DNA adducts and mutations in occupational and environmental biomonitoring.

The methods applied for DNA adduct determination in humans have become more reliable. Yet there is a need to characterize the adducts studied better and when possible, to identify them with the help of the available standard compounds. Use of standard compounds also allows quantification of adduct levels. There is a lack of knowledge on the adduct levels and their half-lives in target and surrogate tissues. Most adduct studies have been carried out on occupational populations exposed to complex mixtures. White blood cells have been the most common source of DNA. Other exposures and tissues should be a subject of study. Notably, dietary exposures have been largely neglected. Biomonitoring of mutations is a relatively new field and a few exposures have so far been investigated. The results have been promising but logistics of the studies have to be improved to make large field studies possible. Future biomonitoring studies should make an effort to combine many end points, with emphasis on adducts, mutations, and constitutional metabolic factors.


Introduction
Application of DNA adducts and mutations in human biomonitoring is recent and the related technologies are continuously improving. Since the 1980s when DNA adduct studies became possible in humans (1,2) the question about their significance has been asked. There is no direct answer to the question but much of circumstantial evidence points to their mechanistic role in cancer (3). Yet direct evidence is lacking and it would be simplistic to assume that one measurement in time would tell much about the risk. However, if a particular exposure has lasted for an extended time, or if it has been excessive, such as anticancer chemotherapy or an accident involving exposure to radioactive material, some increase in risk may be predicted. Some, but by no means all, of the reasons DNA adducts are important are the following: * Adducts cause mutations: e.g., construction of a specific mutation in a phage or virus leads to a specific mutation at a site of adduct; DNA repair defects. * Mutations in viral and cellular oncogenes, tumor suppressor genes; transgenic animals can cause cancer. * Adducts, cross-links in particular, are cytotoxic and thus mitogenic; cell proliferation may contribute to cancer. * Adducts show individual exposure, metabolic capacity, and DNA repair capacity. * Adducts indicate present DNA damage (cf. epidemiology). * Adduct-forming chemicals-e.g., many anticancer agents-are potentially dangerous. * Adducts can show active ingredients in complex mixtures. Although first emphasized in the "initiation" phase of carcinogenesis, the recent evidence of genetic lesions in multiple steps of cancer development, suggests that DNA damage plays a role in many stages of oncogenesis. Among the available methods, only 32P-postlabeling is discussed here because of its wide application in the occupational and environmental studies.
The role of mutations has not been questioned to the same extent as that of DNA adducts because mutations play an important role in the development of cancer (4). However, the appearance of mutations in surrogate tissues of healthy individuals may not be directly informative of the events in target tissues. Additionally, all the mutational systems available for human biomonitoring have their own features and limitations (5). Only one mutational system, based on hypoxanthine-guanine phosphoribosyl transferase (HPRT), is discussed here.

Development of the 32P-Postlabeling Method
The 32P-postlabeling method, introduced about 15 years ago, made it possible for the first time to analyze DNA-adducts existing in DNA in vivo. The method has been used extensively to compare adduct patterns in various tissues and in various exposures. Most studies have focused on unidentified aromatic adducts because the original technique selects for these types of adducts (6). More recently, standard compounds have been used in the identification and quantification of adducts (2). Although this is widely accepted now, it was long thought by a large section of the postlabeling community that labeling of all adducts was complete. By now it has been demonstrated with tens of different synthetic postlabeling standards that, depending on the adducts and conditions of labeling, the recoveries vary between 0 and 100%. Even diastereomers can label differently. In an illustrative experiment, DNA adducts of a number of 3H-labeled polycyclic aromatic hydrocarbons (PAH) were prepared in a microsomal system and used for optimization and measurement of recoveries in the postlabeling assay. The optimal labeling conditions for all tested compounds were very similar. The recoveries varied from 3 to 60% among different PAHs, indicating that the levels of these adducts could be considerably Environmental Health Perspectives * Vol 105, Supplement 4 * June 1997 ' I ' l . . underestimated in analyses of human samples from PAH-exposed populations. Because we have found similar results with an entirely different group of compounds, it generally can be concluded that different adducts require different conditions for optimal labeling (7). Thus the absence of proper standards, or analysis of unknown adducts, impedes quantitative interpretation of the postlabeling results. The most important parameters for optimal postlabeling are that the adduct must be known; the standard must be synthesized and its stability tested; the standard's labeling efficiency must be tested; and the total recovery of the tested DNA adduct must be determined. Below some illustrative examples of the application of the postlabeling method are given. Extensive surveys of the literature are available (1,6).

Surrogate versus Target Tissues
In biomonitoring of almost any end point, surrogate rather than target tissues have to be used. The information on the applicability of surrogate tissues in humans is scanty. In some animal experiments the question of the correlation of adducts in surrogate and target tissue has been addressed [(8) and references therein]. Rats were exposed by inhalation to individual alkenes from ethene to octene; DNA adducts in liver and lymphocytes and hemoglobin adducts were measured ( Figure 1). For all these adducts the levels decreased from ethene to octene. However, the decrease was 5-fold in liver, 30-fold in lymphocytes, and 2000-fold in hemoglobin. This was interpreted as indicating a complex interplay Ge Cs r Figure 1. Liver and lymphocyte DNA (adducts/107 normal nucleotides) and hemoglobin N-(2-hydroxy)valine adducts (x 500 pmol/g) of rats exposed to 300 ppm of alkenes from ethene (C2) to octene (C8) for 12 hr on 3 consecutive days (8). DNA react., reactivity of alkene epoxides with DNA in vitro; Hb, hemoglobin. of tissue uptake, metabolism, and diffusion out of organs.
In humans, smoking has been the main model of exposure used. Smoking is a known risk factor of laryngeal cancer. Aromatic adducts of laryngeal tissue obtained from surgery patients were analyzed and a relationship to smoking was found. Both tumor and normal laryngeal tissues showed a correlation of about 0.9 to the total white blood cells (9).
Tobacco smoke contains methylating and hydroxyethylating principles, originating, for example, from tobacco-specific nitrosamines and ethene, respectively. The levels of 7-methylguanine were highest in the bronchial DNA of smokers, exceeding the level in nonsmokers by almost 4 times (10). In a small number of smokers, both target (bronchial) and surrogate (lymphocyte) DNA were available, showing a correlation of 0.8. [We now know that the 7-methylguanine adduct spot also contains 7-(2-hydroxyethyl)guanine adducts, which, however, were poorly labeled and contributed little to the radioactivity of the spot (11)]. Larynx tissue samples obtained from surgery patients were also assayed for 7-methylguanine DNA adducts. There was a relationship to smoking, and larynx adduct levels exceeded those of white blood cells by twice (12). There was a modest correlation only between 7-alkylguanines and aromatic adducts (above). The smokers had higher aromatic adduct levels in lymphocytes than in granulocytes, which indicated that in chronic exposure the main focus should be on lymphocytes (13). Smokers also had elevated levels of 7-methylguanine, particularly in their lymphocytes as compared to the granulocyte DNA (14). However, taking into consideration the very different half-lives of lymphocytes and granulocytes, the isolation of cells only clears away the "noise" caused by granulocytes and improves the precision, as only about 25% of the DNA in total white blood cells is from lymphocytes. This should be an important principle applied in biomonitoring studies.

Aromatic Adducts
In spite of problems in interpretation of postlabeling results of complex mixtures, most published literature concerns exposures in which PAHs are of primary concern. Many of the groups studied have been at risk of cancer according to epidemiological reports that reflect exposures a few decades ago. The main questions posed have been: a) do the exposed groups show higher, work-related adduct levels than the controls; b) is there a correlation between exposure measures (air concentration or urinary 1-hydroxypyrene) and adducts; and C) how large are the individual variations and metabolic genotypes on the level of adducts?

Occupational Populations
The study populations have included foundry and coke workers, aluminum and electrode workers, and chimney sweeps. Additionally, some other occupational groups have been studied as reference groups in environmental studies, discussed below. Examples are given on some collaborative studies in which our laboratory has participated. The foundry study has involved blood and urine sampling of the same individuals each December for four years. The last sampling was done in December 1993. It is a multi-end point study, including some 15 parameters. Only results from the first 2 years on certain outcomes have been published but an intense compilation of the total material is underway. The first published papers showed elevated total white blood cell DNA adduct levels as measured by immunoassay (15) and postlabeling (16,17), relating to exposure. Among the other occupational groups, coke workers had higher levels of aromatic adducts than the local controls (18,19). Somewhat elevated but not statistically significant differences were seen in electrode and aluminum workers even though air concentrations of PAHs and urinary 1-hydroxypyrene levels indicated excessive exposure (20,21). Adduct levels were also slightly increased in total white blood cell DNA of chimney sweeps. However, the difference to a control group became significant only after adjustment for the CYPIAJ and glutathione S-transferase (GSTMI) genotype (22). Yet each of the genotypes alone had a rather small effect on DNA adduct levels. In all the studies cited, the interindividual variation in the levels of adducts has been large, over 10-fold. The variation is usually larger in the exposed than in the control populations, suggesting that exposures as well as constitutional factors contribute to such a variation.

Environmentaily Exposed Populations
Two series of environmental studies have been conducted, one in Poland and the other mainly in Sweden. The Polish study was initiated several years ago in response to the alarming reports of environmental Environmental Health Perspectives * Vol 105, Supplement 4 -June 1997 pollution in Silesia, a heavily industrialized area. The first study on the Silesian population showed an elevated level of adducts, by postlabeling and immunoassay, in the total white blood cells of the residents (23). This was followed by reports of seasonal differences in adduct levels, which matched the air concentrations of PAHs (18,24). The effects were mainly seen in DNA of the long-lived lymphocytes, while granulocytes showed no clear effect. Sampling in both summer and winter allowed a rough estimation of the half-lives of aromatic adducts in lymphocytes of 1 to 2 months (24). Also, cytogenetic damage was seen in the Silesian population (25).
The nature of the adducts detected by postlabeling has been studied in more detail by comparing nuclease P1, butanol extraction, and immunoaffinity purification of the adducts. Adduct recovery was approximately equal by the P1 and butanol techniques, suggesting that the adducts are of PAH-type. For immunoaffinity chromatography an antibody raised against benzo[a]pyrene diol epoxide-DNA was used. Only about 25% of the adducts were bound by the antibody, indicating that most of the adducts in DNA are not dosely related to benzo[a]pyrene. However, in winter, a time of high air pollution, the relative binding by the immunoaffinity column was higher than in the summer (19). In HPLC analysis using flow-through radioactivity detectors, typical seasonal adduct peaks were noted and they were particularly prominent in lymphocyte DNA collected in the winter. They eluted in the area of PAH-DNA adducts, giving additional support to the conclusion that the adducts are PAH-like ( Figure 2) (26).
A study of bus drivers from central Stockholm and from the city's outskirts compared these subjects to a nonoccupational control group of fine mechanics. All participants were nonsmokers. Aromatic DNA adducts in lymphocytes, PAH adducts in albumin and ethene, and propene adducts in hemoglobin were not elevated in the urban bus drivers (27). A similar type of study was conducted in Milan, Italy. The study subjects were newspaper vendors from busy streets and from the outskirts of Milan. The levels of DNA adducts did not differ in these populations. Some occupational groups, exposed to car and diesel exhaust, were positive controls in the above study. They included garage workers who overhauled diesel buses and inhaled diesel exhaust gases, car mechanics exposed to spilled engine oils C',3 but not to exhaust, and truck workers, unloading and rtloadi trucks. All these groups had i levels of lymphocyte DNA add highest levels being correlated concentrations of diesel exhausts. mated air benzo[a]pyrene leN below 10 ng/m3 (28). GSTM acetyltransferase (NAT2) genot) determined in the study subjects viduals with a combined genotyF acetylation but lacking the GST the adduct levels were sign increased (29). Neither genoty had an effect on the level of addu Styrene Genotoxicity of styrene has been worldwide because it is one of th pected mutagenic compounds cause daily exposures in gram c (30). Styrene is an example of chemical characterization, fasci itself, leads to production of compounds for postlabeling (31, quantification of 06-guanine a white blood cell DNA of lan workers (32). Further studies b carried out to sample the same tion workers periodically in orde sure lymphocyte and granulocy Again it was shown that adducts tially only found in lymphoc) repair of 06-adducts appeared tc since no essential decrease in add was noted after workers took vacation (33).

Ki
The adduct studies suggested that the xnl factory controls, in fact, are not completely unexposed. Strand breaks, measured by the Comet assay, were also increased in the lamination workers. There was a correlation between strand breaks and 06-guanine adducts, but neither correlated with HPRT mutant frequency (34). In vitro data on the 13 effect of styrene oxide (the main metabolite of styrene) on cultured human lymphocytes confirmed the relatively long half-1415 lives of 06-guanine DNA adducts and the 16 induction of strand breaks (35).

80 HPRTMutations
Mutations in the HPRTgene in human lymphocytes have been studied extensively, al nonsmokbut only a few occupational studies and, as C DNA from far as we know, no environmental studies assayed and on chemical exposures have been carried ndard DNA out (5). The nonoccupational studies include those on smoking-, radiation-, chemotherapy-, and disease-induced mutation rates. The limited number of occupaterminal tional studies partially reflects logistic ing diesel problems because cell separation from increased blood has to be carried out within hours of lucts, the blood collection, and living cells have to be with air delivered to the analyzing laboratory. The esti-Moreover, large interindividual differences vels were and dependence of the mutation rates on I and Nage and smoking may discourage attempts ypes were to distinguish small differences between i. In indi-the exposed and control populations. )e of slow Among the occupational groups studied, Ml gene, workers producing the anticancer agent ificantly cyclophosphamide have elevated levels ype alone of lymphocyte HPRT mutations (36).

Lcts.
Exposures to ethylene oxide (37) and styrene/dichloromethane have also caused increases in mutation frequency (38). In of interest our studies on lamination workers (above) e few sus-the HPRT mutation frequency was elethat may vated in workers exposed to styrene but the quantities increase reached statistical significance only 'how the when compared to an external rather than Lnating in an in-house, factory control group (34). standard Induction of HPRTmutations in cultured ) and to a human lymphocytes exposed to styrene dducts in oxide was considered weak (35). nination We have also measured mutant freiave been quencies in occupational populations e laminaexposed to PAHs. In the foundry study, .r to mea-HPRTcorrelated with exposure and adduct rte DNA. levels, while glycophorin A NO mutations are essen-had a moderate, but statistically not signifiytes. The cant trend with exposure (16,17). The be slow, HPRTmutant frequency was not increased [uct levels in garage workers, but at an individual level a 2-week there was a highly significant correlation between adducts and mutant frequency Adducts, x10 Figure 3. Aromatic DNA adduct levels of workers exosed to diesel exhaust (28) as related to HPRTmutation frequency in peripheral lymphocytes (29). The correlation for the total study population, as shown in the figure, or for the exposed workers was approximately 0.35. MF, mutant frequency.
(r-0.35), shown in Figure 3. GSTMI and NA T2 genotypes were determined in the study subjects. The genotypes, alone or combined, had no effect on HPRTmutant frequency (29).

Conclusions
Biomonitoring by means of DNA adducts or mutations must meet requirements of technical validity and unbiased selection and comparison of the study populations. Although individual studies have demonstrated technical reproducibility, a limited number of interlaboratory studies have been conducted (1,5). The study populations have often been small and selected by easy availability rather than by comparability. The elementary requirement of assaying coded samples has frequently been neglected.
DNA adduct studies in humans are becoming more quantitative and therefore offer chances for quantitative risk estimation. Since epidemiological studies always show the risks of exposure decades back, DNA adduct studies can be used for current risk estimation. DNA adduct studies are likely to give clues to individual risks and may therefore be useful in protecting sensitive populations. Assays of point mutations, such as those in the HPRTlocus have been used to a limited extent in biomonitoring of chemical exposure. The studies published so far are promising. Mutations as compared to DNA adducts are mechanistically closer to the cancer end point and would provide a valuable addition to risk assessment of chemical exposures.