Identification of polycyclic aromatic hydrocarbon metabolites in

Surface-Enhanced Raman Analysis of Benzo[A]Pyrene-DNA Adducts on Silver-Coated Cellulose Substrates. Tuan Vo-Dinh , Mayo Uziel , Alan L. Morrison...
0 downloads 0 Views 630KB Size
1148

Anal. Chem. 1985, 57, 1148-1152

Identification of Polycyclic Aromatic Hydrocarbon Metabolites in Mixtures Using Fluorescence Line Narrowing Spectrometry Matthew J. Sanders, R. Scott Cooper, a n d Gerald J. Small* Ames Laboratory- USDOE and Department of Chemistry, Iowa State University, Ames, Iowa 50011 Volker Heisig a n d Alan M. Jeffrey Cancer Centerllnstitute for Cancer Research, Columbia University, 701 West 168th Street, New York, New York 10032

The direct identification of ail SIXcomponents in a laboratory mixture of polycyclic aromatic hydrocarbon (PAH) metabolites using fluorescence line narrowlng (FLN) spectrometry Is described. Metabolites are identifled by comparison to standard spectra of the pure compounds. I n addition, it Is demonstratedthat the sensltivlty of the technlque is adequate for the identlficatlon of PAH metabolite-DNA adducts at a level of -5 adducts per I O ' bases. Typically, using cells in culture, DNA damage levels from PAH carclnogens are 1 adduct per 105-106 base pairs. The potential for the analysis of complex mixtures of DNA adducts is discussed.

-

The concern which has developed over the levels to which individuals may be exposed to genotoxic environmental contaminants has resulted in the development of a variety of methods for the monitoring of such concentrations (1). However, the responses of individuals to equivalent ambient concentrationsof a particular compound may not be identical, resulting from differences in the way in which such compounds are absorbed, translocated, or metabolized. In addition, other compounds which are present may have synergistic or antagonistic effects on genotoxicity. Attempts to quantitate better the potential biological effects of exposure to environmental contaminants have led to the development of new immunological assay and 32Ppostlabeling autoradiography approaches, in which the levels of those compounds which become bound to DNA are measured ( 2 4 ) . Although there still may be differences between individuals in terms of how much damage is actually expressed, these approaches reduce the number of variables involved when compared to measurements of ambient concentrations. In addition, analysis can be limited to only those compounds which actually bind to the DNA, eliminating the need to separate complex mixtures and attempt to identify the active compounds within the resulting fractions. In attempting to develop alternative methods which are more direct and structurally informative, it is important to note that many of the compounds of genotoxic interest, or their DNA adducts, are highly fluorescent. These include polycyclic aromatic hydrocarbons (PAHs) and their derivatives, microbial products such as the aflatoxins, and adducts formed from the reaction of compounds such as chloracetaldehyde with adenine. However, conventional broad-band fluorescence techniques cannot resolve the emission from closely related compounds, e.g., substitutional isomers, in mixtures. To overcome this limitation, several high-resolution fluorescence techniques have been developed which provide spectra consisting of narrow vibronic lines which are sufficiently characteristic for specific identification. Laser-excited Shpol'skii spectrometry (5)and matrix isolation spectroscopy (6, 7) have been shown to be applicable to many PAHs and their derivatives and have been success-

fully used to directly distinguish between structurally very similar species in real complex mixtures. However, these solid-state fluorescencetechniques cannot readily be applied to large biomolecules like DNA-PAH adducts because of the limitations imposed by either solubility or vapor pressure. The same is not true for fluorescence line-narrowing (FLN) spectrometry using water-containing glasses. Small and coworkers have shown that FLN spectrometry can be used to identify amino PAHs (8)and to analyze for nonpolar PAHs in both laboratory mixtures and a solvent-refined coal sample (9,lO). More recently, Heisig et al. showed that FLN spectrometry was applicable to active metabolites of benzo[a]pyrene and the adducts they form with DNA (11). The FLN spectrum obtained for the DNA adduct was similar to that obtained for the nonadducted metabolite. In addition, a mixture of two benzo[a]pyrene metabolites was analyzed by using FLN. These results demonstrate that FLN spectrometry in glasses has exciting potential for the direct high-resolution analysis of nucleic acid damage resulting from chemical carcinogens. A thorough discussion of FLN has recently been given by Personov (12),but we will present a brief summary here. Solute molecules in amorphous glasses or polymers at 4.2 K exhibit vibronic absorption bandwidths of several hundred inverse centimeters, due to site inhomogeneous broadening. The solute molecules adopt essentially an infinite number of sites, with a broad distribution of energies for each vibronic transition. Classical broad-band excitation of fluorescence thus results in broad emission spectra. However, excitation using a narrow-line laser selects out a comparably narrow subset of sites (isochromat). At 4.2 K and at low solute concentration, where energy transfer cannot affect solute site randomization, only this isochromat fluoresces, and as a consequencethe fluorescencespectrum exhibits line narrowing. Several important questions were left unanswered by previous research. One relates to the generality of FLN spectrometry. That is, can FLN spectrometrygenerally be applied to the polar metabolites of PAHs and the DNA adducts they form? Furthermore, does FLN spectrometry possess the selectivity to identify the PAH metabolite components of reasonably complex mixtures? This will be important if synergistic and antagonistic effects on DNA damage are to be studied in the future. Finally, does FLN spectrometry possess the sensitivity to detect DNA damage at the very low levels required for practical applications? The intent of this paper is to explore these questions. To this end, the results of FLN studies on six metabolites derived from benz[a]anthracene, 5-methylchrysene,chrysene, and benzo[a]pyrene are presented and discussed. For the cases so far studied (11, 13) when FLN has been observed for polar PAH metabolites, the corresponding DNA adducts have also yielded FLN spectra. Furthermore, the vibronic structures observed in the fluorescence spectra of a metabolite, its parent PAH chro-

0003-2700/85/0357-1148$01.50/00 IgS5 Amerlcan Chemical Society

ANALYTICAL CHEMISTRY, VOL. 57, NO. 6, MAY 1985 I

1

p

N 2 RT: : :1 LASER

, GATED AMP

$+

OH

CRYOSTAT

,!,

HO&OH

HO

&OH

PRISM

PMT

1149

Hi

BC a1 AT-1

BE a1 AT-8

OH

R=CH, 5Me-CT R = H CT Block diagram of instrumentation for FLNS in glasses at 4.2 K. Shown for N,-pumped dye laser system. Flgure 1.

mophore, and its DNA adduct are quite similar.

EXPERIMENTAL SECTION Instrumentation. A block diagram of the instrumentation is shown in Figure 1. Two excitation laser systems were employed the frequency-doubled (by an Inrad Model 5-12 Autotracker) output of a tunable dye laser (Quanta-Ray PDL-1) with LDS-698 dye (Exciton), pumped with the second harmonic of a Nd:YAG laser (Quanta-Ray DCR-1) for excitation between 335 and 360 nm, and a N2-pumped tunable dye laser (Molectron DL-200 pumped by a UV-14 N2laser) using PBD and BBQ dyes (Exciton) for excitation at energy lower than 360 nm. The frequencydoubled dye laser provided -5-11s pulses at 10 Hz, while the Nz-pumpeddye laser pulses were 10 ns at 25 Hz. The excitation line widths of the two systems were -0.4 and -2 cm-', respectively. Power to the sample was below 2 mW for all measurements. The excitation laser beam was gently focused into the horizontal capillary sample tube by a 50-cm lens to ensure maximum excitation of the samples. For efficient collection of the fluorescence at the vertical monochromator slits, the horizontal fluorescence image from the sample was rotated 90' by a dove prism (Rolyn Optics) prior to arrival at the entrance slit. A 1-m focal length McPherson 2061 monochromator (F7.0) provided a reciprocal linear dispersion of 0.4 nm/mm (2400 grooves/mm of grating) for the fluorescence spectra and was operated with a band-pass of 0.11 nm or -8 cm-' for the spectra presented here. The zero-phonon line widths of the spectra are, therefore, monochromator-limited. The fluorescence was detected with an Amperex XP-2232 photomultiplier specially wired (14) for pulsed applications, and the current was measured by a Quanta-Ray DGA-1 gated amplifier with a fixed gate width of 100 ws. A small portion of the excitation beam was split off and detected by a fast photodiode (United Detector Technology PIN-6D) for use as the reference for normalization of the fluorescence signal. Spectra were recorded in real time with a Heath recorder. Compounds. The PAH metabolites studied are shown in Figure 2 along with their names and acronyms (to be used hereafter). We are gratefulto the following for supplying the PAH derivatives: R. G. Harvey, Ben May Laboratory for Cancer Research, University of Chicago, Chicago, IL (B[a]P-diol);s. s. Hecht and A. Melikian, Naylor Dana Institute for Disease Prevention, American Health Foundation, Valhalla, NY (5-MeCT); and Midwest Research Institute, Kansas City, MO, through the NCI Depository, for the rest. The procedure for preparation of the BPDE-DNA adduct has been described (15). The degree of modification of the DNA was determined by radiography to be -5 in lo6. All compounds were purified by HPLC and used without further treatment. Examination of the FLN spectra and comparison with those of the parent compounds showed no contaminants. The concentrations in the glasses employed for recording the FLN spectra of the pure metabolites are given in Table I. Included also are the concentrations for the six components in the mixture chosen for FLN analysis. Glasses. The same glass-forming solvent, a mixture consisting of 45% glycerol, 35% water, and 20% ethanol by volume, was used for all samples. The liquid helium Dewar and sample cool-down procedure have been described previously (10).

-

HO

6H

BE a1 PT

BCalP diol

Structures and acronyms of PAH metabolites studied: B[a]AT-1, 1,2,3,4-tetrahydro-1,2,3,4-tetrahydroxybenz [a]anthracene; B[a ]AT-8, 8,9,10,1l-tetrahydr0-8,9,10,1l-tetrahydroxybenz[a1anthracene; 5-MeCT, 1,2,3,4-tetrahydro-l,2,3,4-tetrahydroxy-5methylchrysene; CT, 1,2,3,4-tetrahydro-l,2,3,4-tetrahydroxychrysene; B[a ]P-diol 4,5-dihydro-4,5dihydroxy-9-methoxybenro[a ] pyrene; B[a]PT 7,8,9,1O-tetrahydro-7,8,9,1O-tetrahydroxybenzo[a] pyrene. Flgure 2.

Table I. Partial Vibronic Analysis of Components Following (0,O) Excitation compound X,,(O,O), nm

concentration (wm) pure mixture

vibration frequencies," cm-'

1187 (w), 1265 (m), 1411 (s), 1558 (m) 1358 (s), 1441 (m), 1607 (m) B[a]AT-8, 351.7 109 41 5-MeCT, 356.9 170 26 648 (w), 887 (w), 1364 (s), 1440 (m),1594 (m) CT, 348.7 104 16 872 (w), 1366 (s), 1443 (m), 1601 (m) 329 (m), 492 (w), 536 (w), 878 4 B[a]P-diol,373.5 308 (w),1391 (9) 597 (w), 792 (m),856 (m), 1116 B[a]PT, 377.1 220 46 (m), 1249 (s), 1291 (m), 1414 (81, 1563 (4 nAssignmentsare f10 cm-'. Intensities are designated (w), 0.5 of major peak. B[a]AT-l, 386.1

103

9

However, for the present experiments, short (-2.5 cm) lengths of melting point capillary tube (-0.12 cm i.d.) were used to contain the glasses (sample volume of 20-30 wL). The capillarytubes are held in a horizontal position in the Dewar. Two advantages are obtained the small volume allows for higher concentrations when only small amounts of sample are available, and the laser beam prop&ating parallel to and through the open-ended capillary tube excites the entire sample. Their sole disadvantage, compared to the large (- 1 cm i.d.) polymer centrifuge tube used previously (IO),is that glasses contained in them tend to incorporate bubbles and crack more easily, thereby enhancing scattered light. Careful masking significantly suppressed the scatter. In the present experiments,no temporal gating was used. Addition of a gateable diode array and optical multichannel analyzer (OMA) is planned in the near future and will further suppress interferences due to laser scatter.

RESULTS AND DISCUSSION Previous papers on the analytical applications of FLN spectrometry have discussed the underlying physics of the phenomenon (9,16).Excitation into the origin (0,O) band of the fluorescent state with a narrow-line laser selects out a

1150

ANALYTICAL CHEMISTRY, VOL. 57, NO. 6, MAY 1985

B

tI

[ a ] AT-I

W

P

JC

,

390

395

400

405

410

415

X(nm)

8

2

I'

I

, 350

355

111 I

A

I

365

I

370

375

(nm)

Figure, 4. FLN spectrum of mixture of CT (60 Mm) and 5-MeCT (73 wm), following excitation of the (0,O)transition of CT.

L

JL,

350

360

L

355

360

365

370

375

Atnm)

Figure 3. Standard FLN spectra of B[a]AT-1 (top)and CT (bottom) following (0,O)excltation at 4.2 K. L labels laser peak. Vibrational lines are labeled in energy units of inverse centimeters relative to the excitation energy. See Table I for partial vibratlonal analysis of alt SIX metabolites. comparably narrow isochromat, and the fluorescence spectrum is line-narrowed. Not only is (0,O) excitation useful for FLN, but excitation of low-lying vibronic bands, referred to hereafter as (1,O) bands, also affords FLN. For many molecules a low-lying (1,O) absorption band, several hundred inverse centimeters above the (0,O) band, which is due to a single vibration can be found (IO). Excitation of such a (1,O) band affords a FLN spectrum which matches in simplicity the spectrum obtained with (0,O) excitation. However, for (1,O) excitation, the origin or (0,O) fluorescence band is displaced to lower energy of the laser frequency by an amount equal to the frequency of the excited vibration. This is important because the (0,O) fluorescenceband is now a useful analytical line. For (0,O) excitation, the fluorescence origin is typically contaminated by laser light scatter. The complexity which arises when the vibronic absorption feature being pumped is comprised of several overlapping and unresolvable (1,O) bands has been considered (9). Laser excitation of such a vibrationally congested feature pumps as many different isochromats as there are overlapping bands. In the extreme, where the number of bands is very large, FLN is lost and the fluorescence spectrum has the broad and disappointing appearance of the spectrum obtained with a broad-band excitation source which uniformly excites all sites. Importantly, however, when the number of bands is small (-3), the analytical utility of FLN spectrometry is enhanced when (1,O) rather than (0,O) excitation is utilized. Fluorescence excitation spectra, not given here, were used as a guide for the determination of several suitable excitation wavelengths (Aex) for each of the PAH metabolites shown in

Aex= J C

342.2 n m

,

345

L

350

355

360

365

370

A(nm)

Figure 5. FLN spectrum of mixture of six PAH metabolites. Peaks are labeled as being due to B[a]AT-8, 5-MeCT, or CT. Inset: partial spectrum of pure CT, showing lines used for Identification. Figure 2. Earlier work (16) emphasized the importance for analyses of generating standard FLN spectra for at least two excitation wavelengths, and this will be underscored again in what follows. The fluorescent parent chromophores associated with the metabolites are either anthracene, chrysene, pyrene, or phenanthrene. It was important for assessing sample purity to first obtain FLN spectra with (0,O) excitation so that the vibrational frequencies and relative intensities could be compared with those of the parent chromophores. Fluorescence line narrowing was exhibited by all metabolites studied; typical of the spectra obtained are those shown in Figure 3 for B[a]AT-1 and CT, while a partial vibrational analysis for each of the metabolites is given in Table I. On the basis of the close agreement between the metabolite FLN spectra and those of their respective parent chromophores,the sample purities are judged to be high. We note from Figure 2 that B[a]AT-8, CT, and 5-MeCT possess the phenanthrene chromophore and, therefore, that their spectra are quite similar. However, these three metabolitescan be distinguished on the basis of selective excitation of their (0,O) absorption bands. In Table I, Figures 3-7, and throughout the text, laser excitation wavelengths will be given in nanometers. However,

ANALYTICAL CHEMISTRY, VOL. 57, NO. 6, MAY 1985

L

J

I, 310

375

300

385

390

395

A (nml

Figure 8. FLN spectrum of six-component mixture. Peaks are labeled as B[a]AT-1, B[a]Pdiol, and B[a]PT. Inset: partial spectrum of B[a ]P-diol, showing lines used for identification.

L

.Aex= 319.0 nm

380

385

390

395

400

x (nm) Figure 7. FLN spectrum of DNA isolated from mouse fibroblast 10T1/, cells exposed to 1 pg/mL benzo[a]pyrene for 24 h. Level of modification is - 5 per lo6 bases. Structure shown is the adduct (in the 10 position) between DNA and BPDE (see text).

the lines in the FLN spectra are due to vibrations of the molecule and are thus labeled in units of vibrational energy, wavenumbers (cm-l). This is the difference in energy between the excitation and emission wavelengths. For (0,O)excitation, as in Table I and Figure 3, the energies therefore may be directly interpreted as ground-state vibrational frequencies. For any low-temperature high-selectivity fluorescence technique, the electron-phonon interaction is a potential problem for selectivity (17,18). If sufficiently strong, it produces broad low-frequency phonon sidebands which build on the sharp zero-phonon vibronic features of the fluorescence spectrum. For very strong coupling, the zero-phonon features are lost, as is the high selectivity of FLN. The metabolite which exhibits the strongest coupling, moderately strong, is B[a]AT-l, which possesses the anthracene chromophore. This observation is consistent with earlier results on anthracene and substituted anthracenes (10). The moderate electronphonon coupling for B[a]AT-1, Figure 3, posed no problem for the analysis of the six-component metabolite mixture. The benzo[alpyrene metabolites were previously shown to exhibit weak coupling (11). For a fluorescence-based technique, the distinction between the unsubstituted chrysene metabolite, CT, and its analogue,

1151

5-MeCT7is a formidable challenge and important since 5methylchrysene is a potent carcinogen while chrysene is not (19,20).However, the (0,O) absorption band of 5-MeCT is significantly red-shifted (-600 cm-') relative to CT so that in a binary mixture of the two metabolites, 5-MeCT can be determined without any interference from CT when origin excitation for 5-MeCT is used. The key question is whether a A,, suitable for FLN from CT will allow for its characterization since this excitation will excite 5-MeCT with considerable excess vibrational energy. The concentrations of the two components are low enough to preclude any effects of interaction on the spectra. But because 5-MeCT is excited with excess vibrational energy, the utility of FLN for analysis may be negated if the fluorescence of 5-MeCT is broad enough to obscure narrow lines arising from CT. This will be an ever-present problem in doing direct mixture analysis, and different mixtures may have to be characterized at different excitation wavelengths. Figure 4 shows the FLN spectrum of the binary mixture for a A,, coincident with the (0,O) absorption band of CT. All four main peaks for the CT spectrum are evident; see Table I. Interestingly, the set of four line-narrowed bands near 355.5 nm comprises the (0,O) fluorescence multiplet structure for 5-MeCT arising from the excitation of a vibrationally congested vibronic absorption band, as described previously. The energy differences between these bands and the laser excitation frequency provide the frequencies of the overlapped excited state vibrations which are 426,495,546, and 632 cm-l. The multiplet structure provides a fingerprint for identification which may be more useful than the pattern arising from (0,O) excitation. The broad fluorescence at -373.5 nm in Figure 4 results from the overlap of fluorescence lines from each of the four excited-state vibrational isochromats excited by the laser. Note that the peak at 1901 cm-l in the spectrum of Figure 4 is separated from the highest intensity 5-MeCT (0,O) fluorescence multiplet component at 546 cm-I by -1356 cm-l, which is the major ground-state fluorescence line; see Table I. We turn now to the analysis of a six-component PAH metabolite mixture, as given in Table I. In view of earlier work on FLN of DNA-PAH metabolite adducts (11),resolution of the six metabolites would augur well for the resolution of a comparable mixture of adducts. Previous applications of FLN have emphasized the use of a A,, which excites a single isochromat (10,16).Indeed, for a binary mixture of B[a]PT and B[a]P-diol, excitation of the two (0,O) absorption bands resulted in each case in a single-component FLN spectrum, with no interference from the second component (11). However, mixtures generally do not lend themselves to such convenient analysis. Interference from broad fluorescence from other components can obscure the FLN spectrum of the component of interest. For example, if in the analysis of 5-MeCT7A,, = 356.9 nm is used, providing (0,O) excitation for 5-MeCT7then serious interference from the fluorescence of B[a]P-diol occurs. This A,, excites B[a]P-diol in a vibrationally congested region with 1300 cm-l of excess vibrational energy, cf. Table I. Because of interferences of this nature, (LO) excitation for the components of interest proved to be the key to resolving the mixture. In this regard, the multiplet origin structure afforded by (1,O) excitation was particularly helpful. In the process of analysis of the mixture, about 30 fluorescence spectra were obtained, and, except for B[a]AT-8, every component was identified by its FLN spectra at two or more excitation wavelengths. As examples Figures 5 and 6 are presented in order to discuss the analysis procedure. The FLN spectrum in Figure 5 for the mixture was obtained with A,, = 342.2 nm which provides (1,O) excitation for CT with

-

1152

ANALYTICAL CHEMISTRY, VOL. 57, NO. 6, MAY 1985

-550 cm-' of excess vibrational energy. The inset spectrum was obtained for CT by itself and with the same excitation wavelength. Thus, the 517-, 547-, and 635-cm-* bands represent the origin multiplet structure for CT, and these frequencies correspond to the excited-sta$e vibrations being pumped. For completeness, the FLN spectra for A,, = 342.2 nm were obtained for pure B[a]AT-8 ana 5-MeCT and were used to assign the remaining bands in Figure 5. This would not be necessary if one is interested in analyzing only for CT. In Figure 6, the A,, = 366.1 nm was chosen to provide (1,O) excitation for B[a]P-diol, and the inset spectrum obtained for the pure diol establishes its assignments in the mixture spectrum. The multiplet origin structure for B[a]P-diol is particularly rich. The assignments in the mixture spectrum for B[a]PT and B[a]AT-1 were made on the basis of the A,, = 366.1 nm FLN spectra of the pure compounds. The procedure just described was followed kor the analysis of the mixture for each of the metabolites. Finally we turn to the question of sensitivity raised earlier. The FLN spectrum in Figure 7 is that of DNA isolated from mouse fibroblast 10T1/, cells exposed to 1FglmL of benzo[alpyrene for 24 h. The major adduct, shown in the figure, is that formed by the reaction of DNA with 7,tl-dihydroxy9,10-epoxy-7,8,9,10-tetrahydrobenzo[a]pyrene (BPDE), which is an active metabolite of benzo[a]pyrene (15). Its epoxide occupies the 9- and 10-positions on the bay region and is responsible for DNA damage via attack on the guanine base of DNA (21,22). The modification of the DNA used to obtain the spectrum in Figure 7 is low, -5 adducts per lo6base pairs, as determined by radiography. The bands shown can be assigned to the DNA-BPDE adducts on the basis of the previously reported spectrum of the adduct obtained with a modification level of 1 adduct per 250 base pairs (11). The S I N ratio in Figure 7 is expected to be markedly enhanced when an intensified diode array and OMA rather than a PMT is utilized as a detector. Utilization of gating in detection will likely suppress the broad and unidentified fluorescence background in Figure 7. In addition, (1,O) excitation rather than (0,O) excitation, as was used to obtain the spectrum of Figure 7, is expected to lower the detection limits even further.

CONCLUDING REMARKS That FLN spectrometry with water-containing glasses can be applied to some polar metabolites of benzo[a]pyrene and the DNA-BPDE adduct has previously been established (11). In this paper it is established that FLN spectrometry appears to be generally applicable to polar metabolites of PAHs. Very recent and preliminary result# from our laboratory demonstrate that FLN is also operative for DNA adducts formed from the epoxide precursors of B[a]AT-1, B[a]AT-8, CT, and 5-MeCT (23). With this in mind, the successful resolution of the six-component metabolite mixture bodes well for the analysis of comparably complex mixtures of DNA adducts. ,

It is hoped, therefore, that FLN spectrometrywill prove useful for fundamental studies of antagonistic and synergistic effects in DNA damage resulting from exposure to mixtures of PAH. And last but not least, results presented here establish that FLN spectrometry has the requisite sensitivity for high-resolution analysis of DNA damage at the level of 1 adduct per 105-106 base pairs, with the potential for lower limits with improvements in the instrumentation. Registry No. B[a]AT-l, 78326-53-1;B[a]AT-8,53760-22-8; 5-MeCT, 88184-17-2; CT, 79356-06-2; B[a]P-diol, 91913-70-1; B[a]PT, 59957-91-4.

LITERATURE CITED (1) Lee, M. L.; Novotny, M. V.; Bartle, K. D. "Analytical Chemistry of Polycyclic Aromatic Hydrocarbons"; Academic Press: New York, 198 1. (2) Gupta. R. C.; Reddy, M. V.; Randerath, K. Carcinogenesis (London) 1982, 3, 1081-1092. (3) Perera, F. P.; Pokier, M. C.; Yuspa, S. H.; Nakayama, J.; Jaretzki, A,: Curnen, M. M.; Knowles, D. M.; Weinstein. I.B. Carclnogenesls (London)1982, 3 , 1405-1410. (4) Maugh, T. H., 11. Science (Washington, D . C . ) 1984, 226, 1183- 1184. (5) D'Sihra, A. P.; Fassel, V. A. Anal. Chem. 1984, 5 6 , 985A-1000A and references therein. (6) Wehry, E. L.; Mamantov, G. Anal. Chem. 1979, 51, 643A-658A and references therein. (7) Howell, H. E.; Mamantov. G.; Wehry, E. L. Anal. Chem. 1984, 56, 821-823. (8) Chiang, I.; Hayes, J. M.; Small, G. J. Anal. Chem. 1982, 54, 315-318. (9) Brown, J. C.; Edelson, M. C.; Small, G. J. Anal. Chem. 1978, 50, 1394-1397. (10) Brown, J. C.; Duncanson, J. A., Jr.; Small, G. J. Anal. Chem. 1980, 52, 1711-1715. (11) Heislg, V.; Jeffrey, A. M.; McGlade, M. J.; Small, G. J. Science (Washlngton, D. C.) 1984, 223, 289-291. (12) Personov, R. I.In "Spectroscopy and Excitation of Condensed Molecular Systems"; Agranovich, V. M., Hochstrasser, R. M., Eds.; NorthHolland: New York. 1983. Hecht, S. S.; Helslg, V.; Jeffrey, A. M.; Sanders, M. J.; Small, G. J., unpublished results. Harrls, J. M.; Lytle, F. E.; McCain, T. C. Anal. Chem. 1978, 48, 2095-2098. Brown, H. S.; Jeffrey, A. M.; Weinstein, I. 8. Cancer Res. 1979, 39, 1673-1677. Brown, J. C. Ph.D. Dissertation, Iowa State University, Ames, I A , 1982. Burland, D. M.; Haarer, D. IBM J . Res. Develop. 1979 23, 534-546. Friedrich, J.; Haarer, D. Angew. Chem., Int. Ed. Engl. 1984, 23, 113-140. Hoffmann, D.; Bondinell, W. E.; Wynder, E. L. Science (Washington, D . C . ) 1974, 183, 215-216. Hecht, S . S.; Bondinell, W. E.; Hoffmann, D. J . Natl. Cancer Inst. 1974, 53, 1121-1133. Jeffrey, A. M.; Weinstein, I. B.; Jennette, K. W.; Grzeskowiak, K.; Nakanishi, K.; Harvey, R. G.; Autrup, H.; Harris, C. Nature (London)1977, 269, 348-350. Jeffrey, A. M.; Jennette, K. W.; Blobstein, S. H.; Weinstein, I.B.; Beland, F. A,; Harvey, R. G.; Kasal. H.; Miura, I.; Nakanishi, K. J . Am. Chem. SOC. 1976, 98, 5714-5715.

RECEIVED for review December 11,1984. Accepted February 4,1985. This research was supported by the Office of Health and Environmental Research, Office of Energy Research, and by NCI Grant CA021111.