Detection of multiple polycyclic aromatic hydrocarbon-DNA adducts by a high-performance liquid chromatography-32P-postlabeling method.

A 32P-postlabeling procedure for identifying and quantifying hydrophobic DNA adducts was developed (by modifying the method of Randerath and co-workers) in which labeled adducts are separated by high-performance liquid chromatography (HPLC) and quantified by liquid scintillation counting. This method was first developed for fluoranthene-DNA adducts, and methods for optimal detection and quantification of DNA adducts with diol epoxide metabolites of benzo[a]pyrene (BPDE), chrysene (CHDE), and benz[a]anthracene (BADE) have now been established. Analytical conditions slightly different from those adopted for fluoranthene-DNA adducts are required for accurate quantification of BPDE-, CHDE-, and BADE-DNA adducts. In particular, HPLC analysis requires generation of nucleotide 5'-[32P]monophosphate adducts by treatment with nuclease P1, and polycyclic aromatic hydrocarbon adducts demonstrate variable sensitivity to nuclease P1, mediated dephosphorylation. Thus, multiple adducts can be detected in one sample as long as the recovery of adducts under the applied conditions has been determined and chromatographic separation of labeled adducts is achieved. A battery of postlabeling assays can thus make it possible to detect optimally multiple adducts in one DNA sample. Results from these studies indicate that the HPLC- 32P-postlabeling assay is complementary to immunoassays in which related polycyclic aromatic hydrocarbon diol epoxide adducts cross-react for the quantification of adducts.


Introduction
Interspecies and interindividual variability in response to chemical carcinogens may be attributable in part to differences in the dose delivered to the target site. Many carcinogens bind covalently to DNA, and this DNA modification is thought to be a key initial event in mutagenesis and carcinogenesis. Thus, carcinogen-DNA adducts may be biologically important markers of the delivered dose.
Postlabeling assays are useful for quantifying DNA adducts formed in vivo. Recently, a version ofthe "P-postlabeling procedure in which high-performance liquid chromatography (HPLC), instead of thin-layer chromatography, is used for separation of labeled adducts was described (1). This method was used to detect and quantify fluoranthene (FA) adducts in the rat (2). FA, like other polycyclic aromatic hydrocarbons (PAH), is generally produced as a result of incomplete pyrolysis oforganic material, resulting in widespread environmental exposure to complex mixtures of PAH. Therefore, application of the HPLC-32P-posflabeling method in studies for monitoring human exposure to PAH requires chromatographic separation and quantitative recovery of multiple 32P-labeled PAH adducts in one sample. In recent studies, the HPLC 32P-postlabeling procedure has been extended for the detection ofmultiple PAH adducts. We chose to establish detection methods for benzo[a]pyrene (BaP), chrysene (CH) and benz(a)anthracene (BA) diol epoxide (DE) adducts, because these adducts, as well as others, cross-react in a widely applied immunoassay (3)(4)(5). Thus, the postlabeling method described herein would be complementary to that immunosssay.

Methods and Results
The HPLC-32P-Postlabeling Procedure for FA Adducts The HPLC-32P-postlabeling method first developed for quantative analysis of FA adducts (1) is outlined in Figure 1. Briefly, carcinogen-modified DNA is hydrolyzed enzymatically with micrococcal nuclease and spleen phosphodieterase to nucleotide 3 '-monophosphate adducts and nucleotide 3 '-monophosphates. Adducts are separated from unmodified nucleotides by application ofthe hydrolysate to a disposable C18 cartridge. More than 99% ofthe unmodified nucleotides are removed in an aqueous wash, and adducts are eluted with methanol (data not shown).
Complete removal of unmodified nucleotides is essential for reproducible quantification of FADE adducts by postlabeling. Attempts to analyze DNA modified in vitro with diol epoxides of BP, BA, or CH ( Fig. 2) by the HPLC-32P-postlabeling method exactly as previously described (1) i.e., with nuclease P1 pretreatment for 15 min and nuclease P1 post-treatment for 16-18 hr, were unsuccessful; "P-labeled adducts were not detected (data not shown). Because the cornerstone ofthis method is the balance between adduct resistance to briefnuclease P1-mediated dephosphorylation and adduct suceptibility to extensive nuclease P1-mediated dephosphorylation, the sensitivity ofeach of these PAH diol epoxide adducts to nuclease P1-mediated dephosphorylation was investigated.
Chromatographic profiles ofa "P-labeled adducts and the appropriate ultraviolet markers are shown in Figure 3. In each case, a major 32p peak co-eluted with the appropriate authentic nucleotide 5 '-monophosphate adduct. No qualitative difference was observed in any ofthe chromatographic profiles associated with either nuclease P1 pretreatment or the duration of nuclease P1 post-treatment (data not shown).
The effect of nuclease P1 pretreatment on recovery of 32p_ labeled adducts was determined. The improvement in recovery of BPDE, BADE, and CHDE adducts by inclusion of a 10-min nuclease P1 pretreatment step is indicated recovery improved by at least 6-fold as much as 20-fold (BPDE). Analyses for 32p Thus, to eliminate the residual nucleotides as substrates in the labeling reaction, the 3 '-phosphate moiety is selectively removed from unmodified nucleotides by a brief (15-30 min) digestion with nuclease P1 immediately prior to postlabeling. This step only minimally affects recovery of FADE adducts (1).
Transfer ofthe 3"P-label from [_y-32P]ATP to the 5'-hydroxyl position ofadducts is mediated by T4 polynucleotide kinase, producing 3'-[5' -32P]bisphosphates. Reverse-phase HPLC is employed to exploit the hydrophobicity of FAH adducts and thus to separate the multiple adducts that may be found in one sample. Conversion to nucleotide [5 '-32P]monophosophates, which improves reverse-phase HPLC separation of adducts, is accomplished by prolonged treatment, with nuclease P1, to which FADE adducts are sensitive. Finally, adducts are separated by reverse-phase HPLC, and adduct levels are determined by liquid scintillation counting.
Each step in this procedure was optimized previousiy for the detection ofFADE adducts (I); the most important requirement to maximize recovery oflabeled adducts was complete separation ofnormal and modified nucleotides prior to postlabeling. Most ofthe steps inwhich FAadductlossoccurshavebeenidentified (I).
Overall recovery is expected to be 15-35 % ofthe DNA-bound FA at modification levels ofone adduct in 106-107 nucleotides. Since 10-15 % ofthe initial DNA-bound FA is actually recovered, loss ofat most 25 % ofthe DNA-bound FA is unassigned. labeled unmodified nucleotides showed that a 10-min nuclease P1 pretreatment step is indicated in Table 1. Adduct recovery improved by at least 6-fold (BADE) and as much as 20-fold (BPDE). Analyses for 32P-labeled uniodified nucleotides showed that a 10-min nuclease P1 pretreatment is sufficient to dephosphorylate residual unmodified nucleotides (data not shown). Figure 4 shows the dependence of 32P-labeled adduct recovery on the length ofnuclease P1 post-treatment.  (Fig. 5)]. Recovered 32Plabeled BPDE adducts constituted about 10% of the original DNA-bound BaP; thus, approximately 12% ofthe BPDE adducts originally in the DNA were recovered by postlabeling. Losses of BPDE adducts have not been assigned to particular steps in the postlabeling procedure; however, nonspecific losses due to sample manipulation are likely, as in the case ofFADE adducts. Adduct recovery is currently being improved. other detection methods, such as mass spectrometry, which can provide additional structural information. Reverse-phase HPLC has been used for analyzing postlabeled aromatic (9; this work) 500 ; \ as well as alkyl (10,11)  POST-TREATMENT (hrDNA (without intial chromatographic separation) before label-NUCLEASE P1 POST-TREATMENT(hr) ing is applicable for selected adducts (14), whereas butanol extraction is an efficient method for isolating other selected adducts FIGURE 4. Time course for nuclease P1 treatment after 32P-labeling of ben- (15,16). Adduct enrichment by HPLC fractionation is advanzolalpyrene diol epoxide (BPDE)-DNA, chrysene diol epoxide (CHDE)tageous in that the yield ofnucleotides actually in the sample can DNA, and benz[aIanthracene diol epoxide (BADE)-DNA. Each diol be measured and the relative adduct level in the sample can be epoxide-modified DNA (I pg) was mixed with calfthymus DNA (14 ,sg) and postlabeled with a 10-min nuclease PI pretreatment as described (1). The calculated (17)(18)(19)(20). Immunoaffinity chromatography may be mean and range for duplicate samples are indicated.
useful for isolating adducts, although its application is limited by solvent incompatibilities (21) and antibody availability. The Discussion results described above show the value of combining ofmethods for complete elimination of unmodified nucleotides.
Many versions of the 3"P-postlabeling assay have been BPDE, BADE, and CHDE adducts are significantly more sendeveloped to detect adducts of known and unknown structure; sitive to dephosphorylation by nuclease P1 than are FADE admost of these involve thin-layer chromatography for adduct ducts. Similar differences in sensitivity to nuclease P1 has been separation (Z,8). The time-consuming, labor-intensive nature of observed with other classes of adducts, e.g., arylaminethin-layer chromatography, and the fact that quantitative C8-guanine adducts (16). It is notable that adducts of the same estimates ofadduct recovery after this procedure have not been chemical and nucleotide class, i.e., PAH diol epoxide-N2determined rigorously, limits its application in molecular guanine adducts, vary widely in their sensitivity to nuclease P1. epidemiology studies.
These results underscore the fact that the balance between adduct Coupling postlabeling with HPLC offers technical advantages resistance to brief nuclease P1 pretreatment and prolonged 210 I co F.. A battery of postlabeling assays may make it possible to detect optimally multiple adducts in one DNA sample. Alternatively, determination of the recovery of multiple adducts under particular conditions will enable quantitative analysis, albeit sometimes with less than optimal detection. Application ofthe HPLC-32P-postlabeling method to analysis of adducts in human tissues requires the ability to detect and quantify simultaneously multiple adducts, often from the same chemical class. Thus, we undertook to optimize this sensitive method for the detection of multiple PAH adducts as well as to determine the quantitative recovery of each adduct in this procedure. Our results indicate that documentation of the quantitative power ofthis assay is essential for accurate interpretation of adduct levels determined in human samples.