Dealkylation rates of O6-alkyldeoxyguanosine, O4-alkylthymidine and related compounds in an alkyl-transfer system

Chem.-Biol. Interactions, 78 (1991) 23--32 Elsevier Scientific Publishers Ireland Ltd.

23

DEALKYLATION RATES OF O6-ALKYLDEOXYGUANOSINE, O4-ALKYLTHYMIDINE AND RELATED COMPOUNDS IN AN ALKYL-TRANSFER SYSTEM

KOHFUKU KOHDA, MORIYOSHI YASUDA, NORIAKI SAWADA, KEIJI ITANO and YUTAKA KAWAZOE

Faculty of Pharmaceutical Sciences, Nagoya City University, Tanabedori, Mizuho-ku, Nagoya 467 (Japan) (Received August 14th, 1990) (Revision received November 20th, 1990) (Accepted November 21st, 1990)

SUMMARY

Bacterial 06-alkylguanine-DNA alkyltransferase (AGT) removes alkyl group from 06-alkylguanine and O4-alkylthymine residues in DNA, both of which are considered to be DNA damages most related to the induction of cancer and/or mutation. The repair process involves alkyl-transfer of an O-alkyl group to the active site of the enzyme, where an SH-group of cysteine residue plays the role of alkyl accepter. In order to elucidate the chemical characteristics of substrates for this enzyme, dealkylation rates of 06-alkyldeoxyguanosine, 04-alkylthymidine and related compounds were measured using an alkyltransfer system. Thiophenol-triethylamine system was employed as an alkyl acceptor and twenty-one O-alkyl compounds were tested. Dealkylation proceeded with pseudo first order kinetics. The half-life of O6-methyldeoxyguanosine (MedG) was 122 h and no remarkable dependence on N-9 substituents (H, CH 3 and deoxyribose) was observed. A compound lacking 2-NH 2 group underwent demethylation about three times faster than 06-methylguanines did, while, a compound lacking imidazole moiety underwent demethylation about 2.5 times more slowly. The half-life of O4-methylthymidine (MedT) was 38 h and no remarkable dependence on N-1 (H, CH 3 and deoxyribose) and C-5 (H and CH3) substituents was observed. Deethylation proceeded much more slowly than demethylation. Substitution of selenophenol for thiophenol resulted in a 4.5 times faster MedG demethylation rate. Demethylation rates were moderately correlated with values for NMR chemical shift of CH 3 group, an indicator of electron density, although the correlation curves of a series of MedG and MedT derivatives were quite different. This result suggests that some different ratedetermining factors other than electron density are playing a role. These finCorrespondence to: Dr. Kohfuku Kohda, Faculty of Pharmaceutical Sciences, Nagoya City University, Tanabedori, Mizuho-ku, Nagoya 467, Japan. 0009-2797/91/$03.50 © 1991 Elsevier Scientific Publishers Ireland Ltd. Printed and Published in Ireland

24 dings may be of help in resolving the details of the mechanisms of enzymic repair by bacterial and mammalian AGT. Key words: Repair -- O6-Alkylguanine-DNA alkyltransferase -- O6-Methyl deoxyguanosine -- O4-Methylthymidine -- Dealkylation -- Thiol

INTRODUCTION

Simple alkylating carcinogens such as N-alkyl-N-nitrosoureas, N-alkylN'-nitro-N-nitrosoguanidines and N,N'-dialkyl-N-nitrosoamines, etc. are known to alkylate cellular DNA and, as a consequence, to induce cancer through subsequent biological processes in the cells. Alkylation takes place at various nucleophilic sites of DNA bases, in addition to the phosphodiester moieties, resulting in formation of 3-, 7- and O6-alkylguanines, 3- and 7-alkyladenines, O2and O4-alkylthymines, and O2-alkylcytosine, etc. Cells, ranging widely from these in mammals to prokaryotic bacteria, all have the capacity to repair such DNA damage by the action of complex enzymic repair systems. Among them, the ada gene product, 06-alkylguanine-DNA alkyltransferase (AGT), takes part in repairing O6-alkylguanine and 04-alkylthymine moieties [1], both of which have been recognized as DNA damages most probably responsible for the induction of cancer and/or mutation [2,3]. Extensive studies performed with AGT of Escherichia coli revealed that the repair mechanism involves an alkyl-transfer from O-alkylated bases to the SH group of the cysteine (Cys-321) which is located next to the histidine (His-322) of the AGT molecule [4]. The SH group of the cysteine might thereby be activated to a thiolate anion by forming a hydrogen bond with the imidazole moiety of the histidine [5]. From a chemical viewpoint, this alkyl-transfer (dealkylation) proceeds via nucleophilic substitution of O-alkylated bases with SH groups acting as alkylating agents and nucleophiles, respectively. The present report concerns the dealkylation rates of O-methyl and O-ethyl derivatives of nucleic acid components, including O6-methyldeoxyguanosine and 04-metylthymidine in a biomimetic chemical system. Little has so far been revealed regarding chemical dealkylation of O-alkyl bases in connection with enzymic repair by AGT. MATERIALS AND METHODS

Chemicals 06-Methylguanine [6], O6,9-dimethylguanine [7], O6-methyldeoxyguanosine [8], O6,7-dimethylguanine [7], 8-methoxyguanosine [9], O4-methylthymidine [10], 04,1-dimethylthymine [11], O4-methylthymine [11], O2-methylthymine [11], O4,1-dimethyluracil [11], O4-methyluracil [11], 02-methyluracil [11], O2-methylcytosine [12], O4-methylisocytosine [12], O6-ethylguanine [6], 06-ethyldeoxyguanosine [8], 8-ethoxyguanosine [9], 04-ethyluracil [11], and O2-ethyluracil [11] were prepared in this laboratory according to reported procedures. 06-Methylhypoxanthine, 5-methoxyuridine, selenophenol, 2-amino-

25 thiophenol and 2-mercaptoimidazole were purchased from Aldrich Chemical Co. Inc. Wisconsin. Thiophenol, triethylamine, 2-mercaptoethanol, ethanethiol and imidazole were from Tokyo Kasei Co. Ltd., Tokyo. The structures of the tested O-alkyl compounds are shown in Fig. 1 along with their identification numbers.

Conditions of the dealkylation reaction The reaction mixture consisted of O-alkyl compound (8.70 × 10 -3 mmol), thiophenol 90 #l (d = 1.073, 8.78 × 10 -1 mmol), triethylamine 120 ~l (d -0.726, 8.62 × 10-1 mmol) and MeOH 500/~l in a glass tube with a sealed cap, was maintained at 60°C in a dry block bath. OMe

1

OMe

dR

2

OMe

OMe

Me

3

OMe

o .,j 6

7

8

OEt

OEt

.H~N"~"N 11

N H ~ N I'~N

dR

12

O

MeO

10

H 13 NIt 2

0

15

OMe Me

H

9

EtO

14

N

o

0

H

0

MeO " N "

OMe

H

Me

OEt

5

OMe

o ,

Me

OMe

4

OMe

o .,j

dR

OMe

MeO 16

17

0

0

N H ~ N "v"~N Rf

NH21~"-N~ ~N Rf

0

N

NH2""~'-N'r~ N 18

19

20

RI 21

Rf: •-D-ribofurmmsyl dR: 2'-deoxy-6-D-ribofuranosyl Fig. 1. Structures and identification numbers of the O-alkyl compounds employed.

26

Product analysis In every experiment, the formation of dealkyl compound from the corresponding O-alkyl compound was identified by use of normal and/or reversed phase HPLC and authentic samples.

Measurement of deallcylation rates After appropriate times, 10-~1 aliquots of solution were taken out of the reaction mixture and diluted with I ml of MeOH. Ten microliters of the diluted solution were then subjected to analysis of dealkylation using an HPLC apparatus (Twinkle, Jasco or LC-9A, Shimadzu) equipped with a UV detector. Unisil-NH 2 (Jasco) and TSK gel ODS 80TM (Tosoh) columns (4.6 × 250 mm) were used with several solvent systems at a flow rate of 1 ml/min. For the experiments of 'Dependence of SH compounds and bases on O-demethylation' (Tables I and II), a Unisil-NH 2 column (solvent system: CHC13/MeOH = 3/2 (v/v), and wavelength: 250 nm) was used, and the amount of deoxyguanosine formed was measured. For other experiments, the decrease of O-alkyl compound was measured. Column, solvent system and wavelength employed are summarized in Table III. RESULTS

Demethylation of O6-methyldeoxyguanosine was performed with a large excess of an SH-compound as the alkyl accepter in the presence of an amine. In order to ascertain the most suitable conditions, several SH-compounds were tested for alkyl accepting capacity. The results are shown in Table I. In the presence of Et3N , thiophenol and 2-aminothiophenol were most active. In the absence of EtsN, thiophenol was not active even after 500 h incubation (Table II). This suggests that deprotonated thiophenolate anion (pKa 8.0, nucleophilicity constant 9.92) plays a role in the alkyl transfer. As an alternative catalytic base, imidazole, which is a constituent of His-322 next to the alkyl accepting Cys-321 of AGT, was also examined instead of Et3N. Imidazole (pKa 7.2) ef-

TABLE I DEMETHYLATION RATE OF 06-METHYLDEOXYGUANOSINE IN THE PRESENCE OF DIFFERENT SH-COMPOUNDS AND TRIETHYLAMINE OS-Methyldeoxyguanosine (8.70 x 10 -s mmol), EtsN (8.62 x 10-1mmol) and SH-compound (8.78 x 10 -1 mmol) in 5 ml of MeOH were heated to 60°C as described in the Materials and Methods. Compound

Tim e ~

(h) Thiophenol 2-NH2-Thiophenol 2-HS-Imidazole HSCH2CH20H HSCH2CHs

25 25 382 370 Very slowb

~rime (h) required for 10% formation of deoxyguanosine from O6-methyldeoxyguanosine. bAfter 480-h treatment, only 3% demethylation had occurred.

27 TABLE II EFFECT OF BASE ON DEMETHYLATION RATE Oe-Methyldeoxyguanosine was treated with thiophenol in the presence of the indicated base under the reaction conditions described in the legend of Table I.

Base

pK

%a

Et3N Imidazole

11 7.2

50 3

--

--

0

aFormation (%) of deoxyguanosine from Oe-methyldeoxyguanosine after 122-h treatment.

fected demethylation, but to a much less extent than Et3N (Table II). From these results, a combination of thiophenol and EtsN was adopted for subsequent experiments.

Rate of O-dealkylation The disappearance of O-alkyl nucleosides and bases was measured under the dealkylation conditions employed, and the amount of O-alkyl compound remaining in the reaction mixture was plotted versus the reaction time on a semi-log coordinate. All graphs obtained were linear, indicating that the dealkylation proceeded with pseudo-first order kinetics. Curves for O6-methyldeoxyguanosine and O4-methylthymidine are depicted in Fig. 2A. The hag-life (tl/2) and rate constant (kobs) data for all O-alkyl compounds tested are summarized in Table III. Half-lives of N-9 substituted 06-methylguanine derivatives (1--3) fell in the range 122--147 h; no marked dependence on the N-9 substituent (H, CH 3 and deoxyribose) was found. With O6-methylhypoxanthine (4), lacking a 2-NH 2 group, the reaction proceeded 2--3 times faster than the 06-methylguanines.

100 N

8

50 Ned6

"C "0

0

5 . . . .

0

l

•

50

,

I

m

|

i

I

•

•

100

|

. . . .

150

|

200

Reactlon t l ~ (h)

Fig. 2. (A) Time-course of the demethylation of O6-methyldeoxyguanosine (MedG, • ) and O4-methylthymidine (MeriT, • ) under the thiophenol-Et3N system. (B) Time-course of the demethylation of Oe-methyldeoxyguanosine ( • ) under the selenophenol (SePh)-Et3N system.

122 147 145 51 346 38 37.5 43 29 25 811 ¢ - 2000 c e 13.5 9 646 ¢ 168 56 3 180 f

1 2 3 4 5 6 7 8 9 10 11 12d 13 14 15 16 17 18 19 20 21

5.68 4.71 4.78 13.6 2.00 18.2 18.5 16.1 23.9 27.7 0.854 - 0.346 -51.3 77.0 1.07 4.13 12.4 231 3.85 --

( x 10- 3)

k

3.98 3.96 3.95 4.12 3.78 3.85 3.86 3.80 3.84 3.79 ---3.84 3.86 -3.78 4.07 3.97 -3.60

1H

(ppm)

-56.78

-53.51

53.32 ----

53.47

53.08 53.03 52.98 53.76 52.69

laC

NMR ~ chemical shift

aln DMSO-d~. bColumn A; Unisil-NH 2, column B; T S K gel O D S 80TM. eExtrapolated from the kinetic curve. dBecause of low solubility,the non soluble part of compound 12 was removed. eNo demethylation had occurred after 360-h treatment. rNo demethylation had occurred after 221-h treatment.

tl/2

(h)

Compound

No.

A A A A A A B A B A B A B B B B A B B A A

Columnb

HPLC

MeOH/CHCI,(1/9) CHC13 M e O H / C H C 1 3 ( 1) / 9 MeOH/CHC13(1/19) CHC13/n-hexane (3/2) MeOH/CHC13(1/19) MeOH/H20 (1/4) CHC1Jn-hexane (2/3) MeOH/H20 (3/17) MeOH/CHC13(1/19) MeOH/10 mM NaH2PO4(1/4) MeOH/CHC13(7/93) MeOH/H20 (3/17) MeOH/H20 (3/17) MeOH/H20 (1/9) MeOH/H20 (3/17) CHC13/n-hexane (9/1) MeOH/H20 (1/3) MeOH/H20 (3/17) MeOH MeOH

Solvent

280 280 280 250 270 280 280 270 260 270 280 280 270 260 260 270 270 280 280 280 280

(nm)

UV

HALF LIFE AND RATE CONSTANT DATA FOR O-DEALKYLATION, CHEMICAL SHIFT OF O-METHYLGROUP, AND THE CONDITIONS FOR HPLC

TABLE III

t~ oo

29 04-Methylisocytosine (5), lacking the imidazole moiety of the 06-methylguanine nucleus, gave a 2--3 times slower rate. With the series of 04-methylthymines and -uracil, (6--10), the half-lives fell in a shorter range of 25--43 h than those of the 06-methylguanine series and no remarkable dependence on N-1 (H, CH 3 and deoxyribose) or C-5 (H and CH3) substituents was observed. 2-Oxopyrimidines (6--10) underwent demethylation much faster than the 2-NH 2 derivative (5). O-Ethyl derivatives, O6-ethylguanine (12) and 06-ethyldeoxyguanosine (11) underwent deethylation 6--14 times slower than the corresponding O-methyl derivatives. No deethylation occurred with Oa-ethyluracil (13) even after 360 h of treatment. The dealkylation rates of other related O-alkyl compounds were tested under the same dealkylation conditions. The half-lives of 02-methylthymine (14) and -uracil (15) were 2--3 times shorter than those of the corresponding 04-methyl compounds (9,10). O2-Methyl-4-amino derivative (17) underwent demethylation much slower than the corresponding 4-keto compound (15). The deethylation rate of O2-ethyluracil (16) was 72 times slower than that of the corresponding methyl derivative (15). O6,7-Dimethylguanine (18) underwent demethylation about three times faster than its positional isomer, 06,9-dimethylguanine (2). 8-Methoxyguanosine (19) underwent the fastest demethylation (tl/2, 3 h) among the compounds tested. Benzimidazole, a simple model of this compound, showed a similar half-life (4 h). 8-Ethoxyguanosine (20) underwent deethylation much more slowly than the corresponding methyl derivative (19). No demethylation proceeded for 5-methoxyuridine (21) even after 221 h of treatment.

Correlation of demethylation rate with 1U- and 13C-NMR chemical shifts of the O-methyl group The rate of methyl transfer from O-methyl group to thiophenolate is considered to be primarily inversely related to the electron density of the methyl group. The 1H-NMR chemical shift of the methyl proton is also related to the electron density. The demethylation rates of tested O-methyl compounds were therefore plotted against the chemical shift of their O-methyl protons. As shown inFig. 3A, a moderate linear correlation was observed for the series of 06-methylguanine-related derivatives (1--5 and 18). With the series of 04-methylpyrimidine derivatives (5--10, 14, 15 and 17), a moderate linear correlation was also observed, however, the correlation curve was very steep and quite different from that of the 06-methylguanine-related derivatives. Linear correlation was also evident for the plots of demethylation rates of O6-methylguanine-related derivatives (1--5) against 13C-NMR chemical shifts, as shown in Fig. 3B.

Demethylation by selenophenol Sulfur and selenium both belong to column VIA in the periodic table of elements. Seleno groups are generally more polarizable, hence more nucleophilic than the corresponding sulfur groups. When O6-methyldeoxyguanosine was subjected to demethylation using selenophenol, an accelerated time course resulted as shown in Fig. 2B. The half-life of O6-methyldeoxyguanosine was 27 h (k = 25.7 × 10"3) which was 4.5 times faster than was the case with thiophenol.

30 100

100

(A) 1501

-~ -tlo

1H-NI4R

5C

50

(B)

10

13C.NMR

I I

010

90'' 7 ~ l0,

x 4

0.018

o..

-i5C

,

i

1C

101

I t

8

'I

10~.3.

5

50

a ! too @

%\%

-~100

~,0 ! ,(:93

51

I

2

"Ds "~)s .

1

4.2

i

4.1

I

I

I

n

4.0

3.9

3.8

3.7

Ghemlcal shift (ppe)

500 11

54

i

I

J

53

52

Chemlca] sh11"t (PI~)

Fig. 3. Correlation of demethylation rate (rate constant and half-life)with (A) *H-NMR and (B) 13CN M R chemical shifts of the O-methyl group. Identification number (see Fig. 1) of compounds is shown in figure. CBI 2881 JC

DISCUSSION

Extensive studies on isolation and characterization of AGT from Eschevichia coli have been carried out [1]. However, little is known about the chemicophysical properties of this enzyme's substrates, including O6-methylguanine and O4-methylthymine derivatives. We earlier reported the chemical reactivity [7] and the conformation [13] of O6-methylguanine derivatives. The present investigation of dealkylation rates with a biomimetic alkyl-transfer system revealed clear differences between MedG and MedT as well as methyl and ethyl groups. Analysis of the products formed from O-alkyl nucleic acid components under the treatment conditions of our thiophenol-EtsN system unequivocally revealed corresponding dealkylated compounds in all cases. This result demonstrated that the employed dealkylation system involved an SN2 type nucleophilic substitution of O-methyl bases with a nucleophile, as is effected in the AGT-repair system. This conclusion is further supported by the reaction profiles indicating that (i) triethylamine is required in preference to the less basic imidazole, (ii) more ionizable thiophenols are more reactive than less acidic aliphatic sulfhydryls and (iii) the reaction with selenophenol is faster than with the less nucleophilic thiophenol. The fact that O-ethyl derivatives underwent dealkylation much slower than the corresponding O-methyl derivatives is also in agreement with the substrate-dependence generally found in SN2 substitution processes. Based on the reaction mechanism described, alkyl carbon bonding to the oxygen to be cleaved should be electronically positive to facilitate alkyl transfer. The finding

31 that 1H-NMR chemical shift of O-methyl protons was reasonably correlated with demethylation rate in a series of O6-methylguanines and their related derivatives (1--5 and 18), and in O-methylpyrimidine derivatives (5--10, 14, 15 and 17), is therefore important. The marked difference in the correlation curves of O-methylpurines and -pyrimidines may suggest that some different ratedetermining factors other than the electronic deficiency of the CH 3 carbon concerned, are also playing a role. From synthetic chemistry considerations, alkyl phenyl ethers can generally be cleaved by acid hydrolysis using a mineral acid, acid halide and/or Lewis acid. However, acidic conditions are not applicable to the dealkylation of O-alkyl nucleosides and nucleotides because of the ready cleavage of the involved glycosidic bond. For mild O-dealkylation, the Lewis acid-sulfide system is reported to be effective according to the pull-and-push mechanism [14], but is not applicable because of the complex formation between Lewis acid and hetero atoms other than ethereal oxygen to be cleaved. In fact, treatment of anisole with an A1C18-EtSH system resulted in rapid formation of phenol, while with 2-methoxypyridine, the reaction proceeded very slowly and did not go to completion (data not shown). Finally, a brief comment is necessary on the reported substrate specificity in enzymic AGT-repair of O-methyl lesions. Enzymic in vitro demethylation of O6-methylguanine moieties in DNA is reduced by the preincubation of AGT with a large excess of monomeric O6-methylguanine [15]. This fact indicates that monomeric 06-methylguanine can be a substrate of AGT. The rate of repair, however, is crucially dependent on polymeric structure, since the lesions formed in DNA are repaired much faster than monomeric 06-methylguanine. In addition, the repair rates for 06-methylated lesions are delicately dependent on the species of AGT, i.e., whether the AGT was isolated from bacteria or mammals. With bacterial AGT, 06-methylguanine residues in DNA can be repaired 50--100 times faster than O6-ethylguanine residues, whereas, with rat liver AGT, the former lesion is repaired only 3--4 times faster than the latter [16,17]. In addition, whereas the rat liver AGT is unable to repair 04-methylthymine residues in DNA [18], the bacterial AGT can repair both this lesion and a S diastereoisomer of methyltriester [1,18]. The repair rate of O6-methyldeoxyguanosine residues by bacterial AGT is reported to be much faster than that of O4-methylthymidine residues in DNA [19]. This difference from the present results obtained in our biomimetic chemical system requires explanation. This and further elucidation of substrate specificity and molecular mechanisms involved in the removal of O-alkyl lesions awaits further comparative studies with the presently described model system. ACKNOWLEDGEMENT The authors would like to express their gratitude to Professor Sekiguchi of Kyushu University for his encouragement.

32 REFERENCES 1 T. Lindahl, B. Sedgwick, M. Sekiguchi and Y. Nakabeppu, Regulation and expression of the adaptive response to alkylating agents, Annu. Rev. Biochem., 57 (1989) 133--157. 2 A. Loveless, Possible relevance of 0-6 alkylation of deoxyguanosine to the mutagenicity and carcinogenicity of nitrosamines and nitrosamides, Nature, 223 (1969) 206--207. 3 T.P. Brent, M.E. Dolan, H. Frankel-Conrat, J. Hall, P. Karran, F. Laval, G.P. Margison, R. Montesano, A.E. Pegg, P.M. Potter, B. Singer, J.A. Swenberg and D.B. Yarosh, Repair of o-alkylpyrimidinesin mammalian cells: a present consensus, Proc. Natl. Acad. Sci. U.S.A., 85 (1988) 1759--1762. 4 I. Teo, B. Sedgwick, M.W. Kilpatriek, T.V. McCarthy and T. Lindahl, The intracellular signal for induction of resistance to alkylating agents in E. coli, Cell, 45 (1986) 315--324. 5 B. Demple, B. Sedgwick, P. Robins, N. Totty, M.D. Waterfield and T. Lindahl, Active site and complete sequence of the suicidal methyltransferase that counters alkylation mutagenesis, Proc. Natl. Acad. Sci. U.S.A., 82 (1985) 2688--2692. 6 R.W. Balsiger and J.A. Montgomery, Synthesis of potential aniticancer agents. XXV. Preparation of 6-alkoxy-2-aminopurines, J. Org. Chem., 25 (1960) 1573--1575. 7 K. Kohda, K. Baba and Y. Kawazoe, Chemical reactivity of alkylguanines. I. Methylation of 06-methylguanine derivatives, Tetrahedron Lett., 28 (1987) 6285--6288. 8 P.B. Farmer, A.B. Foster, M. Jarman and J. Tisdale, The alkylation of 2'-deoxygnanosine and thymidine with diazoalkanes, Biochem. J., 135 (1973) 203--213. 9 M. Ikehara and K. Muneyama, Studies of nucleosides and nucleotides. XXX. Syntheses of 8-substituted guanosine derivatives, Chem. Pharm. Bull., 14 (1966) 46--49. 10 J.T. Kusmierek and B. Singer, Reaction of diazomethanes with 1-substituted 2,4-dioxopyrimidines. Formation of 02, N a and 04-alkyl products, Nucleic Acids Res., 3 (1976) 989--1000. 11 W. Szer and D. Shugar, in: W.W. Zorbach and R.S. Tipson (Eds.), Synthetic Procedures in Nucleic Acid Chemistry, John Wiley and Sons, New York, 1968, pp. 58--62. 12 G.E. Hilbert and T.B. Johnson, Researches on pyrimidines. CXIII. An improved method for the synthesis of cytosine, J. Am. Chem. Soc., 52 (1930) 1152--1157. 13 Y. Yamagata, K. Kohda and K. Tomita, Structural studies of O~-methyldeoxyguanosine and related compounds: A promutagenic DNA lesion by methylating carcinogens, Nucleic Acids Res., 16 (1988) 9307--9321. 14 M. Node, K. Nishida, K. Fuji and E. Fujita, Hard acid and soft nucleophile system. 2. Demethylation of methyl ethers of alcohol and phenol with an aluminum halide-thiol system, J. Org. Chem., 45 (1980) 4275--4277. 15 M.E. Dolan, K. Morimoto and A.E. Pegg, Reduction of O6-alkylgnanine-DNA alkyltransferase activity in HeLa cells treated with O6-alkylgnanines, Cancer Res, 45 (1985) 6413--6417. 16 A.E. Pegg, M.E. Dolan, D. Scicchitano and K. Morimoto, Studies of the repair of O6-alkylguanine and O4-alkylthymine in DNA by alkyltransferases from mammalian cells and bacteria, Environ. Health Perspect., 62 (1985) 109--114. 17 A.E. Pegg, D. Scicchitano and M.E. Dolan, Comparison of the rates of repair of O~-alkylgnanines in DNA by rat liver and bacterial O6-alkylguanine-DNA alkyltransferase, Cancer Res., 44 (1984) 3806--3811. 18 M.E. Dolan and A.E. Pegg, Extent of formation of O4-methylthymidine in calf thymus DNA methylated by N-methyl-N-nitrosourea and lack of repair of this product by rat liver O6-alkylguanine-DNA alkyltransferase, Carcinogenesis, 6 (1985) 1611--1614. 19 R.J. Graves, B.F.L. Li and P.F. Swarm, Repair of 06-methylgnanine, O%thylgnanine, D6-isopropylguanine and O4-methylthymine in synthetic oligodeoxynucleotides by Escherichia coli ada gene O~-alkylgnanine-DNA alkyltransferase, Carcinogenesis, 10 (1989) 661--666.