- Email: [email protected]
Oxazolidonylethyl adducts to hemoglobin and DNA following nornitrogen mustard exposure
ELSEVIER
Chemico-BiologicalInteractions 99 (1996) 263-275
Oxazolidonylethyl adducts to hemoglobin and DNA following nornitrogen mustard exposure H. Thulin”, V. Zorcecbyc, D. Segerb5ckd, A. Sundwall”, M. Tiirnqvist* bs aPharmacia AB. Box 941, 2Sl 09 Helsingborg, Sweden bDepartment of Environmental Chemistry, Stockholm University, 106 91 Stockholm. Sweden ‘Department of Radiobiology, Stockholm University, 106 91 Stockholm, Sweden ‘Center for Nutrition and Toxicology, Karolinska Institute, Novum, 141 57 Huddinge. Sweden
Received 24 February 1995;revision received 3 November 1995;accepted 6 November
1995
Abstract Formation of adducts to hemoglobin (Hb) and DNA of nomitrogen mustard (NNM) was studied with the aim of developing a method for monitoring exposure to NNM. Adducts to N-terminal valines in Hb were studied by the N-alkyl Edman method using pentafluorophenyl isothiocyanate (PFPITC) as the derivatizing reagent. In preliminary studies live major Hb adducts were shown to be formed in reaction of NNM with red cell hemolysate in vitro. Following treatment with PFPITC three of these were found to be pentafluorophenylthiohydantoins (PFPTHs) of iV-alkylated valines and the fourth probably originates from NNM esters in which PFPITC had reacted with the nitrogen of N-chloroethylaminoethyl. A PFPTH was found to originate from N-2-(3-oxazolidonyl)thylvahne, Val-OZ. Val-OZ is formed in reaction, with ring closure to oxazolidone, of CO* with the 2-chloroethylamino group in the primary valine-N adduct. Resides a few other adducts, Val-OZ was also observed in mouse Hb following injection of NNM, and also after injection of cyclophosphamide. Following reac-
Abbreviations: EI, electron impact ionization; ESI, electrospray ionization; mustard; Val- (or Gua-) NNMCI, N- (or 7-) [N’-(2-chloroethyl)-2-aminoethyl]valine
NNM, nomitrogen (or guanine, respec-
tively); Val- (or Gua-) NNMOH, N- (or 7-) [N’-(2-hydroxyethyl)-2-aminoethyl]valine (or guanine, respec3-(2_chloroethyl)-oxazolidone; 2-(3-oxazolidonyl)ethyl; OZCI, tively); OZ in adduct names, Gua-NNM-Gua, bis[2-(7guaninyl)ethyl]amine; PFPITC, pentafluorophenyl isothiocyanate; PFPTH, pentafluorophenylthiohydantoin; NCI, negative ions, chemical ionization; PCI, positive ions, chemical ionization. l Corresponding 000!9-2797/96/515.00
author. 0
1996 Elsevier Science Ireland
SSDI 0009-2797(95)03674-B
Ltd. All rights reserved
264
H Thulin el al. / C‘hemrco-Biologicrl
Interactions 99 (1996) 263-275
tion in vitro of NNM with DNA, three major adducts to guanine-N-7 were observed; one of them, 7-[N’-(2-chloroethyl)-baminoethyll-guanine (NNMCI), was converted by carbonate to 7-[2-(3-oxazolidonyl)ethyl]guanine (Gua-OZ). In mice treated with NNM, Gua-OZ was the only DNA adduct observed. Val-OZ is a chemically stable Hb adduct, potentially useful for monitoring exposures to NNM and cyclophosphamide. Keywords: Nornitrogen mustard; Cyclophosphamide; Hemoglobin adducts; DNA adducts; N-(oxazolidonylethyl)valine; 7-(0xazolidonylethyl)guanine
1. Introduction For the purpose of monitoring exposure to nornitrogen mustard (NNM) by macromolecule adducts, products of reaction of NNM with the N-termini, valines, of hemoglobin (Hb) were studied by mass-spectrometric (MS) methods. The method
used, the N-alkyl Edman method, has previously been shown to be useful for monitoring human exposures to alkylating agents or their precursors [ 1,2], including cytostatic drugs [3]. To our knowledge Hb adducts from NNM or cyclophosphamide, the metabolism of which is suggested to give rise to NNM [4,5] have so far not been studied. In studies of DNA adducts formed from NNM in vitro, 7-[N’-(Zchloroethyl)2-aminoethyllguanine and 7-[N’-(2-hydroxyethyl)-2-aminoethyllguanine (GuaNNMCl and Gua-NNMOH, respectively) and the guanine dimer, bis[2-(7guaninyl)ethyl]amine (Gua-NNM-Gua), have been identified [6,7]. The same adducts have been identified in reaction of DNA in vitro with phosphoramide mustard, the reactive species formed after metabolism of cyclophosphamide [8]. The major DNA adduct formed from cyclophosphamide in vivo has been putatively identified as NNMOH [9]. This report summarizes the results of preliminary studies of reaction products formed in vivo and in vitro of nornitrogen mustard with Hb and DNA. In particular, this report deals with 2-(3-oxazolidonyl)ethyl (OZ) adducts to Hb and DNA, which were unexpectedly observed to be major adducts in vivo. The OZ adducts are formed in reaction of carbon dioxide with 2-chloroethylamino groups, a reaction that has been described for nornitrogen mustard [lo-121. This reaction has been shown to be both chemical and enzymatic [ 131. The same adduct was studied also in mice treated with this drug. Formation of chloroethyl-oxazolidone (OZCl) in vivo from 4-hydroxycyclophosphamide, the first metabolite of cyclophosphamide, has been demonstrated [ 141. 2. Materials and methods 2.1. Materials NNM. HCl (with >95% purity) and OZCl, synthesized at Pharmacia, [3H]NNM.HCl (26.3 MBq/mg, radiochemical purity 97.7%, custom synthesis from Amersham), (*Hs)NNM (99 atom % deuterium, custom synthesis from MSD
H. Thulin et al. / Chemico-Biological Interactions 99 (19%)
263-275
265
Isotopes, Montreal), cyclophosphamide monohydrate (Sendoxan@, Asta Medica), reference Gua-NNMCl, Gua-NNMOH and Gua-NNM-Gua (provided by Professor K. Hemminki, Huddinge, Sweden) were used. Reference Gua-OZ was prepared by incubating Gua-NNMCl in carbonate buffer. 2.2. Animal experiments and in vitro reactions Female NMRI mice, 9- 10 weeks old, received 8 mg/kg (44.8 pmol/kg) NNM . HCl in saline by i.v. injection. Blood, pooled from five animals, was collected from v. cava under ether anaesthesia in heparinized tubes after 24 h and 14 days. On the same occasion groups of five animals received 8 mg/kg OZCl or 80 mg/kg cyclophosphamide or 10 ml/kg saline (control). The erythrocytes were isolated by centrifugation at 1000 x g, washed with saline, stored at -20°C and used for analysis of Hb adducts. In other experiments mice were given 8 mg/kg [3H]NNM. HCl daily for five consecutive days and on day 9, in order to increase the levels of adducts. Blood and liver were collected 5 h after the last injection. For the identification of adducts, materials with high degree of alkylation were prepared by reaction in vitro of hemolysate of murine or human erythrocytes or calf thymus DNA with [3H]NNM, (2Hs)NNM or OZCl (details in figure and table legends). 2.3. Analysis of Hb adducts Globin was isolated according to Mowrer et al. [ 15,161and adducts to N-terminal valine were isolated as pentafluorophenylthiohydantoins (PFPTHs) after derivatization with pentafluorophenyl isothiocyanate (PFPITC) according to the N-alkyl Edman method using globin treated with (2H4)ethylene oxide as internal standard [16,17]. For the determination of structures of adducts, PFPITC-derivatives of adducts from the in vitro reaction of Hb and [3H]NNM were isolated through HPLC separation of radioactive fractions (for details, see legend of Fig. 1). Adducts in the isolated fractions were studied by gas chromatography-mass spectrometry (GUMS), PCI, NC1 (positive and negative ions, respectively, chemical ionization, with methane at 4.8 Torr as reagent gas) and EI (electron impact ionization), using a Finnigan TSQ 700 instrument. Quantification of the levels of OZ adducts in Hb from treated animals was carried out by GUMS/MS, NC1 analysis through selected reaction monitoring in the daughter scan mode. Collision gas was argon at 1.2 mTorr and collision energy 15 eV. Calibration curves were established from a globin prepared by in vitro treatment of hemolysate with [3H]NNM. Within the range of adduct levels studied these calibration curves were shown to be linear. The level of OZ adduct to valine in the standard globin was quantified by radioactivity counting after isolation by HPLC of the peak corresponding to the PFPTH of Val-OZ after derivatization of the globin with PFPITC. 2.4. DNA adducts Calf thymus DNA was incubated in vitro (100 mM sodium phosphate, pH 7, room temperature, 16 h) with [3H]NNM and DNA was isolated from livers of
266
H. Thulin et al. / Chemico-Biological Inleractions 99 (1996) 263-275
13H]NNM-treated mice (8 mg/kg). After several ethanol precipitations, guanine-7 alkylation products were liberated from the DNA by incubation at pH 6.0 and 100°C for 20 min. Alkylated guanine standards were added and the evaporated supematant was separated by reversed phase HPLC (for details, see figure legends). The eluate was monitored by UV detector and the collected fractions were analysed for radioactivity. The structure of Gua-OZ used as reference was verified by HPLCIMSIMS ES1 (electrospray ionization); in analysis of parent ion m/z 265 (M+H), the major daughter ion m/z 114 (= OZ) was observed at collision energy 15 eV. 3. Results 3. I. Hemoglobin adducts Derivatization with PFPITC of the globin obtained from red cell hemolysate treated in vitro with 13H]NNM, followed by HPLC separation of the PFPITCderivatives of adducts, showed the formation of five major products (B, C, D, E, F in Fig. 1). Conclusions regarding structures of peaks B, C, E and F were drawn from
I
----3-in
vitro
250
200
150
E
0
4
5 100
50
20 fraction
30 number
Fig. 1. HPLC separation of PFPITC derivatives of globin from human Hb (in erythrocyte hemolysate) treated with [‘HINNM in vitro (U-. scale to the left) or from mice administered [‘H]NNM by i.v. injection (0. . * . - ), scale to the right). Reaction conditions: see Table I. Analytical conditions: Adducts were separated on a 5-p Kromasil 100-5 Cl8 reversed phase column (4.6 x 250 mm), eluted at a flow rate of 1 ml/min with a linear gradient of acetonitrile in water increasing from 30% (v/v) to 100% in 40 min. followed by 100% acetonitrile for IO min. Mol. weights of identified derivatives (Table I): B, 312; C, 437; E, 411; F, 618.
H. Thulin et al. / Chemico-Biological
Interactions
99 (19%)
261
263-275
Table 1 Main Hb adducts obtained after derivatization of globin from a reaction of [3H]NNM (5 mM) with hemolysate of human red cells (final volume 4 times the cell volume, 5 mM in phosphate buffer, pH 7.4) Mol. weight Peak in Fig. 1
Assumed structure
HO-CHpCH,
-/“-C\“z
312 (B) S =C,
Assumed Hb-adduct
Yield (“%
-COO-NNMCI
1.1
of total counts in globin)
,CH, N I CsF5
411 (E)
(CH&CH
ox!
- CH-N
- CH2CH2NHCH2CH2-OH
Val-NNMOH
2.9
Val-OZ
0.7
Val-NNMCI (addition of 2 PFPITC)
0.4
\N,!C=S
I c6F5
431 (C)
OhhCH o=d
-CH-N-CH,CH,-N-C=0 \
,‘c=s
I
\
HZC\
N
,o
C"2
I c6F5
(CH3)2CH
618 (F; E?, G?)
o=d
- CH-N \
- CH2CH2 - N-CH,
,\c=s
N
I
c6F5
SA
,,JCH2
I C6hJ
Reaction time 62 h at 37°C. The yields are given as a percentage of the radioactivity bound to globin (27% of the radioactivity added); in all, 6% of the bound level was isolated after derivatization by PFPITC according to the N-alkyl Edman procedure [16,17].
MS techniques, with complementary information from parallel analysis of the products from Hb treated with (‘Hs)NNM. The products B, C, E, F with molecular weights and assumed chemical structure, are listed in Table 1. The product in peak F is also indicated to occur as part of the material in peaks E and G. The in vitro yields of the products (in percent of the total radioactivity in globin) given in Table 1 were inferred from the radioactivity of the HPLC fractions. As illustrated in Fig. 1, analysis of Hb from mice treated with [3H]NNM showed formation of products B, C, D and F, according to retention times and MS data identical with the in vitro products. Instead of the in vitro product E, a peak (E’) with slightly shorter retention time was observed. This peak is indicated to contain more than one product but not ‘adduct 411’ (peak E) identified in the in vitro material (Table 1). In addition the HPLC chromatogram of the in vivo material contained
H. Thulin et al. / Chemico-Biological Interactions 99 (1996~
268
Table 2 Major fragments from PFPTH derivatives of Val-02 and daughter ions of m/z 417 and 425, respectively, pretation [20] m/z (Natural compound)
263-275
and (‘Hs)Val-02 obtained in MS-NC1 and MS-EI, with relative intensities (in %) and suggested inter-
Interpretation
(M = 437)
m/z ( 2HsCompound) (M = 445)
MS-NC1 417 (loo)
425 (90)
409 (37) 334 (loo) 225 (5) MS-EI
417 (28) 342 (100) 225 (3)
Loss of HF Loss of CO from Val residue Probably loss of [F+(CH&(CH),CO] C,F,NCS
350 324 308 225
353 325 311 225
Loss of oxazolidone (natural or ‘H,) Loss of (OZ’H) or (OZ-*H), i.e. Val-PFPTH Probably additional loss of (CH,)& from Val residue As in NC1 above
(IO) (50) (43) (IO)
(22) (72) (63) (25)
I14 (65) 122 (68) 120 (loo) 113 (loo) MS/MS-NCI, daughter ions of parent ions m/z 417 or 425 334 303 283 268
(35) (56) (100) (77)
Treatment
342 303 283 276
(41) (loo) (100) (74)
with (‘Hs)NNM
02 (OZ-‘H)
from Val residue
or (OZ-‘H)
As in MS-NC1 above Loss of 02 Loss of (OZ + HF) Loss of (C,F,-F + H)? in vitro as given in Table
I
one early (A) and one late (G) peak. These peaks, like peak D observed both in vitro and in vivo, have so far not been identified. For the PFPTH of the Val-OZ adduct (C in Fig. l), with molecular weights (M) 437 and 445 after treatment in vitro with NNM or (‘Hs)NNM, respectively, the MS fragmentation patterns, supporting the suggested structure (Table l), are summarized in Table 2. Also data on daughter ions of m/z 417 (M-20) in MS/MS analysis are given in Table 2. From samples collected after in vivo treatment with [3H]NNM, the adduct isolated (peak C in Fig. 1) gave the same retention times by HPLC and GC as the in vitro product C and the same daughter ions of parent ion 417 (M-20) in MS/MS analysis (NCI), corresponding to the PFPTH of Val-OZ (Fig. 2). The relative abundances of the measured mass fragments 334,303,283 and 268 are the same in vivo and in vitro. In control animals without NNM treatment the amount of this adduct was below the detection level (0.1 pmol g-l). In Table 3 data are collected showing the levels, determined by GUMS/MS (NCI), of the Val-OZ adduct in Hb different times after administration of NNM to mice. Table 3 also contains data on this adduct following treatment in vivo or in vitro with OZCl (chloroethyloxazolidone, the primary product of CO, and NNM), or following injection of cyclophosphamide.
H. Thulin et al. / Chemico-Biological Interactions 99 (1996) 263-27s
269
E+OZ 9.44:
1PC -
_ E+03 1.304
I
_
Ecol:
6.74C
I “1
::ri;
1
1387
_
E+OZ 5.340
,‘~‘~~““1~“‘(“~~,~~~~1~~..,..~~1~~~.,1.~.,1...,....
11:40
ln:00
16:X
16:40
17:oo
17:20
Fig. 2. Val-OZ PFPTH from Hb in mice treated with NNM (sampling of blood 24 h after i.v. injection of 44.8 rmolikg body wt.): Abundances of daughter ions of parent ion m/z 417 in analysis by GC/MS/MS, NC1 of 97 mg globin. Compare with in vitro data in Table 2.
H. Thulin et al. / Chemico-Biological Interactions 99 (19%)
270
263-275
Table 3 Oxazolidonylethylvahne in mouse globin following i.v. injection of nornitrogen mustard related compounds or treatment of globin in vitro with chloroethyloxazolidone (OZCI) Time to blood sampling (days)
Sample and treatment
b. c. d. e.
mustard
44.8 smoUkg
Nomitrogen mustard 44.8 pmollkg 3-(2-Chloroethyl)-2-oxazolidone (OZCl) 54 pmollkg Cyclophosphamide 287 pmollkg Control
In vitro f. Hemolysate of mouse erythrocytes 19.8 mM, pH 7.4, 37°C g. Human globin, untreated Analysis
by CC/MS/MS
treated
after derivatization
with OZCI
or
Level of oxazolidinonylethylvaline
In vivo a. Nomitrogen
(NNM)
1
4.2
14
2.9 0.2 8.5
I I 1
0.0 (
22 h
2 x IO4
(pmohg)
0.0 (
to the N-alkyl
Edman method.
.
dpm-!
uv
’
-. 1200 1
1000
800
E
600
u”
Fig. 3. HPLC separation of DNA hydrolysate obtained from in vitro treatment of calf thymus DNA with [3H)NNM, followed by treatment with 50 mM NaHCO,. The adducts were separated on a 5-p Nucleosil C-18 reversed phase column (10 x 250 mm). The eluate was monitored by UV to detect the trace of the four added standards. The flow rate was 2 ml/mm. The gradient program used was running from 10 to 40% methanol in 50 mM ammonium formate, pH 4.65, over 40 min. Fractions and analysed for radioactivity in a liquid scintillation counter.
of
I mm were collected
H. Thulin et al. / Chemico-Biological Interactions 99 (19%)
263-275
271
3.2. DNA adducts
Following HPLC separation of thermally hydrolysed calf thymus DNA treated with [3H]NNM, three major radioactive peaks coeluting with the UV trace of the three added standards (Gua-NNMCl, Gua-NNMOH and the dimer Gua-NNMGua) were observed (data not shown), in agreement with earlier observations [4]. If this DNA sample was subsequently treated with sodium carbonate, no radioactive peak coeluting with Gua-NNMCl was found, but a new product coeluting with the UV-trace of Gua-OZ was predominant (Fig. 3). Analysis of liver DNA from NNMtreated mice showed only one major product, with the same retention time (HPLC) as Gua-OZ (data not shown). 4. Discussion The structure of the OZ adduct to N-terminal valine in Hb was confirmed by MS fragmentation patterns as given for the PFPTH derivative of Val-OZ in Table 2. The structure was further supported by the fact that Val-OZ was obtained, although with relatively low yields, by treatment of Hb in vitro and in vivo with OZCl (Table 3, f and c, respectively). The structure of OZCl has been confirmed by elementary analysis, IR spectrometry and t3C and ‘H NMR (in the tiles of Pharmacia AB, Helsingborg). The relatively fast formation of OZ adducts to Hb is evidently due to the high concentration of C02/HC03- in blood (total concentration about 22 mmol/l), and the favoured formation of five-atom rings. The formation also in the in vitro studies of OZ adducts (Fig. 1) is explained by the fact that these experiments were carried out by adding NNM to red cells hemolysed with distilled water, i.e. in a medium that is still rich in carbonate. In the treatments with NNM of DNA in vitro, i.e. in water with a carbonate concentration maximally corresponding to a > 100 times lower Pcoz, Gua-OZ is not expected to be, and was not found as a major adduct. As for the in vivo mechanism of formation of Val-OZ in Hb, measurements (a, b in Table 3) 1 and 14 days after single administration of NNM to mice show that a chemically stable OZ adduct is formed shortly after the injection, the decrease of its level in the 13 days between (a) and (b) being close to expected considering the zero-order kinetics of the decay of the level [18] at the life span of 40 days of mouse erythrocytes [ 191.The hydrolysis of the chlorine of an NNMCl-adduct is expected to be practically complete after 14 days. (The hydrolysis of the chlorine of the valine adduct is expected to proceed at approximately the same rate as that of the corresponding DNA adduct, Gua-NNM, which has been shown to have a half-life of about 4 h at 37°C l41.1 This would indicate that the OZ adduct measured in (b) (Table 3) is not formed by reaction with CO, after collection of blood or during the preparation of samples. Furthermore, although OZCl was shown in the in vitro experiment to react with valine-N (f), the formation in vivo of Val-OZ in Hb is indicated, from the in viva experiments with OZCl (c), to proceed only to a small extent via OZCl formed from free NNMCl. It seems, rather, that the reaction with COz occurs soon after alkylation of valine-N to Val-NNMCl and before hydrolysis of the chloride. Also
H. Thulin et al. / Chemieo-Biological interactions 99 ( I9961 263-275
272
CICH,CH,
\
CICH,CH,
’
nornlfrogen
N-H
mustard
CICH2CH2 \+ N-H / CH2 -
H,;~QJ
\ CHI
,OH
_
CICH2CH2 \ CICH$H, m=
CICH&H,
2/’
\
N-H HbCH$H,
\CH
\
0
HzC
N-H
’
/
HOCH2CH2
\
3-(2-chloroethyl)-2-oxazolidone (OZCI)
HbCH$H,
, N-H
N-C=0
I
1
PFPITC
Adducis
618,
312
Adduct
437
PFPITC
Adduct
411
(in vitro)
Fig. 4. Suggested of the PFFTHs.
reactions
leading to Hb adducts
from NNM,
identified
here by the molecular
weights
cyclophosphamide was found to give rise to the same adduct, although in a 2-3 times lower yield per pmol/kg injected (d). A scheme suggesting the reactions leading to observed valine adducts is given in Fig. 4. Similar conclusions were drawn with respect to the formation of Gua-OZ in DNA.
H. Thulin et al. / Chemico-Biological Interactions 99 (19%) 263-275
273
Out of the three peaks shown in Fig. 3, the peak coeluting with Gua-NNMCl is formed first in the reaction of DNA with NNM in vitro. This adduct is then converted by hydrolysis or further reaction to peaks coeluting with Gua-NNMOH and Gua-NNM-Gua, respectively (data not shown). Incubation of this DNA with sodium carbonate converted all remaining Gua-NNMCl to Gua-OZ (Fig. 3). Whereas no Gua-OZ was observed to be formed in vitro without addition of carbonate, the adduct coeluting (HPLC) with the Gua-OZ standard was the only one observed in liver DNA from mice treated with [3H]NNM. This indicates that GuaOZ is formed rapidly, before conversion of Gua-NNMCl to Gua-NNMOH and the dimer. After reaction of OZCl with deoxyguanosine, even under violent conditions (boiling for several hours of a solution with 20 mM OZCl), Gua-OZ could not be detected by HPLC analysis. From the data it was estimated that the reactivity of OZCl towards guanine-N-7 is by orders of magnitude slower than that of NNM. Gua-OZ seems, therefore, to be formed mainly by the same mechanism as Val-OZ, viz, addition of CO*, followed by cyclization, to the initial product of reaction of NNM with Gua in DNA (Gua-NNMCl). The data in Table 1 and Fig. 1 were obtained from globin samples precipitated within hours after the treatment in vitro or in vivo with NNM. The PFPTH derivatives with M 3 12 (peak B) and 618 (peak F) are probably formed from NNMCl adducts (to Hb carboxyl-0 and valine-N, respectively, cf. Fig. 4) by reaction of PFPITC with the mustard nitrogen followed by ring closure through alkylation of the isothiocyanate nitrogen. Considering the short life span of NNMCl adducts it is therefore expected that the levels of these adducts would be much lower in the analysis of blood samples drawn some time after treatment. In agreement herewith, adduct 618, a low level of which is still seen in samples 24 h after exposure (sample a in Table 3), is practically non-detectable ( < 0.1 pmol/g) after 14 days (sample b). The expectedly stable Val-NNMOH (‘adduct 41 l’, peak E), which is a predominant adduct in vitro (Table I), was not found following treatment in vivo. The peak E’ in the in vivo sample (Fig. 1) has a somewhat shorter retention time than peak E, confirmed by repeat analysis. In addition, preliminary MS analysis of peak E’ has shown the presence of a couple of compounds, which are, however, not main component(s). The data thus indicate that stable oxazolidone adducts to Hb and DNA are formed in vivo after treatment with NNM. The absence of a detectable background of Val-OZ in Hb is favourable to monitoring at low exposures. This adduct should therefore be studied further for the development of a monitoring technique for NNM and cyclophosphamide. In NNM-treated mice, Val-OZ constitutes about 1% of the total level of adducts to Hb as judged from radioactivity (Table 1). Such a low percentage of Val-OZ adducts is certainly sufficient for monitoring of cyclophosphamide in treated patients. Dose monitoring for the purpose of hygienic surveillance of persons occupationally exposed to this drug or NNM may require larger Hb samples with enrichment, e.g. by means of antibodies, of the adduct prior to MS analysis.
274
H. Thulin et al. / Chemieo-Biofogicd lnterafiions 99 { 1996) 263-278
Acknowledgements
We are indebted to K. Hemminki, Huddinge, for kindly providing reference compounds, to L. Ehrenberg for fruitful discussions, to Riitta Partanen, Huddinge, for skilful technical assistance, to L. Svensson, Helsingborg, for mass-spectrometric identification of OZ and derivatives and N. Lundh and Lillemor Halvarsson, Helsingborg, for animal experiments. The study was supported financially by Pharmacia AB, the Swedish Environmental Protection Agency and the Swedish Cancer Fund. References [I]
A.D. Tates, T. Grummt, M. Tomqvist, P.B. Farmer, F.J. van Dam, H. van Mossel, H.M. Schoemaker, S. Osterman-Golkar, Ch. Uebel, Y.S. Tang, A.H. Zwinderman, A.T. Natarajan and L. Ehrenberg, Biological and chemical monitoring of occupational exposure to ethylene oxide, Mutat. Res., 250 (1991) 483-497. [2] M. Tomqvist, Current research on hemoglobin adducts and cancer risks, an overview, in: CC. Travis (Ed.), Use of Biomarkers in Assessing Health and Environmental Impacts of Chemical Pollutants, NATO ARW, Luso, Portugal, Plenum, New York, 1993, pp. 17-30. [3] E. Bailey. P.B. Farmer, Y-S. Tang, H. Vangikar, A. Gray, D. Slee, R.M.J. Ings, D.B. Campbell, J.G. McVie and R. Dubbelman, Hydroxyethylation of hemoglobin by I-(2chloroethyl)-lnitrosoureas, Chem. Res. Toxicol. (1991) [4] I. Jardine, C. Fenselau, M. Appler, M.-N. gas chromatography-chemical ionization nomitrogen mustard in the plasma and
462-466. Kan, R.B. Brundrett and M. Colvin, Quantification by mass spectrometry of cyclophosphamide mustard and urine of patients receiving cyclophosphamide therapy,
Cancer Res., 38 (1978) 408-415. [5] P.J. Cox and L. Levin, Novel metabolic products of cyclophosphamide in human urine, Biochem. Pharmacol., 24 (1975) 1233-1235. [6] J. Doskocil and Z. Sormovl, The reaction of DNA with mustards 1. The reaction products, Collect. Czech. Chem. Commun., 30 (1965) 481-491. [7] K. Hemminki, DNA-binding products of nomitrogen mustard, a metabolite of cyclophosphamide, Chem.-Biol. Interact., 61 (1987) 126-130. [8] K. Hemminki, Nitrogen mustards and ethylene imines, in: K. Hemminki, A. Dipple, D.E.G. Shuker, F.F. Kadlubar, D. Segerback and H. Bartsch (Eds.), DNA Adducts: ldentilication and Biological Significance, IARC Scientific Publication No. 125, International Agency for Research on Cancer, Lyon, 1994, pp. 313-321. [9] J. Benson, C.N. Martin and R.C. Garner, N-(2-Hydroxyethyl)-N_[2-(7-guaninyl)ethyl]amine, the putative major DNA adduct of cyclophosphamide in vitro and in vivo in the rat, Biochem. Pharmacol., 37 (1988) 2979-2985. [IO] H. Arnold and H. Bekel, Die Oxazolidon-Reaktion Aminen, Arzneim.-Forsch., I4 (1964) 750-752.
von Z-chlorlthyl-substituierten
sekundaren
[Ill
C.B. Robinson and H.F. Herbrandson, Kinetics and mechanism of the reaction of 2,2’dichlorodiethylamine with aqueous carbon dioxide, J. Am. Chem. Sot. (1972) 7883-7887. 1121 E. Balbinder, C.I. Reich, D. Shugarts, J. Koegh, R. Fibiger, T. Jones and A. Banks, Relative mutagenicity of some urinary metabolites of the antitumor drug cyclophosphamide, Cancer Res., 41 (1981) 2967-2972. (131 C.E. Williamson, J.G. Kirby, J.I. Miller, S. Sass, S.P. Kramer, A.M. Seligman and B. Witten, The enzymatic transformation of some nitrogen mustards in the presence of carbon dioxide: implications in respiration, Cancer Res., 26 (1966) 323-330. [I41 P.S. Hong, A. Srigritsanapol and K.K. Chan, Pharmacokinetics of 4hydroxycyclophosphamide and metabolites in the rat, Drug Metab. Dispos., I9 (1991) l-7. [IS]
J. Mowrer,
M. Tiimqvist,
S. Jensen, and L. Ehrenberg,
Modified
Edman degradation
applied
to
H. Thulin et al. / Chemico-Biological
Interactions
99 (19%)
263-275
275
hemoglobin for monitoring occupational exposure to alkylating agents, Toxicol. Environ. Chem., 11 (1986) 215-231. [I61 M. TBmqvist, Epoxide adducts to N-terminal valine of hemoglobin, Methods Enzymol., 231B (1994) 650-657. [I71 M. Tiimqvist, J. Mowrer, S. Jensen and L. Ehrenberg, Monitoring of environmental cancer initiators through hemoglobin adducts by a modified Edman degradation method, Anal, B&hem., I54 (1986) 255-266. [I81 F. Granath, L. Ehrenberg and M. Tiimqvist, Degree of alkylation of macromolecules in vivo from variable exposure, Mutat. Res., 284 (1992) 297-306. 1191 N.I. Berlin, Life span of the red cell, in: C. Bishop and D.M. Surgenor (Eds.), The Red Blood Cell, Academic Press, New York, 1964, pp. 423-450. (201 P. Rydberg, B. Liining, CA. Wachtmeister and M. Tiimqvist, Synthesis and characterization of N-substituted valines and their phenyl- and pentafluorophenyl-thiohydantoins, Acta Chim. &and., 47 (1993) 813-817.
Chemico-BiologicalInteractions 99 (1996) 263-275
Oxazolidonylethyl adducts to hemoglobin and DNA following nornitrogen mustard exposure H. Thulin”, V. Zorcecbyc, D. Segerb5ckd, A. Sundwall”, M. Tiirnqvist* bs aPharmacia AB. Box 941, 2Sl 09 Helsingborg, Sweden bDepartment of Environmental Chemistry, Stockholm University, 106 91 Stockholm. Sweden ‘Department of Radiobiology, Stockholm University, 106 91 Stockholm, Sweden ‘Center for Nutrition and Toxicology, Karolinska Institute, Novum, 141 57 Huddinge. Sweden
Received 24 February 1995;revision received 3 November 1995;accepted 6 November
1995
Abstract Formation of adducts to hemoglobin (Hb) and DNA of nomitrogen mustard (NNM) was studied with the aim of developing a method for monitoring exposure to NNM. Adducts to N-terminal valines in Hb were studied by the N-alkyl Edman method using pentafluorophenyl isothiocyanate (PFPITC) as the derivatizing reagent. In preliminary studies live major Hb adducts were shown to be formed in reaction of NNM with red cell hemolysate in vitro. Following treatment with PFPITC three of these were found to be pentafluorophenylthiohydantoins (PFPTHs) of iV-alkylated valines and the fourth probably originates from NNM esters in which PFPITC had reacted with the nitrogen of N-chloroethylaminoethyl. A PFPTH was found to originate from N-2-(3-oxazolidonyl)thylvahne, Val-OZ. Val-OZ is formed in reaction, with ring closure to oxazolidone, of CO* with the 2-chloroethylamino group in the primary valine-N adduct. Resides a few other adducts, Val-OZ was also observed in mouse Hb following injection of NNM, and also after injection of cyclophosphamide. Following reac-
Abbreviations: EI, electron impact ionization; ESI, electrospray ionization; mustard; Val- (or Gua-) NNMCI, N- (or 7-) [N’-(2-chloroethyl)-2-aminoethyl]valine
NNM, nomitrogen (or guanine, respec-
tively); Val- (or Gua-) NNMOH, N- (or 7-) [N’-(2-hydroxyethyl)-2-aminoethyl]valine (or guanine, respec3-(2_chloroethyl)-oxazolidone; 2-(3-oxazolidonyl)ethyl; OZCI, tively); OZ in adduct names, Gua-NNM-Gua, bis[2-(7guaninyl)ethyl]amine; PFPITC, pentafluorophenyl isothiocyanate; PFPTH, pentafluorophenylthiohydantoin; NCI, negative ions, chemical ionization; PCI, positive ions, chemical ionization. l Corresponding 000!9-2797/96/515.00
author. 0
1996 Elsevier Science Ireland
SSDI 0009-2797(95)03674-B
Ltd. All rights reserved
264
H Thulin el al. / C‘hemrco-Biologicrl
Interactions 99 (1996) 263-275
tion in vitro of NNM with DNA, three major adducts to guanine-N-7 were observed; one of them, 7-[N’-(2-chloroethyl)-baminoethyll-guanine (NNMCI), was converted by carbonate to 7-[2-(3-oxazolidonyl)ethyl]guanine (Gua-OZ). In mice treated with NNM, Gua-OZ was the only DNA adduct observed. Val-OZ is a chemically stable Hb adduct, potentially useful for monitoring exposures to NNM and cyclophosphamide. Keywords: Nornitrogen mustard; Cyclophosphamide; Hemoglobin adducts; DNA adducts; N-(oxazolidonylethyl)valine; 7-(0xazolidonylethyl)guanine
1. Introduction For the purpose of monitoring exposure to nornitrogen mustard (NNM) by macromolecule adducts, products of reaction of NNM with the N-termini, valines, of hemoglobin (Hb) were studied by mass-spectrometric (MS) methods. The method
used, the N-alkyl Edman method, has previously been shown to be useful for monitoring human exposures to alkylating agents or their precursors [ 1,2], including cytostatic drugs [3]. To our knowledge Hb adducts from NNM or cyclophosphamide, the metabolism of which is suggested to give rise to NNM [4,5] have so far not been studied. In studies of DNA adducts formed from NNM in vitro, 7-[N’-(Zchloroethyl)2-aminoethyllguanine and 7-[N’-(2-hydroxyethyl)-2-aminoethyllguanine (GuaNNMCl and Gua-NNMOH, respectively) and the guanine dimer, bis[2-(7guaninyl)ethyl]amine (Gua-NNM-Gua), have been identified [6,7]. The same adducts have been identified in reaction of DNA in vitro with phosphoramide mustard, the reactive species formed after metabolism of cyclophosphamide [8]. The major DNA adduct formed from cyclophosphamide in vivo has been putatively identified as NNMOH [9]. This report summarizes the results of preliminary studies of reaction products formed in vivo and in vitro of nornitrogen mustard with Hb and DNA. In particular, this report deals with 2-(3-oxazolidonyl)ethyl (OZ) adducts to Hb and DNA, which were unexpectedly observed to be major adducts in vivo. The OZ adducts are formed in reaction of carbon dioxide with 2-chloroethylamino groups, a reaction that has been described for nornitrogen mustard [lo-121. This reaction has been shown to be both chemical and enzymatic [ 131. The same adduct was studied also in mice treated with this drug. Formation of chloroethyl-oxazolidone (OZCl) in vivo from 4-hydroxycyclophosphamide, the first metabolite of cyclophosphamide, has been demonstrated [ 141. 2. Materials and methods 2.1. Materials NNM. HCl (with >95% purity) and OZCl, synthesized at Pharmacia, [3H]NNM.HCl (26.3 MBq/mg, radiochemical purity 97.7%, custom synthesis from Amersham), (*Hs)NNM (99 atom % deuterium, custom synthesis from MSD
H. Thulin et al. / Chemico-Biological Interactions 99 (19%)
263-275
265
Isotopes, Montreal), cyclophosphamide monohydrate (Sendoxan@, Asta Medica), reference Gua-NNMCl, Gua-NNMOH and Gua-NNM-Gua (provided by Professor K. Hemminki, Huddinge, Sweden) were used. Reference Gua-OZ was prepared by incubating Gua-NNMCl in carbonate buffer. 2.2. Animal experiments and in vitro reactions Female NMRI mice, 9- 10 weeks old, received 8 mg/kg (44.8 pmol/kg) NNM . HCl in saline by i.v. injection. Blood, pooled from five animals, was collected from v. cava under ether anaesthesia in heparinized tubes after 24 h and 14 days. On the same occasion groups of five animals received 8 mg/kg OZCl or 80 mg/kg cyclophosphamide or 10 ml/kg saline (control). The erythrocytes were isolated by centrifugation at 1000 x g, washed with saline, stored at -20°C and used for analysis of Hb adducts. In other experiments mice were given 8 mg/kg [3H]NNM. HCl daily for five consecutive days and on day 9, in order to increase the levels of adducts. Blood and liver were collected 5 h after the last injection. For the identification of adducts, materials with high degree of alkylation were prepared by reaction in vitro of hemolysate of murine or human erythrocytes or calf thymus DNA with [3H]NNM, (2Hs)NNM or OZCl (details in figure and table legends). 2.3. Analysis of Hb adducts Globin was isolated according to Mowrer et al. [ 15,161and adducts to N-terminal valine were isolated as pentafluorophenylthiohydantoins (PFPTHs) after derivatization with pentafluorophenyl isothiocyanate (PFPITC) according to the N-alkyl Edman method using globin treated with (2H4)ethylene oxide as internal standard [16,17]. For the determination of structures of adducts, PFPITC-derivatives of adducts from the in vitro reaction of Hb and [3H]NNM were isolated through HPLC separation of radioactive fractions (for details, see legend of Fig. 1). Adducts in the isolated fractions were studied by gas chromatography-mass spectrometry (GUMS), PCI, NC1 (positive and negative ions, respectively, chemical ionization, with methane at 4.8 Torr as reagent gas) and EI (electron impact ionization), using a Finnigan TSQ 700 instrument. Quantification of the levels of OZ adducts in Hb from treated animals was carried out by GUMS/MS, NC1 analysis through selected reaction monitoring in the daughter scan mode. Collision gas was argon at 1.2 mTorr and collision energy 15 eV. Calibration curves were established from a globin prepared by in vitro treatment of hemolysate with [3H]NNM. Within the range of adduct levels studied these calibration curves were shown to be linear. The level of OZ adduct to valine in the standard globin was quantified by radioactivity counting after isolation by HPLC of the peak corresponding to the PFPTH of Val-OZ after derivatization of the globin with PFPITC. 2.4. DNA adducts Calf thymus DNA was incubated in vitro (100 mM sodium phosphate, pH 7, room temperature, 16 h) with [3H]NNM and DNA was isolated from livers of
266
H. Thulin et al. / Chemico-Biological Inleractions 99 (1996) 263-275
13H]NNM-treated mice (8 mg/kg). After several ethanol precipitations, guanine-7 alkylation products were liberated from the DNA by incubation at pH 6.0 and 100°C for 20 min. Alkylated guanine standards were added and the evaporated supematant was separated by reversed phase HPLC (for details, see figure legends). The eluate was monitored by UV detector and the collected fractions were analysed for radioactivity. The structure of Gua-OZ used as reference was verified by HPLCIMSIMS ES1 (electrospray ionization); in analysis of parent ion m/z 265 (M+H), the major daughter ion m/z 114 (= OZ) was observed at collision energy 15 eV. 3. Results 3. I. Hemoglobin adducts Derivatization with PFPITC of the globin obtained from red cell hemolysate treated in vitro with 13H]NNM, followed by HPLC separation of the PFPITCderivatives of adducts, showed the formation of five major products (B, C, D, E, F in Fig. 1). Conclusions regarding structures of peaks B, C, E and F were drawn from
I
----3-in
vitro
250
200
150
E
0
4
5 100
50
20 fraction
30 number
Fig. 1. HPLC separation of PFPITC derivatives of globin from human Hb (in erythrocyte hemolysate) treated with [‘HINNM in vitro (U-. scale to the left) or from mice administered [‘H]NNM by i.v. injection (0. . * . - ), scale to the right). Reaction conditions: see Table I. Analytical conditions: Adducts were separated on a 5-p Kromasil 100-5 Cl8 reversed phase column (4.6 x 250 mm), eluted at a flow rate of 1 ml/min with a linear gradient of acetonitrile in water increasing from 30% (v/v) to 100% in 40 min. followed by 100% acetonitrile for IO min. Mol. weights of identified derivatives (Table I): B, 312; C, 437; E, 411; F, 618.
H. Thulin et al. / Chemico-Biological
Interactions
99 (19%)
261
263-275
Table 1 Main Hb adducts obtained after derivatization of globin from a reaction of [3H]NNM (5 mM) with hemolysate of human red cells (final volume 4 times the cell volume, 5 mM in phosphate buffer, pH 7.4) Mol. weight Peak in Fig. 1
Assumed structure
HO-CHpCH,
-/“-C\“z
312 (B) S =C,
Assumed Hb-adduct
Yield (“%
-COO-NNMCI
1.1
of total counts in globin)
,CH, N I CsF5
411 (E)
(CH&CH
ox!
- CH-N
- CH2CH2NHCH2CH2-OH
Val-NNMOH
2.9
Val-OZ
0.7
Val-NNMCI (addition of 2 PFPITC)
0.4
\N,!C=S
I c6F5
431 (C)
OhhCH o=d
-CH-N-CH,CH,-N-C=0 \
,‘c=s
I
\
HZC\
N
,o
C"2
I c6F5
(CH3)2CH
618 (F; E?, G?)
o=d
- CH-N \
- CH2CH2 - N-CH,
,\c=s
N
I
c6F5
SA
,,JCH2
I C6hJ
Reaction time 62 h at 37°C. The yields are given as a percentage of the radioactivity bound to globin (27% of the radioactivity added); in all, 6% of the bound level was isolated after derivatization by PFPITC according to the N-alkyl Edman procedure [16,17].
MS techniques, with complementary information from parallel analysis of the products from Hb treated with (‘Hs)NNM. The products B, C, E, F with molecular weights and assumed chemical structure, are listed in Table 1. The product in peak F is also indicated to occur as part of the material in peaks E and G. The in vitro yields of the products (in percent of the total radioactivity in globin) given in Table 1 were inferred from the radioactivity of the HPLC fractions. As illustrated in Fig. 1, analysis of Hb from mice treated with [3H]NNM showed formation of products B, C, D and F, according to retention times and MS data identical with the in vitro products. Instead of the in vitro product E, a peak (E’) with slightly shorter retention time was observed. This peak is indicated to contain more than one product but not ‘adduct 411’ (peak E) identified in the in vitro material (Table 1). In addition the HPLC chromatogram of the in vivo material contained
H. Thulin et al. / Chemico-Biological Interactions 99 (1996~
268
Table 2 Major fragments from PFPTH derivatives of Val-02 and daughter ions of m/z 417 and 425, respectively, pretation [20] m/z (Natural compound)
263-275
and (‘Hs)Val-02 obtained in MS-NC1 and MS-EI, with relative intensities (in %) and suggested inter-
Interpretation
(M = 437)
m/z ( 2HsCompound) (M = 445)
MS-NC1 417 (loo)
425 (90)
409 (37) 334 (loo) 225 (5) MS-EI
417 (28) 342 (100) 225 (3)
Loss of HF Loss of CO from Val residue Probably loss of [F+(CH&(CH),CO] C,F,NCS
350 324 308 225
353 325 311 225
Loss of oxazolidone (natural or ‘H,) Loss of (OZ’H) or (OZ-*H), i.e. Val-PFPTH Probably additional loss of (CH,)& from Val residue As in NC1 above
(IO) (50) (43) (IO)
(22) (72) (63) (25)
I14 (65) 122 (68) 120 (loo) 113 (loo) MS/MS-NCI, daughter ions of parent ions m/z 417 or 425 334 303 283 268
(35) (56) (100) (77)
Treatment
342 303 283 276
(41) (loo) (100) (74)
with (‘Hs)NNM
02 (OZ-‘H)
from Val residue
or (OZ-‘H)
As in MS-NC1 above Loss of 02 Loss of (OZ + HF) Loss of (C,F,-F + H)? in vitro as given in Table
I
one early (A) and one late (G) peak. These peaks, like peak D observed both in vitro and in vivo, have so far not been identified. For the PFPTH of the Val-OZ adduct (C in Fig. l), with molecular weights (M) 437 and 445 after treatment in vitro with NNM or (‘Hs)NNM, respectively, the MS fragmentation patterns, supporting the suggested structure (Table l), are summarized in Table 2. Also data on daughter ions of m/z 417 (M-20) in MS/MS analysis are given in Table 2. From samples collected after in vivo treatment with [3H]NNM, the adduct isolated (peak C in Fig. 1) gave the same retention times by HPLC and GC as the in vitro product C and the same daughter ions of parent ion 417 (M-20) in MS/MS analysis (NCI), corresponding to the PFPTH of Val-OZ (Fig. 2). The relative abundances of the measured mass fragments 334,303,283 and 268 are the same in vivo and in vitro. In control animals without NNM treatment the amount of this adduct was below the detection level (0.1 pmol g-l). In Table 3 data are collected showing the levels, determined by GUMS/MS (NCI), of the Val-OZ adduct in Hb different times after administration of NNM to mice. Table 3 also contains data on this adduct following treatment in vivo or in vitro with OZCl (chloroethyloxazolidone, the primary product of CO, and NNM), or following injection of cyclophosphamide.
H. Thulin et al. / Chemico-Biological Interactions 99 (1996) 263-27s
269
E+OZ 9.44:
1PC -
_ E+03 1.304
I
_
Ecol:
6.74C
I “1
::ri;
1
1387
_
E+OZ 5.340
,‘~‘~~““1~“‘(“~~,~~~~1~~..,..~~1~~~.,1.~.,1...,....
11:40
ln:00
16:X
16:40
17:oo
17:20
Fig. 2. Val-OZ PFPTH from Hb in mice treated with NNM (sampling of blood 24 h after i.v. injection of 44.8 rmolikg body wt.): Abundances of daughter ions of parent ion m/z 417 in analysis by GC/MS/MS, NC1 of 97 mg globin. Compare with in vitro data in Table 2.
H. Thulin et al. / Chemico-Biological Interactions 99 (19%)
270
263-275
Table 3 Oxazolidonylethylvahne in mouse globin following i.v. injection of nornitrogen mustard related compounds or treatment of globin in vitro with chloroethyloxazolidone (OZCI) Time to blood sampling (days)
Sample and treatment
b. c. d. e.
mustard
44.8 smoUkg
Nomitrogen mustard 44.8 pmollkg 3-(2-Chloroethyl)-2-oxazolidone (OZCl) 54 pmollkg Cyclophosphamide 287 pmollkg Control
In vitro f. Hemolysate of mouse erythrocytes 19.8 mM, pH 7.4, 37°C g. Human globin, untreated Analysis
by CC/MS/MS
treated
after derivatization
with OZCI
or
Level of oxazolidinonylethylvaline
In vivo a. Nomitrogen
(NNM)
1
4.2
14
2.9 0.2 8.5
I I 1
0.0 (
22 h
2 x IO4
(pmohg)
0.0 (
to the N-alkyl
Edman method.
.
dpm-!
uv
’
-. 1200 1
1000
800
E
600
u”
Fig. 3. HPLC separation of DNA hydrolysate obtained from in vitro treatment of calf thymus DNA with [3H)NNM, followed by treatment with 50 mM NaHCO,. The adducts were separated on a 5-p Nucleosil C-18 reversed phase column (10 x 250 mm). The eluate was monitored by UV to detect the trace of the four added standards. The flow rate was 2 ml/mm. The gradient program used was running from 10 to 40% methanol in 50 mM ammonium formate, pH 4.65, over 40 min. Fractions and analysed for radioactivity in a liquid scintillation counter.
of
I mm were collected
H. Thulin et al. / Chemico-Biological Interactions 99 (19%)
263-275
271
3.2. DNA adducts
Following HPLC separation of thermally hydrolysed calf thymus DNA treated with [3H]NNM, three major radioactive peaks coeluting with the UV trace of the three added standards (Gua-NNMCl, Gua-NNMOH and the dimer Gua-NNMGua) were observed (data not shown), in agreement with earlier observations [4]. If this DNA sample was subsequently treated with sodium carbonate, no radioactive peak coeluting with Gua-NNMCl was found, but a new product coeluting with the UV-trace of Gua-OZ was predominant (Fig. 3). Analysis of liver DNA from NNMtreated mice showed only one major product, with the same retention time (HPLC) as Gua-OZ (data not shown). 4. Discussion The structure of the OZ adduct to N-terminal valine in Hb was confirmed by MS fragmentation patterns as given for the PFPTH derivative of Val-OZ in Table 2. The structure was further supported by the fact that Val-OZ was obtained, although with relatively low yields, by treatment of Hb in vitro and in vivo with OZCl (Table 3, f and c, respectively). The structure of OZCl has been confirmed by elementary analysis, IR spectrometry and t3C and ‘H NMR (in the tiles of Pharmacia AB, Helsingborg). The relatively fast formation of OZ adducts to Hb is evidently due to the high concentration of C02/HC03- in blood (total concentration about 22 mmol/l), and the favoured formation of five-atom rings. The formation also in the in vitro studies of OZ adducts (Fig. 1) is explained by the fact that these experiments were carried out by adding NNM to red cells hemolysed with distilled water, i.e. in a medium that is still rich in carbonate. In the treatments with NNM of DNA in vitro, i.e. in water with a carbonate concentration maximally corresponding to a > 100 times lower Pcoz, Gua-OZ is not expected to be, and was not found as a major adduct. As for the in vivo mechanism of formation of Val-OZ in Hb, measurements (a, b in Table 3) 1 and 14 days after single administration of NNM to mice show that a chemically stable OZ adduct is formed shortly after the injection, the decrease of its level in the 13 days between (a) and (b) being close to expected considering the zero-order kinetics of the decay of the level [18] at the life span of 40 days of mouse erythrocytes [ 191.The hydrolysis of the chlorine of an NNMCl-adduct is expected to be practically complete after 14 days. (The hydrolysis of the chlorine of the valine adduct is expected to proceed at approximately the same rate as that of the corresponding DNA adduct, Gua-NNM, which has been shown to have a half-life of about 4 h at 37°C l41.1 This would indicate that the OZ adduct measured in (b) (Table 3) is not formed by reaction with CO, after collection of blood or during the preparation of samples. Furthermore, although OZCl was shown in the in vitro experiment to react with valine-N (f), the formation in vivo of Val-OZ in Hb is indicated, from the in viva experiments with OZCl (c), to proceed only to a small extent via OZCl formed from free NNMCl. It seems, rather, that the reaction with COz occurs soon after alkylation of valine-N to Val-NNMCl and before hydrolysis of the chloride. Also
H. Thulin et al. / Chemieo-Biological interactions 99 ( I9961 263-275
272
CICH,CH,
\
CICH,CH,
’
nornlfrogen
N-H
mustard
CICH2CH2 \+ N-H / CH2 -
H,;~QJ
\ CHI
,OH
_
CICH2CH2 \ CICH$H, m=
CICH&H,
2/’
\
N-H HbCH$H,
\CH
\
0
HzC
N-H
’
/
HOCH2CH2
\
3-(2-chloroethyl)-2-oxazolidone (OZCI)
HbCH$H,
, N-H
N-C=0
I
1
PFPITC
Adducis
618,
312
Adduct
437
PFPITC
Adduct
411
(in vitro)
Fig. 4. Suggested of the PFFTHs.
reactions
leading to Hb adducts
from NNM,
identified
here by the molecular
weights
cyclophosphamide was found to give rise to the same adduct, although in a 2-3 times lower yield per pmol/kg injected (d). A scheme suggesting the reactions leading to observed valine adducts is given in Fig. 4. Similar conclusions were drawn with respect to the formation of Gua-OZ in DNA.
H. Thulin et al. / Chemico-Biological Interactions 99 (19%) 263-275
273
Out of the three peaks shown in Fig. 3, the peak coeluting with Gua-NNMCl is formed first in the reaction of DNA with NNM in vitro. This adduct is then converted by hydrolysis or further reaction to peaks coeluting with Gua-NNMOH and Gua-NNM-Gua, respectively (data not shown). Incubation of this DNA with sodium carbonate converted all remaining Gua-NNMCl to Gua-OZ (Fig. 3). Whereas no Gua-OZ was observed to be formed in vitro without addition of carbonate, the adduct coeluting (HPLC) with the Gua-OZ standard was the only one observed in liver DNA from mice treated with [3H]NNM. This indicates that GuaOZ is formed rapidly, before conversion of Gua-NNMCl to Gua-NNMOH and the dimer. After reaction of OZCl with deoxyguanosine, even under violent conditions (boiling for several hours of a solution with 20 mM OZCl), Gua-OZ could not be detected by HPLC analysis. From the data it was estimated that the reactivity of OZCl towards guanine-N-7 is by orders of magnitude slower than that of NNM. Gua-OZ seems, therefore, to be formed mainly by the same mechanism as Val-OZ, viz, addition of CO*, followed by cyclization, to the initial product of reaction of NNM with Gua in DNA (Gua-NNMCl). The data in Table 1 and Fig. 1 were obtained from globin samples precipitated within hours after the treatment in vitro or in vivo with NNM. The PFPTH derivatives with M 3 12 (peak B) and 618 (peak F) are probably formed from NNMCl adducts (to Hb carboxyl-0 and valine-N, respectively, cf. Fig. 4) by reaction of PFPITC with the mustard nitrogen followed by ring closure through alkylation of the isothiocyanate nitrogen. Considering the short life span of NNMCl adducts it is therefore expected that the levels of these adducts would be much lower in the analysis of blood samples drawn some time after treatment. In agreement herewith, adduct 618, a low level of which is still seen in samples 24 h after exposure (sample a in Table 3), is practically non-detectable ( < 0.1 pmol/g) after 14 days (sample b). The expectedly stable Val-NNMOH (‘adduct 41 l’, peak E), which is a predominant adduct in vitro (Table I), was not found following treatment in vivo. The peak E’ in the in vivo sample (Fig. 1) has a somewhat shorter retention time than peak E, confirmed by repeat analysis. In addition, preliminary MS analysis of peak E’ has shown the presence of a couple of compounds, which are, however, not main component(s). The data thus indicate that stable oxazolidone adducts to Hb and DNA are formed in vivo after treatment with NNM. The absence of a detectable background of Val-OZ in Hb is favourable to monitoring at low exposures. This adduct should therefore be studied further for the development of a monitoring technique for NNM and cyclophosphamide. In NNM-treated mice, Val-OZ constitutes about 1% of the total level of adducts to Hb as judged from radioactivity (Table 1). Such a low percentage of Val-OZ adducts is certainly sufficient for monitoring of cyclophosphamide in treated patients. Dose monitoring for the purpose of hygienic surveillance of persons occupationally exposed to this drug or NNM may require larger Hb samples with enrichment, e.g. by means of antibodies, of the adduct prior to MS analysis.
274
H. Thulin et al. / Chemieo-Biofogicd lnterafiions 99 { 1996) 263-278
Acknowledgements
We are indebted to K. Hemminki, Huddinge, for kindly providing reference compounds, to L. Ehrenberg for fruitful discussions, to Riitta Partanen, Huddinge, for skilful technical assistance, to L. Svensson, Helsingborg, for mass-spectrometric identification of OZ and derivatives and N. Lundh and Lillemor Halvarsson, Helsingborg, for animal experiments. The study was supported financially by Pharmacia AB, the Swedish Environmental Protection Agency and the Swedish Cancer Fund. References [I]
A.D. Tates, T. Grummt, M. Tomqvist, P.B. Farmer, F.J. van Dam, H. van Mossel, H.M. Schoemaker, S. Osterman-Golkar, Ch. Uebel, Y.S. Tang, A.H. Zwinderman, A.T. Natarajan and L. Ehrenberg, Biological and chemical monitoring of occupational exposure to ethylene oxide, Mutat. Res., 250 (1991) 483-497. [2] M. Tomqvist, Current research on hemoglobin adducts and cancer risks, an overview, in: CC. Travis (Ed.), Use of Biomarkers in Assessing Health and Environmental Impacts of Chemical Pollutants, NATO ARW, Luso, Portugal, Plenum, New York, 1993, pp. 17-30. [3] E. Bailey. P.B. Farmer, Y-S. Tang, H. Vangikar, A. Gray, D. Slee, R.M.J. Ings, D.B. Campbell, J.G. McVie and R. Dubbelman, Hydroxyethylation of hemoglobin by I-(2chloroethyl)-lnitrosoureas, Chem. Res. Toxicol. (1991) [4] I. Jardine, C. Fenselau, M. Appler, M.-N. gas chromatography-chemical ionization nomitrogen mustard in the plasma and
462-466. Kan, R.B. Brundrett and M. Colvin, Quantification by mass spectrometry of cyclophosphamide mustard and urine of patients receiving cyclophosphamide therapy,
Cancer Res., 38 (1978) 408-415. [5] P.J. Cox and L. Levin, Novel metabolic products of cyclophosphamide in human urine, Biochem. Pharmacol., 24 (1975) 1233-1235. [6] J. Doskocil and Z. Sormovl, The reaction of DNA with mustards 1. The reaction products, Collect. Czech. Chem. Commun., 30 (1965) 481-491. [7] K. Hemminki, DNA-binding products of nomitrogen mustard, a metabolite of cyclophosphamide, Chem.-Biol. Interact., 61 (1987) 126-130. [8] K. Hemminki, Nitrogen mustards and ethylene imines, in: K. Hemminki, A. Dipple, D.E.G. Shuker, F.F. Kadlubar, D. Segerback and H. Bartsch (Eds.), DNA Adducts: ldentilication and Biological Significance, IARC Scientific Publication No. 125, International Agency for Research on Cancer, Lyon, 1994, pp. 313-321. [9] J. Benson, C.N. Martin and R.C. Garner, N-(2-Hydroxyethyl)-N_[2-(7-guaninyl)ethyl]amine, the putative major DNA adduct of cyclophosphamide in vitro and in vivo in the rat, Biochem. Pharmacol., 37 (1988) 2979-2985. [IO] H. Arnold and H. Bekel, Die Oxazolidon-Reaktion Aminen, Arzneim.-Forsch., I4 (1964) 750-752.
von Z-chlorlthyl-substituierten
sekundaren
[Ill
C.B. Robinson and H.F. Herbrandson, Kinetics and mechanism of the reaction of 2,2’dichlorodiethylamine with aqueous carbon dioxide, J. Am. Chem. Sot. (1972) 7883-7887. 1121 E. Balbinder, C.I. Reich, D. Shugarts, J. Koegh, R. Fibiger, T. Jones and A. Banks, Relative mutagenicity of some urinary metabolites of the antitumor drug cyclophosphamide, Cancer Res., 41 (1981) 2967-2972. (131 C.E. Williamson, J.G. Kirby, J.I. Miller, S. Sass, S.P. Kramer, A.M. Seligman and B. Witten, The enzymatic transformation of some nitrogen mustards in the presence of carbon dioxide: implications in respiration, Cancer Res., 26 (1966) 323-330. [I41 P.S. Hong, A. Srigritsanapol and K.K. Chan, Pharmacokinetics of 4hydroxycyclophosphamide and metabolites in the rat, Drug Metab. Dispos., I9 (1991) l-7. [IS]
J. Mowrer,
M. Tiimqvist,
S. Jensen, and L. Ehrenberg,
Modified
Edman degradation
applied
to
H. Thulin et al. / Chemico-Biological
Interactions
99 (19%)
263-275
275
hemoglobin for monitoring occupational exposure to alkylating agents, Toxicol. Environ. Chem., 11 (1986) 215-231. [I61 M. TBmqvist, Epoxide adducts to N-terminal valine of hemoglobin, Methods Enzymol., 231B (1994) 650-657. [I71 M. Tiimqvist, J. Mowrer, S. Jensen and L. Ehrenberg, Monitoring of environmental cancer initiators through hemoglobin adducts by a modified Edman degradation method, Anal, B&hem., I54 (1986) 255-266. [I81 F. Granath, L. Ehrenberg and M. Tiimqvist, Degree of alkylation of macromolecules in vivo from variable exposure, Mutat. Res., 284 (1992) 297-306. 1191 N.I. Berlin, Life span of the red cell, in: C. Bishop and D.M. Surgenor (Eds.), The Red Blood Cell, Academic Press, New York, 1964, pp. 423-450. (201 P. Rydberg, B. Liining, CA. Wachtmeister and M. Tiimqvist, Synthesis and characterization of N-substituted valines and their phenyl- and pentafluorophenyl-thiohydantoins, Acta Chim. &and., 47 (1993) 813-817.