Structural Dependence of Oligonucleotide Photooxidation STEVEN E. ROKITA, BERNARD MU, and LORRAINE ROMERO-FREDES

Department of Chemistry, State University of New York at Stony Brook, Stony Brook, New York 11794-3400

SY NOPSlS

Oxidative photosensitization was used to characterize the conformational-dependent reactivity of various structures fonned by oligonucleotides 14-15 nucleotides in length. The rate and product composition from a single hit process was analyzed using quantitative ion exchange chromatography under native and denaturing conditions. The primary damage incurred under aerobic acetone sensitization was base oxidation that, in turn, would induce strand scission upon a secondary treatment with piperidine. The reactive intermediates of this process were not consistent with diffusible radical species or singlet oxygen, as indicated by isotope and quenching studies. Derivatization was most likely initiated through a type I photoprocess with a direct interaction between DNA bases and excited state acetone preceding an irreversible oxidation step. This dominant reaction demonstrated no obvious sequence or site specificity for initial modification; the relative reactivity among the oligonucleotides did not correspond to any simple trend of base composition or near neighbor analysis. Likewise, the steric requirements of base modification allowed for similar rates of oxidation for single-strand, helical, and aberrant forms of DNA. Hybridization of the most reactive oligonucleotides, however, did suppress their relative single-strand v s double-strand reactivity by as much as fourfold.

INTRODUCTION Conformational studies on proteins and nucleic acids have developed from quite distinct perspectives due to the obvious differences in their chemistry, yet a few popular misconceptions have also guided their growth. Characterization of the secondary and tertiary structures of proteins has always been central to the study of polypeptides. Biological activity is rarely expressed by amino acid sequence alone; instead, function most often follows the formation of intra- and intermolecular associations that establish the higher orders of organization. Attention focused on DNA has been quite different. Once the standard double helix of DNA was identified, most polynucleotide studies centered on the linear array of bases that encode or regulate genetic information. Only recently have a

0 1990 John Wiley & Sons, Inc. CCC 0006-3525/90/010069-09 $04.00 Biopolymers, Vol. 29, 69-77 (1990)

number of laboratories demonstrated that the variability in DNA secondary structure rivals the diversity of polypeptide folding.', Both well-defined model systems and direct macromolecular analysis have been required to characterize the critical features of secondary conformations. For example, Kaiser and his colleagues utilized a range of synthetic peptides to demonstrate that secondary structural features often dictated the biological response to certain peptide hormones far better than that controlled by primary ~equence.4~ Inspired by this, our laboratory is developing an oligonucleotide-based system to identify the extent to which DNA reactivity is also governed by conformation rather than linear sequence. The structure and reactivity of mismatched, misaligned, and unusual duplex DNA is now the subject of intense scrutiny through the application of n ~ n r ,~~r y, s~t a l l o g r a p h yand ~ ~ ~chemical characteri~ation.~-"The motivation behind this rapidly expanding field originates from the preeminent role DNA secondary structure is thought to play in the molecular basis of site-specific DNA recognition 69

70

ROKITA, LAU, AND ROMERO-FREDES 5'd( CACGGGMCGCATG) 3 ' d(GTGCCC2TGCGTAC) regular duplex I 3 + 51

S'd(CACGGGMCGCATG) 3'd(GTGCCCTTGCGTAC) 3

S'd(CACGGGTGCGCATG) 3'd(GTGCCCACGCGTAC) regular duplex i l + 2)

5'd(CACGGGTGCGCATG) 3'd(GTGCCCaGCGTAC) double mismatch ( 1 + 31

S'd(CACGGGTGCGCATG) 3'd(GTGCCCACGCGTAC) 2

5 ' d(CACGGGT-GCGCATG ) 3'd(GTGCCCA+XCGTAC) insert

(1 + 4) 3 d (GTGCCCATCGCGTAC) 4

Scheme 1

and reactivity.'.'' Our particular interest lies in the structural elucidation of sites within duplex DNA that are hyperreactive or hypermutable under exposure to ~ u n l i g h t . ' ~ In , ' ~this initial comparison between duplex structure and reactivity, the oxidative processes and rates of reaction were described for a related series of sequences. Conformation-specific reactivity was characterized under irradiation > 290 nm in order to mimic most closely the solar excitation process in vivo. The well-described photochemistry of nucleotides and DNA exposed to 254-nm light15 has identified numerous pathways for reaction, many of which are still viable under exposure to higher wavelengths. However, the relative quantum yield of each process, and hence the relevance of each to an environmental model, depends on the wavelength of In addition, photosensitization becomes increasingly important in vivo and in vitro as photolysis is shifted away from the absorption maximum of DNA a t 260 nm and toward the spectrum of sunlight filtered through earth's atmosphere.I6 Acetone has been used in ths study as a general model of many cellular components that could act as photosensitizers. The photochemical properties of acetone allow for induction of both direct DNA-sensitizer reactions (type I processes) and indirect oxygen-mediated reactions (type I1 processes) without requiring acetone to bind specifically onto DNA. This small and rather featureless compound should therefore delineate the reactivity inherent in specific conformations of DNA as opposed to that induced from specific sensitizer- DNA complexation. The oligonucleotides 1-5 were characterized under defined conditions both individually and annealed to form two fully complemented duplexes and two modified structures (Scheme 1). These oligomers were anticipated to react in much the same manner as would a similar sequence embed-

''

ded in a large polymorphic DNA duplex.'" Only minimal sequence modifications were designed to break the typical Watson-Crick base pairing structure in order to avoid introducing extraneous variables into the analysis.21 Thymine was the only base introduced to create aberrant sequences and this was only placed in the core of the duplexes. Such modifications were expected to produce asymmetric distorted helices.', 22,'3 Sequence vs conformation controlled reactivity of these structures are now reported here in terms of initial rates and product profiles of photochemical derivatization. MATERIALS A N D METHODS Materials All oligonucleotides were synthesized via standard solid phase cyanoethyl phosphoramidite chemistry on a DuPont Coder 300 and purified by C-18 reverse phase chromatography. The 5'-dimethoxytrityl was then removed with 80% acetic acid and the fully deprotected oligomer further purified by the anion exchange Chromatography described below if not already deemed homogeneous by this procedure. All aqueous solutions were made with purified water (Nanopure, Sybron/Barnstead); 2-pentanone was distilled before use. Acetone (high performance liquid chromatography grade, Fisher Scientific) and all other reagents were obtained a t the highest grade commercially available, and used without further purification (except when noted). Procedures DNA concentrations were measured per mol of oligonucleotide from the corresponding c 360 values as calculated from the sum of nucleotides with regard to adjacent bases.24 For duplex studies,

OLIGONUCLEOTIDE PHOTOOX1l)A'l'lON

equimolar concentrations of complementary strands were annealed by slowly cooling ( > 3 h) the DNA prior to photolysis. Photolytic degradation of DNA was monitored by discontinuous assay of remaining starting materials. Each time point (minimally done in duplicate) derived from a 100-pL air-saturated solution typically containing ca. 0.5 absorbance units/mL oligomer (4.2 p M per oligomer), 100 m M NaCl [chelex (BioRad) treated], 10 mM potassium phosphate, pH 7, and 7 m M acetone. Samples were exposed to wavelengths of greater than 290 nm by using a simple Pyrex filter a t the focal point of a 150 W high-pressure xenon arc lamp (Photochemical Research Associates). Reaction progress and product formation were quantified by integrating the elution profile of irradiated DNA separated on a Mono Q [Pharmacia] anion exchange column under denaturing conditions (pH 12). This procedure allowed for simultaneous detection of each parent oligonucleotidefrom single-strand or duplex studies and identification of products arising from strand scission and oxidative base modification. Samples were either analyzed directly or after treatment with piperidine (90°C, 30 min) to induce strand scission a t sites of base oxidatiort."~ Only initial rates (0-30 s) were monitored to ensure that oligonucleotide derivatization originated from the starting structures and not from heterogeneous mixtures of partially degraded DNA. Detector output ( A254) was standardized with the measured absorption of 6 under denaturing conditions.

RESULTS AND DISCUSSION Modification Under Aerobic lrradiation

Aerobic photolysis sensitized by acetone was chosen to begin the analysis of the model system in order to measure the relative propensity for DNA modification in the most general manner. Hyper-

sensitivity to the widest variety of processes could be surveyed simultaneously under these conditions. The first triplet state of acetone lies a t high enough energy to transfer its excitation to oxygen and all of the DNA bases, and consequently, acetone can induce both type I and type I1 photoprocesses26 (Scheme 2). Type I excitation transfer to DNA most often results in thymine derivatization since this base has the lowest triplet level of DNA, and can be populated directly by acetone or by a secondary process from neighboring bases. Both energy transfer and electron transfer lead to distinct sets of oxidation products in the presence of oxygen.27Other potential reactions during the photolysis include type 11 activation of oxygen, forming singlet oxygen that primarily combines with guanine.26 Electron transfer to oxygen, another likely event, generates a number of reduced species, among which hydroxyl radical has been noted to be most reactive.B This radical attacks both the bases and the sugar backbone of DNA.30 Deoxyribose oxidation, the result of the latter reaction, produces strand scission and has become the basis for an important method of DNA footprinting:" In the initial development of this system and its controls, the overall progress of photooxidation was confirmed to be dependent on duration of irradiation and concentration of DNA. In every case, the parent oligonucleotides were consumed a t a characteristic and constant rate under irradiation (for example, see Figure 2), and no detectable derivatization occurred after irradiation ceased. This suggested that all of the major processes required a continual flux of photons. When the concentration of 3 was varied over the range of 1.6-10 pM, product formation increased linearly with DNA concentration, consistent with an overall reaction that was first order in DNA. Loss of starting material, however, did not fit an exponential decay and rather remained linear well past a single half-life. Initial rate conditions persisted since the decrease in reaction sites, total nucleotides, remained insignificant after a single hit degradation of DNA

acetone

I

I

*[acetone) 0

oxidation 2I DLA J

disproportionation and reaction

-

I

.+

71

[DLA) 02

+

4 .-

I O2

O2

Scheme 2

oxidation

direct reaction and formation of oxygen radicals

72

ROKITA, LAU, AND ROMERO-PREDES

was recorded. Complete consumption of the parent oligonucleotide constituted only a single modification at 1 of 14 possible nucleotides; thus only 3.6% of the total nucleotides would have reacted after 50% loss of the original DNA. Photochemically induced reactions monitored in this model also required the presence of sensitizer and oxygen. Initial rate of DNA degradation depended linearly on acetone concentration over the range studied (3-17 mM). Acetone concentration was then set and held constant at 7 mM for all other studies to provide a convenient time scale for monitoring the relative reactivities of DNA described in this report. To demonstrate the role of oxygen in these photochemical experiments, 2-pentanone (22 mM) was substituted for acetone in order to maintain a detectable level of sensitizer through the degassing procedure that required cycling samples between vacuum and argon six times.

The photochemistry of 2-pentanoneis quite similar to that of acetone32 and shown here to degrade 75% of 1 under aerobic irradiation (20 s). When oxygen was excluded, the total loss of 1 was less than 10%. The photochemistry of duplex DNA (1 + 2) demonstrated a similar dependence for oxygen. Therefore, all of the major pathways that resulted in detectable loss of starting material required activation dependent on sensitizer and oxidation dependent on molecular oxygen. Chromatographic Analysis of DNA Products Further description of acetone sensitization of DNA was initiated to discriminate between the processes that would dominate the analysis of sequence and conformation-dependent reactivity. The photolysis products were characterized by their chromatographic behavior under neutral and denaturing

0

Time (min)

D

I

A254 I

0

25

50

Time (mi.)

Figure 1. Anion exchange analysis for oligonucleotide products formed under acetone sensitized irradiation. For each chromatogram, a 100 pL solution of 3 (0.5 AU/mL, 4 p M ) , 10 m M potassium phosphate, pH 7, 3 0 0 m M NaCl, and 7 mM acetone was irradiated (> 290 nm), treated as indicated, and finally eluted through a Mono Q (Pharmacia) column. (A) The control sample of DNA was not irradiated but only treated with 0.2M pipendine (90°C, 30 min) followed by high vacuum evaporation, resuspension in water, and characterization by separation under a NaCl gradient (0.0-560 m M over 30 min at 1 mL/min) in 10 m M potassium phosphate pH 7. Under the same chromatographic conditions, DNA irradiated for 30 s yielded (B) without and (C) with pipendine treatment. A photolyzed sample (30 s) without pipendine treatment produced (D) under denaturing chromatography (0.0-5sO mM NaCl over 50 min in 11.5 m M NaON).

OLIGONUCLEOTIDE PHOTOOXIDATION

Photolytic Base Modification

Figure 2. Chromatographic detection and rate determination of duplex (1 + 2) oxidation. Standard aliquots of annealed DNA were irradiated for (A) 0.0 s, (B) 10 s, (C) 20 s, and (D) 30 s, and analyzed under denaturing conditions (see Figure ID). Integrated absorption (254 nm) areas of these elutions profiles were converted to % remaining starting material of 2 and 1, plotted against time, and fit by linear regression to yield (E) and (F),respectively.

73

74

ROKITA, LAU, AND ROMERO-FREDES

modified DNA since pH 12 had no effect on samples later analyzed a t pH 7. Furthermore, the products detected under pH 12 conditions were still susceptible to piperdine induced hydrolysis (see Figure 2). Only denaturing chromatography was available to monitor the rate of strand modification when duplex DNA was studied since only these conditions would allow for distinct isolation of each oligonucleotide strand. Kinetic measurements were equivalent whether or not irradiated DNA was treated with piperidine since both scission products and base-modified strands eluted prior to the parent oligomers a t pH 12. Figure 2 represents a typical series of chromatograms used to measure the reaction rates throughout this report. Only the initial velocities of starting material consumption were compiled to ensure that the data originated from the primary DNA structures and not from those already modified. In addition, stoichiometric quantities of each strand were used in duplex studies to avoid any competing reactions based on single strand forms. A major accumulation of piperidine scission fragments 6-8 nucleotides in length would have indicated that the irregular duplex structures (1 + 3 and 1 + 4) induced a highly localized hyperreac-

conditions on an ion exchange column (Mono Q, Pharmacia). This provided for isolation, quantification, and distinct identification of each oligomer independent of its complementary strand. The primary basis for separation was oligonucleotide length but base composition also affected retention time. Strand scission produced compounds that eluted much before starting material and concurrent with oligonucleotide fragments; more subtle shifts in retention were attributed to the effects of base oxidation as discussed below. When 3 was irradiated and then directly fractionated by ion exchange under neutral conditions, no DNA fragments were detected; rather, the retention of starting material appeared to broaden and tail as if the sample had become heterogeneous due to base modifications (Figure 1B). Scission products were only formed when the irradiated DNA was first treated with piperidine and then analyzed chromatographically (Figure 1C). This treatment also eliminated the heterogeneous material eluting with 3 and thus indicated that most all base modified products were also labile to alkaline conditions. Base derivatization was more clearly distinguished from starting material under denaturing chromatography (pH 12, Figure 1D). The denaturing conditions alone did not hydrolyze

Table I Effect of Various Additives on "he 4probic Photolysis of Duplex DNA (1 + 3) Sensitized by Acetone Half-life of Each Strand (s)"

Addition of Quencher

Solvent

46 i 5 78 i 20 76 & 16 28 & 4 2 280

-

H ,O D,O (99.9%) H,O H,O H2O

3

1

1.OM mannitol 1.OM tBuOH 10%glycerol

41 54 66 36 82

i-4 & 18 i 10 i9 i7

'Half-life values were extrapolated from the linear decrease of initial oligonucleotide v s time of s noted. Remaining starting oligomer was irradiation (0-30 s) under standard conditions except a measured from integrated ion exchange chromatography (for example, see Figure 2). Uncertainties were derived from the standard error of the reaction velocity.

Table I1 Photosensitized Oxidation of Single-Strand DNA Composition Oligomer 1 2

3 4

5

T C

A

G

2 2 4 3 1

2 2 1 2 4

6 4 4 4 5

4 6 5 6 4

Nearest Neighbor (5'-3') Pu-Pu Pu-Py Py-Pu Py-Py 2 0 0 0 4

5 6 5 6 4

6 5 4 5 5

0 2 4 3 0

Initial Rate of Scission + Baseoxidation" (pmol/s) 4.24 i 0.92 4.79 0.46 5.50 i 0.92 6.51 & 0.76 7.22 i 0.61

"Remaining oligomer w a s quantified by integrated chromatographic anaIysis (A,,, ) of samples irradiated for 0-30 s and treated with piperidine. Velocities were then calculated from the average loss of oligomer vs time at each determination; uncertainties were derived from the standard error.

OLIGONUCLEOTIDE PHOTOOXIDATION

75

Table I11 Photosensitized Oxidation of Duplex Oligomersa

Initial Rate of Scission

+

Base Oxidation

Structure

Composition

T, ("C)

Regular Duplex

( i}

60-62

4.16 f 0.21 4.45 i 0.55

58-60

1.89 f 0.46 4.20 k 0.50

48-50

4.37 ? 0.38 5.08 i 0.50

53-55

3.24 i 0.67 2.65 k 0.42

Duplex

Regular Duplex Two base

Mismatch Base Insert

'1 (t) \3

(PmOl/S)

"Oligonucleotide pairs were annealed to form the appropriate duplex structures and subject to irradiation for 0-30 s. Initial rates of modification were then monitored in the same manner as in Table 11. Optical (260 nm) melting temperatures were measured in duplicate under conditions similar to those used in photolysis.

tivity. However, no one photoproduct ever dominated the product profiles illustrated in Figures 1 and 2 and summarized in Tables 1-111. Neither sequence nor conformation seemed to induce a unique or localized reactivity. Global effects of Conformation would not have been easily identified by this type of product analysis but would be apparent in the overall rate of DNA modification (discussed below). The parameters affecting local vs global structures have only recently become the subject of intensive i n v e ~ t i g a t i o n .As ~ ~shown - ~ ~ by nmr, delocalized effects are most often associated with a homopolymeric region in which a bulge may migrate along the duplex ~ t r u c t u r e . ~ ~ ~ ~ Major Oxidative Pathways induced Photochemically

Solvent isotope and quenchng studies were used to describe the primary mode of DNA degradation since multiple processes and intermediates could still have been generated in the confines of oxidative base modification. The most likely sites of reaction, guanine and thymine, were distinguishable by their independent mechanisms of reaction. Guanine is the major target of singlet oxygen that forms by excitation transfer from sensitizer to oxygen.",." The intermediacy of this species is most often implicated by the enhancement of singlet oxygen-hsed reactions as H,O is replaced by D,0.38 Such a solvent isotope effect was detected during photosensitized modification of DNA by thiouracil;'"' rose bengal,:'() and methylene blue."' In contrast, D,O was found not to stimulate photolysis in tht> presence of rib~flavin,~'2-methyl-1,4-

naphthoquinone,4' or as summarized in Table I, acetone. D,O even slightly protected against base modification during acetone sensitization. This type of inhibition has also been noted during the modification of DNA mediated by ionizing radiati~n.~' If singlet oxygen still contributed to the oxidation of DNA, it could only have participated indirectly by generating oxygen-based radicals for later combination with DNA. Another oxidant of DNA, hydroxyl radical, can be generated under ground state or photochemical condition^.^',^^ In previous studies, this radical was shown to modify DNA more efficiently than singlet oxygen by almost three orders of magnitude") and therefore could have potentially contributed to the rates measured here. Hydroxyl radical is also known to react quite quickly with various reagents that serve as standard quenching material^.^'^ Among the most commonly employed are mannitol, glycerol, and t - b ~ t a n o lThe . ~ ~ ability of these to inhibit the overall transformation detected in this model were mixed a t best. Even high concentrations of mannitol only reduced the rate of DNA photochemical oxidation by less than twofold (Table I). The effect of glycerol appears anomalous when compared to that of mannitol and might be attributable to solvent perturbations instead of or in addition to quenching of radicals.45 Most surprising was an activation of base modification in the presence of t-butanol. Secondary alcohols are known to react with the excited states of DNA46 but there is little precedence for a tertiary alcohol to enhance photosensitized degradation of DNA. The major path for light-induced reaction of oligonucleotides in this study likely involved a type

76

HOKITA, LAU, AND ROMERO-FREDES

I process. Either energy and electron transfer between the excited state of acetone and DNA likely initiated the oxidative processes that ultimately depended on irreversible interactions with molecular oxygen. This would best explain the product profiles and the relative insensitivity to singlet oxygen and hydroxyl radical effectors. Rate determinants for the individual oligomers and their duplexes must therefore draw from one or more of the critical interactions between sensitizer DNA and oxygen DNA.

Conformation vs Rate of Oxidative Modification

The rate of single-strand photooxidation was used to represent the likely limit in reactivity of this model system. Uncomplemented strands were expected to adopt a range of random structures that would be quite accessible for the approach of sensitizer, ground state molecular oxygen, and oxygenbased radicals. For all oligonucleotidesstudied, 1-5, the initial rate of derivatization varied only slightly ( k 28%) from the average of 5.65 pmol (oligomer)/s (Table 11). Even though a type I process might be expected t o react primarily a t thymine residues, the initial rate of base modification displayed no dependence on the number of thymines per strand. Moreover, the reactivity of these oligonucleotides was not obviously contingent on the sum of any particular base or its adjacent neighbors. No single reaction then dominated the photochemical reactivity of this model of unstructured polynucleotides. Much of the conformational flexibility of singlestrand DNA is lost as two complementary strands hybridize to form a helical structure. This duplex formation drastically reduces base modification by such ground state oxidants as KMn0,,47 and OsO,.’ Consequently, perturbations of the complementary hydrogen bohding network that binds the two strands would adversely affect not only the thermal stability but also the reactivity of duplex DNA. The photochemical lability of DNA in our model would then have been expected to increase in the order (1 2) = (3 + 5) < (1 + 4) < (1 + 3) if the decrease in T,,, values had foretold of an increase in chemical reactivity. However, the initial rates of scission (Table 111) clearly show that this is not the case. Under acetone-sensitized photooxidation, these strands exhibited initial rates of modification that ranged from 25% (5) to almost 100% (1, 2) of those associated with the random structures of single-strand DNA.

+

The oligonucleotides that demonstrated greatest change in reactivity on conversion from single- to double-strand DNA were those most easily modified in the single-strand form. Duplex formation markedly inhibited oxidation, but only for 4 and 5. The aberrant structures, clearly indicated by a reduction in T,, did not facilitate an initial modification event; base insertion and mispairing neither affected the rate nor site of oxidative attack. The reactive intermediates generated photochemically did not then differentiate between the various conformations of oligonucleotides. The insensitivity toward structure could have resulted from the dynamic nature of oligonucleotides in solution4’ since this might tend to suppress or compensate for anomalous behavior. Yet polynucleotide motion did not prevent bulky oxidants KMnO, and OsO, from selective r e a c t i ~ n . ~ ,Only ,~ the much smaller reagents such as the hydroxyl radical seem to modify DNA without preference to sequence or struct ~ r e The . ~ ~critical parameter that most likely determines the conformational dependence of a reaction must then be the steric requirements inherent in derivatization of DNA. The photochemistry described here proceeded without appreciable selectivity under initial reaction, and would suggest that photochemically excited electron abstraction, energy transfer and subsequent reactions with oxygen had few steric constraints.

CONCLUSION Oligonucleotide reactivity under aerobic photosensitization was characterized to determine the predominant mechanism of modification in a model that could sustain many of the processes initiated on exposure to sunlight. The small and nonspecific sensitizer, acetone, was used in these studies in order to most generally survey a conformational bias in DNA reaction under the possibility of both type I and I1 oxidative processes. Derivatization of DNA, as detected by chromatography, required oxygen and sensitizer participation, and primarily yielded nucleic acid base modification with the concomitant induction of alkaline labile sites. The response of this reaction to various effectors suggested that nucleotide derivatization was controlled by an initial interaction between DNA bases and the excited state of acetone. This pathway for DNA activation was insensitive to steric perturbation since the initial rates of single-strand, helical, and aberrant structures of DNA were similar. The hyperreactivity of DNA during photochemical oxi-

OLIGONUCLEOTIDE PHOTOOXIDATION

dation in vivo is then likely to originate from preassociation of reactive cornp~nents,'~ and not merely from the innate solvent accessibility or geometry of specific DNA conformations. Anaerobic studies are currently underway to assess the structural dependence of base dimerization, a localized and preassociated reaction by definition. This work is dedicated to the memory of Dr. E. T. Kaiser who brought his charm and spirit to numerous areas of science. We thank Robert Rieger for much of the oligonucleotide synthesis and the National Cancer Institute (CA43593) for their generous support.

REFERENCES 1. Barton, J. (1988) Chem. Engin. News 66,30-42. 2. Wells, R. (1988) J. Bwl. Chem. 263, 1095-1098. 3. Kaiser, E. & Kezdy, F. (1983) Proc. Natl. Acad. Sci. [JSA 80,1137-1143. 4. Kaiser, E. & Kezdy, F. (1984) Science 223, 249-255. 5. Patel, D., Shapiro, L. & Hare, D. (1987) Ann. Rm. Biophys. Chem. 16, 423-454. 6. Woodson, S. & Crothers, D. (1987) Biochemistry 26, 904-91 2. 7. Prive, G., Heinemann, U., Chandrasegaran, S., Kan, L.-S., Kopka, M. & Dickerson, R. (1987) Science 238, 498-504. 8. Hunter, W., Kneale, G., Brown, T., Rabinovich, D. & Kennard, 0. (1986) J . Mol. Bwl. 190, 605-618. 9. Johnston, B. & Rich, A. (1985) Cell 42, 713-724. 10. Novack, D., Casna, N., Fischer, S. & Ford, J. (1986) Proc. Natl. Acad. Sci. USA 83, 586-590. 11. Kohwi-Shigematsu, T., Manes, T. & Kohwi, Y. (1987) R o c . Natl. Acad. Sci. USA 84, 2223-2227. 12. Ripley, I,. & Glickman, B. (1983) in Cellular Responses to DNA Damage, Friedberg, E. C. & Bridges, B. A., Eds., A. R. Liss, New York, 1983, p. 521. 13. Todd, P. & Glickman, B. (1982) Proc. Natl. Acad. Sci. USA 79, 4123-4127. 14. LeClerc, J. & Istock, N. (1982) Nature 297, 596-598. 15. Wang, S., Ed. (1976) Photochemistry and Photobwlogy of Nucleic Acids, Vols. I and 11, Academic Press, New York. 16. Jagger, J. (1985) Solar-UV Actions on Living Cells, Praeger, New York. 17. Umlas, M., Franklin, W., Chan, G. & Haseltine, W. (1985) Photochem. Photobiol. 42,265--273. 18. Gill, R. & Coohill, T. (1987) Photochem. Photobiol. 45, 264-271. 19. Mei, H.-Y. & Barton, J. (1988) Proc. Natl. Acad. Sci. USA 85, 1339-1343. 20. Yoon, C., Kuwabara, M., Law, R., Wall, R. & Sigman, D. (1988) J . BWL. Chem. 263, 8458-8463. 21. Shakked, 2. & Rabinovich, D. (1986) Prog. Bwphys. M o ~ c Bwl. . 47,159-195.

77

22. Hare, D., Shapiro, L. & Patel, D. (1986) Biochemistry 25, 7445-7456. 23. Rabinovich, D., Haran, T., Eisenstein, M. & Shakked, S. (1988) J . Mol. Biol. 200, 151-161. 24. Fasman, G., Ed. (1975) CRC Handbook of Biochemistry and Molecular Biology-Nucleic Ac&, Vol. I, 3rd ed., CRC Press, Boca Raton, FL, p. 175. 25. Maxam, A. & Gilbert, W. (1980) Methods in Enzymol. 65, 499-560. 26. Rahn, R. & Patrick, M. (1976) in Photochemistry and Photobiology of Nucleic Acids, Wang, S., Ed., Vol. 11, Academic Press, New York, pp. 97-145. 27. Piette, J., Merville-Louis, M.-P. & Decuyper, J. (1986) Photochem. Photobwl. 44,793-802. 28. Cadet, J. & Teoule, R. (1978) Photochem. Photobiol. 28, 661-667. 29. Imlay, J. & Linn, S. (1988) Science 240, 1302--1309. 30. Epe, B., Mutzel, P. & Adam, W. (1988) Chem-Biol. Interactwns 67, 149-165. 31. Tullius, T. (1987) Trends Biochem. 12, 295-300. 32. Wu, K. & Trozzolo, A. (1979) J . Photochem. 10, 407-410. 33. Joshua-Tor, L., Rabinovich, D., Hope, €I.,Frolow, F., Appella, E. & Sussman, J. (1988) Nature 334, 82-84. 34. Kalnik, M., Norman, D., Zagorski, M., Swann, P. & Patel, D. (1989) Biochemistry 28, 294-303. 35. Furlong, J., Sullivan, K., Murchie, A., Gough, G. & Lilley, D. (1989) Biochemistry 28, 2009--2017. 36. Sullivan, K., Murchie, A. & Lilley, D. (1988) J . Biol. Chem. 263, 13074-13082. 37. Friedmann, T. & Brown, D. (1978) Nucleic Aczd Res. 5, 617-622. 38. Foote, C. (1984) Methods Enzymol. 105, 36-47. 39. Peak, J., Peak, M. & Foote, C. (1986) Photochem. Photobiol. 44,111-116. 40. Cadet, J., Decarroz, C., Wang, S. & Midden, W. (1983) Israel J . Chem. 23, 420-429. 41. Wagner, J., Cadet, J. & Fisher, G. (1984) Photochem. Photobiol. 40, 589-597. 42. Cadet, J., Berger, M., Decarroz, C., Wagner, J., van Lier, J., Ginot, L. & Vigny, P. (1986) Biochimie 68,813-834. 43. Greenwald, R. (1985) CRC Handbook of Methods for Oxygen Radical Research, CRC Press, Boca Raton, FL. 44. Sagripanti, J.-L., & Kraemer, K. (1989) J . Biol. Chem. 264, 1729-1734. 45. Saenger, W., Hunter, W. & Kennard, 0. (1986) Nuture 324, 385-388. 46. Livneh (Noy), E., Tel-Or, S., Sperling, J. & Elad, D. (1982) Biochemistry 21, 3698-3703. 47. Hayatsu, H. & Ukita, T. (1967) Biophys. Biochem. Res. Commun. 29, 556-561. 48. Kintanar, A., Huang, W.-C., Schindele, D., Wemmer, D. & Drobny, G. (1989) Biochemistry 28, 282-293.

Received April 11, 1989 Accepted June 13, 1989