Identification of polycyclic aromatic hydrocarbon metabolites and DNA

Matthew J. Sanders,1 R. Scott Cooper, Ryszard Jankowiak, and Gerald J. Small* ... modifications of DNA (Intact adducts) formed from dlol ep- oxides of...
0 downloads 0 Views 626KB Size
816

Anal. Chem. 1986, 58, 816-820

Identification of Polycyclic Aromatic Hydrocarbon Metabolites and DNA Adducts in Mixtures Using Fluorescence Line Narrowing Spectrometry Matthew J. Sanders,' R. Scott Cooper, Ryszard Jankowiak, and Gerald J. Small* Ames Laboratory-USDOE

and Department of Chemistry, Iowa State University, Ames, Iowa 50011

Volker Heisig' and Alan M. Jeffrey Cancer Centerllnstitute for Cancer Research, Columbia University, 701 West 168th St., New York, New York 10032

Fluorescence line narrowlng spectrometry is applied to five modlflcatlons of DNA (intact adducts) formed from diol epoxldes of benzo[a]pyrene, chrysene, 5-methylchrysene, and benz[a]anthracene. The dlrect identification of all flve adducts In a laboratory mixture is accompllshed. I n addltlon, a mixture of six correspondlng metabolites plus the DNA adducts from benzo[a Ipyrene and 5-methyichrysene Is resolved. Each adduct can be distlngulshed from its correspondlng tetroi metabolite. Utlllzatlon of an intensifled diode array-optical multichannel analyzer provldes detection of the adduct from benzo[a]pyrene with a S I N 150 for a damage level of -5 bases in lo6.

-

In the last several years there has been considerable interest in the development of new analytical methods for the detection of very low concentrations of chemical carcinogens that have become bound to DNA ( I ) . The formation of an adduct between DNA and carcinogens, including many of the polycyclic aromatic hydrocarbons (PAHs), is thought to be the crucial step in the initiation of carcinogenesis (1-3). Techniques for the measurement of PAH in the environment have been developed (4), but a better indication of the amount of exposure and risks for individuals is the actual level of DNA adducts formed in the body. Measurements of individual body burden can take into account adducts formed by many different carcinogens from multiple sources. For many years, studies of carcinogenesis have been carried out using radioactively labeled compounds (5). Although this was adequate for studies of carcinogenesis in laboratory animals and cell cultures, present studies are aimed at measuring existing levels of adducts in humans. Radioactivity measurements cannot be used directly to determine body burden, and furthermore, radioactively labeled compounds are available for only a small fraction of potential carcinogens and are unsuited for the analysis of complex mixtures. Recent attempts a t identifying DNA-PAH adducts have led to the development of several ultrasensitive detection techniques (1-5). Synchronous fluorescence spectroscopy (6-8), in which the excitation and emission frequencies are scanned together to maintain a constant energy or wavelength difference between them, takes advantage of the sensitivity of fluorescence detection to allow the measurement of a single adduct in at least lo7DNA bases ( I ) . However, the line widths of the measured fluorescence bands are too broad to allow the distinction

-

between chromophores that have severely overlapped SI So absorption origins. In a second approach, which relies on fluorescence detection (9, l o ) ,DNA adducts are hydrolyzed to release the carcinogen moiety (in the case of benzo[a]pyrene, B[alP, as free tetrols), and these are then extracted from the residual DNA and separated by HPLC. A modification level of 1 adduct in lo7 bases, for a DNA sample of 100 mg, has been detected (9, 10). A different technique relies upon antibodies that selectively bind to specific DNA-PAH adducts (1,11,12). Adduct levels of -1 in lo8 bases have been measured. Although these immunological methods are frequently used, they require the development of a new specific antibody for each adduct of interest. Potentially the most sensitive techniques yet developed involve the digestion of DNA and the labeling of the individual nucleotides with 32P(13, 14). Adducts, separated from unmodified nucleotides by TLC (14),have been detected by autoradiography a t levels corresponding to 1 in 1O'O bases of DNA. In addition, adducts labeled with electrophores (15) have been separated in a gas chromatograph and detected with a sensitivity of about 1 adduct in lo8 bases ( I ) . Because the nucleotides are labeled and detection depends on the labeling tag rather than on the carcinogen fragment, these techniques are broadly applicable. However, because of the high sensitivity specific adducts can only be identified by cochromatographic techniques. The method is, however, compatible with complex mixtures containing adducts from different carcinogens. An important point is that none of the above methodologies is readily applicable to the direct analyses of the complex mixtures that will result from studies of cocarcinogenesis or competition between carcinogens where structural information on the adducts is also needed. We have recently shown that fluorescence line narrowing (FLN) spectrometry was applicable to the direct analysis of mixtures of PAH tetrol metabolites (16). Each component in a mixture of six tetrols, including tetrols of the potent carcinogens 5-methylchrysene and benzo[a]pyrene, could be identified directly from FLN spectra of the mixture. It was found that FLN spectrometry is also applicable to a DNA adduct of benzo[a]pyrene and that the spectra of the adduct are quite similar to those of the tetrol N

-

(17-19).

The theory of FLN spectrometry has been discussed in detail previously (20),so only a brief description will be given here. Solute molecules in amorphous glasses a t 4.2 K have inhomogeneously broadened vibronic absorption bandwidths of several hundred wavenumbers. The molecules adopt essentially an infinite number of sites so that there is a broad distribution of transition energies for each vibronic transition. Excitation with a spectrally broad source therefore results in many transitions (sites) being excited, and fluorescence is

@ 1986 American Chemlcal Society 0003-2700/86/0358-0816$01.50/0

ANALYTICAL CHEMISTRY, VOL. 58, NO. 4, APRIL 1986

broad and often featureless. A narrow-line laser, however, selects only the narrow subset of molecules (isochromat) that has a transition a t the laser frequency. At 4.2 K and a t low solute concentrations where energy transfer cannot cause solute site randomization, only the selected isochromat fluoresces. The fluorescence spectrum consists of narrow lines, corresponding to vibrational frequencies of the molecule, SO a “fingerprint” of the molecule is obtained. Previous applications of FLN spectrometry to the analysis of mixtures of PAH metabolites demonstrated that individual compounds could easily be identified (16-19). Results similar to these were predicted for the DNA adducts. In this paper we present results that verify this prediction, in addition to results of studies on mixtures containing both unbound metabolites and DNA adducts. The term “DNA adduct” as used here means the DNA modified by adduction with a metabolite, i.e., not the individual modified nucleotides excised from the

817

HO HO I

Ill

OH

Ho@ HO

HO

IV

DNA. EXPERIMENTAL SECTION Instrumentation. The FLN spectrometry instrumentation and method have been described in detail previously (16). TWO pulsed excitation sources, a frequency-doubled NdYAG-pumped dye laser and a N2-pumped dye laser, provided excitation wavelengths from 335 to 390 nm. All samples were dissolved in a glass mixture consisting of 45% glycerol, 35% water, and 20% ethanol. This proved to be an excellent solvent for the DNA adducts. The glasses were contained in open-ended capillary sample tubes suspended horizontally in a liquid helium immersion Dewar. All spectra were obtained with the samples at 4.2 K. Following excitation of the samples, the fluorescence was dispersed in a 1-m monochromator. Line widths in the spectra are monochromator limited at -8 cm-I for all spectra presented here. Previous FLN studies of PAH metabolites and adducts (16-19) employed a photomultiplier tube (PMT) with a gated current amplifier as the detector. Most of the results presented in this paper were acquired by using the PMT. Recently we have begun using a detector consisting of a Tracor Northern TN-6134gateable intensified diode array and a TN-1710 optical multichannel analyzer (OMA). The gate pulse can turn the array on for 5-100 ns and can be delayed with respect to the excitation laser pulse. This gated operation allows time discrimination in the fluorescence spectra to eliminate scattered light or to separate fluorescence from interfering components with different fluorescence lifetimes. Further, the array detector is more sensitive than the PMT, and the OMA offers advantages of signal averaging and magnetic disk storage. Representative spectra that illustrate the utility of the diode array-OMA system are presented. Materials and Methods. We consider here only those materials and methodologiesthat have not been previously described (16-19). Samples of trans-3,4-dihydroxy-anti-1,2-epoxy-1,2,3,4tetrahydrobenz[a]anthracene,trans-8,9-dihydrodiol-unti-l0,11epoxy-8,9,10,ll-tetrahydrobenz[a]anthracene, and trans-l,2-dihydroxy-anti-3,4-epoxy-1,2,3,4-tetrahydrochrysene were obtained from the NCI Carcinogen Standard Reference Repository and were used directly. Samples of 5-methylchrysene derivative (21) were kindly provided by S. Hecht. Calf thymus DNA was obtained from Sigma Chemical Co. and repurified by phenol extraction and ethanol precipitation prior to use. DNA (2 mg in 4 mL of 10 mM cacodyate buffer pH 7.0) was modified by the addition of the appropriate diol epoxide (2 mg dissolved in 150 WLof tetrahydrofuran followed by 3.85 mL of acetone). After reaction for 4 h the solution was extracted 4 times with ethyl acetate and, after the addition of sodium acetate pH 5.4 (200 fiL of 2 M solution), the DNA was precipitated 3 times with 2 volumes of ethanol. Tetrols were obtained from the ethyl acetate extracts and purified by HPLC. The modified DNA was enzymatically digested to monodeoxyribonucleosidederivatives (22)and the modified deoxyribonucleosidederivatives separated by HPLC. The HPLC separations were obtained with an 850 system (Du Pont, Wilmington, DE) with an exponential gradient from 30 to 60% methanol in water over 90 min and at 50 OC on a Waters Associates (Milford, MA) WBondapak C-18 column. Samples were detected by their absorption at 254 nm. The extent

:I& V

DNA Adducts Figure 1. Structures of the five DNA adducts. The acronyms for adducts I-V are, respectlvely, 5-MeCDE-DNA, CTDE-DNA, B[a 1PDE-DNA, B[a]ADE-8-DNA, and B[a]ADE-l-DNA.

of modification of adduct I11 in Figure 1was determined by the Randerath procedure (14, 15) and was 0.4%. Each of the ethyl acetate extracts gave one major tetrol compound which was collected. Minor components were discarded. DNA samples digested after treatment with the 8,9-diol 10,llepoxide of benz[a]anthracene or the chrysene diol epoxide each gave one major adduct, while that from the 3,4-diol1,2-epoxide of benz[a]anthracene gave two major adducts in a 3.4:l ratio. Samples were collected from the HPLC and the solvents removed by evaporation under reduced pressure. The Randerath procedure was not used to determine the modification levels of adducts 11, IV, and V in Figure 1since UV absorption measurements showed that they are comparable to that of adduct 111, vide supra. Further, quantitation by FLN was not an objective of this work. For the FLN studies of a mixture of the five adducts shown in Figure 1, approximately equally concentrated (- 100 pg/mL) solutions of the adducts were used to prepare =50-pL sample volumes for analysis (16). The observed FLN intensities from the mixture are consistent with the above adducts having comparable modification levels. The same is true for adduct I. Compound Acronyms. The five DNA adducts studied are shown in Figure 1and are labeled as I-V. Their full names are not given here since they are lengthy and can be deduced from the names of the corresponding tetrols and diol epoxides (DE) given in ref 16. The acronyms used here for adducts I-V are, respectively, 5-MeCDE-DNA, CTDE-DNA, B[a]PDE-DNA, B[a]ADE-8-DNA, and B[a]ADE-l-DNA. The acronyms used for the tetrols corresponding to adducts I-V are, respectively, 5-MeCT, CT, B[a]PT, B[a]AT-8, and B[a]AT-1. The sixth metabolite studied is 4,5-dihydro-4,5-dihydroxy-9-methoxybenzo[a]pyrene (17),designated as B[a]P-diol.

RESULTS AND DISCUSSION Previous papers have discussed the physics of FLN spectrometry (23,24)and its use for mixture analysis (26).Excitation into the origin (0,O)band or a low-lying, 51600 cm-’, vibronic band of a fluorescent molecule results in a fluorescence spectrum consisting of narrow lines. For excitation of a (0,O) transition, the energy separation between the spectral lines and the excitation energy gives values for ground-state

818

ANALYTICAL CHEMISTRY, VOL. 58, NO. 4, APRIL 1986

vibrational frequencies. Resonant (0,O) fluorescence is generally obscured by laser scatter. The separation between the laser and (0,O) fluorescence lines following ( 1 , O ) excitation gives the frequencies of excited-state vibrations. The choice of (0,O) or (1,O) excitation depends upon relative absorption coefficients and the presence of interfering metabolites. In either event, knowledge of the excitation frequency and the vibrational spacings is sufficient to identify many metabolites of interest. The (1,O) type excitation was used for all of the results on mixture analysis presented in this paper since it optimizes selectivity (16). The five DNA adducts used in this study are shown in Figure 1. The parent chromophores are phenanthrene, anthracene, and pyrene. As in our study of mixtures of PAH metabolites, which preceded this (161, the DNA adducts were individually characterized, and several excitation wavelengths were selected for each. As expected, fluorescence line narrowing was exhibited by each of the adducts. However, for adduct V, there is significant intensity in the phonon sidebands (PSBs), and the zero-phonon lines (ZPLs), although sharp, are weak. The ratio of ZPL intensity to that of PSB of the corresponding metabolite was also fairly low (16). It was previously observed (25) that the PSB intensity increased for substituted anthracenes with increasing substitution, so the intetense PSBs for adduct V were not unexpected. Although this did not prove to be an obstacle in analyzing mixtures of DNA adducts, it did make identification of the adduct in the presence of the tetrol more difficult. None of the other adducts or tetrols studied exhibited significant PSB intensity. It is interesting to note that although there is good correlation between the vibrational frequency spacings for the DNA adducts, as compared to the unbound tetrol, there is also a small but consistent red shift (about 100-300 cm-I depending on the adduct) in the SI So absorption spectrum of the adduct. The red shifts observed for all the DNA adducts were important in allowing adducts to be spectroscopically distinguished from their corresponding tetrols. Standard FLN spectra have been generated for the DNA adducts shown in Figure 1. For each adduct spectra were obtained for at least two excitation wavelengths and vibrational analyses completed. Because of space limitations they are not given here. Instead attention is focused on the resolution of a mixture of the five adducts, Figure 1. This is not a trivial problem since the B[a]ADE-DNA (I), CTDE-DNA (11),and B[a]ADE-8-DNA (IV) adducts each possess phenanthrene as the parent fluorescent chromophore. Nevertheless, Figure 2 demonstrates that with an excitation wavelength of 343.1 nm these three adducts are readily distinguished. This figure contains portions of the standard FLN spectra for each of the adducts that are readily correlated with the peaks which appear in the mixture spectrum. The above excitation wavelength provides (1,0) type excitation for each adduct, and the FLN structure (peaks labeled with excited-state vibrational frequencies in cm-l) corresponds to multiplet origin structure (16). In part the resolution of adducts I, 11, and IV depends on their S1 So absorption spectra being shifted relative to each other. It is adduct I whose absorption lies furthest to the red, consistent with its labeled vibrational frequencies being the highest (1040 and 1348 cm-l). Figure 3 shows a portion of the FLN mixture spectrum obtained for an excitation wavelength of 369.5 nm. With this spectrum B[a]PDE-DNA (111) and B[a]ADE-1-DNA (V) are readily distinguished. Again (1,O) type excitation for both adducts is provided by A,, = 369.5 nm. Sharp zero-phonon lines for B[a]ADE-1-DNA a t 1181 and 1391 cm-l are apparent superimposed on the broad phonon side profile centered near 395 nm. It is possible that different host glasses may lead to

-

-

1

B

350

A'

355

A.(nrn)

-

360

uooQ@ ti0

HO

Figure 2. FLN spectrum from a mixture of the five DNA adducts obtained with (1, 0) type excitation at 3 4 3 . 1 nm, T = 4.2 K. Peaks labeled A, B, and C are due to the three individual adducts pictured. See text for further details.

L / 370-575-36-$2

390 395

h(nm)

Flgure 3. FLN spectrum from a mixture of the five DNA adducts obtained with (1, 0) type excitation, T = 4.2 K. Peaks are labeled as being due to either B[a]PDE-DNA (111) or B[a]AED-1-DNA (V) as indicated by the insert spectra for pure adducts I11 and V.

a diminution of phonon sideband activity for adduct V. Note that this activity for B[a]PDE-DNA is weak with the FLN spectrum dominated by zero-phonon lines (e.g., at 581 cm-l, which is an excited-state vibration). Thus, as was the case (16) for a mixture of the tetrols associated with the adducts of Figure 2, it is shown that (1, 0) type vibronic excitation affords facile resolution for a comparably complex adduct mixture. A further test of the selectivity of FLN spectrometry was the identification of DNA adducts in the presence of the

ANALYTICAL CHEMISTRY, VOL. 58, NO. 4, APRIL 1986

819

w Ae,=342.2 nm

i 345

350

355

345

heX=342 2 n m

360

365

370

hinrn)

Figure 4. FLN spectra of a mixture of a DNA adduct and corre-

sponding tetrol. Top spectrum (pure 5-MeCT) and bottom spectrum (pure 5-MeCDE-DNA) show lines labeled in cm-l used for identification. See text for further discussion. Results, not glven, have been obtalned which establish that each and every adduct of Figure 1 can be distinguished from its corresponding tetrol. corresponding tetrob. For each of the five pairs of compounds tested, the DNA adduct could be identified in mixtures containing approximately equal amounts of each. The basis for this identification is the small red shift in the absorption spectra of the DNA adducts, as compared with the tetrols. Excitation into a region of origin or (0,O) absorption for both the bound and unbound metabolites yields a fluorescence spectrum with lines from both generally overlapped. The ground-state vibrational energy differences between the bound and unbound metabolite determine the displacements between overlapped lines. Typical displacements are small, 5 5 cm-'. However, the red shift results in different vibrations being excited with higher energy (1, 0) excitation. Because the vibronic isochromats excited by A, = 342.2 nm are different for the tetrol and adduct, the distinction between the two is trivial based on origin multiplet structure, Figure 4. The zero-phonon features are labeled with excited-state vibrational frequencies. Both the 5-MeCT and corresponding adduct (I) exhibit weak electron-phonon coupling. Finally, a complex mixture of all six metabolites (see Experimental Section) and the adducts 5-MeCDE-DNA (I) and B[a]PDE-DNA (111)was analyzed. Resolution by FLN was straightforward when wavelengths (Aex) were used that provided the best separation of the two adducts from their corresponding tetrols. Some of the results are shown in Figures 5 and 6. For A,, = 342.2 nm, peaks from three tetrols and 5-MeCDE-DNA are observed, Figure 5. The most important features are labeled with the name of the compound and excited-state vibrational frequencies. When A,, = 366.1 nm is used, zero-phonon features from the remaining three metabolites are observed, lower spectrum of Figure 6. This spectrum was obtained with a PMT detector and without gating. The feature a t 1110 cm-l, which is a shoulder riding on the much more intense emission from the B[a]AT-1 tetrol is due to the B[a]PDE-DNA adduct. The parent chromophore of the interfering tetrol is anthracene. The upper spectrum of Figure 6 was obtained with the gateable diode array-OMA system. The array can be gated for periods of 5-100 ns and can be delayed with respect to the excitation pulse (10-ns width). The fluorescent parent chromophore of

350 355 Ahm)

360

Figure 5. FLN spectrum of an eight-component mixture including six metabolites (see Experimental Section) and the 5-MeCDE-DNA and B[a]PDE-DNA adducts. L labels the laser peak (Aex = 342.2 nm), T = 4.2 K. Bands are labeled as 5-MeCDE-DNA, CT, 5-MeCT, and B[a]AT-8 and assigned on the basis of standard spectra for the pure individual compounds.

v-

1 _

I

u

Figure 6. FLN spectra of a eight-component mixture (see Figure 5 caption) obtained for excitation wavelength of 366.1 nm. The lower spectrum was obtained ungated with a PMT, and only bands due to three metabolites (B[a]P-diol,B[a]PT, B[a]AT-1) are readily apparent based on standard spectra. The upper spectrum (insert)was obtained with a gateable intensified diode array-OMA (80-ns gate delay) and proves that the weak feature at 1110 cm-' in the lower spectrum is dud to the B[a]PDE-DNA (see text).

the B[a]PDE-DNA adduct is pyrene, Figure 1. Since the lifetime of pyrene is -400 ns (26) and that of anthracene is -5 ns, delay of the gate pulse will lead to a marked reduction of the B[a]AT-1 emission intensity. The upper (inset) spectrum of Figure 6 was obtained with a gate delay of 80 ns, and all features can be assigned to B[a]PDE-DNA (16). Thus, temporal discrimination greatly enhances the versatility of FLN for analysis of complex DNA-PAH adduct mixtures. The gating electronics of the array can also be used to eliminate contributions to the measured fluorescence intensity from scatter of the excitation pulse. Further, the OMA used with the diode array allows signals to be averaged so that spectra with greatly improved signal-to-noise ratios may be

820

ANALYTICAL CHEMISTRY, VOL. 58, NO. 4, APRIL 1986

B[c]PDE-DNA

he,:369 5 nm PMT

inary studies from this laboratory have shown that it can be applied to metabolites in urine and globin-metabolite adducts (globin can act as a surrogate for DNA). Further work in these areas is in progress. Future work will also include the analysis of damaged DNA from animal and human cells and urine for DNA adduds that have been excised during the repair process.

ACKNOWLEDGMENT The valuable technical assistance of John M. Hayes throughout the course of this work is gratefully acknowledged. We are indebted to S. S. Hecht of the Naylor Dana Institute for Disease Prevention for the samples of 5-MeCT and 5MeCDE-DNA.

LITERATURE CITED

370

375 380 h(nm1

Flgure 7. FLN spectra of a B[a]PDE-DNA adduct at a modification level of 510' base pairs. The adduct was isolated from mouse fibroblast 10 T cells exposed to 1 bg/mL for 24 h: upper spectrum obtained with PMT, lower with intensified diode array-OMA. The wavelength scale in nanometers corresponds to the top spectrum. The lower spectrum is expanded as indicated by solid lines.

acquired. Figure 7 shows the FLN spectrum of the B[a]PDE-DNA adduct at a modification level of -5 in lo6 base pairs. The adduct can be detected with a S I N of -150. Improvements in the collection and excitation optics of the existing system should be able to improve this to 1 adduct in lo8 detection limit, competitive with several of the existing methods. The addition of an excimer-pumped dye laser system in the near future should allow even lower limits to be achieved.

CONCLUDING REMARKS The FLN methodology developed earlier (16) for the resolution of PAH-metabolite mixtures has been shown to be applicable to comparably complex mixtures of DNA-metabolite adduct mixtures. As employed here the selectivity derives from the FLN phenomenon itself, selective excitation, time resolution, and vibronic excitation. There are two principal reasons why vibronic or ( 1 , O ) excitation is particularly useful: the vibrational dynamics (including mode mixing) in the SI state are markedly more sensitive to structural modifications of the metabolite than in the So ground state, and the same modifications produce significant shifts in the energy of the SIstate. With vibronic or ( 1 , O ) type excitation the principal zero-phonon bands in the FLN spectrum possess vibrational frequencies associated with the SI rather than the So state. The results presented here and elsewhere (16-19) indicate that FLN may have a wide variety of applications for the study of chemically initiated carcinogenesis. For example, prelim-

(1) Maugh, T. H., 11. Science (Washington, D . C . ) 1984, 226, 1183-1184. (2) Miller, E. C. Cancer Res. 1978, 3 8 , 1479-1496. (3) Nicollnl, C., Ed.; "Chemical Carcinogenesis"; Plenum Press: New York, 1982. (4) Lee, M. L.; Novotny, M. V.; Bartle, K. D. "Analytical Chemistry of Polycyclic Aromatic Hydrocarbons"; Academic Press: New York, 1981. (5) Baird, W. M. I n "Chemical Carcinogens and DNA"; Grover, P. L., Ed.; CRC Press: Boca Raton, FL, 1979; Vol. I. (6) Vo-Dinh, T.; Martinez, P. R. Anal. Chim. Acta 1981, 725,13-19. (7) Rahn, R. 0.; Chang, S. S.; Holland, J. M.; Stephens, T. J.; Smith, L. H. J . Biochem . Blophys . Methods 1980, 3 , 285-29 1. (8) Vo-Dinh, T.; Gammage, R. B.; Hawthorne, A. R.; Thorngate, J. H. Environ. Sci. Technoi. 1978, 72,1297-1302. (9) Shugart, L.; Holland, J. M.; Rahn, R. 0. Carcinogenesis (London)1983, 4 , 195-198. Chang, S. S.; Holland, J. M.; Shugart. L. R. Blochem. (10) Rahn, R. 0.; Biophys. Res. Commun. 1982, 109. 262-268. (11) Santella, R. M.; Lin, C. D.; Cleveland, W. L.; Weinstein, I.B. Carcinogenesis (London) 1982, 5 , 373-377. (12) Perara, F. P.; Pokier, M. C.; Yuspa, S. H.; Nakayama, J.; Jaretzki, A,; Curnen, M. M.; Knowles, D. M.; Weinstein, I.B. Carclnogenesis (London)1982, 3 , 1405-1410. (13) Reddy, M. V.; Gupta, R. C.; Randerath, E.: Randerath, K. Carcinogenesis (London) 1984, 5 , 231-243. (14) Gupta, R. C.; Reddy, M. V.; Randerath, K. Carcinogenesis (London) 1982, 3 , 1081-1092. (15) Sentlssl, A,; Joppich, M.; O'Connell, K.; Nazareth, A,; Giese, R. W. Anal. Chem. 1984, 56, 2512-2517. (16) Sanders, M. J.; Cooper, R. S.; Small, G. J.; Heisig, V.; Jeffrey, A. M. Anal. Chem. 1985, 57, 1148. (17) Heisig, V.; Jeffrey, A. M.; McGlade, M. J.; Small, G. J. Science (Washington, D . C . ) 1984, 223, 289-291. (18) Heisig, V.; Jeffery, A. M.; McGlade, M. J.; Small, G. J. R o c . Int. Symp. Polynucl. Aromat. Hydrocarbons, 8th 1984. (19) Heisig, V.; Jeffrey, A. M.; McGlade, M. J.; Small, G. J. Anal. Chem. Symp. Ser. Anal. Spectrosc. 1984, 19, 31-36. (20) Personov, R. I. I n "Spectroscopy and Excitation of Condensed Molecular Systems"; Agranovlch, V. M., Hochstrasser, R. M., Eds.; NorthHolland: New York, 1983. (21) Mellklan, A. A.; Amln, S.; Hecht. S. S.; Hoffmann, D.; Pataki, J.; Harvey, R. G. Cancer Res. 1984, 4 4 , 2524-2529. (22) Brown, H. S.; Jeffrey, A. M.; Weinstein, I.B. Cancer Res. 1979, 3 9 , 1673-1677. (23) Brown, J . C.; Edelson, M. C.; Small, G. J. Anal. Chem. 1978, 5 0 , 1394-1397. (24) Brown, J. C. Ph.D. Dissertation, Iowa State University, Ames, IA, 1982. (25) Brown, J. C.; Duncanson, J. A., Jr.; Small, G. J. Anal. Chem. 1980, 52, 1711-1715.

RECEIVED for review September 4,1985. Accepted December 2,1985. This research was supported by the Office of Health and Environmental Research, Office of Energy Research, and by NCI Grant CA 02111.