Chem. Res. Toxicol. 1991,4, 311-317
311
Dependence of Conformations of Benzo[ a Ipyrene Diol Epoxide-DNA Adducts Derived from Stereoisomers of Different Tumorigenicities on Base Sequence Camille J. Roche,?**Alan M. Jeffrey,! Bing Mae,? Anna Alfano,?Seog K. Kim,fJ Victor Ibanez,?and Nicholas E. Geacintov*?t Chemistry Department and Radiation and Solid State Laboratory, New York University, New York, New York 10003,and Cancer Center, Institute of Cancer Research, Columbia University, New York, New York 10032 Received October 17,1990 The conformations of covalent adducts derived from the binding of the highly tumorigenic stereoisomer (+)-trans-7,8-dihydroxy-anti-9,10-epoxy-7,8,9,10-tetrahydrobenzo[a]pyrene [ (+)-anti-BPDE] and its nontumorigenic (-)-anti-BPDE isomer with poly[(dG)-(dC)], poly[ (dG-dC).(dG-dC)], poly[ (dT-dC)*(dG-dA)],and poly[ (dA-dC)*(dG-dT)]were investigated by employing W absorbance and linear dichroism methods. The degrees of orientation of the BPDE residues (bound covalently to N2 of deoxyguanosine), relative to the DNA bases, are most pronounced in the alternating and nonalternating (dG)*(dC)polymers and decrease in polymers with neighboring dA.dT base pairs. The tumorigenic (+)-anti-BPDE isomer gives rise predominantly to external (solvent-exposed) site I1 adducts, while the (-)-enantiomer gives rise predominantly to site I adducts with significant carcinogen-nucleoside interactions. In the mixed (dA-dC)-(dG-dT) and (dT-dC)*(dG-dA) copolymers, the (+)-anti-BPDE isomer also binds predominantly to N2 of deoxyguanosine, but the adducts are weakly oriented with respect to the DNA bases. T h e incidence of site I1 adducts is considerably reduced as compared t o the (dG)-(dC) and (dG-dC)*(dG-dC)polymers, and there is a greater proportion of site I adducts; the presence of a significant proportion of unordered adduct forms is also suggested from the diffuseness and broadness of the absorption spectra in the dA-dT base pair containing polymers. The preference of formation of site I1 adducts in dG-rich sequences in the case of the biologically highly active (+)-anti-BPDE isomer is discussed in terms of the known binding and mutation spectra.
Introduction The differences in the tumorigenicities and mutagenicities of the two enantiomers of trans-7,8-dihydroxyanti-9,10-epoxy-7,8,9,10-tetrahydrobenzo[a]pyrene (antiBPDE) have been well documented (1-4).Both enantiomers bind predominantly to the exocyclic amino group of deoxyguanosine in DNA, although the (-)-enantiomer binds significantly to deoxyadenosine as well. Brookes and Osborne (2) concluded that the differences in biological activities are related to spatial differences in adduct conformations, and to differences in the processing of these lesions in a cellular environment. There is a significant difference in the conformations of adducts derived from the binding of these two enantiomers to DNA (5-12).The adduct conformations can be classified according to the site I/site I1 hypothesis (13). Site I1 adducts are characterized by an external, solventexposed adduct conformation, while site I complexes involve considerable carcinogen-base stacking interactions (7,9,11). Physicochemical experiments with the (+)- and (-)-anti-BPDE enantiomers bound covalently to DNA have shown that the more tumorigenic (+)-enantiomer prefers a site I1 conformation, while the (-)-enantiomer exhibits predominantly site I type conformations (9).The
preference of tumorigenic isomers of different polycyclic aromatic diol epoxides to bind to DNA preferentially as site I1 adducts, and the preferred site I adduct orientations of the less active isomers, has been noted (9, 12). Sequence effects may also be important in distinguishing tumorigenic from nontumorigenic polycyclic aromatic carcinogen-DNA lesions. Boles and Hogan (14)showed that BPDE exhibits a preference for binding to sequences of guanines. Kootstra et al. (15)showed that racemic anti-BPDE, in the Chinese hamster ovary aprt gene, binds preferentially in the 5' flanking region of the gene in a segment containing two consecutive GC box consensus sequences; this GGGCGG sequence is an important eukaryotic promoter element found in the 5' flanking region of many genes. The mutation spectra of racemic antiBPDE in the supF tRNA gene of the pZ189 shuttle vector was investigated by Yang et al. (16);in the progeny of BPDE-treated plasmids, a majority of the observed base substitutions were G-C T.A transversions. Two hot spots were found at the middle base pair of GGG triplets, with six other hot spots located at G-C base pairs of the supF gene (15).Carothers and Grunberger (17)found that most of the mutations (9190)induced by BPDE in the dihydrofolate reductase gene in Chinese hamster ovary cells also involved G C T-A transversions; base substitutions were favored at guanines flanked by adenines and other guanines. Reardon et al. (18)found that both anti-BPDE enantiomers bind to codon 12 (GGA sequence) of the rat c-Hi-rasprotooncogene sequence incorporated into a plasmid. Taken together, these studies suggest that the binding of BPDE to sequences rich in G-C base pairs, particularly those containing runs of G's, may be of particular significance.
-
-+
~~~
~
* Correspondence should be addressed to this author at the Chemistry Department, New York University, New York, NY 10003. 'New York University. *Present address: Department of Chemistry, Yale University, New Haven, CT 06520. 1 Columbia University. Present address: Department of Physical Chemistry, Chalmers University of Technology, S-41296 Cijteborg, Sweden. 0893-228x/91/2704-0311$02.50/0
0 1991 American Chemical Society
312 Chem. Res. Toxicol., Vol. 4, No. 3, 1991
Roche et a1.
I n this work we have studied the base sequence dependence of the binding of (+)-anti-BPDE and (-)-anti-BPDE to dG and t h e site I/site I1 adduct conformational heterogeneity employing the synthetic polynucleotides poly[ (dG).(dC)], p~ly[(dG-dC)*(dC)],p~ly[(dA-dC).(dG-dT)], and poly[ (dT-dC)-(dG-dA)]. Absorbance and flow linear dichroism techniques were used to distinguish between site I and site I1 a d d u c t conformations.
I
I
( A ) Site
I
I
I
1
I !I
Experlmental Section Both (+)-anti-BPDE and (-)-anti-BPDE were purchased from the National Cancer Institute Chemical Carcinogen Reference Standard Repository (lot nos. 83-344-49-5 and 83-344-99, respectively). Caution: BPDE is hazardous and should be handled with great care, avoiding contact with the skin; any spillage should be immediately treated with a dilute solution of acid in order t o hydrolyze the diol epoxide. Stock solutions (about 1 mM BPDE) were prepared in 1 9 1 tetrahydrofuran-triethylamine solutions. The polynucleotides were purchased from Pharmacia (Piscataway, NJ). Poly[(dG-dC).(dG-dC)],poly[(dA-dC).(dG-dT)], and poly[(dT-dC)-(dG-dA)]were directly dissolved in 5 mM sodium cacodylate, 0.1 M NaCl, and l mM EDTA, pH 7.1, buffer solution. These solutions were then exhaustively dialyzed against the buffer solution used in all subsequent experiments (5 mM sodium cacodylate, 20 mM NaCl, pH 7.1). The method of Wilson et al. (19) was employed to dissolve the nonalternating poly[(dG).(dC)];this method involves dissolving the polymer is basic solution (pH 12) and subsequent extensive dialysis against the cacodylate buffer solution. The integrity of all the polymers was verified by standard methods. Circular dichroism spectra were recorded for all polymer solutions and compared to those published in the literature (20). The polymer concentrations were determined by using the published molar extinction coefficients for the different polynucleotides (20). The covalent adducts were prepared by adding small aliquots of the BPDE-tetrahydrofuran stock solution to solutions of polymers (0.1-0.2mM nucleotide concentrations). The initial concentration of BPDE was typically 2-10 pM, and the tetrahydrofuran concentration did not exceed 1 % by volume. The reactions were allowed to proceed to completion a t room temperature (1h), and then the solutions were extensively dialyzed against the sodium cacodylate-20 mM NaCl buffer solution in order to remove the tetraol hydrolysis products (7,8,9,10-tetrahydroxytetrahydrobenzo[a]pyrene,or BPT). The extent of covalent modification was estimated spectroscopically as described previously (5). The linear dichroism spectra were determined by partialy orienting the polynucleotides in a Couette cell consisting of a rotating (300-600 rpm) inner quartz cylinder (22 mm 0.d.) and a stationary outer cylinder (23 mm i.d.). The degree of alignment of the nucleic acid bases and the pyrenyl residues attached covalently to the polynucleotides is probed with linearly polarized light oriented either parallel or perpendicular with respect to the flow direction. The linear dichroism (LD) is defined as LD = AI, - A ,
(1)
where A,, and A, are the absorbances measured with the electric vector of the polarizer oriented parallel and perpendicular, respectively, with respect to the flow direction (21). The apparatus was calibrated by measuring the magnitude of the LD signal with a polarizing crystal instead of the sample, and thus the LD values are reported in absolute units. The nature of the covalent adducts was examined by the usual methods of enzyme digestion of the polynucleotides to the nucleoside level (22). The digested adducts were removed from the proteins by use of a Sephadex LH-20 column, and the adducts were analyzed on an HPLC system (LKB, Pharmacia, Piscataway, NJ) and a Whatman Partisill0 ODS-2 reversephase column. The column was eluted with the following methanol in water gradient linear 45-50%, 20 min, isocratic 30 min, and then 50-60% over 50 min (23).The adducts were detected by their absorbance at 280 nm, and by the fluorescence of the pyrenyl residues (245-nm
11CY
,
I
I
300
320
340
360
380
400
Wavelength, nm
Figure 1. Model absorption spectra for sites I and 11. (A) Site I Absorption spectrum of noncovalently intercalated (+)-BPDE (2 pM) in native DNA (1.5 mM), pH 8.6. Absorption spectrum measured immediately after mixing. The absorbance a t the 354-nm maximum is 0.03. (B) Site 11: Dotted line, absorption spectrum of tram-anti-BPDE-@dG adduct in buffer solution; solid line, linear dichroism of a covalent adduct derived from the covalent binding of (+)-anti-BPDEto poly[ (dG-dC).(dG-dC)](the linear dichroism signal is proportional to the absorbance; see text). The two spectra are normalized to one another a t the maxima. excitation, 370-nm emission cutoff filter) using a Kratos FS 970 fluorescence detector (Kratos Analytical Instruments, Inc., Ramsey, NJ). The retention times of the adducts were compared to those of the covalent adducts formed by the modification of single-stranded poly(dG), poly(dA), and poly(dC) with BPDE, using the same reaction conditions and protocols as described above.
Results Site I and Site I1 Absorption Spectra. Typical site I and site I1 absorption spectra are depicted in Figure 1. The site I spectrum is that of (+)-BPDE noncovalently bound to native double-stranded DNA. This absorption spectrum was measured within 90 s after adding the BPDE (3 pM) to an aqueous buffer solution at pH 8.6 containing a n excess of DNA (1.5 mM) as described elsewhere (6); under these conditions 95% of t h e BPDE is bound to t h e DNA by an intercalation mechanism. T h e absorption maxima are located at 337 and 354 nm, which represents a red shift of 10-11 n m with respect to free BPDE and BPT. Typical site I1 type spectra are shown in Figure 1B. The absorption spectrum of (+)-anti-BPDE bound covalently via its C'O position to t h e exocyclic amino group of deoxyguanylic acid (dGMP) is shown in Figure 1B (dotted line). This adduct, which has trans stereochemistry (4,24) a n d is denoted by BPDE-dGMP, was prepared as described by Cheng e t al. ( 4 ) a n d was characterized by its HPLC elution time a n d C D spectrum. As in free BPDE, t h e vibronic bands at 330 a n d 346 nm a r e rather sharp, although t h e maxima are red-shifted by 2-3 nm. The linear dichroism spectrum of a covalent adduct derived from t h e binding of (+)-anti-BPDE to poly[ (dG-dC).(dG-dC)] is also shown in Figure 1B (solid line) a n d almost coincides with
Conformations of BPDE-DNA Covalent Adducts
lB
(-I-EPDE
I
P
S ! c o o
v, .E c
(r
2-025,-
2 s 0 .-
n
;-0.5c .-I 0
-0.75-
I
,
)
,
I
T
Chem. Res. Toxicol., Vol. 4, No. 3, 1991 313
(+I- BPDE
I
-*15
q .* :*.. . e.:
,
I
, I
-
i
- +0.5
*.
,
,
0.0
8-
I
1
Wavelength, nm Figure 2. Absorption and linear dichroism spectra of (+)anti-BPDE and (-)-anti-BPDE covalently bound to poly[ (dG).(dC)]. (A) and (C): (-)-adduct, nucleotide concentration ([N]) = 0.23 mM, covalently bound BPDE residues ([BPDQB]) = 1.1 pM. (B) and (D): (+)-adduct,[Nl = 0.21 mM, [BPDEcBI = 2.7 pM. Solid lines, experimental; dotted lines, simulated spectra (see text). that of the BPDE-dGMP adduct. The BPDE-polynucleotide adduct was illuminated with UV light (346 f 6 nm) as described by Zinger et al. (7)to reduce the relatively small fraction of photolabile site I lesions which are present in these adducts as well (see below). The good correspondence between the absorption spectrum of the BPDE-dGMP adduct and the linear dichroism spectrum of the (+)-anti-BPDE-poly[(dG-dC)*(dG-dC)] adduct (the LD spectrum has the same shape as the absorption spectrum) is consistent with the solvent exposure of site I1 adducts (5, 7). Characteristics of BPDE-Poly[ (dG).(dC)] and BPDE-Poly[ (dG-dC).(dG-dC)] Adducts. The absorption and linear dichroism spectra of covalent adducts derived from the binding of (+)-anti-BPDE and (-)-antiBPDE to the poly[(dG)-(dC)]homopolymer are shown in Figure 2. The (-)-adduct displays an absorption spectrum with broad maxima at 338 and 353 nm (Figure 2A); the linear dichroism spectrum is negative in sign throughout the 310-380-nm wavelength region and resembles an inverted absorption spectrum (Figure 2C). These characteristics are typical of site I adducts. The (+)-adducts exhibit somewhat sharper maxima at 330 and 345 nm (Figure 2B). The linear dichroism spectrum is predominantly positive, with corresponding maxima at 329 and 345 nm (Figure 2D), and is thus predominantly of the site I1 type. However, a small negative LD signal with a minimum at about 360 nm is also evident (it disappears upon UV illumination), which suggests that some site I adducts are also present (7). The characteristics of the (-)-anti-BPDE- and (+)anti-BPDE-poly[(dG-dC)-(dG-dC))] adducts are similar to those obtained with the homopolymer poly[(dG).(dC)]. The only prominent difference is in the shape of the LD spectrum of the (-)-adduct (Figure 3 0 . There is a sharp valley between the negative LD minima at 338 and 353 nm; this suggests a more prominent positive contribution of site I1 conformations to the overall LD spectrum in the (-1-anti-BPDE-poly[(dG-dC)*(dG-dC)] adducts.
1.0
320 340 360 300
320
340
360 300
Wavelength, n m Figure 3. Absorption and linear dichroism spectra of (+)anti-BPDE and (-)-anti-BPDE bound covalently to poly[(dGdC)-(dG-dC)].(A) and (C): (-)-adduct, [N]= 0.11 mM, [ B P D b ] = 1.6 pM. (B) and (D): (+)-adduct,[N] = 0.09 mM, [BPDEcB] = 0.45 pM. Solid lines, experimentalspectra; dotted lines, simulated spectra. 7
4
0
.
0
3
0.01
'..
Wavelength, nm Figure 4. Absorption and linear dichroism spectra of (+)anti-BPDE and (-)-anti-BPDE bound covalently to poly[(dTdC).(dG-dA)]. (A) and (C): (-)-adduct, [N] = 0.14 mM, [BPDEcB]= 1.5 pM. (B)and (D): [N] = 0.17 mM, [BPDEcB] = 0.75 pM. Solid lines, experimental spectra; dotted lines, sim-
ulated spectra.
Characteristics of BPDE-Poly[ (dA-dC)*(dG-dT)] and BPDE-Poly[ (dT-dC)-(dG-dA)]Adducts. In the case of the nonalternating purine-pyrimidine copolymer poly[ (dT-dC).(dG-dA)],the (-)-enantiomer seems to give rise predominantly to site I adducts and the spectroscopic characteristics (Figure 4A,C) are similar to those of the nonalternating homopolymer (-)-anti-BPDE-poly[ (dG). (dC)] adduct (Figure 2A,C). In the case of the (+)-antiBPDE-poly[(dT-dC).(dG-dA)] adducts, both the absortion (Figure 4B)and the LD spectra (Figure 4D) suggest that
Chem. Res. Toxicol., Vol. 4, No. 3, 1991 I
I
I
'
I
I
'
1
'
1
'
1
I
B
0.30-A
I
(-1-
I
C
1
I
Roche et al.
'
I
BPDE
'
I
'
(+)-BPDE
I
ID
1
'
1
'
1
'
I
I
I
1
2
3
4
5
-o.30
1
I
A
+0.20
0.20
+0.10
+0.10
S
0
0
10
20
30 Time lminl
40
50
60
Figure 6. HPLC profiles of enzyme digests derived from (+)-
anti-BPDE covalently bound to poly[ (dT-dC).(dG-dA)]and poly[(dA-dC).(dG-dT)].Solid line, detection via absorbance at 280 nm; dotted line, fluorescence detection (see text). Wavelength, nm
Figure 5. Absorption and linear dichroism spectra of (+)anti-BPDE and (-)-anti-BPDEbound covalently to poly[(dAdC).(dGdT)I. (A) and (C): (-)-adduct, [Nl = 0.27 mM, [ B P D b ] = 6.9 rM. (B)and (D):(+)-adduct,[N] = 0.3 mM,[BPDEcB] = 6.9 rM. Solid lines, experimental spectra; dotted lines, simu-
lated spectra.
there is a much more significant contribution of site I adducts than in the case of the (+)-anti-BPDE-poly[(dG).(dC)] adducts. This is especially evident from the prominent negative LD spectrum with a minimum at 353 nm (Figure 4D). Adducts of (-)-anti-BPDE with the alternating purine-pyrimidine copolymer poly[(dA-dC).(dG-dT)] are characterized by site I type absorption spectra since the broad absorption maxima are located at 337 and 352 nm (Figure 5A). However, within the indicated noise level, there is no LD signals (Figure 5C). The (+)-anti-BPDE-poly [ (dA-dC)-(dG-dT)]adducts are clearly heterogeneous in nature. Site I adducts appear to be dominant since broad and weakly discernible absorption maxima appear near 335 and 352 nm (Figure 5B), while the LD minima are located at 338 and 358 nm (Figure 5D). HPLC Analysis of Covalent Adducts. The higher incidence of site I conformations in the (+)-anti-BPDEpoly[(dA-dC).(dG-dT)]adducts might, in principle, be due to differences in the chemical nature of the adducts. We have therefore compared the adduct distributions in (+)-anti-BPDE-poly[(dA-dC).(dG-dT)] and (+)-antiBPDE-poly[ (dT-dC).(dG-dA)]adducts. The HPLC profiles of the enzyme digests are shown in Figure 6. Both the absorbance and the fluorescence signals are shown, but the absorbance provides a better indication of the relative quantities of the adducts than does the fluorescence. The elution profiles for the two different polymers are quite similar. The major peaks (3 in Figure 6) in both cases coelute with the trans-(+)-BPDE-ZP-dG adduct standard (24). Peaks 1 and 2 are due to tetraols, while peaks 5 and 6 are trans- and cis-BPDE-dA adducts, respectively (25). Adduct 4 might be a deoxycytidine adduct, but its relative abundance is small in these two polymers. Because of the low binding ratios, and the cost of the polymers and of (-)-anti-BPDE, the HPLC profiles of the corresponding covalent adducts derived from this enantiomer were not examined. Average Orientation Angles of Pyrenyl Residues in Different Base Sequences. The angle of orientation
6 between a given transition moment and the reference axis in the LD experiment (in this case the flow direction within the Couette cell) is given by LD(X)= ( 3 / 2 ) ~ ( ~ )cos2 ( 3 e - I)F(G) (2) where LD(h) and A(h) are the magnitudes of the linear dichroism and absorbances at a given wavelength h and 0 I F(G) I1.0 is a factor that expresses the degree of alignment of the polynucleotide in the flow field of the Couette cell. The average orientation of the long axes of the pyrenyl residues relative to the average orientations of the DNA bases can be estimated (5) by comparing the reduced dichroism LD/A due to the DNA bases (at their absorption maxima) and the BPDE residue absorption band (345-360 nm): - 3 cos2 OBpDE - 1 = 1- 3 cos2 dBPDE A' = (LD/A)BPDE (LD/A)DNA 3 COS^ eDNA- 1 (3) where the angle 8DNA = go', Le., perpendicular alignment of the DNA bases with respect to the flow direction, is assumed (5). The range of possible values for A'are -2.0 IA ' S +1.0; A ' = 1.0 for intercalative conformations (A' > 0 for site I conformations in general, corresponding to eBpDE between -55' and goo), while for site I1 conformations A' < 0 and 0 < 55' ( 5 ) . The experimental values of A', and the apparent orientation angles eBpDE deduced from these A'values, are summarized in Table I. In poly[ (dG-dC).(dG-dC)] and poly[(dG).(dC)], the site I1 binding of (+)-anti-BPDE is characterized by angles of 39-43'; these values are close to those determined for (+)-anti-BPDE-DNA adducts (5, 27) but somewhat larger than the value of 26' estimated for (+)-anti-BPDE-poly[ (dG-dC).(dG-dC)] adducts by Eriksson et al. (26). The site I adducts are oriented with apparent angles of 57-64' in the alternating and nonalternating dG-dC polynucleotides; similar results were obtained with native DNA (5). Analysis of Absorption a n d Linear Dichroism Spectra i n Terms of Linear Combinations of Site I and Site 11 Adduct Spectra. How well can the experimentally observed LD and absorption spectra be represented as sums of different proportions of the site I and site I1 model absorption spectra shown in Figure l? Simulated absorption and LD spectra can be constructed by simply summing different proportions of the site I and site I1 absorption spectra.
Chem. Res. Toxicol., Vol. 4, No. 3, 1991 315
Conformations of BPDE-DNA Covalent Adducts Table I. Estimated Average Orientation Angles and Site I/Site I1 Distributions (Ratios of Absorbances Evaluated at the Site I/Site I1 Absorption Maxima) of BPDE Bound Covalently to Different Polynucleotides’ polynucleotide A’ BnPna, deg site I’ site 11’ X (dG)-(dC) (+)
(-1
-0.61b 0.44‘
43 64
0.12 0.90
0.88 0.10
-0.022 -0.015
-0.806
39 57
0.10 0.76
0.90 0.24
-0.022
0.10d -0.216 0.04d
51 4 5
0.30 =l.O
0.70
-0.0056
-0.056 no LD
=55
0.70 ==LO
0.30
-0.014
(dG-dC).(dG-dC) (+) (-)
-0.0060
Calculations of LD(X) and A(X) according to eqs 4 and 5, using the constants K and K’as adjustable scale factors, require the use of only two unknown parameters ( X and f I or fII, since f I + fII = 1.0). The same values of f I and f n were used to construct the LD and absorbance spect,ra in each case. Only fairly restrictive sets of fI and X values are found to approximate the LD and absorbance spectra simultaneously; these values are listed in Table I. The corresponding simulated LD and absorbance spectra are shown in Figures 2-5 (dotted lines).
(dT-dC).(dG-dA) (+)
(-1
(dA-dC)*(dG-dT)
(+I (-)
“All A‘ values within *lo%; the corresponding OBDpe values have been evaluated for the central A’ value. *Evaluated at 346 nm. CEvaluatedat 350 nm. dEvaluated at 355 nm. (Fractional absorbances, f I and fiI (eq 6a,b).
The LD signal for a given adduct species is proportional to the absorbance (eq 2). Therefore, the contributions of the individual site I and site I1 components to the overall linear dichroism and absorbance spectra at any wavelength are (4) and
The variables t’I(A) and dII(X) are the molar extinction coefficients of site I and site I1 normalized to unity at the absorption maxima of 354 and 346 nm, respectively; the wavelength dependences of these quantities are depicted in Figure 1. The terms K and K’ are wavelength-independent scaling factors with K = (3/2)(3cos2On - l)F(G)K’ + tn(346)C~The actual concentrations and K’= CI(~U)CI of chromophores with site I or site I1 conformations are denoted by CIand CII, respectively; the actual molar extinction coefficients at the 354- and 346-nm site I and site I1 absorption maxima are denoted by q(354) and en(346), respectively. The fractional absorbances, evaluated at the absorption maxima of the site I and site I1 components, are f~= ~ I ( ~ ~ ~ ) C I I / K ’ (6b) f 1 1 = d346)C11/K’ (6b) The factor X reflects the ratio of orientation factors for sites I and I1 and is therefore always negative in sign: 3 COS^ eI - 1 X = 3 cos2 eII - 1 (7) The f~and fII ratios would be equal to the fractions of BPDE residues at site I and site I1 only if their molar extinction coefficients were equal. For noncovalently bound BPDE molecules, the molar extinction coefficient at 354 nm is 15OOO f 1OOO M-’, while for covalently bound BPDE residues qI = 29000 M-’ cm-’ (unpublished observations). The value of q(354) for covalently bound residues is not known exactly; however, because of hypochromicity effects due to significant site I pyrenyl-DNA base interactions, it is reasonable to assume that q(354) I ~ ~ ( 3 4 6Therefore, ). the concentrations of bound pyrenyl residues provided in the legends to Figures 2-5 tend to be lower estimates because t values of =29 OOO M-’ cm-’ were used to estimate the extent of binding (5).
Discusslon Base Sequence Dependence of Site I/Site I1 Adduct Distributions. Before discussing the data shown in Figures 2-5, a few points should be made concerning the fits of both the simulated LD and absorption spectra to the relevant experimental curves. Unordered adducts, if present, contribute to the absorbance but not to the linear dichroism spectra and thus lead to smaller experimentally observed A’values. The LD signal and A’values are also equal to zero when the orientation angle is -55’; therefore, when A’ is close to zero, the possibility that all of the adduct species are fortuitously oriented at the magic angle of =55O cannot be ruled out; however, a random orientation seems to be a more ikely explanation whenever the LD signal is small or equal to zero. In general, good fits of the simulated LD spectra, but poor fits of the simulated absorption spectra to the experimental data, reveal the presence of poorly oriented or unordered adduct species. Poly[ (dG).(dC)]. Since the differences between the experimental and simulated LD and absorbance curves are relatively modest, the (-)-anti-BPDE-poly [ (dG)-(dC)] adducts can be closely approximated by the site I/site I1 model; unordered adduct forms, if present, constitute a minor fraction of all adduct forms. In the case of the (+)-anti-BPDE adducts, there is considerably more absorbance beyond 350 nm than is suggested by the LD spectrum; because of the large red shifts, these additional long-wavelength-absorbingforms probably involve pyrenyl chromophores with significant base-stacking interactions. I t was not possible to simultanously fit the negative LD portion of the spectrum near 360 nm and the trough near 337 nm in Figure 2D by using the same sets of f I and fII parameters. This suggests that the absorption spectra of the site I and site I1 forms in this polymer are somewhat broader than the model spectra shown in Figure 1. The fractions of site I1 absorbances (fII) obtained from these fits are 0.88, while in the case of the (-)-adducts, the site I conformations are more abundant (fI = 0.90,Table I). Because 4354) It(346), the fractions of adducts with site I conformations, CI/(CI + CII), 1 f i . Therefore, the fII values represent upper limits of the fraction of adducts with site I1 conformations. Poly[ (dG-dC).(dG-dC)]. In the (-)-enantiomer case, the simulated LD and absorbance spectra deviate quantitatively from the experimental spectra. There is little structure in the observed absorbance spectrum, although a relatively sharp maximum occurs in the simulated absorption spectrum at 353 nm. This result indicates that the site I and site I1 absorbance spectra in this polymer are also somewhat broader than the model spectra (Figure 1). The agreement between the simulated and experimental LD spectra are fairly reasonable (Figure 3D), while the deviations in the absorbance spectra are significant (Figure 3B). The magnitude of the experimental LD signal beyond 350 nm is much smaller than suggested from the rather prominent absorbances in this long-wavelength region, indicating that unordered long-wavelength-ab-
Roche et al.
316 Chem. Res. Toxicol., Vol. 4, No.3, 1991
sorbing adduct forms are present. The site II/site I absorbance ratios (eq 6) are 91 in the case of the (+)-adduct and about 1:3 in the case of the (-)-adduct. On the basis of a different approach, Eriksson et al. (26) also concluded that (+)-anti-BPDE-poly[ (dG-dC).(dG-dC)] adducts are heterogeneous, with an estimated 80% site I1 type and 20% site I type adducts. Poly[ (dT-dC)-(dG-dA)].The (-)-anti-BPDE adducts can be approximated by a homogeneous site I distribution. The experimental absorbance spectrum is broader than the simulated one, while the reverse appears to apply in the case of the LD spectra (Figure 4, panels A and C, respectively). In the case of the (+)-adducts, the agreement between the simulated and experimental LD spectra is quite good (Figure 4D), while the experimental absorbance spectrum is significantly broader than the simulated one, especially in the long-wavelength region (>350 nm). The ratio of oriented site II/site I (+)-BPDE adduct absorbances is only 2.3:l. Poly[(dA-dC)-(dG-dT)].There is no LD signal in the case of the (-)-adduct. In the case of the (+)-adduct, the simulated spectra shown in Figure 5B,D were obtained with a site II/site I adduct absorbance ratio of only 0.4. This is the only polynucleotide (or DNA) that we have studied until now in which the (+)-anti-BPDE isomer appears to give rise to more site I than site I1 adducts. Our HPLC data show that this effect cannot be attributed to a different chemical adduct distribution and therefore must be due to conformational differences. Influence of dA.dT Base Pairs on Apparent Degrees of Adduct Orientations. When the experimentally observed A’values tend to be large, it can be surmised that the presence of poorly oriented adducts tends to be small; low values of A’ can provide indications that a sizable fraction of the adducts are poorly or not at all oriented. In all polymers, the magnitudes of A’are higher in the case of adducts derived from (+)-BPDE than those obtained from (-)-BPDE. Furthermore, the degree of orientation of the pyrenyl residues, as judged from the values of A’, is highest in poly[(dG).(dC)] and decreases in the order poly[ (dG-dC)*(dG-dC)]> poly[(dT-dC)*(dG-dA)]> poly[ (dA-dC).(dG-dT)]. Furthermore, adducts derived from the binding of both anti-BPDE enantiomers to the nonalternating purine-pyrimidine copolymers are somewhat better oriented than in the alternating purine-pyrimidine copolymers (Table I). In general, the presence of dA.dT base pairs is accompanied by lower apparent degrees of orientation of pyrenyl residues; for example, A’ values of zero are observed in adducts derived from the binding of (+)-anti-BPDE to the polynucleotides poly[ (dA-dT).(dA-dT)] and poly[ (dA). (dT)] (27), in which the BPDE residues are bound to the N6-position of adenosine (4, 25). In polynucleotides in which dA.dT base pairs are present adjacent to the covalent BPDE-W-dG lesions, the apparent degrees of orientation of the pyrenyl residues are significantly lower than in poly[ (dG)-(dC)]and poly[ (dG-dC)-(dG-dC)](Table I). The apparent lower degrees of orientation of BPDE residues covalently bound to dG residues with adjacent dA.dT base pairs may be due to the higher apparent flexibilities, decreased stiffness, and lower torsional rigidities of polymers rich in dA-dT base pairs (28-30). Furthermore, the base-pair opening or “breathing” rates are about 10 times greater for dA.dT than for dG.dC base pairs (31). All of these factors suggest that the BPDEM-dG adducts reside in less rigid microenvironments when they are located in the vicinities of dA-dT base pairs, thus accounting for the lower apparent degrees of orientations
in the mixed copolymers poly[ (dT-dC).(dG-dA)] and poly[ (dA-dC)*(dG-dT)]. Sequence Dependence of Adduct Conformations and Mutational Hot Spots. It is interesting to note that binding and mutational hot spots occur in sequences that are rich in dG.dC base pairs, especially those containing runs of dG’s (16-18). In alternating and nonalternating dG.dC base pair sequences, the highly mutagenic (+)anti-BPDE gives rise to well-oriented and well-defined site I1 adducts; the formation of these types of adducts is more closely correlated with the tumorigenic activities of different PAH diol epoxides than the formation of adducts with site I conformations (9,12). When dA-dT base pairs surround the BPDE-dG adducts, the average degree of orientation (as judged from the lowering of the reduced linear dichroism values) is decreased, and a higher proportion of site I adducts and unordered species is apparent, particularly in poly[ (dA-dC).(dG-dT)]. One might speculate that BPDE lesions in runs of guanines may hinder the appropriate interactions of these DNA segments with repair enzymes and polymerases because of the relative rigidities of such sequences. When neighboring dA-dT base pairs are present, the sequences containing the BPDEmodified dG residue may be less stiff than those occurring in runs of (dG).(dC) sequences, thus allowing for better interactions with various enzymes, perhaps leading to more efficient repair and fewer errors during replication. In principle, these hypotheses could be tested by monitoring the repair kinetics of PBDE adducts in different sequences, and by designing appropriate site-directed mutagenesis experiments (32).
Summary and Conclusions The covalent binding of the two BPDE enantiomers to synthetic polynucleotides tends to be heterogeneous, thus confirming some previous observations by Chen (33). The site I/site I1 model, in all cases, provides a better quantitative approximation of the linear dichroism spectra than of the absorption spectra; this is attributed to the presence of unordered or poorly oriented adducts, which are more abundant in polymers containing dA-dT base pairs. A detailed analysis of the absorption and linear dichroism spectra indicates that, in the alternating and nonalternating dG.dC polymers, the orientation of pyrenyl residues relative to the planes of the bases is better defined than in polynucleotides containing dA.dT base pairs. Adjacent dA.dT base pairs lower the apparent degrees of orientations. The highly mutagenic and tumorigenic (+)-antiBPDE isomer perferentially gives rise to site I1 adduct conformations in both the alternating (dG-dC).(dG-dC) and nonalternating (dG).(dC) sequences. These results lead to the hypothesis that the occurrence of mutation hot spots in sequences of G’s may be correlated with the relative stiffness and lower degrees of lateral and/or torsional flexibilities of such sequences relative to those which are rich in dA.dT base pairs. Acknowledgment. This work was supported by the Office of Health and Environmental Research, Department of Energy, Grant DE-FG02-88ER60674. Partial support by the U.S. Public Health Service, Grant CA 20851, awarded by The National Cancer Institute is also acknowledged. The Radiation and Solid State Laboratory at NYU is supported by the Department of Energy, Grant DE-FG02-86ER60405. References (1) Conney, A. H. (1982) Induction of microsomal enzymes by for-
eign chemicals and carcinogenesis by polycyclic aromatic hydrocarbons. Cancer Res. 42, 4875-4917.
Conformations of BPDE-DNA Covalent Adducts (2) Brookes, P., and Osborne, M. R. (1982) Mutation in mammalian cells by stereoisomers of anti-benzo[a]pyrene diol epoxides in relation to the extent and nature of the DNA reaction products. Carcinogenesis 3,1223-1226. (3) Stevens, C. W., Bouck, N., Burgess, J. A., and Fahl, W. E. (1985) Benzo[a]pyrene diol epoxides: different mutagenic efficiency in human and bacterial cells. Mutat. Res. 152, 5-14. (4) Cheng, S. C., Hilton, B. D., Roman, J. M., and Dipple, A. (1989) DNA adducts from carcinogenic and noncarcinogenic enantiomers of benzo[a]pyrene dihydrodiol epoxide. Chem. Res. Toxicol. 2, 334-340. (5) Geacintov, N. E., Ibanez, V., Gagliano, A. G., Jacobs, S. A., and Harvey, R. G. (1984) Stereoselective covalent binding of antibenzo[a]pyrene diol epoxide to DNA. Conformation of enantiomer adducts. J. Biomol. Struct. Dyn. 1, 1473-1484. (6) Roche, C. J., Geacintov, N. E., Ibanez, V., and Harvey, R. G. (1989) Linear dichroism properties and orientations of different ultraviolet transition moments of benzo[a]pyrene derivatives bound noncovalently and covalently to DNA. Biophys. Chem. 33, 277-288. (7) Zinger, D., Geacintov, N. E., and Harvey, R. G. (1987) Conformations and selective photodissociation of heterogeneous benzo[alpyrene diol epoxide enantiomer-DNA adducts. Biophys. Chem. 27, 131-137. (8) Jernstrom, B., Lycksell, P.-O., Griislund, A., and NordBn, B. (1984) Spectroscopic studies of DNA complexes formed after reaction with anti-benzo[a]pyrene-7,&dihydrodiol-9,lO-oxide enantiomers of different carcinogenic potencies. Carcinogenesis 5, 1129-1 135. (9) Harvey, R. G., and Geacintov, N. E. (1988) Intercalation and binding of carcinogenic hydrocarbon metabolites to nucleic acids. Acc. Chem. Res. 21, 66-73. (10) Grblund, A., and Jernstrom, B. (1989) DNA-carcinogen interaction: covalent DNA-adducts of benzo[a]pyrene 7,8-dihydrodiol9,lO-epoxides studied by biochemical and biophysical techniques. Q.Reo. Biophys. 22, 1-37. (11) Kolubayev, V., Brenner, H. C., and Geacintov, N. E. (1987) Stereoselective covalent binding of enantiomers of anti-benzo[alpyrene diol epoxide to DNA as probed by optical detection of magnetic resonance. Biochemistry 26, 2638-2641. (12) Geacintov, N. E. (1988) Mechanisms of reaction of polycyclic aromatic epoxide derivatives with nucleic acids. In Polycyclic Aromatic Hydrocarbon Carcinogenesis: Structure-Activity Relationships (Yang, S. K., and Silverman, B. D., Eds.) Vol. 11, pp 181-206, CRC Press, Boca Raton, FL. (13) Geacintov, N. E., Gagliano, A. G., Ibanez, V., and Harvey, R. G. (1982) Spectroscopic characterizations and comparisons of the structures of the covalent adducts derived from the reactions of 7,8-dihydroxy-7,8,9,10-tetrahydrobenzo[a]pyrene-9,10-oxide, and the 9,lO-epoxides of 7,8,9,10-tetrahydrobenzo[a]pyreneand 9,10,11,12-tetrahydrobenzo[e]pyrenewith DNA. Carcinogenesis 3, 247-253. (14) Boles, T. C., and Hogan, M. E. (1986) High-resolution mapping of carcinogen binding sites on DNA. Biochemistry 25,3039-3043. (15) Kootstra, A., Lew, L. K., Nairn, R. S., and MacLeod, M. C. (1989) Preferential modification of GC boxes by benzo[a]pyrene7,8-diol-9,10-epoxide. Mol. Carcinog. 1, 239-244. (16) Yang, J.-L., Maher, V. M., and McCormick, J. J. (1987) Kinds of mutations formed when a shuttle vector containing adducts of (f)-7~,8a-dihydroxy-9a,10ol-epoxy-7,8,9,lO-tetrahydrobenzo[a]-
pyrene replicates in human cells. Proc. Natl. Acad. Sci. U.S.A. 84, 3787-3791. (17) Carothers, A. M., and Grunberger, D. (1990)DNA base changes
Chem. Res. Toxicol., Vol. 4, No. 3, 1991 317 in benzo[a]pyrene diol epoxide-induced dihydrofolate reductase mutants of Chinese hamster ovary cells. Carcinogenesis 11, 189-192. (18) Reardon, D. B., Bigger, A. C. H., Strandberg, J., Yagi, H., Jerina, D. M., and Dipple, A. (1989) Sequence selectivity of optically active hydrocarbon dihydrodiol epoxides with Rat H-ras DNA. Chem. Res. Toxicol. 2, 12-14. (19) Wilson, W. D., Wang, Y.-H., Krishnamoorthy, C. R., and Smith, J. C. (1986) Intercalators as probes of DNA conformation: propidium binding to alternating and non-alternating polymers containing guanine. Chem.-Biol. Interact. 58, 41-56. (20) Wells, R. D., Larsen, J. E., Grant, R. C., Shortle, B. E., and Cantor, C. R. (1970) Physicochemical studies on polydeoxyribonucleotides containing defined repeating nucleotide sequences. J. Mol. Biol. 54, 465-497. (21) Geacintov, N. E., Ibanez, V., Rougee, M., and Bensasson, R. V. (1987) Orientation and linear dichroism characteristics of porphyrin-DNA complexes. Biochemistry 26, 3087-3092. (22) Jeffrey, A. M. (1985) Polycyclic Aromatic Hydrocarbon-DNA Adducts. Formation, Detection, and Characterization. In Polycyclic Hydrocarbons and Carcinogenesis (Harvey, R. G., Ed.) ACS Symposium Series 283, pp 187-208, American Chemical Society, Washington, DC. (23) Panthananickal, A., and Marnett, L. J. (1981) Comparison of commercial reverse-phase high-performance liquid chromatography columns for the separation of benzo[a]pyrene diol epoxidenucleic acid adducts. J. Chromatogr. 206, 253-265. (24) Jeffrey, A. M., Weinstein, I. B., Jennette, K. W., Grzeskowiak, K., Nakanishi, K., Harvey, R. G., Autrup, H., and Harris, C. (1977) Structures of benzo[a]pyrene-nucleic acid adducts formed in human and bovine bronchial explants. Nature (London) 269, 348-350. (25) Jeffrey, A. M., Gneskowiak, K., Weinstein, I. B., Nakanishi, K., Roller, P., and Harvey, R. G. (1979) Benzo[a]pyrene-7,8-dihydrodiol9,lO-epoxide adenosine and deoxyadenosine adducts: structure and stereochemistry. Science 206, 1309-1311. (26) Erikssen, M., NordBn, B., JernstrBm, B., and Griislund, A. (1988) Binding geometries of benzo[a]pyrene diol epoxide isomers covalently bound to DNA. Orientational distribution. Biochemistry 27, 1213-1221. (27) Roche, C. J. (1987) The Physical and Covalent Interactions of Polycyclic Aromatic Hydrocarbon Epoxide Derivatives with Nucleic Acids. Dissertation, New York University. (28) Hogan, M. E., and Austin, R. H. (1987) Importance of DNA stiffness in protein-DNA binding specificity. Nature (London) 329, 263-266. (29) Sarai, A., Mazur, J., Nussinov, R., and Jernigan, R. L. (1989) Sequence dependence of DNA conformational flexibility. Biochemistry 28, 7842-7849. (30) Hard, T., Fan, P., Magde, D., and Kearns, D. R. (1989) On the flexibility of DNA: time-resolved fluorescence polarization of intercalated quinacrine and 9-amino-6-chloro-2-methoxyacridine. J. Phys. Chem. 93,4338-4345. (31) Kochoyan, M., Leroy, J. L., and GuBron, M. (1987) Proton exchange and base pair lifetimes in a deoxy-duplex containing a purine-pyrimidine step and in the duplex of inverse sequence. J. Mol. Biol. 196, 599-609. (32) Basu, A. K., and Essigman, J. M. (1988) Site-specifically modified oligonucleotides as probes for the structural and biological effects of DNA-damaging agents. Chem. Res. Toxicol. 1, 1-18. (33) Chen, F.-M. (1986) Binding of enantiomers of trans-7,a-dihydroxy-9,10-epoxy-7,8,S,lO-tetrahydro-benzo[a]pyrene to polynucleotides. J. Biomol. Struct. Dyn. 3, 401-418.
