J. Phys. Chem. 1992,96, 1983-1989 results seem to suggest that the e x p l a n a t i o r ~ ~of~the - ~ ~observed -~~ temperature effects may not require extensive conformational changes of the Ej heads with temperature. Instead, the explanation may be associated with changes in the extent of interaction of the Ej heads with water (for example, through hydration) upon increasing temperature at a relatively fixed average Ej conformation.25s26J0In order to verify this possibility, careful measurements of conformational characteristics of PEO chains in water as a function of temperature need to be conducted. Within the context of our approach, our main conclusion is that polymer scaling laws are applicable within an uncertainty of a few percent for fairly short isolated PEO chains in water. Specifically, the shortest free chain that satisfies scaling with a 2% accuracy has 12 bonds (four ethylene oxide units), while the shortest attached chain has 30 bonds (10 ethylene oxide units). For more details, see section I11 and Table 11. Our results suggest that some systems containing relatively short PEO chains, suspected of being only marginally long enough to be described using polymer science concepts,l may, in fact, be described in that way to a reasonable approximation. An important example are aqueous solutions of some of the CiEj nonionic surfactants containing a relatively short PEO block with j ethylene oxide units. If the surfactant is part of a micelle, the Ej head is often a short PEO chain attached to the interface between the (25) Mitchell, D. J.; Tiddy, J. T.; Waring, L.; Bostock, T.; McDonald, M.
P.J . Chem. Soc., Faraday Trans. I 1983,79,915 and references cited therein. (26) (27) 2053. (28) (29)
Nilsson, P. G.; Lindman, B. J. Phys. Chem. 1983, 87, 4756. Kjellander, R.; Florin, E. J. Chem. Soc., Faraday Trans. I 1981, 77, Goldstein, R. E. J. Phys. Chem. 1984, 80, 5340. K a r l s t r h , G. J . Phys. Chem. 1985, 89,4962.
1983
oillike core of the micelle and the aqueous solution. With an understanding that our conclusions are strictly valid for an isolated terminally attached PEO chain, it is nevertheless tempting to suggest that as long as the Ej head has 10 or more ethylene oxide units (that is, for j 1 lo), its conformational characteristics may be described reasonably well using polymer scaling laws. This conclusion could aid considerably in the development of micellization models aimed at exploiting polymer science concepts to describe the contribution of the E. polymeric heads to the free energy of micellization.'**Jo In addition, our results could also shed light on the interpretation of recent experiments aimed a t measuring forces3 between two surfaces coated with CiE, nonionic surfactants and immersed in aqueous solution, with the Ej heads exposed to the aqueous solvent. In some of these experiments4 values of j = 20, 40, and 50 were utilized, while in others5j = 5 . We suggest that in the former case polymer scaling laws may be applicable in the analysis of the experimental results.
Acknowledgment. This research was supported in part by the National Science Foundation (NSF) Presidential Young Investigator (PYI) Award to D.B. and an NSF Grant DMR-84-18718 administered by the Center for Materials Science and Engineering at MIT. D.B. is grateful for the support by the Texaco-Mangelsdorf Career Development Professorship at MIT. He is also grateful to the following companies for providing PYI matching funds: BASF, British Petroleum America, Exxon, Kodak, and Unilever. Some of the calculations in this paper were performed on the MIT Cray-2. Finally, the authors would like to acknowledge very useful discussions with Nicholas Abbott on polymer scaling laws. RWhy NO. PEO,25322-68-3.
Relative Abundance and Reactivity of Primary Ion Radicals in ?-Irradiated DNA at Low Temperatures. 2. Single- vs Double-Stranded DNA Mengyao Yan, David Becker, Stephen Summerfield, Paul Renke, and Michael D. Sevilla* Department of Chemistry, Oakland University, Rochester, Michigan 48309 (Received: August 20, 1991; In Final Form: October 18, 1991)
An electron spin resonance investigation of the free-radical distribution in y-irradiated frozen samples of single stranded DNA (ssDNA) and double stranded DNA (dsDNA) is reported. Computer analysis of the ESR spectra of y-irradiated ssDNA in D 2 0 at 100 K suggests a more uniform distribution of the radical ions on the DNA bases than found for dsDNA, with the approximate ssDNA composition: thymine anion, T'-(30-35%), cytosine anion, C+ (20-28%), guanine cation, G'+(26-28%), and adenine cation, A'+ (8-17%), with small amounts of purine anions or pyrimidine cations. Upon annealing y-irradiated dsDNA we find that the electron transfers from C'*- to T and the T'-formed protonates at C-6 to produce the 5,6-dihydrothymin-S-y1radical (TH', TD'). The rate of deuteration of T'-is shown to be 20-fold less than the rate of protonation. Electron transfer in ssDNA appears to be less facile than in dsDNA. Hole transfer from adenine to guanine in dsDNA is nearly complete at 100 K, but appears less complete in ssDNA at 100 K. The results suggest that DNA strandedness, which affects DNA conformation, base stacking and cross strand interactions, is important in determining the initial ion radical sites as well as the subsequent ion radical transfer and reactions. A calculationof the ion radical abundance as a function of base transfer, starting from a random distribution in a model DNA strand, suggests base pairing greatly augments hole and electron transfer through the DNA strand and that on average ion transfer over only ca. 2 bases is sufficient to produce the ion distribution found in ssDNA and dsDNA.
Introduction
Knowledge of the chemical nature and the distribution of the radiation induced damage sites in DNA is important to the understanding of the effect of radiation on biological systems. Recent investigations have taken a new interest in the ion radical composition in yirradiated DNA a t low temperatures.'-* Until ( I ) Sevilla, M. D.; Becker, D.; Yan, M.; Summerfield, S. R. J . Phys. Chem. 1991, 95, 3410.
0022-365419212096-1983$03.00/0
recently most investigations suggested that the thymine anion and guanine cation were the principal species found in y-irradiated (2) (a) Steenken, S. Chem. Reu. 1989,89,503. (b) Steenken, S. Absrracrs, 38th Meeting of the Radiation Research Society, 1990; p 55. (c) Steenken, S. Free Radical Res. Commun., in press. (3) Candeias, L. P.;Steenken, S. J . Am. Chem. SOC.1989, I l l , 1094. (4) (a) Bernhard, W. A. J. Phys. Chem. 1989, 93, 2187. (b) Bernhard, W. A. In The Early Effects of Radiation on DNA; Fielden, E. M.,and O'Neill, P., Eds.; Springer-Verlag: Berlin, 1991; pp 141-154.
0 1992 American Chemical Society
1984 The Journal of Physical Chemistry, Vol. 96, No. 4, 1992
DNA at low temperatures.E-10 The fact that the final reaction product found in DNA is mainly the 5,6dihydrothymin-S-yl radical, TH', was a major piece of evidence in support of the thymine anion as the likely anionic primary radical. However, Bernhard in work with oligonucleotides in aqueous glasses found that the cytosine anion was the dominant anion radical; he suggested cytosine anion may also be the dominant anion in DNA as well!-" In previous work the cytosine anion and thymine anion were difficult to distinguish in both DNA model systemsI2-l6and since the spectra of both are doublets of about equal coupling. Recently we found that the cytosine and thymine anion spectra were considerably more resolved and distinguishable in D 2 0 (vs H20) matrices and we concluded that cytosine anion is the dominant anion in y-irradiated DNA.' Recent ab initio calculations17and measurements of ion radical pK's2,3suggest that the cytosine anion is stabilized by protonation from its complementary base guanine, so that it is better designated as a neutral species, C-(H+)'. For simplicity we will use the symbol C'*-for both the anion and the protonated species. Similar arguments might suggest the guanine cation has deprotonated to its complementary base, although this is not as ~ l e a r . * - ~ J ~ In this work, anion radical spectra of C" and T'-derived from polynucleotides, and radical cation spectra of G'+and A'+ derived from mononucleotides from our past effort,' are used to analyze the relative radical ion composition of y-irradiated ssDNA and dsDNA at 100 K and higher temperatures. The TH' and TD' radical ESR spectra, isolated from DNA experiments in this work, are employed together with the ionic radical spectra to analyze the radical abundances of dsDNA and ssDNA on annealing. Through the use of a computer-assisted analysis, we find that at 100 K, the ion radicals are more uniformly distributed among the DNA bases in ssDNA than in dsDNA. Structures. The relevant primary and intermediate DNA radical structures are
(5) Graslund, A.; Ehrenberg, A.; Rupprecht, A.; Strom, G. Biochim. Biophys. Acta 1971, 254, 172. (6) Gregoli, S.; Olast, M.; Bertinchamps, A. Radiat. Res. 1982,89, 238. (7) Boon, P. J.; Cullis, P. M.; Symons, M. C. R. J . Chem. Soc., Perkin Trans. 2 1984, 1393. (8) (a) Graslund, A.; Ehrenberg, A.; Rupprecht, A. Int. J . Radiat. Biol. 1977, 31, 145. (b) Hiittermann, J.; Voit, K.; Oloff, H.; Kohnlein, W.; Grfislund, A.; Rupprecht, A. Faraday Discuss. Chem. Soc. 1984,No. 78,135. (c) Zell, I.; Hiittermann, J.; Grfislund, A.; Rupprecht, A,; Kohnlein, W. Free Radical Res. Commun. 1989, 6 , 105. (9) Sevilla, M. D.; D'Arcy, J. B.; Morehouse, K. W.; Englehardt, M. L. Photochem. Photobiol. 1978, 28, 37. (10) Graslund, A.; Ehrenberg, A.; Rupprecht, A.; Strom, G.;Crespi, H. Int. J . Radiat. Biol. 1975, 28, 313. (11) Barnes, J.; Bernhard, W. A.; Mercer, K. R. Radiat. Res. 1991, 126, 104. (12) Sevilla, M. D. In Excited States in Organic Chemistry and Biochemistry; Pullman, B., Goldblum, N., Eds.; D. Reidel: Dordrecht, Holland, 1977. (13) Greaoli. S.: Olast. M.: Bertinchamm A. Radiat. Res. 1979. 77.417. (14) Sevila, M. D.; Failor, R.; Clark, 5.;Holroyd, R. A,; Pettei, M. J . Phys. Chem. 1976,80, 353. (15) Cullis, P. M.; Podmore, I.; Lawson, M.; Symons, M. C. R.;Dalgarno, B.; McClymont, J. J . Chem. Soc., Chem. Commun. 1989, 1003. (1 6) Symons, M. C. R.Int. J . Radiat. Biol. 1990, 58, 93. (17) Colson, A.-0.; Besler, B.; Close, D. M.; Sevilla, M. J . Phys. Chem. 1992, 96, 661.
Yan et al.
V
V
Figure 1. The spectra of the individual ion radicals used in reconstructions of experimental spectra: (A) C" from polydCpolydG; (B)T' from poly[dAdT].poly[dAdT]; (C) G'+ from d G M P in 8 M LiCl; (D) A'+ from dAMP in 8 M Lick (E)G'- from dGMP in 10 M LiCI. The small wing components are due to small amounts a hydrogen addition radical (GH,') which does not interfere with our analysis. (F) A'- from dAMP in 10 M LiC1. All spectra in this figure and subsequent figures were recorded at 100 K and 6.3 pW power unless otherwise noted. The three markers are 13.09 G apart and are centered at g = 2.0056.
Experimental Section Materials and Methods. Salmon testes DNA (type 111, sodium salt, base composition: 57.4% AT and 42.6% GC), deuterium oxide (D20, 99.996 atom % D), and DNA base nucleotides (2'-deoxyadenosine 5'-monophosphate, d A M P 2'-deoxycytidine 5'-monophosphate, d C M P 2'-deoxyguanosine 5'-monophosphate, d G M P and thymidine S'-monophosphate, T M P all sodium salts) were purchased from Sigma Chemical Co. and were used without further purification. Sample Preparation. Anoxic samples of ssDNA were prepared by addition of 10 mL of 99.9% D 2 0 to 200 mg of dsDNA. After the DNA was fully dissolved, it was denatured by heating to 98 OC for 10 min.'EJ9 The denatured DNA was immediately cooled, first to 0 OC by emersion in a stirred ice water bath, then to 77 K in liquid nitrogen. After lyophilization the DNA fluff was equilibrated with D 2 0 vapor at 100% humidity for 1-6 days at 4 OC or at room temperature under N2. The 4 OC equilibration results in a sample which is 9 1 wt % D20;the room temperature equilibration yields 55 wt % D20. The sample is prepared by rolling the DNA into an approximately 4 mm X 15 mm cylinder under nitrogen. These denatured DNA samples (ssDNA) likely contain regions of base pairing between strands. Anoxic samples of dsDNA at 100% D 2 0 relative humidity were prepared by addition of 5 mL of 99.9% D 2 0 to 200 mg of DNA. The fully hydrated DNA was lyophilized and then equilibrated with D 2 0 under a nitrogen atmosphere for periods of from 1 to 7 days and samples were prepared as described above. Anoxic samples of dsDNA in D 2 0 ice were prepared by addition of 5 mL of 99.9% D 2 0 to 200 mg of DNA followed by lyophilization of the DNA. For samples in which complete exchange was desired this process was performed a second time. After lyophilization 1 mL of 99.996% or 99.9% D 2 0 was added to the DNA under N2 atmosphere and a sample prepared by drawing the viscous fluid into a tube, freezing, and finally removing the ice sample. Samples of deoxynucleotides (dAMP, dGMP) in 10 M LiCl were prepared by addition of 5 mg of the deoxynucleotide to 1 mL of 10 M LiCl followed by cooling to 77 K.20 For an initial ESR spectrum the DNA samples were annealed to ca. 120-130 K for 1-2 min t o remove the signal from any trapped 'OH radical in ice. The 'OH radical appeared only in (1 8) Boyer, R. F. Modern Experimental Biochemistry; Addison-Wesley: Reading, MA, 1986; pp 437-444. (19) Lindahl, T.; Nyberg, B. Biochemistry 1972, 11, 3610. (20) Sevilla, M. D.; Mohan, P. In?. J . Radiat. Biol. 1974, 25, 635.
The Journal of Physical Chemistry, Vol. 96, No. 4, 1992 1985
?Irradiated DNA
TABLE I: Apparent Primary Ion Radical Distribution in y-Irradiated DNA at 100 K
wt % DzO"
sample
91 91
I I I1 I1
ssDNA
55 55 57 57
dsDNA
111 111
[mass DzO/(mass vacuum-dried DNA
% CY'20.8 25.2 25.0 28.2 42.1 40.3
+ mass DzO)] X 100.
% T34.2 32.2 31.9 31.0 16.3 16.7
% A'+
% G'-
16.7 9.5 15.2 7.8 3.1
0.0
R
76 A'-
b
b
5.4
0.9
b
b
4.9
0.0
b
b
0.0
6.7
0.054 0.045 0.055 0.045 0.048 0.040
bBasis spectrum not used in this analysis.
ice samples and no significant change in the DNA radical spectra occurred on annealing to 120-1 30 K. ESR spectra were recorded at 100 K after annealing the sample to the desired higher temperature for 4 min. All spectra were recorded at 45 dB to avoid power saturation.' Each spectrum was stored in a 1000-point array on a microcomputer with field calibration marks from the three Fremy's salt ESR lines (AN = 13.09 G, g = 2.0056). Computer Analysis. Spectra of the anion and cation radicals of each DNA base (C'*-, T-, Go+,A'+, Figure 1) were used in a computer linear least-squares reconstruction of the experimental spectrum which yielded the ionic composition at low temperatures. These spectra correspond to basis set I in our previous work.' In this basis set the cation radical spectra (G'+ and A*+) are the dGMP'+ and dAMP" nucleotide cations in 8 M LiCl/D20. The C'*- and T'- radical spectra were derived from experiments on irradiated poly[dG].poly[dC] and poly[dAdT].poly[dAdT], by subtracting the Go+or A'+ spectrum from the appropriate polynucleotide spectrum. The TD' radical spectrum was obtained by annealing DNA samples to high temperatures at which TD' is the major component (>90%). All spectra were recorded at 100 K and 6.3 pW (45 dB). The spectra of G*-, A'-, TD', and TH' also employed in our analyses are discussed later. The goodness-of-fit of the reconstructed spectrum to the experimental spectrum is given by R.4
R=
% G'+ 28.3 27.1 28.8 28.3 38.5 36.3
' CIYi,expl i
RWultS
The primary Ionic Radical Distribution in =DNA and &DNA at 100 K. In Figure 2A we present an overlay of the ESR spectra of ssDNA sample I (solid line) and dsDNA sample I11 (dotted line). Both DNA samples were equilibrated with D 2 0 and were y-irradiated to a dose of 0.5 Mrad. The doublet in ssDNA is less resolved and has a ca. 1.0 G narrower splitting than the doublet in the &DNA spectrum. The smaller splitting and lower resolution are expected from a substantialincrease in the fraction of thymine anion present. The analyses for the primary ion radical distribution in dsDNA and ssDNA are shown in Table I; computer reconstructions based on these parameters are shown in Figure 2B for dsDNA and 2C for ssDNA. Samples of ssDNA at lower hydration levels (sample 11) showed slightly less T'-and more C'*than more highly hydrated ssDNA (sample I). The analysis indicates that the anions are more evenly distributed between the pyrimidine bases and the cations between the purine bases in ssDNA than in dsDNA. These results suggest that electron transfer from thymine to cytosine and hole transfer from adenine to guanine is less efficient in ssDNA than in dsDNA. In both cases, anions are about 5 5 4 0 % of the radical population and cations are about 40-45%. Exbtence of tbe Radicals T",C', A-, or G' in =DNA. Since ion transfer between strands is limited in ssDNA, it is reasonable and C'+ may be to expect that species such as A*-, G*-, T+, present. The spectra of the purine anions A'- and G* are singlets20 similar to those of A" and G" but distinguishable by line width, line shape, and g value. Reconstructions which included adenine anion radical from dAMP (Figure 1E) and guanine anion radical from dGMP in 10 M LiCl (Figure 1F) as well as the four usual basis spectra were performed. The results in ssDNA suggest that presence of only small amounts of the purine anions, ca. 5% G'-
B
\,I
-
dsDNA
V
---
SSDNA sim
I / \ \I
I
V Figure 2. Comparison and analyses of the initial ESR spectra of =DNA and dsDNA (100% DzO RH) at 100 K: (A) Solid line, the ESR spectrum of ssDNA; dotted line, the ESR spectrum of dsDNA. (B) Solid line, the ESR spectrum of dsDNA; dotted line, the corresponding computer reconstruction from experimental spectra of the individual ions, G'+, and A'+. (C) Solid line, the ESR spectrum of ssDNA; C/O-, T-,
dotted line, the corresponding computer reconstruction using the same functions as in B. The values of the ion compositions are given in Table I. All DNA samples received 0.5 Mrad y-irradiation. and 1% A'- (Table I). There is an 8% decrease in the A'+ contribution when the purine anions are included in the set of basis spectra. For dsDNA, G- is rejected and about 6% A' is suggested (Table I). The small contributions of purine anions found in ssDNA are within the *5% uncertainty in the determined percentage compositions. The thymidine cation, T'+, produces a characteristic spectrum spread over 85 GZ1and it is not detectable in any of our DNA spectra. The cytosine cation, C*+, possesses a 15 G a-proton coupling, a 15 G &proton coupling, and smaller nitrogen couplings?2 A spectrum was simulated23using these cytosine cation singlecrystal parameters and was included in the analysis of both ssDNA and dsDNA. We found that it made no substantial contribution to either spectrum ( T > A > G. Thus, using C as the energetically stable anion site, only a 1.5 base jump (Le., N = 1.5) is required to produce the experimentally observed dsDNA distribution. Since electron transfer over greater distances than 1.5 bases is likely,27an energetic competition could exist between the pyrimidines for the electron in dsDNA, rather than cytosine alone being the low-energy trapping site. Our observed anion composition in ssDNA is close to a random distribution of the pyrimidine anions but does not include significant amounts of purine anions. This is consistent with a model in which the electron transfers from a purine to nearby pyrimidine, but in which little net transfer occurs from one pyrimidine to another in ssDNA. We should note that there are complexities to the analysis for relative ion abundances. There are likely regions of double strandedness in our ssDNA, which would increase our observed amounts of C'- and G'+;there may also be regions of single (27) (a) Gray, H. B.; Malmstrom, B. G. Biochemistry 1989.28, 7499. (b) McLendon, G . Arc. Chem. Res. 1988, 21, 160. (c) Isied, S. S.; Vassilian, A,; Wishart, J. F.; Creutz, C.;Schwarz, H. A.; Sutin, N. J . Am. Chem. Soc. 1988, 110, 635. (28) Orlov, V. M.; Smirnov, A. N.; Varshavsky, Y . M. TefrahedronLetf. 1976, 48, 4317.
Yan et al.
.-
H
o6 ,! cO.5' 1
"
2
"
3
"
4
"
"
5
6
N (Number of Bases)
Figure 8. Plot which shows the fraction of cytosines within N bases of a thymine in ssDNA and in dsDNA vs the distance from the thymine in bases. Since the probability for access to a thymine is lower in ssDNA
than dsDNA electron transfer between the pyrimidines is predicted to be more difficult in ssDNA. This is a consequence of the fact that the model allows for intrastrand transfer in ssDNA but allows both intraand interstrand transfer in dsDNA. strandedness in our dsDNA, which would increase the amount of T'-and A'+. It seems most reasonable to assume that our observed ion abundances reflect base ion transfer as modeled in our computer analysis and the presence of a fraction of the DNA of the incorrect strandedness in our samples. The large amount of TD'/TH' observed in our dsDNA samples after annealing is good evidence for electron transport from C'*to T. We can then ask, how many bases must the electron transfer to result in the near complete transfer from C to T we observe? An analysis, using on our DNA data base, of the fraction of cytosines having a thymine within N bases is shown in Figure 8. Note that 95%of C have a T within 2 bases in dsDNA. Clearly, long-range electron transfer is not necessary in dsDNA in order for the reductive path to lead to TH' radical. For ssDNA it is clear from the figure that electron transport from T to C is less facile than in dsDNA. Summary and Conclusions In this work we have found results that suggest that the initial ion radical distribution in irradiated DNA is sensitive to the structure of the DNA. In hydrated dsDNA, cytosine anion is the dominant anion' whereas in hydrated ssDNA a near-random distribution of the pyrimidine anions is found. These results can be explained by interstrand base pairing and base stacking effects. In ssDNA base pairing has been largely disrupted and base stacking has been partially disrupted, especially for adjacent pyrimidi11es.2~ This reduces charge transfer between DNA strands as well as along the DNA strand and as a consequence ion localization occurs in a more random fashion in ssDNA than in dsDNA. A similar stacking effect on charge transfer has been suggested in work on dinucleoside anions.I4 In addition, the mechanism which is believed to operate to localize the electron on cytosine, i.e., proton transfer to cytosine anion from its base pair, guanine, is thwarted in ssDNA because no complementary guanine is present. Bernhard in work with oligonucleotideanions in aqueous glasses also has been lead to suggest base pairing has a significant effect on the initial anion radical d i ~ t r i b u t i o n . ~ ~ Our modeling of the ion composition as a function of an N-base electron and hole transfer in both ds and =DNA shows that large differences in ion distribution can be expected for these two forms of DNA. The N-base transfer is to be distinguished from the initial localization of charge on DNA. The electron ejected by ionization can easily travel over a considerable number of bases before l o ~ a l i z a t i o n . * These ~ * ~ ~ 'hot electrons" may be trapped preferentially by certain DNA bases; however, we have shown that even with random trapping of electrons and random hole formation, subsequent charge transfer over 1-3 bases will yield the ion ~
~~~
(29) Gregoli, S.;Olast, M.: Bertinchamps, A. Int. J . Radiaf.Biol. 1970, 17, 39.
The Journal of Physical Chemistry, Vol. 96, No. 4, 1992 1989
7-Irradiated DNA distributions found in this work. Ab initio molecular orbital calculations predict that the electron affinity of the pyrimidines are about 1 eV more than the purines, with thymine about 0.2 eV higher than cytosine." However, in dsDNA the calculation^^^ as well as experiment2a3predict that C' anion readily protonates from its complementary base G. The calculations further predict that T'-formed in dsDNA shifts to cytosine as the C*- anion protonates.I7 The remaining T'-in hydrated dsDNA (1 5%) may reflect an energetic competition between C and T for the electron and/or small regions of single-stranded DNA in the double stranded DNA. The reactions of the initial anionic species as samples are annealed result in protonation only at thymine to produce the 5,6dihydro-5-thymyl radical, TH' (or TD' in DzO). The formation of TD'/TH' in dsDNA clearly supports a model for ion transport in the DNA. In dsDNA, the thymine anion begins as only 15% of the original radical intensity but TD'/TH' ends at 35% of the original radical intensity. Since considerable radical decay takes place during annealing, the fraction of anion which converts to TD'/TH' is more than 2.3 times the original thymine anion concentration. Thus cytosine anion must convert to thymine anion on annealing. Our modeling suggests that a ca. 2-base average migration of the anionic site is sufficient to transfer 95% of the C"- to T. This transfer is followed by irreversible protonation of T'- to form TH' or TD': C"- + y-+ TD'/TH'
In ssDNA we begin with 32% thymine anion but end with 25% of original radical concentration as THO; the smaller relative amount of TH' formed is in accord with the lower degree of ion transport in ssDNA expected from our modeling. On annealing, the cation decays by charge recombination and, probably, by conversion to other species; it has been suggested, for example, that G'+converts to sugar radicals which decay more quickly than they are f ~ r m e d . ~ ~ ? ~ ~ The formation of TH' in DNA has been well investigated, as it was the first DNA radical identified by ESR.32-37 In these (30) Cullis, P. M.; Symons, M. C. R.; Wren, B. W. J. Chem. Soc., Perkin Trans. 2 1985, 1819. (31 1 Cullis. P. M.: Svmons. M. C. R. In Mechanism o f D N A Damape and Repaik Simic,'M. G.,'G&smann, L., Upton, A. C., Eds.;>lenum: New'York, 1986; pp 29-37.
previous literature it was suggested that the proton that attached to T'-to form TH' in DNA (D20) originated from nonexchangeable sites on a s ~ g a r . ~ Our ~ " ~work using dsDNA in D 2 0 / H 2 0mixtures shows that protonation of T'-to form TH' is favored at least 20-fold over deuteration to form TD' at low temperatures. Our results suggest that only in fully exchanged DNA can the deuterated species be observed free from the dominant eight-line spectrum of TH' and that only exchangeable sites contribute to TH' formation. Approximately 50% of the mass of DNA is in the sugar phosphate backbone and about 50% of the ionizations occur there. It is, therefore, puzzling why sugar radicals have not yet been identified in irradiated DNA by ESR, especially since they have been observed in nucleotide single crystals.38q Electrons formed in the ionization of the sugars would be likely to migrate to the more electron-affinic bases; whether there is hole migration from the sugars to the bases is not as clear. We find an anion/cation ratio greater than 1, which is consistent with the presence on the sugar phosphate backbone of hole radicals whose spectra are too broad and poorly resolved to be observed. However, the uncertainties in our analyses cannot rule out the possibility that the hole formed on the sugar phosphate backbone transfers to the bases. Distinguishing between these two possibilities is a task for future efforts.
Acknowledgment. We thank the Office of Health and Environmental Research of the US.Department of Energy (Grant No. DEFG0286ER60455) and the National Cancer Institute of the National Institutes of Health (Grant No. ROlCA54524) for support of this research. The donors of the Petroleum Research Fund, administered by the American Chemical Society, and the National Science Foundation REU are gratefully acknowledged for support for undergraduate research students P.R. and S.S. (32) Ormerod, M. G. Int. J. Radiat. Biol. 1965, 9, 291. (33) Holroyd, R. A.; Glass, J. Int. J. Radiat. Biol. 1968, 14, 445. (34) Cook, J. B.; Wyard, S.J. Nature 1966, 219, 526. (35) Cook, J. B.; Wyard. S.J. Int. J . Radiat. Biol. 1966, 11, 357. (36) Salovey, R.; Shulman, R. G.; Walsh, W. M. J . Chem. Phys. 1963, 39, 839. (37) Ehrenberg, A.; Ehrenberg, L.; Loforth, G. Nature 1963, 200, 376. (38) Hole, E. 0.; Sagstuen, E. Radiat. Res. 1987, 109, 190. (39) Hole, E. 0.; Nelson, W. H.; Close, D. M.; Sagstuen, E. J. Chem. Phvs. 1987. 88. 5218. 140) Bemhard, W. A.; Close, D. M.; Mercer, K. R.; Corelli, J. C. Radiat. Res. 1976, 55, 19.
