Thiols Alter the Partitioning of Calicheamicin-Induced Deoxyribose 4

528

Chem. Res. Toxicol. 2001, 14, 528-535

Thiols Alter the Partitioning of Calicheamicin-Induced Deoxyribose 4′-Oxidation Reactions in the Absence of DNA Radical Repair Daniel M. Lopez-Larraza,† Kenneth Moore Jr.,‡ and Peter C. Dedon* Division of Bioengineering and Environmental Health, 56-787, Massachusetts Institute of Technology, 77 Massachusetts Avenue, Cambridge, Massachusetts 02139 Received January 12, 2001

Cellular thiols have been proposed to play a protective role in oxidative DNA damage by quenching radical species in solution and by repairing deoxyribose and nucleobase radicals. There is also evidence for participation of thiols in reactions after formation of the DNA radical. Previous studies with neocarzinostatin, a thiol-dependent DNA-cleaving enediyne, revealed that the structure and charge of the activating thiol influence the partitioning of deoxyribose 4′-oxidation reactions between a 3′-phosphoglycolate residue and the alternative 4′-keto-1′aldehyde abasic site [Kappen, L. S., et al. (1991) Biochemistry 30, 2034-2042; Dedon, P. C., et al. (1992) Biochemistry 31, 1917-1927]. However, interpretation of these results is confounded by the formation of a neocarzinostatin-thiol conjugate that could alter the position of the activated drug in the minor groove and quench drug radicals. Using the DNA-cleaving enediynes calicheamicin γ1I and Ø, which are identical except for their trigger moieties, we now present a more definitive study of the role of thiol structure in the partitioning of the deoxyribose 4′-oxidation reaction. In the absence of thiols, calicheamicin Ø, which can undergo hydrolytic or reductive activation, generated 4′-oxidation products consisting of 26% 3′-phosphoglycolate residues, 33% 3′-phosphate-ended fragments, and 41% abasic sites (determined as the 3′-phosphopyridazine derivative). Using a series of thiols of varying size and charge, we found that, at concentrations that do not quench drug or DNA radicals, the negatively charged thiols glutathione and thioglycolate did not significantly affect the baseline proportions of the 4′oxidation products. However, neutral thiols (O-ethylglutathione, methyl thioglycolate, 2-mercaptoethanol, and dithiothreitol) and, to a greater extent, the positively charged aminoethanethiol inhibited the production of 3′-phosphoglycolate residues with a proportional increase in the number of abasic sites. The effect of the thiols on the quantities of single- and doublestranded DNA lesions produced by calicheamicin γ1I was also investigated since 3′-phosphoglycolate residues produced by calicheamicin exist only in double-stranded DNA lesions, and the thiol effects could have resulted from quenching of drug or DNA radicals. These studies revealed that, at thiol concentrations found to alter deoxyribose 4′-oxidation reactions, there was no apparent quenching of drug radicals or repair of DNA radicals. Thus, the effects of the thiols on the deoxyribose 4′-oxidation chemistry are due to reactions with a key intermediate in the phosphoglycolate- and abasic site-generating pathways. These results also suggest that cellular glutathione plays a relatively minor role in the chemistry of deoxyribose 4′-oxidation, which has implications for other oxidative reactions occurring in the minor groove of DNA (e.g., deoxyribose 5′- and 1′-oxidation).

Introduction Cellular thiols affect oxidative DNA damage at several levels ranging from chemical repair of DNA radical centers to scavenging of solvated radicals or activated oxygen species (1). Given the high rate of addition of oxygen to carbon-centered radicals in DNA to form peroxylradicals (0.3-3 × 109 M-1 s-1; see refs 2 and 3) and the relatively slow rate of glutathione-mediated repair of DNA radicals (1-2 × 105 M-1 s-1 for doublestranded DNA and ∼2 × 106 M-1 s-1 for single-stranded DNA; see refs 3-6), it is reasonable to assume that thiols may also play a role in the reactions that occur after * To whom correspondence should be addressed. Telephone: (617) 253-8017. Fax: (617) 258-0225. E-mail: [email protected]. † Present address: Instituto Multidisciplinario de Biologı´a Celular (IMBICE), 526 e/10 y11. C., C. 403, 1900 La Plata, Argentina. ‡ Present address: Pfizer Pharmaceuticals, Groton, CT 06378.

formation of the peroxyl radical. In support of this hypothesis, studies with the DNA-cleaving enediynes neocarzinostatin and esperamicin suggest a role for thiols in defining the spectrum of products arising from deoxyribose oxidation (7-9). Interpretation of the results of these studies, however, is hampered by formation of a thiol-drug conjugate with neocarzinostatin that could alter the identity of the abstracted hydrogen atom (7, 8) and by thiol-mediated quenching of drug and DNA radicals for both neocarzinostatin and esperamicin (9, 10). Furthermore, activation of these enediynes is not possible in the absence of a reducing agent, so there is no benchmark by which to assess the effects of thiols on the oxidation chemistry. We now present the results of studies with the enediynes calicheamicin γ1I and Ø that provide clear evidence of a role for thiols in defining the spectrum of deoxyribose oxidation products, without

10.1021/tx0100082 CCC: $20.00 © 2001 American Chemical Society Published on Web 04/04/2001

Thiols and Deoxyribose Oxidation Chemistry

Chem. Res. Toxicol., Vol. 14, No. 5, 2001 529

Figure 2. Structures of the thiols used in the studies presented here.

Figure 1. Structure and mechanism of activation of calicheamicins Ø and γ1I. Hydrolysis or thiol-mediated reduction of the thioacetate group of calicheamicin Ø leads to the formation of the same diradical intermediate formed by reduction of calicheamicin γ1I by thiols.

interference from quenching of drug radicals or repair of DNA radicals. The enediyne family of antibiotics, including neocarzinostatin, calicheamicin, and esperamicin (Figure 1), presents a unique opportunity to define the role of thiols in the chemistry of deoxyribose oxidation. As shown in Figure 1 for the calicheamicin/esperamicin class, enediynes produce oxidative DNA damage by a common mechanism involving reductive activation to form a diradical species that, when positioned in the minor groove, abstracts hydrogen atoms from deoxyribose on each DNA strand (11, 12). The resulting carbon-centered radicals then undergo a series of reactions, analogous to radiation-induced deoxyribose oxidation, that lead to the formation of a spectrum of products unique to each position in deoxyribose (reviewed in ref 11). This is illustrated in Scheme 1 for enediyne-mediated abstraction of the 4′-hydrogen atom of deoxyribose, which results in at least two sets of products at ambient oxygen levels: a strand break with 3′-phosphoglycolate and 5′-phosphateended fragments or a 4′-keto-1′-aldehyde abasic site. 4′Chemistry associated with enediynes is unusual in that 3′-phosphoglycolate residues arise only in double-stranded lesions (13, 14). With calicheamicin, bistranded lesions represent >95% of the DNA damage and the 4′-chemistry on one strand is accompanied by products of 5′-hydrogen

atom abstraction on the complementary strand (13, 15, 16). The effect of thiols on DNA oxidation has been the subject of several studies addressing the role of thiols in repair of radiation- and drug-induced DNA radicals (35, 8, 9, 17-20). Several general rules have arisen from studies of DNA oxidation produced by ionizing radiation (3-5, 17, 18). Positively charged thiols concentrate on the negatively charged phosphate backbone of DNA and are thus more effective at quenching DNA radicals than neutral or negatively charged thiols. An analogous electrostatic argument can be invoked to explain the relatively weak effect of negatively charged thiols such as glutathione on repair of DNA radicals relative to neutral and positively charged thiols (3-5, 17, 18). Epstein et al. observed that these same rules apply to the effect of thiols on quenching of drug or deoxyribose radicals in bistranded lesions produced by the enediyne esperamicin A1 (9). The ability of the thiol to reduce the proportion of drug-induced double-stranded lesions decreased in the following order: positive charge > neutral > negative charge (9). The charge and other physical properties of the activating thiol were similarly found to affect the rate of neocarzinostatin activation and the relative quantities of neocarzinostatin-mediated single- and double-strand breaks in plasmid DNA (7, 8, 19, 21). These studies further revealed that thiols alter the proportions of deoxyribose 4′-oxidation products (e.g., phosphoglycolate residues and abasic sites). However, the activating thiol forms a covalent bond with the drug, and the presence of an adducted thiol could influence the damage chemistry by altering the position of the activated drug in the minor groove or by intramolecular hydrogen atom transfer to quench drug radicals (10). It is thus unclear from these studies how the structure of the thiol influenced its interaction with deoxyribose oxidation intermediates and the subsequent spectrum of oxidation products. To more clearly define the role of thiols in deoxyribose oxidation chemistry, we used a series of thiols (Figure 2)

Scheme 1. Reactions Involved in the Oxidation of the 4′-Position of Deoxyribose by Enediynes

530

Chem. Res. Toxicol., Vol. 14, No. 5, 2001

to examine the effects of thiol structure on the partitioning of the deoxyribose 4′-oxidation reactions initiated by the DNA-cleaving enediynes calicheamicin γ1I (11) and Ø (22). As shown in Figure 1, the structures of the two calicheamicins are identical except for their trigger moieties. While calicheamicin γ1I is activated by reduction of a methyl trisulfide, calicheamicin Ø can be activated by either hydrolysis or reduction of its thioacetate moiety. Because the two calicheamicins are identical in all respects except activation in forming the diradical, the ability of calicheamicin Ø to undergo simple hydrolytic activation permits a determination of the chemistry of deoxyribose oxidation unperturbed by thiols and other reducing agents. We now present a more definitive study of the role of thiol structure in the partitioning of deoxyribose 4′-oxidation reactions between strand breaks with 3′-phosphoglycolate residues and 4′-keto-1′-aldehyde abasic sites.

Experimental Procedures Chemicals. Calicheamicins γ1I and Ø were provided by G. Ellestad (Wyeth-Ayerst Research, Pearl River, NY) and K. C. Nicolaou (Scripps Research Institute, La Jolla, CA), respectively. γ-Glutamylglycine, glutathione, glutathione O-ethyl ester, 2-mercaptoethanol, 2-aminoethanethiol hydrochloride, thioglycolic acid, dithiothreitol, and methyl thioglycolic acid were purchased from Sigma Chemical Co. (St. Louis, MO). Plasmid pUC19 was prepared as described elsewhere (23). Sequencing Gel Analysis of Deoxyribose 4′-Oxidation Chemistry. A 143 bp DNA fragment was prepared by 5′-32P end labeling of HindIII-digested pUC19 by standard procedures (24) followed by digestion with PvuII. The DNA fragments were then resolved on a 10% polyacrylamide gel, and the 143 bp fragment was isolated by the crush-and-soak method (24). For thiol-containing reaction mixtures, DNA cleavage was initiated by adding 2 µL of a 5 µM solution of calicheamicin Ø in methanol (100 nM final concentration) to 96 µL of a solution of (final concentrations after subsequent addition of thiol) endlabeled DNA (50 000 cpm), 30 µg/mL sonicated and exonucleasetreated calf thymus DNA (25), 1 mM EDTA, and 50 mM 4-(2hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES; pH 7.4). The reaction was started by addition of 2 µL of the thiol as a freshly prepared aqueous solution (except methanolic solutions of methyl thioglycolate) in which the pH was adjusted to 7.4 with NaOH or HCl. In the absence of thiol, the concentration of calicheamicin Ø ranged from 1 to 100 µM, under the conditions described above, to account for the slower rate of hydrolytic activation. To control for ionic strength contributions by glutathione, γ-glutamylglycine was added to several control reaction mixtures with calicheamicin Ø. However, γ-glutamylglycine did not affect the quantity or chemistry of oxidation products (data not shown). In all reaction mixtures, the final methanol concentration was adjusted to 7%. Reactions that included thiol were allowed to proceed for 1 h at 37 °C, while incubation times for reactions with calicheamicin Ø alone ranged from 1 to 24 h depending on the drug concentration (i.e., higher concentrations required shorter reactions). The fragments were then treated with 100 mM hydrazine for 1 h at ambient temperature to convert the 4′-keto-1′-aldehyde abasic site to a 3′-phosphopyridazine-ended fragment (26). Sequencing gel analysis was employed to quantify the proportions of 3′-phosphoglycolate, 3′-phosphate, and 3′-phosphopyridazine residues at the AGGATC cleavage site in the 5′-32Plabeled HindIII-PvuII fragment of pUC19 (see Figure 3). Drugtreated DNA samples were precipitated with 70% ethanol and 0.3 M sodium acetate (pH 7), resuspended in 3 µL of a formamide-based, nonalkaline sequencing gel loading buffer (24), boiled for 1 min, and immediately chilled in an ice/water mixture. The samples were resolved on a 60 cm long and 0.4

Lopez-Larraza et al.

Figure 3. Representative example of sequencing gel resolution of calicheamicin Ø-mediated DNA damage chemistry at the AGGATC site in the 142 bp 5′-32P-labeled HindIII-PvuII fragment of pUC19. The DNA sequence surrounding this cleavage site is shown in the diagram above the gel image, with the drug binding site underlined and the cleavage site marked with an arrow. The DNA was treated with calicheamicin Ø in the absence (lanes 1 and 2) or presence (lanes 3-6) of a thiol, subjected to derivatization with hydrazine, and resolved on a 25% polyacrylamide sequencing gel: lanes 1 and 2, a 1 h reaction with 100 and 20 µM calicheamicin Ø, respectively, in the absence of thiol; lanes 3 and 4, 100 nM calicheamicin Ø with 10 mM glutathione (duplicate reactions); and lanes 5 and 6, 100 nM calicheamicin Ø with 10 mM glutathione O-ethyl ester (duplicate reactions). (A) 3′-Phosphopyridazine-ended fragment representing the hydrazine derivative of the 4′-keto-1′-aldehyde abasic site. (B) 3′-Phosphate-ended fragment. (C) 3′-Phosphoglycolate-ended fragment. mm thick, prewarmed, 20% denaturing polyacrylamide gel at a constant power of 75 W for 18-20 h. The gel was then subjected to phosphorimager analysis (Molecular Dynamics) directly or as a dried gel following fixation in a solution of 10% acetic acid and 10% methanol to prevent cracking of the gel. Plasmid Topoisomer Analysis of Single- and DoubleStranded DNA Lesions. DNA cleavage was initiated by adding aliquots of thiol (final concentrations ranging from 1 µM to 30 mM) to a solution containing 30 µg/mL of supercoiled pUC19, 3 nM calicheamicin γ1I, 1 mM EDTA, and 50 mM HEPES (pH 7.4; 50 µL total reaction volume). Reactions were allowed to proceed for 1 h at 37 °C. Samples were then treated with 100 mM putrescine dihydrochloride for 1 h at 37 °C to convert all abasic sites to strand breaks (27, 28). DNA samples were resolved on a 1% agarose gel (Tris-borate-EDTA; 24) at 5 V/cm. As described elsewhere, DNA bands in dried agarose gels were visualized by in situ hybridization with the 32P-labeled probe prepared from pUC19 and the plasmid topoisomer species quantified by phosphorimager analysis (23, 29). The quantity of each plasmid topoisomer was expressed as a percentage of the total signal for supercoiled (form I), nicked (form II), and linear (form III) DNA molecules.

Results Effect of Thiol Structure on the Partitioning of Deoxyribose 4′-Oxidation Reactions. We previously demonstrated that, when activated by a thiol, calicheamicin γ1I oxidizes the 4′-position of deoxyribose at the C of the AGGATC sequence in the 5′-32P-labeled HindIII-PvuII fragment of pUC19 (13). This damage occurs as part of a bistranded lesion accompanied by 5′oxidation of the 5′-most C of the complementary GATCCT sequence, with the bistranded lesion well isolated from other damage sites to allow clear definition of the 4′-oxidation chemistry. Control experiments confirmed that, with all of the thiols studied, both calicheamicins Ø and γ1I produced identical cleavage patterns in the 5′-



Table 1. Thiols and Partitioning of 4′-Chemistrya no. of 3′-residues (%)b thiol

3′-PGc

3′-Pc

3′-PPc

ne

no thiol 10 mM glutathione 10 mM O-ethylglutathione 1 mM thioglycolate 10 mM thioglycolate 10 mM methyl thioglycolate 10 mM 2-mercaptoethanol 1 mM dithiothreitol 10 mM dithiothreitol 100 mM dithiothreitol 0.01 mM aminoethanethiol 0.1 mM aminoethanethiol 1 mM aminoethanethiol 10 mM aminoethanethiol

26 ( 4 22 ( 8 13 ( 4 27 ( 4 22 ( 1 9(2 12 ( 3 26 14 ( 5 0 23 16 6 0

33 ( 5 32 ( 7 35 ( 5 27 ( 4 28 ( 2 29 ( 8 29 ( 7 74d 28 ( 11 100d 77d 84d 94d 57 ( 13

41 ( 7 46 ( 4 52 ( 8 46 ( 6 50 ( 3 62 ( 9 59 ( 9 58 ( 14 43 ( 13

9 9 7 6 5 5 5 1 6 1 1 1 1 5

a

Quantitation of deoxyribose 3′-residues was carried out as described in Experimental Procedures. b Data for each residue are expressed as a percentage of total damage at the AGGATC site ((SD). c 3′-PG, 3′-phosphoglycolate; 3′-P, 3′-phosphate; 3′-PP, 3′phosphopyridazine (hydrazine derivative of the abasic site). d Samples treated with putrescine instead of hydrazine. e n is the number of independent experiments. 32P-labeled HindIII-PvuII fragment as well as identical proportions of deoxyribose oxidation products (data not shown). These results confirm earlier observations of the identical nature of the DNA damage produced by calicheamicins Ø and γ1I (22). The advantage of using calicheamicin Ø is the ability to identify the deoxyribose oxidation products in the absence of thiols. Calicheamicin Ø alone undergoes hydrolytic activation to oxidize the 4′-position of the C of the AGGATC sequence, with the resulting products well resolved on a 25% polyacrylamide sequencing gel as shown in Figure 3. The identification of sugar fragmentation products was carried out on the basis of the previously defined sequencing gel migration characteristics of DNA fragments possessing 4′-oxidation products (7, 11, 14, 19, 30-37). A 3′-phosphoglycolate-ended fragment (band C in Figure 3) migrates ∼0.25 nucleotide (nt) faster than a 3′-phosphate-ended fragment (band B in Figure 3; defined by comigration with Maxam and Gilbert sequencing reactions; 38), while a 3′-phosphopyridazine-ended fragment, derived from a reaction of hydrazine with the 4′-keto-1′-aldehyde abasic site, migrates 1-3 nt slower than the 3′-phosphate-ended fragment (band A in Figure 3). The presence of thiols in the calicheamicin Ø damage reaction was observed to alter the quantity of the 3′-phosphoglycolate and 3′-phosphopyridazine products. A representative example is shown in Figure 3, and Table 1 summarizes the effects of the various thiols on the quantities of 4′-oxidation products. Damage produced by calicheamicin Ø alone served as a benchmark for defining thiol effects and consisted of 26% 3′-phosphoglycolate residues, 33% phosphate-ended fragments, and 41% abasic site derivatives. The presence of negatively charged glutathione did not cause a significant shift in the partitioning of 4′-chemistry, while the neutral glutathione O-ethyl ester reduced the proportion of 3′-phosphoglycolate to 13% of the total lesions with a concomitant increase in the level of abasic site derivatives (Table 1). Similar results were observed with the negatively charged thioglycolate and its neutral methyl ester. With the latter thiol, however, there was a larger reduction in the level of 3′-phosphoglycolate residues (and increase in the level

Figure 4. Plot of the quantity of 3′-phosphate-ended DNA fragments (b) or 3′-phosphopyridazine-ended DNA fragments [abasic site derivative (O)] as a function of the quantity of 3′phosphoglycolate-ended DNA fragments.

of abasic site derivatives) than that observed with glutathione O-ethyl ester. The neutral thiols 2-mercaptoethanol and dithiothreitol also reduced the level of 3′phosphoglycolate residues, while the greatest reduction was observed with the positively charged aminoethanethiol, which completely inhibited formation of 3′-phosphoglycolate at thiol concentrations exceeding 1 mM (Table 1). Table 1 also illustrates the concentration dependence of the thiol effects on 4′-chemistry for dithiothreitol and aminoethanethiol. As shown in Figure 4, a plot of the quantities of phosphate- and phosphopyridazine-ended fragments versus phosphoglycolate-ended fragments reveals an important relationship. The quantity of phosphoglycolate residues was inversely related to the quantity of the abasic site derivatives (slope ) -1), which suggests that thiols interact with an intermediate directly involved in the partitioning of the 4′-oxidation reaction. Furthermore, there is no relationship between the quantity of phosphateended fragments and either the phosphoglycolate residues or the abasic site derivatives (slope ∼ 0 in Figure 4). This latter observation suggests that the phosphateended fragments are not derived from hydrolysis of the abasic sites during processing of the DNA and that they likely arise from an intermediate, prior to the partitioning point, that is not affected by the thiol. Role of Thiols in the Repair of DNA Radicals and Quenching of Drug Radicals. An alternative explanation for the thiol effects lies in the observation that enediyne-induced 3′-phosphoglycolate residues exist in only double-stranded DNA lesions (13, 14, 19). It is possible that the thiols act to quench one of the calicheamicin radicals or repair the deoxyribose radical arising on the strand opposite the AGGATC sequence. Either event would convert the lesion from bistranded to single-stranded and thus reduce the quantity of 3′phosphoglycolate residues. To test this model, we used plasmid topoisomer analysis to define the effect of the various thiols on the quantity of calicheamicin γ1I-induced single- and double-stranded DNA lesions. The experiment involved treating supercoiled plasmid pUC19 with calicheamicin γ1I (3 nM) and a thiol, followed by resolution of the damaged DNA on an agarose gel. As shown in the example in Figure 5, single-stranded lesions convert the supercoiled (form I) plasmid to a nicked, closed-circular molecule (form II), while a double-stranded lesion linearizes the plasmid (form III).

532


Figure 5. Representative example of the topoisomer analysis used to define the effect of thiols on calicheamicin γ1I-mediated double- and single-stranded DNA lesions. Samples of supercoiled pUC19 were treated with 3 nM calicheamicin γ1I and O-ethyl glutathione followed by cleavage of abasic sites with putrescine and resolution of plasmid topoisomers on an agarose gel. Plasmid topoisomers are identified in the right margin.

The quantities of single- and double-stranded DNA damage as a function of thiol concentration are depicted graphically in Figure 6. For all thiols except aminoethanethiol, the level of double-stranded lesions increased as a function of thiol concentration and reached a maximum at 1-10 mM, while the level of single-stranded lesions remained relatively constant at background levels. The optimal concentration for all thiols resulted in similar quantities of double-stranded lesions; 17-22% of the plasmid DNA was linearized. This is equivalent to a concentration of plasmid molecules of ∼3 nM with doublestranded lesions, which is consistent with a nearly quantitative reaction of 3 nM calicheamicin. As shown in Figure 6, the optimal concentration of aminoethanethiol for double-stranded lesions was 0.1 mM, and at higher concentrations, there was a greater reduction in the level of double-stranded lesions and an increase in the level of single-stranded lesions than occurred with the other thiols.

Discussion Of the many roles of thiols in radiation- and druginduced oxidative DNA damage, including quenching of solvent radicals and repair of DNA radicals, we have focused on the effect of thiols on deoxyribose oxidation chemistry. The specific reaction studied, that of deoxyribose 4′-oxidation, starts with a common pathway in which addition of molecular oxygen to the carboncentered radical eventually leads to a partitioning of the reaction to form either of two product sets, one containing a 3′-phosphoglycolate residue and one containing an oxidized abasic site, as shown in Scheme 1 (11). Previous studies with enediynes as model systems for studying thiol effects on 4′-oxidative chemistry have been difficult to interpret due to the multiple roles played by the thiol in drug activation and drug structure (8, 14, 19, 39). We have used a thiol-independent enediyne, calicheamicin Ø (22), to define the baseline deoxyribose oxidation chemistry and then to determine how thiols affect this chemistry. Our results are entirely consistent with previous studies involving radiation (3-5, 18, 40) and enediyne antibiotics (8, 14, 19, 20), and they highlight the importance of charge in the interaction of thiols with DNA. Negative charge appears to limit the interaction of thiols with intermediates in 4′-oxidation, as indicated by the lack of an effect of glutathione and thioglycolate on partitioning of 4′-chemistry and the substantial reduction in the level of phosphoglycolate residues when these thiols are present as their neutral esters. In a similar


manner, the neutral thiols 2-mercaptoethanol and dithiothreitol shift the 4′-chemistry to abasic sites. Relative to the other thiols, the positively charged aminoethanethiol produced the most substantial shift in 4′-oxidation products, which is consistent with the concentration of the thiol on or around the negatively charged backbone of DNA. The observed effects of thiols on 4′-chemistry can be explained by at least two mechanisms: (1) quenching of drug or DNA radicals by the thiols or (2) reaction of the thiols with an intermediate in the degradation of the deoxyribose 4′-radical. In the first scenario, the inhibition of the formation of phosphoglycolate residues could result either from quenching of the calicheamicin radical responsible for 4′-hydrogen atom abstraction from deoxyribose or from quenching of the carbon-centered radical at the 4′-position in deoxyribose. Either of these repair events would convert a double-stranded lesion to a singlestranded event. Given the fact that enediyne-induced phosphoglycolate residues form in only double-stranded lesions, repair of drug or DNA radicals would lower the observed levels of phosphoglycolate. However, this model cannot account for the observed increase in the level of abasic site derivatives accompanying decreases in the level of phosphoglycolate. Furthermore, the plasmid topoisomer experiments definitively rule out drug and DNA radical repair. With all of the thiols that were studied, the level of single-stranded lesions does not increase, and the level of double-stranded lesions does not decrease, at thiol concentrations observed to alter the partitioning of 4′-chemistry (Figure 6). There is evidence for radical repair for aminoethanethiol at concentrations exceeding 0.5 mM (Figure 6), with a sharp reduction in the level of double-stranded lesions and a concomitant increase in the level of single-stranded damage. This phenomenon is likely due to concentration of the positively charged thiol on DNA due to electrostatic binding to the phosphate backbone (17, 18, 40). Though a similar repair effect appears to occur with methyl thioglycolate (Figure 6), we did not extend the studies with the other thiols to concentrations high enough to produce detectable radical repair. Thus, the observed effects of thiols on the partitioning of 4′-oxidation chemistry are not due to a shift from double- to single-stranded lesions caused by chemical repair of drug or deoxyribose radicals. The lack of observed quenching of calicheamicin or DNA radicals by thiols stands in contrast to the results of previous studies with esperamicin (9) and neocarzinostatin (8, 10, 19). With both enediynes, neutral and positively charged thiols caused a reduction in the level of double-stranded DNA lesions relative to that of negatively charged thiols. This was due presumably to thiol-mediated repair of drug or DNA radicals. Direct evidence for quenching of a neocarzinostatin radical was obtained in the studies performed by Chin and Goldberg (10). They observed that a hydrogen atom was transferred from a non-sulfur position in the thiol to one of the drug radicals. While esperamicin and calicheamicin share a common DNA-cleaving enediyne moiety, they differ substantially in their binding modes. Esperamicin binds in the minor groove by intercalation of its deoxyfucose anthranilate moiety (36), while the nonintercalative groove-binding mode of calicheamicin is mediated by its extended carbohydrate side chain (Figure 1) (41-43). Different binding modes or binding kinetics may expose esperamicin or its DNA damage site to more frequent



Figure 6. Effect of thiol concentration on calicheamicin γ1I-mediated double- and single-stranded DNA lesions. Plasmid topoisomers in gels such as that shown in Figure 5 were quantified by phosphorimager analysis. The quantities of DNA forms II [single-stranded lesions (O)] and III [double-stranded lesions (b)], expressed as a percentage of total plasmid signal, are plotted as a function of thiol concentration. Each point represents the mean and standard error for three to six replicates.

interactions with thiols, thus permitting a higher degree of radical repair. The second and more likely explanation for the observed effects of thiols on the spectrum of deoxyribose

4′-oxidation products involves a reaction of the thiols with an oxygen-derived intermediate lying at a critical point in the partitioning of the 4′-oxidation reaction. The details of the reaction mechanism after addition of molecular

534


oxygen to the deoxyribose radical have not been firmly established. However, previous studies with neocarzinostatin suggest one possible scenario. Under anaerobic conditions, the nitroaromatic radiation sensitizer, misonidazole, substitutes for oxygen and causes an increase in the level of 3′-phosphoglycolate residues in both double- and single-stranded lesions (7, 19), the latter having no detectable glycolate under aerobic conditions. The known mechanism for the reaction of misonidazole with the deoxyribose radical suggests that the enhanced production of glycolate involves formation of an oxyl radical intermediate at the 4′-position (Scheme 1). This immediately suggests an opportunity for a thiol to reduce the oxyl radical to produce the 4′-hydroxyl nucleoside that rearranges to the 4′-keto-1′-aldehyde abasic site. Limited access of glutathione and other negatively charged thiols would thus preclude reduction of the oxyl radical and permit the formation of higher levels of phosphoglycolate residues. Though speculative, such a mechanism warrants further study, as does the formation of phosphoglycolate residues in single-stranded lesions produced by γ-radiation (44). The observed lack of an effect of glutathione on deoxyribose 4′-oxidation chemistry raises questions about the extent of the role played by this ubiquitous cellular thiol in oxidative DNA damage in vivo. Previous studies of radioprotection by thiols suggest a model in which anionic thiols such as glutathione cannot compete effectively with oxygen for reaction with DNA radicals and that the major effect of glutathione involved scavenging of solvent radicals (5, 17, 18, 40). Our results place a narrower provision on the access of glutathione to reactions in DNA, with a limited access to intermediates in the 4′-oxidation reactions. On the basis of studies from the Tullius laboratory, the deoxyribose 4′-hydrogen atom occupies a relatively solvent-exposed position in the structure of DNA, with hydroxyl radical reactivity decreasing in the following order: 5′ > 4′ > 3′ ∼ 2′ ∼ 1′ (45). While further studies are needed to define the effects of thiols on the more exposed 5′-hydrogen atom, previous studies with neocarzinostatin point to a similar charge effect in the influence of thiols on partitioning of 5′oxidation reactions (19). This suggests that the charge of the thiol influences even the most solvent-exposed positions in DNA. Having made this argument, we must now exercise considerable care in interpreting the results of our studies with calicheamicins γ1I and Ø and acknowledge that the presence of a drug molecule bound to the site of oxidative damage could and probably does influence subsequent chemical reactions. Precedent for this phenomenon has been established with bleomycin, an iron-chelating antibiotic that cleaves DNA by 4′-hydrogen atom abstraction from deoxyribose (11, 46). Povirk and co-workers have proposed that the second cleavage event in doublestranded lesions produced by bleomycin results from reduction of the tethered Fe3+ to Fe2+ by the peroxyl radical formed initially on one DNA strand (47). The persistence of a bound enediyne at the damage site may similarly explain differences in the 4′-chemistry produced by radiation and enediynes, particularly the lack of phosphoglycolate residues in single-stranded lesions produced by enediynes (reviewed in ref 11) and the presence of phosphoglycolate as a major product of singlestranded breaks produced by γ-radiation (44). These examples serve to illustrate the complexities of drug-


induced deoxyribose oxidation and to temper our interpretation of the observed effects of thiols on the partitioning of 4′-chemistry. Participation of a bound drug molecule in the deoxyribose degradation reactions may also be responsible for the formation of a constant quantity of 3′-phosphateended fragments in our studies (Figure 4). The 3′phosphate residues are unlikely to arise by adventitious hydrolysis of abasic sites, since this would result in an increase in the level of 3′-phosphate as the proportion of abasic sites increases. It is also unlikely that the phosphate residues arise from 5′-chemistry since we did not detect nucleoside 5′-aldehyde residues at the AGGATC damage site (data not shown). One explanation for the presence of the 3′-phosphate residues lies in the recent observation by Giese and co-workers of the reversibility of the addition of molecular oxygen to the deoxyribose radical (6). If the drug remains bound at the damage site, it may shield the resulting peroxy radical from reaction with the thiol and allow oxygen to dissociate from the peroxy radical. Giese and co-workers have shown that, under anaerobic conditions, the 4′-radical will degrade to yield a 3′-phosphate-ended fragment (6). Proof of this mechanism will require detection of the residual nucleoside fragment. In summary, we have shown that, compared to their neutral counterparts, negatively charged thiols are limited in their access to reactions in the minor groove of the DNA. Furthermore, we have shown that thiols alter the partitioning of calicheamicin-induced deoxyribose 4′oxidation reactions in the absence of chemical repair of DNA or drug radicals. The results support the view that glutathione, despite its ubiquitous presence in cells and its ability to scavenge radicals in solution, may have little effect on DNA oxidation reactions.

Acknowledgment. We thank Dr. George Ellestad and Prof. K. C. Nicolaou for providing calicheamicins γ1I and Ø, respectively. This work was supported by National Institutes of Health Grants ES09980 and CA72936.

References (1) Alper, T., and Howard-Flanders, P. (1956) Role of oxygen in modifying the radiosensitivity of E. coli B. Nature 178, 978-979. (2) Willson, R. L. (1971) Pulse radiolysis studies on reaction of triacetoneamine-N-oxyl with radiation-induced free radicals. Trans. Faraday Soc. 67, 3008-3019. (3) Held, K. D., Harrop, H. A., and Michael, B. D. (1984) Effects of oxygen and sulphydryl-containing compounds on irradiated transforming DNA. III. Reaction rates. Int. J. Radiat. Biol. 45, 627636. (4) Prise, K. M., Gillies, N. E., Whelan, A., Newton, G. L., Fahey, R. C., and Michael, B. D. (1995) Role of charge in the radioprotection of E. coli by thiols. Int. J. Radiat. Biol. 67, 393-401. (5) Fahey, R. C., Prise, K. M., Stratford, M. R. L., Watfa, R. R., and Michael, B. D. (1991) Rates of repair of pBR322 DNA radicals by thiols as measured by the gas exploson technique: evidence that counter-ion condensation and co-ion depletion are significant at physiological ionic strength. Int. J. Radiat. Biol. 59, 901-907. (6) Dussy, A., Meggers, E., and Giese, B. (1998) Spontaneous cleavage of 4′-DNA radicals under aerobic conditions: Apparent discrepancy between trapping rates and cleavage products. J. Am. Chem. Soc. 120, 7399-7403. (7) Kappen, L. S., Goldberg, I. H., Frank, B. L., Worth, L. J., Christner, D. F., Kozarich, J. W., and Stubbe, J. (1991) Neocarzinostatin-induced hydrogen atom abstraction from C-4′ and C-5′ of the T residue at a d(GT) step in oligonucleotides: shuttling between deoxyribose attack sites based on isotope selection effects. Biochemistry 30, 2034-2042.

Thiols and Deoxyribose Oxidation Chemistry (8) Dedon, P. C., and Goldberg, I. H. (1992) Influence of thiol structure on neocarzinostatin activation and expression of DNA damage. Biochemistry 31, 1909-1917. (9) Epstein, J. L., Zhang, X., Doss, G. A., Liesch, J. M., Krishnan, B., Stubbe, J., and Kozarich, J. W. (1997) Interplay of hydrogen abstraction and radical repair in the generation of single- and double-strand DNA damage by the esperamicins. J. Am. Chem. Soc. 119, 6731-6738. (10) Chin, D.-H., and Goldberg, I. H. (1992) Internal hydrogen abstraction by activated neocarzinostatin: Quenching of the radical at C2 by hydrogen atom transfer from the R carbon of the adducted thiol. J. Am. Chem. Soc. 114, 1914-1915. (11) Dedon, P. C., and Goldberg, I. H. (1992) Free-radical mechanisms involved in the formation of sequence-dependent bistranded DNA lesions by the antitumor antibiotics bleomycin, neocarzinostatin, and calicheamicin. Chem. Res. Toxicol. 5, 311-332. (12) Doyle, T. W., and Borders, D. B. (1995) Enediyne antitumor antibiotics. In Enediyne Antibiotics as Antitumor Agents (Borders, D. B., and Doyle, T. W., Eds.) pp 1-15, Marcel Dekker, New York. (13) Dedon, P. C., Salzberg, A. A., and Xu, J. (1993) Exclusive production of bistranded DNA damage by calicheamicin. Biochemistry 32, 3617-3622. (14) Dedon, P. C., and Goldberg, I. H. (1990) Sequence-specific doublestrand breakage of DNA by neocarzinostatin within a staggered cleavage site. J. Biol. Chem. 265, 14713-14716. (15) Zein, N., Poncin, M., Nilakantan, R., and Ellestad, G. A. (1989) Calicheamicin γ1I and DNA: molecular recognition process responsible for site-specificity. Science 244, 697-699. (16) Hangeland, J. J., De Voss, J. J., Heath, J. A., Townsend, C. A., Ding, W.-d., Ashcroft, J. S., and Ellestad, G. A. (1992) Specific abstraction of the 5′(S)- and 4′-deoxyribosyl hydrogen atoms from DNA by calicheamicin γ1I. J. Am. Chem. Soc. 114, 9200-9202. (17) Fahey, R. C., Vojnovic, B., and Michael, B. D. (1991) The effects of counter-ion condensation and co-ion depletion upon the rates of chemical repair of poly(U) radicals by thiols. Int. J. Radiat. Biol. 59, 885-899. (18) Zheng, S., Newton, G. L., Ward, J. F., and Fahey, R. C. (1992) Aerobic radioprotection of pBR322 by thiols: Effect of thiol net charge upon scavenging of hydroxyl radicals and repair of DNA radicals. Radiat. Res. 130, 183-193. (19) Dedon, P. C., Jiang, Z.-W., and Goldberg, I. H. (1992) Neocarzinostatin-mediated DNA damage in a model AGT‚ACT site: mechanistic studies of thiol-sensitive partitioning of C4′ DNA damage products. Biochemistry 31, 1917-1927. (20) Kappen, L. S., and Goldberg, I. H. (1992) Neocarzinostatin acts as a sensitive probe of DNA microheterogeneity: Switching of chemistry from C-1′ to C-4′ by a G‚T mismatch 5′ to the site of DNA damage. Proc. Natl. Acad. Sci. U.S.A. 89, 6706-6710. (21) McAfee, S. E., and Ashley, G. W. (1992) Modulation of neocarzinostatin-mediated DNA double-strand damage by activated thiol: Deuterium isotope effects. Nucleic Acids Res. 20, 805-809. (22) Nicolaou, K. C., Li, T., Nakada, M., Hummel, C. W., Hiatt, A., and Wrasildo, W. (1994) Calicheamicin ØI1: A rationally designed molecule with extremely potent and selective DNA cleaving properties and apoptosis inducing activity. Angew. Chem., Int. Ed. 33, 183-186. (23) LaMarr, W. A., Sandman, K. M., Reeve, J. N., and Dedon, P. C. (1997) Differential effects of DNA supercoiling on radical-mediated DNA strand breaks. Chem. Res. Toxicol. 10, 1118-1122. (24) Ausubel, F. M., Brent, R., Kingston, R. E., Moore, D. D., Seidman, J. G., Smith, J. A., and Struhl, K. (1989) Current Protocols in Molecular Biology, John Wiley and Sons, New York. (25) Dedon, P. C. (2000) Protocols for the determination of binding mode: intercalation. In Current Protocols in Nucleic Acid Chemistry (Bergstrom, D., Ed.) John Wiley and Sons, New York. (26) Sugiyama, H., Xu, C., Murugesan, N., and Hecht, S. M. (1985) Structure of the alkali-labile product formed during iron(II)bleomycin-mediated DNA strand scission. J. Am. Chem. Soc. 107, 4104-4105. (27) Povirk, L. F., and Houlgrave, C. W. (1988) Effect of apurinic/ apyrimidinic endonucleases and polyamines on DNA treated with bleomycin and neocarzinostatin: specific formation and cleavage of closely opposed lesions in complementary strands. Biochemistry 27, 3850-3857. (28) Lindahl, T., and Andersson, A. (1972) Rate of chain breakage at apurinic sites in double-stranded deoxyribonucleic acid. Biochem-

Chem. Res. Toxicol., Vol. 14, No. 5, 2001 535 istry 11, 3618-3623. (29) LaMarr, W. A., Nicolaou, K. C., and Dedon, P. C. (1998) Supercoiling affects the accessibility of glutathione to DNA-bound molecules: Positive supercoiling inhibits calicheamicin-induced DNA damage. Proc. Natl. Acad. Sci. U.S.A. 95, 102-107. (30) De Voss, J. J., Townsend, C. A., Ding, W.-d., Morton, G. O., Ellestad, G. A., Zein, N., Tabor, A. B., and Schreiber, S. L. (1990) Site-specific atom transfer from DNA to a bound ligand defines the geometry of a DNA-calicheamicin γ1I complex. J. Am. Chem. Soc. 112, 9669-9670. (31) Zein, N., Sinha, A. M., McGahren, W. J., and Ellestad, G. A. (1988) Calicheamicin γ1I: an antitumor antibiotic that cleaves doublestranded DNA site specifically. Science 240, 1198-1201. (32) Walker, S., Landovitz, R., Ding, W. D., Ellestad, G. E., and Kahne, D. (1992) Cleavage behavior of calicheamicin γ1 and calicheamicin T. Proc. Natl. Acad. Sci. U.S.A. 89, 4608-4612. (33) Liang, Q., Choi, D.-J., and Dedon, P. C. (1997) Calicheamicininduced DNA damage in a reconstituted nucleosome is not affected by histone acetylation: Role of drug structure in the target recognition process. Biochemistry 36, 12653-12659. (34) Christner, D. F., Frank, B. L., Kozarich, J. W., Stubbe, J., Golik, J., Doyle, T. W., Rosenberg, I. E., and Krishnan, B. (1992) Unmasking the chemistry of DNA cleavage by the esperamicins: Modulation of 4′-hydrogen abstraction and bistranded damage by the fucose-anthranilate moiety. J. Am. Chem. Soc. 114, 87638767. (35) Mathur, P., Xu, J., and Dedon, P. C. (1997) Cytosine methylation enhances DNA damage produced by groove binding and intercalating enediynes: Studies with esperamicins A1 and C. Biochemistry 36, 14868-14873. (36) Yu, L., Golik, J., Harrison, R., and Dedon, P. (1994) The deoxyfucose-anthranilate of esperamicin A1 confers intercalative DNA binding and causes a switch in the chemistry of bistranded DNA lesions. J. Am. Chem. Soc. 116, 9733-9738. (37) Kappen, L. S., Goldberg, I. H., Wu, S. H., Stubbe, A., Worth, L., and Kozarich, J. W. (1990) Isotope effects on the sequence-specific cleavage of dC in d(AGC) sequences by neocarzinostatin: elucidation of chemistry of minor lesions. J. Am. Chem. Soc. 112, 27972798. (38) Maxam, A., and Gilbert, W. (1980) Sequencing end-labeled DNA with base-specific chemical cleavages. Methods Enzymol. 65, 499560. (39) Kappen, L. S., and Goldberg, I. H. (1992) Mismatch-induced switch of neocarzinostatin attack sites in the DNA minor groove. Biochemistry 31, 9081-9089. (40) Zheng, S., Newton, G. L., Gonick, G., Fahey, R. C., and Ward, J. F. (1988) Radioprotection of DNA by thiols: Relationship between the net charge of a thiol and its ability to protect DNA. Radiat. Res. 114, 11-27. (41) Paloma, L. G., Smith, J. A., Chazin, W. J., and Nicolaou, K. C. (1994) Interaction of calicheamicin with duplex DNA: Role of the oligosaccharide domain and identification of multiple binding modes. J. Am. Chem. Soc. 116, 3697-3708. (42) Kumar, R. A., Ikemoto, N., and Patel, D. J. (1997) Solution structure of the calicheamicin γ1I-DNA complex. J. Mol. Biol. 265, 187-201. (43) Walker, S. L., Andreotti, A. H., and Kahne, D. E. (1994) NMR characterization of calicheamicin γ1I bound to DNA. Tetrahedron 50, 1351-1360. (44) Henner, W. D., Rodriguez, L. O., Hecht, S. M., and Haseltine, W. A. (1983) γ-Ray induced deoxyribonucleic acid strand breaks. 3′ Glycolate termini. J. Biol. Chem. 258, 711-713. (45) Balasubramanian, B., Pogozelski, W. K., and Tullius, T. D. (1998) DNA strand breaking by the hydroxyl radical is governed by the accessible surface areas of the hydrogen atoms of the DNA backbone. Proc. Natl. Acad. Sci. U.S.A. 95, 9738-9743. (46) Stubbe, J., and Kozarich, J. W. (1987) Mechanisms of bleomycininduced DNA degradation. Chem. Rev. 87, 1107-1136. (47) Steighner, R. J., and Povirk, L. F. (1990) Bleomycin-induced DNA lesions at mutational hot spots: Implications for the mechanism of double-strand cleavage. Proc. Natl. Acad. Sci. U.S.A. 87, 8350-8354.

TX0100082

Thiols Alter the Partitioning of Calicheamicin-Induced Deoxyribose 4

Recommend Documents