ludeman, S. M., and Egan. W. (1989) 31PNMR studies of the kinetics of bisalkylation

Dirven et al. by iphosphoramide mustard: comparisons with phosphoranide mustard. J . Med. Chem. 32, 1768-1773. (18)Dirven, H. A. A. M., Dictus, E. L. J. T., Broeders, N. L. H. L., van Ommen, B., and van Bladeren, P. J. (1995) The role of human glutathione S-transferase isoenzymes in the formation of glutathione conjugates of the alkylating cytostatic drug thiotepa. Cancer Res. 55, 1701-1706. (19) Tsuchida, S., Sekeni, Y., Shineha, R., Nishihira, T., and Sato, K. (1989) Elevation of the placental S-transferase form (GST z) in tumor tissues and the levels in sera of patients with cancer. Cancer Res. 49, 5225-5229. (20) Ciaccio, P. J., Tew, K. D., and LaCreta, F. P. (1991) Enzymatic conjugation of chlorambucil with glutathione by human glutathione S-transferases and inhibition by ethacrynic acid. Biochem. Pharmacol. 42, 1504-1507. (21) Bolton, M. G., Colvin, 0. M., and Hilton, J . (1991) Specificity of isoenzymes of murine hepatic glutathione S-transferase for the conjugation of glutathione with L-phenylalanine mustard. Cancer Res. 51, 2410-2415. (22) Bolton, M. G., Hilton, J.,Robertson, K. D., Streeper, R. T., Colvin, 0. M. and Noe, D. A. (1993) Kinetic analysis of the reaction of melphalan with water, phosphate, and glutathione. Drug Metab. Dispos. 21, 986-996. (23) Meyer, D. J., Gilmore, K. S., Harris, J . M., Hartley, J. A,, and Ketterer, B. (1992) Chlorambucil-monoglutathionyl conjugate is sequestered by human alpha class glutathione S-transferases. Br. J . Cancer 66, 433-438.

TX9401551