Identification of Phosphoramide Mustard/DNA Adducts Using Tandem Mass Spectrometry Julia R. Cushnir, Stephen Naylor,* John H. Lamb and Peter B. Farmer* MRC Toxicology Unit, Carshalton, Surrey, SMS 4EF, UK
Nigel A. Brown MRC Experimental Embryology and Teratology Unit, St. George's Hospital, University of London, London, SW17 ORE, UK
Philip E. Mirkes Department of Pediatrics, University of Washington, Seattle, W A 98195,USA
The reaction pathway of alkylating agents is often exploited in the design of bifunctional anti-cancer drugs. These drugs form mono-DNA adducts as well as inter- and intra-strand cross-linked adducts, notably by reaction at DNA bases, including the N-7-position of guanine (G). A positive-ion fast-atom bombardment (FAB) mass spectrum of an in uitro preparation of DNA alkylated with phosphoramide mustard (the active metabolite of the anti-cancer drug cyclophosphamide) indicated the presence of the two mono-DNA adducts N-(2-chloroethyl)-N[2-(7-guaninyl)ethyl] amine, designated NOR-G, and N-(2-hydroxyethyl)-N-[2-(7-guaninyl)ethyl]amine, designated N O R - G - O H , (MH+ 257/259 and 239, respectively) but not the presence of the cross-linked adduct N,N-bis-[2-(7-guaninyl)ethyl] amine, designated G - N O R 4 (MH+ 372). Using synthetic standards, daughterion spectra of NOR-G, NOR-G-OH and G - N O R 4 were obtained (matrix 0.2 ~ p - t o l u e n esulphonic acid in glycerol) by positive-ion FAB tandem mass spectrometry (FAB-MSIMS). The daughter-ion spectra of both mono-DNA adducts N O R 4 and NOR-G-OH contained a fragment ion at rnlz 152 [G+H]+, whereas the cross-linked adduct, G-NOR-G, showed an ion at mlz221, [MH-GI+. Evidence for the presence of NOR-G, NOR-G-OH and G-NOR-G in the in uitro preparation was obtained by performing a double parent-ion scan on rnlz 152 and 221. The presence of G - N O R 4 was further supported by performing a single parent-ion scan on m/z 221. The use of this MS/MS technique should eliminate the need for intricate sample purification in the identification of G-NOR-G in biological extracts.
Electrophilic attack can occur at various nucleophilic sites within the body, including DNA and RNA bases and proteins.' A common class of electrophiles is alkylating agents and the reaction pathway of these has been exploited in the design of anti-cancer drugs.2 Bis-(2-chloroethyl)methylamine was the first reported clinically effective anti-cancer agent3 and is a bifunctional alkylating agent, forming both monoDNA adducts as well as inter- or intra-strand crosslinks. A related compound, cyclophosphamide (1) (Fig. 1) is also a bifunctional alkylating agent but unlike bis-(2-chloroethyl)methylamine requires metabolic activation to generate a reactive alkylating agent.4,5 Initial activation of (1) occurs predominantly in the liver via the mixed function oxidase system. The oxidized derivative, 4-hydroxycyclophosphamide, is in equilibrium with the acyclic tautomer, aldophosphamide, which may spontaneously decompose to phosphoramide mustard (2) (Fig. 1). Phosphoramide mustard is believed to be one of the principal active metabolites of cyclophosphamide.6 Alkylation due to (2) occurs primarily at the N-7-position of guanine,"* but the products formed have been found to be very unstable, with half-lives of -2-3 h.93l o Various factors could contribute to this instability including facile cleavage of the N-P bond to give the corresponding nornitrogen adducts (3-5)." The cytotoxicity of bifunctional alkylating agents is often attributed to their ability to form cross-linked adducts.12 This cross-linkage may inhibit DNA strand separation which occurs on replication and hence * Authors to whom correspondence should be addressed.
prevents cell division. Once the mono-DNA adduct has been formed between the phosphoramide-mustard (2) and DNA it is possible that a further reaction may
r
H
Phosphoramide Mustard
Cyclophosphamide 1
2
Nor-G-OH
Nor-G 3
4
G-NOR-G 5
Figure 1. Structures of 1, cydophosphamide; 2, phosphoramide mustard; 3, N-(2-chloroethyl)-N-[2-(7-guaninyl)ethyl]amine,N O R - G ; 4, N-(2-hydroxyethyl)-N-[2-(7-guaninyl)ethyl]amine,N O R - G 4 H ; 5 , N,N-bis[2-(7-guaninyl)ethyl]amine, G-NOR-G.
0951 -4198/90/0410-0414$05 .OO 410
RAPID COMMUNICATIONS IN MASS SPECTROMETRY, VOL. 4, NO 10. 1990
@ John Wiley & Sons Limited, 1990
PHOSPHORAMIDE MUSTARDlDNA ADDUCTS
occur, to give the cross-linked adduct G - N O R 4 ( 5 ) , which is the adduct that has tentatively been attributed to account for the cytotoxicity of the anticancer drug (1). Tentative evidence also exists to suggest that the cross-linked adduct ( 5 ) is responsible for the teratogenic effect of this class of c o m p ~ u n d s . ' ~ In this work, we describe a method for the rapid characterization of N-7-alkylguanine adducts in a complex in uitro mixture, formed by the reaction of (2) with calf-thymus DNA. Both fast-atom bombardment mass spectrometry (FAB-MS) and fast-atom bombardment tandem mass spectrometry (FAB-MS/MS) are used. For the latter technique, particular emphasis was placed on parent-ion scanning.14 EXPERIMENTAL Materials and methods 2'-Deoxyguanosine was obtained from Sigma Chemical Co., Ltd (Poole, UK). Bis-2-chloroethylamine hydrochloride, 2,2,2-trifluoroethanol and triethylamine (Gold Label) were purchased from Aldrich Chemical Co., Ltd (UK). (5) were prepared NOR-G (3) and G-NOR-G by modification of a literature procedure:l5 2'deoxyguanosine (0.3 g) and bis-2-chloroethylamine hydrochloride (0.9 g) in 2,2,2-trifluoroethanol (25 mL) containing triethylamine (300 pL) were refluxed together overnight. The precipitate formed on cooling was filtered, hydrolysed by refluxing in 1~ hydrochloric acid for lOmin, then the solution was neutralized with potassium hydroxide. The resulting reaction mixture was analysed by FAB-MS and found to contain guanine, NOR-G and in low yield, G-NOR-G. A reaction product mixture containing a greater percentage of G-NOR-G and a considerably reduced percentage of NOR-G was prepared by reacting 2'deoxyguanosine (0.3 g) and bis-2-chloroethylamine hydrochloride (70 mg) in 2,2,2-trifluoroethanol (25 mL) containing triethylamine (80 ILL) as before. NOR-G-OH (4) was synthesized from NOR-G (3) (purified by high-performance liquid chromatography (HPLC)) by mild alkaline (pH 8) hydrolysis using sodium hydroxide. The synthetic mixture containing G-NOR-G was purified by dissolving the solid (400 pg) in 1M hydrochloric acid and evaporating the solution to dryness under a stream of nitrogen. The hydrochloride salt was then dissolved in double-distilled water (1 mL), loaded onto a CI8 Sep-Pak column, washed with double distilled water (1 mL) and eluted with methanol (2 mL). Calf-thymus DNA (50 mg, Sigma) was vortex-mixed with double-distilled water and left overnight at 4 "C. Phosphoramide mustard (2) (40 mg) in double-distilled water (250 vL) was added to the DNA solution, vortexmixed and incubated at 37 "C for 90 min. 1 M hydrochloric acid (1.15 mL) was added and the solution was heated at 70 "C for 60 min, followed by centrifuging at 4000rpm at room temperatur? for 10min and the supernatant liquid was then decanted. After neutralization with 1M sodium hydroxide, the supernatant liquid was applied to a 30X400mm Sephadex G-10 column and eluted with 100 mM ammonium formate in doubledistilled water at a flow rate of -2.5mL/min. Three fractions were collected, as shown in Fig. 2, and were
-
0
John Wiley & Sons Limited, 1990
purified as for the synthetic mixture, omitting the initial preparation of the hydrochloride salt. HPLC was performed using an LDC/Milton Roy (Stone, Staffs, UK) or Gilson (Anachem, Luton, UK) instrument, with an Ultrasphere 5 ODS CIS-reversed phase column (4.6 mm x 25 cm). Methanol and 1% trifluoroacetic acid (Sigma) in water were used as the HPLC solvents. The samples were analysed by increasing the methanol concentration from 2.5% (time zero mins) to 25% (time 14 mins) and then from 25%-80% (time 20 mins). Mass spectra All spectra were obtained on a type 70SEQ instrument, with EBQ,Q, geometry (VG Analytical Ltd, Manchester, UK). The samples were ionized by FAB using a beam of xenon atoms (energy 8.5 keV) in the ppositive-ion mode. The matrix used was 0 . 2 ~ toluene sulphonic acid (BDH) in glycerol (Aldrich) .16 The ions were accelerated from the source at 8 keV and the parent ions selected by the two sectors EB (MS1) with resolution of approximately 1000, and subjected to collision-activated dissociation (CAD) using Q1,which is an RF-only quadrupole and is the collision cell. The collision-cell conditions for the daughter-ion were optimized to ensure spectrum of G-NOR-G maximum abundance of the daughter ion at m / z 221 (collision energy of 20 eV, and collision gas (air) pressure of 5 x lopsmBar). Daughter- and parent-ion spectra were acquired by scanning the mass filter quadrupole over the mass range 50-600 u and the scans were collected in the multichannel analysis (MCA) mode. RESULTS AND DISCUSSION Initial studies focused on determining FAB-MS matrix conditions for optimum sensitivity to detect compounds 3-5. In positive-ion FAB-MS, only 3nmol of G-NOR-G (5) could be detected using a matrix of just glycerol, whereas less than 1nmol of G-NOR-G (5) could be detected using 0.2 M p-toluene sulphonic acid in glycerol, indicating a greater than threefold
I I1
I
30
0 0
" " ' J "
130
2 0
'
I
I
330
'
t i m e (mins.)
Figure 2. Chrornatogram of phosphoramide mustard (2)/DNA reaction products from a Sephadex G-10 column; flow rate 2.5 mL/rnin. A = adenine; G = guanine.
RAPID COMMUNICATIONS IN MASS SPECTROMETRY, VOL. 4,
NO.10. 1990
411
PHOSPHORAMIDE MUSTARD/ DNA ADDUCTS
6n
(b)
1
I
'Is
1%
b
?I
8 I88
lsl
dB
68
$$a
KBB
mL?
Figure 3. Positive-ion daughter spectra of: (a) NOR-G (3), MH+ = 257; (b), NOR--G-OH (4), MH+=239; (c), G-NOR-G (5), MH+ = 372.
increase in sensitivity. Positive-ion FAB-MS of the synthetic mixture (containing 3 and 5) exhibited ions at mlz 152, 257 (and 259), 372 and 408 (and 410). These ions correspond to [ G + H ] + , [3+H]+, [5+H]+ and [5 2H + CI]+, respectively. FAB-MSIMS studies were then performed to determine how these compounds plus synthetic 4, fragment when undergoing CAD. The results obtained for the daughter-ion spectra of the synthetic standards 3-5 are shown in Fig. 3. Both positive-ion daughter spectra for rnlz 257 [3 HI+ and rnlz 239 [4 + H]+ afford a daughter ion at mlz 152, corresponding to [G + HI+, indicating the loss of the side chain. However, the positive-ion daughter spectra of mlz 372 [5 H]+ and 408 [S+ 2H + Cl]+ contain ions at mlz 221 and 257 respectively, indicating the loss of guanine in both cases. Isolation of G-NOR-G, 5 , is problematical since it is relatively insoluble in most common solvents. Using FAB-MSIMS parent-ion scanning, which should eliminate the need for extensive sample purification, a method for rapidly quantitating such adducts in humans and animal embryos treated with cyclophosphamide is possible. Initially, a FAB-MS scan of a partially purified sample derived from a Sephadex G-10 column (Fraction 3) from the in uitro alkylation of DNA with phosphoramide mustard, was performed and is shown in Fig. 4. The FAB spectrum contained ions at mlz 239 and 257/ (4) and 259, which correspond to NOR-G-OH NOR-G (3), respectively. However, no ions were + H]+ (5) and 408 detected at rnlz 372 [G-NOR-G [G-NOR-G 2H a ] + . Hence it was decided to apply a more selective technique, namely tandem mass spectrometric parent-ion scanning. The daughter-ion spectra of the standards 3-5 show that there is not a daughter ion common to all three
+
adducts (Fig. 3). Since the quantity of the in uitro sample was limited, a double parent-ion scan was carried out to detect parents of mlz 152 (NOR-G (3) and NOR-G-OH (4)) and rnlz 221 (G-NOR-G (5)) and data stored in a single MCA file. Initially, a double parent-ion scan of the matrix as well as the two Sephadex G-10 fractions 1 and 2, were performed. These spectra contained ions predominantly at m/z 152 and 221, as well as other less abundant ions at mlt 179,233,238,251, 313 and 394, all of which had approximately 10% of the intensity of the ion at mlz 221. No ions were detected in these spectra that corresponded to any of the three adducts 3-5. This method was then applied to the synthetic mixture containing predominantly G-NOR-G (5) (results not shown). An ion at mlz 372 was clearly detected above the background corresponding to compound 5. A further, much less abundant ion, was seen at rnlz 257 attributable to NOR-G (3), but this was barely visible above the background ions. A double parent-ion scan of fraction 3 from the Sephadex G-10 column was performed, and ions at rnlz 239, 2571259 and 372 were (4), detected corresponding to NOR-G-OH NOR-G (3) and G-NOR-G (S), respectively (Figs 4,5), and clearly indicated that all three guanine adducts were preent in the in uitro sample reaction of compound 2 with DNA. A single parent-ion scan of mlz 221 was also carried out on the remaining amount of limited in uitro sample and this is shown in Fig. 6. An ion was still clearly detected above background at mlz 372, whereas ions at mlz 2571259 and 239 were barely visible above background. This confirms that the ion at rnlz 372 is solely derived from mlz 221 and corresponds to G-NOR-G
+
+
+ +
'
' m/r
Figure4. Positive-ion FAB-MS of the Sephadex G-10 fraction 3 derived from reaction of phosphoramide mustard (2) with calfthymus DNA.
412 RAPID COMMUNICATIONS IN MASS SPECTROMETRY, VOL. 4.NO 10, 1990
0
John Wiley & Sons Limited, 1990
PHOSPHORAMIDE MUSTARD/DNA ADDUCTS
15%
m/z
Figure5. Double parent-ion scan of mlz 152 and 221 of the Sephadex (3-10 fraction 3, with data being collected into the same MCA datum file.
(5). Performing parent-ion scanning increases specificity of detection for G-NOR-G ( 5 ) . As previously stated, 1nmol of G-NOR-G (5) can be detected by
m/r
Figure6. Single parent-ion scan of m / z 221 of the Sephadex G-10 fraction 3.
0
John Wiley & Sons Limited, 1990
FAB-MS. However, a lower limit of -80pmol of G-NOR-G ( 5 ) can be detected using parent-ion scanning. In this work we also investigated the origins of NOR-G-OH (4) (MH+ 239), since previous worker^'^^'^ have isolated very different ratios of this adduct relative to NOR-G (3) and G-NOR-G (5). Both Hemminki” and Benson et al. ,17 have investigated the in uitro reaction of (1) with DNA. Hemminki” (3), NOR-G-OH (4) and found NOR-G (5) to be formed, with the ratio of G-NOR-G NOR-G-OH (4) to NOR-G (3) increasing with incubation time. In contrast, Benson et a1.,17found that (4) was the predominant product NOR-G-OH formed, even after using a shorter incubation period compared to Hemminki. In this work NOR-G (3) was dissolved in both 0.1 M hydrochloric acid and sodium hydroxide (pH -8) and both solutions were left for 48 h at room temperature and then analysed by FAB-MS. The acid solution still contained only NOR-G (3) (MH+ 257/259) whereas the alkaline solution now contained an abundant ion at rnlz 239 corresponding to NOR-G-OH (4). Comparison of the incubation conditions used by the two groups reveal that they do differ, but not to the extent that one would predict such a great variation in adduct formation. In the respective work-up procedures, Hemminki” extracts with phenol saturated with water, whereas Benson er al. ,17 extract with phenol saturated with Tris-HC1 (pH 8.0). By comparison with our experiments, it is possible that the high percentage of NOR-G-OH (4), observed by Benson et af.,17 may have resulted from their basic extraction con(4), ditions. However, the increase of NOR-G-OH seen by Hemminki” with increasing incubation time does seem to suggest that some NOR-G-OH (4) is formed in the actual incubation reaction. The increased quantity of NOR-G-OH (4) detected by Benson et a1.,17 may be an artifact of isolation, and not all due to the reaction of (1) with DNA. CONCLUSION A method has been described to detect rapidly, with minimum sample clean-up, the in uitro adducts formed when reacting compound 2, the active metabolite of compound 1, with calf-thymus DNA. By performing a double parent-ion scan of mlz 152 and 221 and a single parent-ion scan of mlz 221, we have obtained evidence that at least three adducts are formed in this in uitro preparation: NOR-G (3), NOR-G-OH (4) and ( 5 ) . However, it is still not clear at this G-NOR-G stage whether the product of hydrolysis, (4), is formed during the in uitro incuNOR-G-OH bation reaction or is an artifact of the isolation procedure. By performing parent-ion scanning, we have identi( 5 ) , although fied the cross-linked adduct G-NOR-G no molecular ion was detected when a FAB-MS scan was performed on the in uitro mixture. Hence, the tandem mass spectrometric method showed an increase in detection limits when analysing G-NOR-G (5) over performing just FAB-MS. By using this method, we have removed the need for isolating pure ( 5 ) , which can be problematical due to G-NOR-G its solubility properties. The rapid identification of this
RAPID COMMUNICATIONS IN MASS SPECTROMETRY, VOL. 4, NO. 10, 1990 413
PHOSPHORAMIDE MUSTARD/DNA ADDUCTS
compound is important since it is believed that the cross-linked adducts are of crucial importance for the cytotoxicity of the anti-cancer drugs, and appear to exert a teratogenic effect. REFERENCES 1. DNA and Protein Adducts: Evaluation of their use in exposure monitoring and risk assessment. ECETOC, Monograph no. 13, Brussels (1989). 2. P. B. Farmer, Pharm. Ther. 35, 301 (1987). 3. A. Gilman and F. S. Philips, Science, 103, 409 (1946). 4. P. J. Cox, P. B. Farmer, A. B. Foster, E. D. Gilby and M. Jarman, Cancer Treatment Rep. 60, 483 (1976). 5. R. F. Struck, M. C. Kirk, M. H. Witt and W. R. Laster, Jr. Biomed. Mass Spectrom. 2, 46 (1975). 6. T. A. Connors, P. J. Cox, P. B. Farmer, A. B. Foster and M. Jarman, Biochem. Pharmacol. 23, 115 (1974).
414
7. V. T. Vu, C. C. Fenselau and 0. M. Colvin, J . A m . Chem. SOC. 103,7362 (1981). 8. J. R. Mehta, M. Przybylski and D. B. Ludlum, Cancer Res. 40, 4183 (1980). 9. S . Kallama and K. Hemminki, Acta. Pharmacol. Toxicol. 54,214 (1984). 10. J. R. Mehta and D. B. Ludlum, Cancer Res. 42, 2996 (1982). 11. 0. M. Colvin, R. B. Brundett, M-N. N. Kan, I. Jardine and C. Fenselau, Cancer Res. 36, 1121 (1976). 12. P. Brookes and P. D. Lawley, Biochem. 3. 80, 496 (1961). 13. S. A. Little and P. E. Mikes, Cancer Res. 47, 5421 (1987). 14. G. C. DiDonato and K. L. Busch, Anal. Chem. 58, 229 (1986). 15. K. Hemminki, Cancer Res. 45, 4237 (1985). 16. H. Moses and G. W. Wood, Biomed. Enuiron. Mass Spectrom. 15, 547 (1988). 17. A. J. Benson, C. N. Martin and R. C. Garner, Biochem. Pharmacol. 31, 2979 (1988). Received 30 July 1990; accepted 10 Auguust 1990.
RAPID COMMUNICATIONS IN MASS SPECTROMETRY, VOL. 4, NO. 10. 1990
@ John Wilcy & Sons Limited, 1990
Nigel A. Brown MRC Experimental Embryology and Teratology Unit, St. George's Hospital, University of London, London, SW17 ORE, UK
Philip E. Mirkes Department of Pediatrics, University of Washington, Seattle, W A 98195,USA
The reaction pathway of alkylating agents is often exploited in the design of bifunctional anti-cancer drugs. These drugs form mono-DNA adducts as well as inter- and intra-strand cross-linked adducts, notably by reaction at DNA bases, including the N-7-position of guanine (G). A positive-ion fast-atom bombardment (FAB) mass spectrum of an in uitro preparation of DNA alkylated with phosphoramide mustard (the active metabolite of the anti-cancer drug cyclophosphamide) indicated the presence of the two mono-DNA adducts N-(2-chloroethyl)-N[2-(7-guaninyl)ethyl] amine, designated NOR-G, and N-(2-hydroxyethyl)-N-[2-(7-guaninyl)ethyl]amine, designated N O R - G - O H , (MH+ 257/259 and 239, respectively) but not the presence of the cross-linked adduct N,N-bis-[2-(7-guaninyl)ethyl] amine, designated G - N O R 4 (MH+ 372). Using synthetic standards, daughterion spectra of NOR-G, NOR-G-OH and G - N O R 4 were obtained (matrix 0.2 ~ p - t o l u e n esulphonic acid in glycerol) by positive-ion FAB tandem mass spectrometry (FAB-MSIMS). The daughter-ion spectra of both mono-DNA adducts N O R 4 and NOR-G-OH contained a fragment ion at rnlz 152 [G+H]+, whereas the cross-linked adduct, G-NOR-G, showed an ion at mlz221, [MH-GI+. Evidence for the presence of NOR-G, NOR-G-OH and G-NOR-G in the in uitro preparation was obtained by performing a double parent-ion scan on rnlz 152 and 221. The presence of G - N O R 4 was further supported by performing a single parent-ion scan on m/z 221. The use of this MS/MS technique should eliminate the need for intricate sample purification in the identification of G-NOR-G in biological extracts.
Electrophilic attack can occur at various nucleophilic sites within the body, including DNA and RNA bases and proteins.' A common class of electrophiles is alkylating agents and the reaction pathway of these has been exploited in the design of anti-cancer drugs.2 Bis-(2-chloroethyl)methylamine was the first reported clinically effective anti-cancer agent3 and is a bifunctional alkylating agent, forming both monoDNA adducts as well as inter- or intra-strand crosslinks. A related compound, cyclophosphamide (1) (Fig. 1) is also a bifunctional alkylating agent but unlike bis-(2-chloroethyl)methylamine requires metabolic activation to generate a reactive alkylating agent.4,5 Initial activation of (1) occurs predominantly in the liver via the mixed function oxidase system. The oxidized derivative, 4-hydroxycyclophosphamide, is in equilibrium with the acyclic tautomer, aldophosphamide, which may spontaneously decompose to phosphoramide mustard (2) (Fig. 1). Phosphoramide mustard is believed to be one of the principal active metabolites of cyclophosphamide.6 Alkylation due to (2) occurs primarily at the N-7-position of guanine,"* but the products formed have been found to be very unstable, with half-lives of -2-3 h.93l o Various factors could contribute to this instability including facile cleavage of the N-P bond to give the corresponding nornitrogen adducts (3-5)." The cytotoxicity of bifunctional alkylating agents is often attributed to their ability to form cross-linked adducts.12 This cross-linkage may inhibit DNA strand separation which occurs on replication and hence * Authors to whom correspondence should be addressed.
prevents cell division. Once the mono-DNA adduct has been formed between the phosphoramide-mustard (2) and DNA it is possible that a further reaction may
r
H
Phosphoramide Mustard
Cyclophosphamide 1
2
Nor-G-OH
Nor-G 3
4
G-NOR-G 5
Figure 1. Structures of 1, cydophosphamide; 2, phosphoramide mustard; 3, N-(2-chloroethyl)-N-[2-(7-guaninyl)ethyl]amine,N O R - G ; 4, N-(2-hydroxyethyl)-N-[2-(7-guaninyl)ethyl]amine,N O R - G 4 H ; 5 , N,N-bis[2-(7-guaninyl)ethyl]amine, G-NOR-G.
0951 -4198/90/0410-0414$05 .OO 410
RAPID COMMUNICATIONS IN MASS SPECTROMETRY, VOL. 4, NO 10. 1990
@ John Wiley & Sons Limited, 1990
PHOSPHORAMIDE MUSTARDlDNA ADDUCTS
occur, to give the cross-linked adduct G - N O R 4 ( 5 ) , which is the adduct that has tentatively been attributed to account for the cytotoxicity of the anticancer drug (1). Tentative evidence also exists to suggest that the cross-linked adduct ( 5 ) is responsible for the teratogenic effect of this class of c o m p ~ u n d s . ' ~ In this work, we describe a method for the rapid characterization of N-7-alkylguanine adducts in a complex in uitro mixture, formed by the reaction of (2) with calf-thymus DNA. Both fast-atom bombardment mass spectrometry (FAB-MS) and fast-atom bombardment tandem mass spectrometry (FAB-MS/MS) are used. For the latter technique, particular emphasis was placed on parent-ion scanning.14 EXPERIMENTAL Materials and methods 2'-Deoxyguanosine was obtained from Sigma Chemical Co., Ltd (Poole, UK). Bis-2-chloroethylamine hydrochloride, 2,2,2-trifluoroethanol and triethylamine (Gold Label) were purchased from Aldrich Chemical Co., Ltd (UK). (5) were prepared NOR-G (3) and G-NOR-G by modification of a literature procedure:l5 2'deoxyguanosine (0.3 g) and bis-2-chloroethylamine hydrochloride (0.9 g) in 2,2,2-trifluoroethanol (25 mL) containing triethylamine (300 pL) were refluxed together overnight. The precipitate formed on cooling was filtered, hydrolysed by refluxing in 1~ hydrochloric acid for lOmin, then the solution was neutralized with potassium hydroxide. The resulting reaction mixture was analysed by FAB-MS and found to contain guanine, NOR-G and in low yield, G-NOR-G. A reaction product mixture containing a greater percentage of G-NOR-G and a considerably reduced percentage of NOR-G was prepared by reacting 2'deoxyguanosine (0.3 g) and bis-2-chloroethylamine hydrochloride (70 mg) in 2,2,2-trifluoroethanol (25 mL) containing triethylamine (80 ILL) as before. NOR-G-OH (4) was synthesized from NOR-G (3) (purified by high-performance liquid chromatography (HPLC)) by mild alkaline (pH 8) hydrolysis using sodium hydroxide. The synthetic mixture containing G-NOR-G was purified by dissolving the solid (400 pg) in 1M hydrochloric acid and evaporating the solution to dryness under a stream of nitrogen. The hydrochloride salt was then dissolved in double-distilled water (1 mL), loaded onto a CI8 Sep-Pak column, washed with double distilled water (1 mL) and eluted with methanol (2 mL). Calf-thymus DNA (50 mg, Sigma) was vortex-mixed with double-distilled water and left overnight at 4 "C. Phosphoramide mustard (2) (40 mg) in double-distilled water (250 vL) was added to the DNA solution, vortexmixed and incubated at 37 "C for 90 min. 1 M hydrochloric acid (1.15 mL) was added and the solution was heated at 70 "C for 60 min, followed by centrifuging at 4000rpm at room temperatur? for 10min and the supernatant liquid was then decanted. After neutralization with 1M sodium hydroxide, the supernatant liquid was applied to a 30X400mm Sephadex G-10 column and eluted with 100 mM ammonium formate in doubledistilled water at a flow rate of -2.5mL/min. Three fractions were collected, as shown in Fig. 2, and were
-
0
John Wiley & Sons Limited, 1990
purified as for the synthetic mixture, omitting the initial preparation of the hydrochloride salt. HPLC was performed using an LDC/Milton Roy (Stone, Staffs, UK) or Gilson (Anachem, Luton, UK) instrument, with an Ultrasphere 5 ODS CIS-reversed phase column (4.6 mm x 25 cm). Methanol and 1% trifluoroacetic acid (Sigma) in water were used as the HPLC solvents. The samples were analysed by increasing the methanol concentration from 2.5% (time zero mins) to 25% (time 14 mins) and then from 25%-80% (time 20 mins). Mass spectra All spectra were obtained on a type 70SEQ instrument, with EBQ,Q, geometry (VG Analytical Ltd, Manchester, UK). The samples were ionized by FAB using a beam of xenon atoms (energy 8.5 keV) in the ppositive-ion mode. The matrix used was 0 . 2 ~ toluene sulphonic acid (BDH) in glycerol (Aldrich) .16 The ions were accelerated from the source at 8 keV and the parent ions selected by the two sectors EB (MS1) with resolution of approximately 1000, and subjected to collision-activated dissociation (CAD) using Q1,which is an RF-only quadrupole and is the collision cell. The collision-cell conditions for the daughter-ion were optimized to ensure spectrum of G-NOR-G maximum abundance of the daughter ion at m / z 221 (collision energy of 20 eV, and collision gas (air) pressure of 5 x lopsmBar). Daughter- and parent-ion spectra were acquired by scanning the mass filter quadrupole over the mass range 50-600 u and the scans were collected in the multichannel analysis (MCA) mode. RESULTS AND DISCUSSION Initial studies focused on determining FAB-MS matrix conditions for optimum sensitivity to detect compounds 3-5. In positive-ion FAB-MS, only 3nmol of G-NOR-G (5) could be detected using a matrix of just glycerol, whereas less than 1nmol of G-NOR-G (5) could be detected using 0.2 M p-toluene sulphonic acid in glycerol, indicating a greater than threefold
I I1
I
30
0 0
" " ' J "
130
2 0
'
I
I
330
'
t i m e (mins.)
Figure 2. Chrornatogram of phosphoramide mustard (2)/DNA reaction products from a Sephadex G-10 column; flow rate 2.5 mL/rnin. A = adenine; G = guanine.
RAPID COMMUNICATIONS IN MASS SPECTROMETRY, VOL. 4,
NO.10. 1990
411
PHOSPHORAMIDE MUSTARD/ DNA ADDUCTS
6n
(b)
1
I
'Is
1%
b
?I
8 I88
lsl
dB
68
$$a
KBB
mL?
Figure 3. Positive-ion daughter spectra of: (a) NOR-G (3), MH+ = 257; (b), NOR--G-OH (4), MH+=239; (c), G-NOR-G (5), MH+ = 372.
increase in sensitivity. Positive-ion FAB-MS of the synthetic mixture (containing 3 and 5) exhibited ions at mlz 152, 257 (and 259), 372 and 408 (and 410). These ions correspond to [ G + H ] + , [3+H]+, [5+H]+ and [5 2H + CI]+, respectively. FAB-MSIMS studies were then performed to determine how these compounds plus synthetic 4, fragment when undergoing CAD. The results obtained for the daughter-ion spectra of the synthetic standards 3-5 are shown in Fig. 3. Both positive-ion daughter spectra for rnlz 257 [3 HI+ and rnlz 239 [4 + H]+ afford a daughter ion at mlz 152, corresponding to [G + HI+, indicating the loss of the side chain. However, the positive-ion daughter spectra of mlz 372 [5 H]+ and 408 [S+ 2H + Cl]+ contain ions at mlz 221 and 257 respectively, indicating the loss of guanine in both cases. Isolation of G-NOR-G, 5 , is problematical since it is relatively insoluble in most common solvents. Using FAB-MSIMS parent-ion scanning, which should eliminate the need for extensive sample purification, a method for rapidly quantitating such adducts in humans and animal embryos treated with cyclophosphamide is possible. Initially, a FAB-MS scan of a partially purified sample derived from a Sephadex G-10 column (Fraction 3) from the in uitro alkylation of DNA with phosphoramide mustard, was performed and is shown in Fig. 4. The FAB spectrum contained ions at mlz 239 and 257/ (4) and 259, which correspond to NOR-G-OH NOR-G (3), respectively. However, no ions were + H]+ (5) and 408 detected at rnlz 372 [G-NOR-G [G-NOR-G 2H a ] + . Hence it was decided to apply a more selective technique, namely tandem mass spectrometric parent-ion scanning. The daughter-ion spectra of the standards 3-5 show that there is not a daughter ion common to all three
+
adducts (Fig. 3). Since the quantity of the in uitro sample was limited, a double parent-ion scan was carried out to detect parents of mlz 152 (NOR-G (3) and NOR-G-OH (4)) and rnlz 221 (G-NOR-G (5)) and data stored in a single MCA file. Initially, a double parent-ion scan of the matrix as well as the two Sephadex G-10 fractions 1 and 2, were performed. These spectra contained ions predominantly at m/z 152 and 221, as well as other less abundant ions at mlt 179,233,238,251, 313 and 394, all of which had approximately 10% of the intensity of the ion at mlz 221. No ions were detected in these spectra that corresponded to any of the three adducts 3-5. This method was then applied to the synthetic mixture containing predominantly G-NOR-G (5) (results not shown). An ion at mlz 372 was clearly detected above the background corresponding to compound 5. A further, much less abundant ion, was seen at rnlz 257 attributable to NOR-G (3), but this was barely visible above the background ions. A double parent-ion scan of fraction 3 from the Sephadex G-10 column was performed, and ions at rnlz 239, 2571259 and 372 were (4), detected corresponding to NOR-G-OH NOR-G (3) and G-NOR-G (S), respectively (Figs 4,5), and clearly indicated that all three guanine adducts were preent in the in uitro sample reaction of compound 2 with DNA. A single parent-ion scan of mlz 221 was also carried out on the remaining amount of limited in uitro sample and this is shown in Fig. 6. An ion was still clearly detected above background at mlz 372, whereas ions at mlz 2571259 and 239 were barely visible above background. This confirms that the ion at rnlz 372 is solely derived from mlz 221 and corresponds to G-NOR-G
+
+
+ +
'
' m/r
Figure4. Positive-ion FAB-MS of the Sephadex G-10 fraction 3 derived from reaction of phosphoramide mustard (2) with calfthymus DNA.
412 RAPID COMMUNICATIONS IN MASS SPECTROMETRY, VOL. 4.NO 10, 1990
0
John Wiley & Sons Limited, 1990
PHOSPHORAMIDE MUSTARD/DNA ADDUCTS
15%
m/z
Figure5. Double parent-ion scan of mlz 152 and 221 of the Sephadex (3-10 fraction 3, with data being collected into the same MCA datum file.
(5). Performing parent-ion scanning increases specificity of detection for G-NOR-G ( 5 ) . As previously stated, 1nmol of G-NOR-G (5) can be detected by
m/r
Figure6. Single parent-ion scan of m / z 221 of the Sephadex G-10 fraction 3.
0
John Wiley & Sons Limited, 1990
FAB-MS. However, a lower limit of -80pmol of G-NOR-G ( 5 ) can be detected using parent-ion scanning. In this work we also investigated the origins of NOR-G-OH (4) (MH+ 239), since previous worker^'^^'^ have isolated very different ratios of this adduct relative to NOR-G (3) and G-NOR-G (5). Both Hemminki” and Benson et al. ,17 have investigated the in uitro reaction of (1) with DNA. Hemminki” (3), NOR-G-OH (4) and found NOR-G (5) to be formed, with the ratio of G-NOR-G NOR-G-OH (4) to NOR-G (3) increasing with incubation time. In contrast, Benson et a1.,17found that (4) was the predominant product NOR-G-OH formed, even after using a shorter incubation period compared to Hemminki. In this work NOR-G (3) was dissolved in both 0.1 M hydrochloric acid and sodium hydroxide (pH -8) and both solutions were left for 48 h at room temperature and then analysed by FAB-MS. The acid solution still contained only NOR-G (3) (MH+ 257/259) whereas the alkaline solution now contained an abundant ion at rnlz 239 corresponding to NOR-G-OH (4). Comparison of the incubation conditions used by the two groups reveal that they do differ, but not to the extent that one would predict such a great variation in adduct formation. In the respective work-up procedures, Hemminki” extracts with phenol saturated with water, whereas Benson er al. ,17 extract with phenol saturated with Tris-HC1 (pH 8.0). By comparison with our experiments, it is possible that the high percentage of NOR-G-OH (4), observed by Benson et af.,17 may have resulted from their basic extraction con(4), ditions. However, the increase of NOR-G-OH seen by Hemminki” with increasing incubation time does seem to suggest that some NOR-G-OH (4) is formed in the actual incubation reaction. The increased quantity of NOR-G-OH (4) detected by Benson et a1.,17 may be an artifact of isolation, and not all due to the reaction of (1) with DNA. CONCLUSION A method has been described to detect rapidly, with minimum sample clean-up, the in uitro adducts formed when reacting compound 2, the active metabolite of compound 1, with calf-thymus DNA. By performing a double parent-ion scan of mlz 152 and 221 and a single parent-ion scan of mlz 221, we have obtained evidence that at least three adducts are formed in this in uitro preparation: NOR-G (3), NOR-G-OH (4) and ( 5 ) . However, it is still not clear at this G-NOR-G stage whether the product of hydrolysis, (4), is formed during the in uitro incuNOR-G-OH bation reaction or is an artifact of the isolation procedure. By performing parent-ion scanning, we have identi( 5 ) , although fied the cross-linked adduct G-NOR-G no molecular ion was detected when a FAB-MS scan was performed on the in uitro mixture. Hence, the tandem mass spectrometric method showed an increase in detection limits when analysing G-NOR-G (5) over performing just FAB-MS. By using this method, we have removed the need for isolating pure ( 5 ) , which can be problematical due to G-NOR-G its solubility properties. The rapid identification of this
RAPID COMMUNICATIONS IN MASS SPECTROMETRY, VOL. 4, NO. 10, 1990 413
PHOSPHORAMIDE MUSTARD/DNA ADDUCTS
compound is important since it is believed that the cross-linked adducts are of crucial importance for the cytotoxicity of the anti-cancer drugs, and appear to exert a teratogenic effect. REFERENCES 1. DNA and Protein Adducts: Evaluation of their use in exposure monitoring and risk assessment. ECETOC, Monograph no. 13, Brussels (1989). 2. P. B. Farmer, Pharm. Ther. 35, 301 (1987). 3. A. Gilman and F. S. Philips, Science, 103, 409 (1946). 4. P. J. Cox, P. B. Farmer, A. B. Foster, E. D. Gilby and M. Jarman, Cancer Treatment Rep. 60, 483 (1976). 5. R. F. Struck, M. C. Kirk, M. H. Witt and W. R. Laster, Jr. Biomed. Mass Spectrom. 2, 46 (1975). 6. T. A. Connors, P. J. Cox, P. B. Farmer, A. B. Foster and M. Jarman, Biochem. Pharmacol. 23, 115 (1974).
414
7. V. T. Vu, C. C. Fenselau and 0. M. Colvin, J . A m . Chem. SOC. 103,7362 (1981). 8. J. R. Mehta, M. Przybylski and D. B. Ludlum, Cancer Res. 40, 4183 (1980). 9. S . Kallama and K. Hemminki, Acta. Pharmacol. Toxicol. 54,214 (1984). 10. J. R. Mehta and D. B. Ludlum, Cancer Res. 42, 2996 (1982). 11. 0. M. Colvin, R. B. Brundett, M-N. N. Kan, I. Jardine and C. Fenselau, Cancer Res. 36, 1121 (1976). 12. P. Brookes and P. D. Lawley, Biochem. 3. 80, 496 (1961). 13. S. A. Little and P. E. Mikes, Cancer Res. 47, 5421 (1987). 14. G. C. DiDonato and K. L. Busch, Anal. Chem. 58, 229 (1986). 15. K. Hemminki, Cancer Res. 45, 4237 (1985). 16. H. Moses and G. W. Wood, Biomed. Enuiron. Mass Spectrom. 15, 547 (1988). 17. A. J. Benson, C. N. Martin and R. C. Garner, Biochem. Pharmacol. 31, 2979 (1988). Received 30 July 1990; accepted 10 Auguust 1990.
RAPID COMMUNICATIONS IN MASS SPECTROMETRY, VOL. 4, NO. 10. 1990
@ John Wilcy & Sons Limited, 1990
