Study of DNA methylation by tobacco-specific N-nitrosamines.

An enzyme-linked immunosorbent assay (BA-ELISA) involving use of biotin-labeled anti-rabbit IgG and avidin-labeled horseradish peroxidase was developed for the measurement of O6-methyl-2'-deoxyguanosine (O6-MedGuo). Up to 5 micrograms of methylated DNA was enzymatically hydrolyzed, and the extent of inhibition of binding of immobilized O6-MedGuo-bovine serum albumin to rabbit anti-O6-MedGuo was measured. Fifty percent inhibition of antigen-antibody binding was achieved with 2.5 pmole of of O6-MedGuo. Separation of O6-MedGuo from unmodified nucleosides by high-performance liquid chromatography (HPLC-BA-ELISA) allowed detection of 700 fmole O6-MedGuo in 1 mg of DNA. Among the tobacco-related carcinogens, 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanone (NNK) is one of the most potent. In F344 rats it induces nasal cavity, lung and liver tumors. Four hours after a single IV injection of NNK to F344 rats (87 mg/kg body weight), O6-MedGuo was present in target organs (mumole O6-MedGuo/mole dGuo) (nasal mucosa, 219; lung, 13.2; and liver, 34.5) but was not detectable in nontarget organs. F344 rats receiving daily IP injections of NNK (40 mg/kg body weight) for 14 days were sacrificed 24 hr after the last injection. The levels of (O6-Medguo/dGuo) were 7.9 and 11.4 mumole/mole in the nasal mucosa and lung, respectively. In the liver no O6-MedGuo was detected, but 1050 mumole of 7-MeGua/mole Gua was measured by HPLC-fluorimetry. No DNA methylation was observed in the nasal mucosa or liver of F344 rats treated with the nicotine-derived carcinogen N'-nitrosonornicotine.(ABSTRACT TRUNCATED AT 250 WORDS)


Introduction
The high incidence of lung cancer among smokers is well documented (1). Studies carried out during the last 10 years have shown that N-nitrosamines present in tobacco smoke are relatively potent carcinogens in animals. The same studies have suggested that N-nitrosamines are important etiological factors in tobacco smoking-related cancer (2). Among the 13 N-nitrosamines detected in cigarette smoke, the tobacco-specific N-nitrosamines are among the most abundant (3). Prospective and retrospective studies have demonstrated a link between snuff exposure and development of oral cancer (4,5). At present, tobacco-specific N-nitrosamines are the only carcinogens to have been isolated in significant amounts from snuff (6). The levels of ab-*Naylor Dana Institute for Disease Prevention, American Health Foundation, Valhalla, NY 10595. " sorption of tobacco components by respiratory tract tissues vary among smokers and intrinsic exposure to Nnitrosamines during tobacco smoking is difficult to assess. The tobacco-specific N-nitrosamines: N'-nitrosonornicotine (NNN) and 4-(methylnitrosamino)-1-(3pyridyl)-1-butanone (NNK) are derived from nicotine and are metabolized rapidly in vivo to metabolites identical to those of nicotine (2). Consequently, an assessment of exposure to nicotine-derived N-nitrosamines by quantification of their metabolites is not feasible.
However, alkylating species are generated only from NNN and NNK, and levels of their persistent adducts in DNA and proteins could be used as an index of human exposure. Alternatively, assessment of repair of promutagenic DNA damage would measure the ability of an individual to respond to carcinogen insults. The objective of the present study was to develop a safe, inexpensive and sensitive method to measure the levels of 06-methyl-2'-deoxyguanosine (06-MedGuo) in DNA damaged by activated NNK.
Culture ofRat Nasal Mucosa. Removal of the nasal septum and its covering mucosa from the rat nasal cavity and the culture technique of the septum have previously been described (8). Each septum was cultured in a 35-mm culture dish (Falcon, Bethesda, MD) with 2 mL of medium containing either 100 jig of NNK or 101 ,ug of NNAI/mL. After 1, 3, or 24 hr of culture, the septa were harvested, and DNA was extracted from three combined explants. The medium was harvested and stored frozen. After filtration of the media, levels of NNK and NNAI were measured by HPLC on an octodecylsilane-bonded phase column (4.6 mm x 12.5 cm) as described elsewhere (9).
Rat Treatment with NNK and NNN. Five F344 rats each weighing approximately 280 g at the start of the experiment were fed an NIH-07 diet (Ziegler Brothers, Inc., Gardners, PA). They were injected IV with a solution of NNK in 0.9% NaClIH2O (87 mg/kg body weight) and decapitated 4 hr later. Livers, lungs, kidneys, esophagi, spleens, and hearts were excised and immediately frozen over dry ice. The nasal cavity was opened as described previously (8), and the mucosa covering the nasal septa and ethmoturbinates of the five rats was scraped off with a scalpel, pooled, and the DNA extracted as described previously (10).
Two F344 rats each weighing about 248 g received each daily IP injections of 10 mg of NNN dissolved in 0.9% NaCl/H20 (20 mg/mL). Each rat received 14 injections. Two other rats each received 10 mg of NNK dissolved in 0.9% NaCI/H2O (20 mg/mL). The four rats were exsanguinated by heart puncture 24 hr after the last injection and the tissues mentioned above were excised and stored frozen until DNA extraction.
DNA samples were dissolved in 50 mM pH 7.2 Tris HCI and DNA concentration was determined by fluorimetry (11). DNA was hydrolyzed according to the method of Muller and Rajewsky (12).
Antibody Production and Purification 06-Methylguanosine (06-MeGuo) was conjugated to bovine serum albumin (BSA) (Sigma, St. Louis, MO) or keyhole limpet hemocyanin (KLH) (Calbiochem, La-Jolla, CA) according to the method of Erlanger and Beiser (13). New Zealand White rabbits were immunized with an emulsion of the conjugate in complete Freuds adjuvant (Calbiochem). The inoculum was injected intradermally at approximately 20 sites on the back of each rabbit. Four booster injections were given at 8-week intervals. The 2-week bleeding after the last booster injection was used in the present study. The anti-O6-MedGuo antibodies were purified by affinity chromatography as previously described for the anti-06-ethyl-2'-deoxyguanosine (12).
BA-ELISA. The biotin-avidin enzyme-linked immunosorbent assay (BA-ELISA) was performed entirely at room temperature in 96-well polystyrene microtiter plates (Costar, Cambridge, MA). Each experimental value was determined in triplicate. The amount of 06-MedGuo in a sample was determined by the use of two standard curves. The first set of standards was used to measure the extent of inhibition of binding of the primary antibody. The primary antibody was diluted 1:1000. Further dilution of 4:5, 3:5 and 2:5 (corresponding to 20, 40 and 60% inhibition) served to construct the standard curve. The second curve consisted of five 6-fold dilutions of 06-MedGuo ranging between 0.1 and 25 pmole per well and was used to equate a given amount of 06-MedGuo with a given degree of inhibition.
All wells were coated with 100 ,uL of 06-MeGuo-BSA at a concentration of 100 ng/mL in 50 mM Tris-HCl buffer, pH 7.2, for 1 hr. After the removal of unbound 06-MeGuo-BSA, unoccupied protein binding sites were saturated by the addition of 100 ,LL of a 1% (w/v) solution of BSA in phosphate-buffered saline (PBS) for 30 min. Wells were then emptied and washed. The standard washing procedure was to rinse twice with PBS-Tween (Sigma) (0.5 mL Tween 20/L PBS) and four times with distilled water.
To set up the standard curves and experimental points on the plate the following procedure was followed. All reagents were diluted in PBS containing 0.1% (w/v) BSA. 06-MedGuo standards or experimental samples as 50 ,uL aliquots were added to the appropriate wells. Wells used for the primary antibody standards received 50 ,uL of diluent. Then wells containing 06-MedGuo standards or experimental samples received 50 ,uL of primary antibody (1:1000 dilution) and wells for the primary antibody standards received 50 pL of the appropriate antibody dilution. After incubation for 1 hr the wells were emptied and washed with PBS as above.
HPLC-BA-ELISA. The HPLC apparatus has been described previously (14). Separation of deoxyribonucleosides obtained by DNA hydrolysis was performed with a pu-Bondapak-C15 column (4.9 mm x 30 cm) (Waters Associated, Milford, MA) and with a gradient of 0 to 35% aqueous methanol in 50 min. The flow rate was 1 mL/min, and 1-mL fractions were collected. The fractions were then lyophilyzed and reconstituted with 0.5 mL of PBS. Individual fractions were then analyzed by BA-ELISA as described above. The recovery of O-MedGuo from HPLC vary between 70 and 100%. HPLCfluorimetry was carried out as described by Herron and Shank (15). Figure 1 demonstrates the specificity and sensitivity of the antibody used in the BA-ELISA for O6-MedGuo. Fifty percent inhibition of antibody binding to immobilized 06-MedGuo-BSA was achieved with 2.5 pmole of O6-MedGuo. If a probability grid was used to plot the inhibition curves, inhibition was linear between 15 and 85%. Fifteen percent inhibition is equivalent to 100 fmole of O6-MedGuo per well. Since, of the compounds tested, those containing a methyl group attached to the O6 position of guanine gave the highest levels of inhibition, the specificity of the antibody was clearly directed towards this group. N6-Methyl-2'-deoxyadenosine (N6-MedAdo) was at least 1000-fold less efficient than O6-MedGuo in inhibiting antigen-antibody binding. This indicates that the antibody may recognize the amino group at the 2 position of the O6-MedGuo. The poor inhibition of bindinf of 7-methyl-2'-deoxyguanosine (7-MedGuo) versus 0 -MedGuo and of 3-methylguanine versus O6-MeGua emphasizes the importance of the methyl group being attached to the O6 position of the guanine moiety. All five nucleosides naturally occurring in DNA were found to inhibit binding only at concentrations 100,000fold higher than those required for O6-MedGuo binding.

Results
The maximum amount of hydrolyzed DNA that could be used per well was 5 ,ug. This maximum limit sets the lower limit of methylation which can be detected for 20 pmole O6-MedGuo/mg DNA. At 25 pg of DNA per cell this level of methylation corresponds to 316,000 O6-MedGuo residues per diploid genome (16). To circumvent this limitation, reverse-phase HPLC was used to separate O6-MedGuo from unmodified and other methylated nucleosides. As shown in Figure 2 this technique was used to measure the level of O6-MedGuo in DNA isolated from the nasal mucosa of NNK treated rats.
Only those fractions having an elution volume equal to that of 06-MedGuo gave substantial inhibition (64%) of antibody binding in the BA-ELISA. By coupling HPLC with the BA-ELISA the amount of DNA analyzed can be increased from 5 ,ug to 1 mg or more. Following this procedure a 1 mg sample containing as little as 700 fmole, which corresponds to 10,500 residues of 06_ MedGuo per diploid genome, can be analyzed in duplicate (six wells). At a concentration of 100,000 pmole per well, 2'-deoxyadenosine gave 39% inhibition, dGuo, adenine, and thymidine gave less than 20% inhibition and 2'-deoxycytosine, 5-methyl-2'-deoxycytosine, 3-methylguanine, and 7-methyladenine gave less than 10% inhibition.
iU S1 IIU N 0 S1 S4 R x e cs cx VOLUME (ml) The metabolic pathways of NNK are illustrated in Figure 3. Carbonyl reduction of NNK gives NNA1. As with most N-nitrosamines, hydroxylation of the carbon a to the N-nitroso group of NNK and NNAI could be considered a crucial and rate-limiting step in their bioactivation to alkylating species. For instance, a-methylene hydroxylation of NNK and NNAI would lead to the generation of methyldiazohydroxide (7). The keto acid and hydroxy acid (compounds 10 and 11 of Fig. 3) are two major urinary metabolites of F344 rats treated with NNK (17).
The methylating capacities of NNK and NNAI were compared by culturing rat nasal mucosa with these two N-nitrosamines. As shown in Table 1, NNK was rapidly reduced to NNAI in this system. Only 14% of the initial amount of NNK remained at the end of the 24-hr culture period. In contrast, NNA1 was not extensively metabolized, and 77% of the initial amount remained unchanged after 24 hr. aThe initial concentration of NNK or NNA1 was 482 nmole/mL. Values are mean ± SE obtained from four cultured media. bND = not detected. Limit of detection is 1.2 nmole/mL. After 24 hr of culture, the total level of NNK + NNAI was 107 nmole/mL and 373 nmole/mL when NNK and NNA1 were used as starting N-nitrosamines, respectively. The extent of DNA methylation was higher when NNK rather than NNA1 was added to the culture medium ( Table 2). With NNA1, DNA methylation after 24 hr was associated with the oxidation of a small fraction of NNA1 to NNK. F344 rats were either treated with a single dose of NNK and sacrificed 4 hr later or were administered 14 daily doses of NNK and sacrificed 24 hr later. As shown in Table 3, 06-MedGuo was found only in nasal mucosa, lung and liver.

Discussion
The BA-E LISA described in this report is a sensitive, rapid, and simple method for the detection of 06-MedGuo in DNA modified by methylating agents. The use of biotin-avidin reagents reduces nonspecific binding of the second antibody and results in lower background readings. The BA-ELISA can easily be performed in one day and does not require the use of radiolabeled tracers. The sensitivity of BA-ELISA was increased considerably by chromatographic separation of methylated and naturally occurring nucleosides obtained by enzyme hydrolysis.
The HPLC-BA-ELISA is more time-consuming than the BA-E LISA, taking 2 days to complete, but is limited in the amount of DNA which can be analyzed only by the amount of hydrolyzed DNA which can be loaded on the HPLC column. In HPLC-fluorimetry it is critical to achieve good separation of the modified nucleobases from other components that could also fluoresce at the particular excitation and emission wavelengths used. Good separation of the nucleoside mixture is not as critical in HPLC-BA-ELISA because of the high specificity of the antibody for the carcinogen modified nucleoside.
As shown in Figure 3, reduction of NNK to NNAI is a major metabolic pathway which was observed in vivo, in organ cultures and during incubation of NNK with rat liver microsomal fraction (3). Interestingly, this reduction was also observed in mouse tissue early during fetal life (18). In human tissues excised from the oral cavity and upper respiratory tract and cultured in vitro, reduction of NNK to NNAI was also the major metabolic pathway. a-Carbon hydroxylation of NNK by cultured human tissues was also observed, but was not as extensive as in cultured animal tissue (19).
The reoxidation of NNAI to NNK was observed with explants of A/J mouse lung (a target tissue of NNK and NNAI), although the equilibrium favored NNAI (20). The ability of NNK and NNAI to generate methylating species was compared in cultured rat nasal mucosa (Table 1). Levels of 06-MeGua and 7-MeGua were higher with NNK than with NNA1, suggesting that NNK is a better substrate for the activating enzymes. This hypothesis was supported by a study of the interconversion of NNK and NNAI by these explants. NNK was efficiently reduced to NNAI but the reverse reaction was not favored. Furthermore, the low level of oxidation of NNAI to NNK paralled a low level of DNA methylation observed with NNAI after 24 hr of culture. These results suggest that NNAI is not associated with the activation of NNK to DNA methylating species.
NNK, along with NNN and N'-nitrosoanatabine, are formed by N'-nitrosation of either nicotine or anatabine during curing and/or smoking of tobacco (21,22). All three N'-nitrosamines induce tumors in F344 rats, but NNK is the most potent. Even at a dose of 1 mmole/kg body weight NNK injected SC to male F344 rats induces a high percentage ofnasal tumors (74%) and lung tumors (85%) but a low percentage of liver tumors (11%). In contrast, NNN induces a lower percentage of nasal tumors (56%), a low percentage of lung tumors (14%), and no liver tumors (23).
After treatment of F344 rats with NNK, methylation of the DNA at the 06-Gua site was observed only in organs that develop tumors (Table 3). In NNN-treated rats no 06-MeGua was observed in nasal mucosa or liver DNA. These results suggest that the relatively higher carcinogenic potency of NNK compared to NNN could be due to the formation of the promutagenic lesion O6_ MeGua by NNK. However, the formation of other adducts could also mediate the carcinogenicity of NNK. A recent mutagenicity study of compounds analogous to NNK supports this hypothesis (24).
As shown in Figure 3, methyl hydroxylation of NNK would lead to the pyridyloxobutyldiazohydroxide (Compound 5). According to a whole-body autoradiographic study of rats treated with NNK, alkylation by compound 5 would take place in the mucosa of the ethmoturbinates, lateral nasal gland, bronchial mucosa, and liver (25). Whether compound 5 (Fig. 3) leads to persistent and mispairing adducts is currently being studied with monospecific antibodies and BA-ELISA.
Methodology developed in the present study will certainly be instrumental in assessing damage to human DNA by NNK and other methylating substances present in tobacco smoke and chewing tobacco.