Origin of malondialdehyde from DNA degraded by Fe(II) x bleomycin.

Ferrous bleomycin is known to break DNA efficiently in vitro in the presence of O2, giving rise to ologonucleotides, bases, and compounds resembling malondialdehyde in their chromogenic reaction with 2-thiobarbituric acid. Chromatography of radiolabeled DNA reaction mixtures resolves three kinds of malondialdehyde-like products, related by sequential conversions. The first chromogenic product is linked to DNA, and its formation does not entail the release of a base. It decomposes readily to the second product, a compound containing the base and deoxyribose carons 1'-3'. Hydrolysis of either product yields the third, which is indistinguishable from authentic malondialdehyde. These findings suggest that the oxygen-dependent cleavage of DNA by Fe(II) x bleomycin can begin with the rupture of the deoxyribose 3'-4'-carbon bond. The initiation of these events is concurrent with the initiation of another mode of DNA degradation, involving the release of free base alone, in a yield similar to that of chromogen.

Material resembling malondialdehyde, which is derived from lipid oxidation, has long provided food chemists with a sensitive assay of rancidity (12) by means of its chromogenic reaction with 2-thiobarbituric acid, yielding an intensely colored adduct (EW = 1.6 X 10' (13)) (Fig. 1). The colored adduct characteristic of malondialdehyde is also formed from products of DNA, degraded by ionizing radiation (9, 10, 14, 15), or by aerobic Fe(I1) solutions (16), as well as by bleomycin. The possibility that the chromogen from DNA might not be malondialdehyde, but a precursor of malondialdehyde, was appreciated by Kapp and Smith (lo), who found that the chromogen precipitated with X-irradiated DNA, while authentic malondialdehyde did not.
In an attempt to determine the chemistry of DNA breakage by ferrous bleomycin, we tried to isolate physically the malondialdehyde reportedly produced. Our recovery of malondialdehyde in distillates of bleomycin-treated DNA reaction mixtures was so inferior to that from model mixtures containing authentic malondialdehyde that we re-examined our reaction mixtures by chromatographic fractionation. Our analysis indicates that in bleomycin reaction mixtures, the chromogen previously considered to be malondialdehyde is not malondialdehyde, but rather consists of two intermediates, each containing the deoxyribose carbons 1'-3' derived from an initial drug-induced cleavage of the 3'-4' carbon bond of the sugar. These intermediates can react with 2-thiobarbituric acid to give an adduct identical with the one produced by malondialdehyde, or they can undergo acid or base hydrolysis to produce malondialdehyde.

EXPERIMENTAL PROCEDURES
Preparation, Assay, and Deriuatization of Malondialdehyde-Malondialdehyde was generated (17) from malonaldehyde bis(dimethy1 acetal) (Aldrich) by treating a 5 mM solution with 0.1 N HC1 a t 50°C for 1 h in a stoppered flask. The solution was then cooled and diluted 5-fold with water. Stock solutions were stored a t 10°C for up to 2 months with no loss, based on colorimetric assay with 2thiebarhituric acid (13), and samples eluted simply from Sephadex G-10 columns (Fig. 2).
We assayed malondialdehyde by heating a sample to 92°C with excess 2-thioharhituric acid at pH 2 to 3 for 20 min, cooling, and measuring A331. Samples (50.3 m l ) were made up to 0.8 ml with a solution containing 42 mM 2-thiobarhituric acid and 1 mM EDTA. Assay mixtures containing 0.05 to 8.0 nmol of rnalondialdehyde obeyed Beer's law. The published value of E M = 1.6 X lo5 M" cm ' (13) agrees with our assay standardization, for which we assumed that malondialdehyde was obtained quantitatively from the bis(dimethy1 acetal). with a Cary model 118 recording spectrophotometer thermostated a t The course of color development was monitored, when appropriate, 86°C. Preheated cuvettes containing 3 ml of 35 mM Z-thiobarhituric acid and 1 mM EDTA received 0.2-ml samples of authentic malondialdehyde or aerobic DNA reaction mixtures containing Fe(II).hleomycin. They were then tightly stoppered, and A5.v~ was measured a t 1-min intervals. First order rate constants were calculated by computer, using least squares criteria, as described (161. The 2-thiobarbituric acid adduct of malondialdehyde was prepared according to Sinnhuber et al. (18) from the bis(dimethy1 acetal) in 1 N HC1. A yield of 84% as washed crystals was obtained, with no spectrally detectable excess of 2-thiobarbituric acid. Aliquots were dissolved in H20 for optical spectroscopy, and in perdeuterated dimethyl sulfoxide (299.5 atom B 'H; Aldrich) for 'H NMR. A similar preparation of crystals was made from material distilled at 80°C under a stream of N2 (18), from a 100-ml reaction mixture of 0.5 mM bleomycin, 2.5 mM DNA, and 1.0 mM Fe(I1) in 20 mM sodium phosphate buffer, pH 7.0, subsequently adjusted with HC1 to pH 1.5 prior to distillation. Optical spectra were obtained using a Cary model 14R recording spectrophotometer with 1-cm pathlength cuvettes. Fourier transform 'H NMR spectra were obtained with a Briiker model WH-360 spectrometer at the University of Pennsylvania Regional NMR Center.
Radiolabeled DNA-Variously thymidine-labeled DNAs were extracted from purified bacteriophage X, grown in a Thy-host, Escherichia coli strain RS15, kindly provided by J. A. Wechsler and R. A. Scalafani, Department of Biology, University of Utah, Salt Lake City (unpublished strain). Its genotype is tonA lamB str thyA deoC (or deoB) (Xrr857 Sam7). This heat-inducible, lysis-defective bacteriophage X lysogen efficiently utilized exogenous thymidine in DNA synthesis.
Reaction mixtures were assembled in unstoppered polypropylene tubes (10 X 75 mm), which were equilibrated to the desired temperatures. Unless otherwise specified, they were incubated at O"C, and  (Table I), reactions were permitted to approach completion (>99%) before subsequent treatments. Some completed reaction mixtures were then heated to 50°C for 10 min; others were exposed to 0.1 N HCI or NaOH at 92°C for 10 min, then cooled and neutralized. For controls, unproductive reaction mixtures of similar composition were obtained by adding the DNA last, instead of the Fe(I1) (16).
The stability of the chromogenic products and their 2-thiobarbituric acid adducts to oxygen was demonstrated by addition of anaerobic 2-thiobarbituric acid solution to a completed reaction mixture that had been equilibrated with argon. No difference in AIa2 was observed between this and the color-forming reaction under air. fractionated by precipitation of DNA with ethanol. received half the Fractionation of Reaction Products-Reaction mixtures to be usual concentrations of bleomycin and Fe(II), in order to diminish the formation of small, ethanol-soluble oligonucleotides. One equivalent of chromogen was formed per 56 eq of oligonucleotide phosphate provided. Using an ethanol ice bath, reaction constituents were equ& ibrated to -5"C, which was measured with a YSI model 42 Sc Telathermometer and a Teflon-sheathed, solution-type thermistor probe. Those few reaction mixtures that froze were discarded. Reaction mixtures (100 pl) contained 18 mM sodium phosphate buffer, pH 7.0, 1 mM DNA containing [6-3H]thymidine (1.6 Ci/mol), 0.1 mM bleomycin, and 0.12 mM Fe(I1). The last was added to initiate the reaction, which was run for 5 min at -5°C before addition of 1 M LiCl and 0.25 mM undigested nonradioactive DNA, followed at once by the addition of 3 volumes of cold (-5'C) ethanol. The mixtures were centrifuged at 18,000 X g for 2 h at -25"C, and the precipitates were redissolved in 50 p1 of 20 mM sodium phosphate buffer, pH 7. Supernatants and precipitates were assayed for radioactivity and P-thiobarbituric acid-reactive material. Each assay mixture was supplemented to contain the same amount of LiCl and ethanol present in the supernatants.
Sephadex G-10 columns served to fractionate some products by gel filtration and others by adsorption chromatography (23). Columns (18 X 1 cm diameter) were equilibrated and eluted with 20 mM sodium phosphate buffer, pH 7.0, at 6°C. DNA eluted at 6.0 ml, and 'Hz0 eluted at 11.5 ml , which were taken to indicate void and included volumes, respectively. The reaction mixtures analyzed were incubated at 0°C. Reaction aliquots (0.6 ml) were applied to the column, either at once, after heating to 50"C, or after base hydrolysis with 0.1 N NaOH at 92'C for 10 min followed by neutralization. Columns were eluted at a rate of -4 ml/h with a hydrostatic pressure head of 12 to 14 cm. Thirty-nine 0.66-ml fractions were collected, followed by 3.3ml fractions. Of each fraction, 0.3 ml was assayed with 2-thiobarbituric acid, and the remainder was mixed with 5.0 ml of TT-21 scintillant (Yorktown) and assayed for radioactivity by scintillation spectrometry. Recovery of radioactivity was always complete (95 to 105'%), but recovery of chromogenic activity gradually decreased (from 90%) as the column was re-used, except that the chromogen in hydrolyzed samples was always completely recovered.
Reversed-phase thin layer chromatography of radioactive DNA reaction products was carried out on Analtech RPS 0.25-mm plates, developed with ascending 4 M ethanol, 15 mM sodium phosphate buffer, pH 7, at 6°C for 4.5 h. The solvent front moved 15 cm. Reaction mixtures contained 0.2 mM [6-'H]thymidine-labeled DNA (30 Ci/mol of nucleotide), 50 ELM bleomycin, 50 p~ Fe(II), and 18 mM sodium phosphate buffer, pH 7.0, plus 5 p~ authentic [methyl-Clthymine (55 Ci/mol) used as an internal standard. They were incubated at 0°C and analyzed either before or after base hydrolysis at 92°C in a sealed glass capillary. Chromatography was started promptly after applying a 1-p1 aliquot to the thin layer at 6°C without drying. Fractions (0.5 cm) were scraped from the support plate for scintillation counting in Aquasol (New England Nuclear). Although these 2-thiobarbituric acid adducts appear identical, several observations lead us to conclude that they arise from different precursors. The most evident difference between authentic malondialdehyde and the chromogen from DNA is in the kinetics of their reactions with 2-thiobarbituric acid. At 86"C, the adduct from malondialdehyde forms homogeneously with tlp = 2.2 min, while the reaction with the DNA products is >90% complete in 2 min, the time of our earliest observations. If the Fe(1I). bleomycin/DNA reaction mixtures are exposed to 0.1 N HC1 or 0.1 N NaOH for 10 min at 92OC before the 2-thiobarbituric acid reaction, their rates of subsequent color development are equal to the rate with authentic malondialdehyde.

2-Thiobarbituric Acid Reaction Products and Kinetics-When Fe(I1) is added to aerobic mixtures of bleomycin and
Another difference, which is similarly nullified upon treatment of reaction products with acid or base, is seen in the stability of chromogenic activity. The DNA-derived chromogen is lost from reaction mixtures at 6°C with t,,P = 70 h, while the chromogenic activity of authentic malondialdehyde added to unproductive control incubations is stable, like our 1 mM malondialdehyde stock solutions.

WAVELENGTH (nm) CHEMICAL SHIFT (ppm)
FIG. 1. Structure and spectra of the 2-thiobarbituric acid adduct with malondialdehyde. The left panel depicts the optical spectra, and the right panel depicts the 'H NMR spectra of the 2thiobarbituric acid adducts with: A , authentic malondialdehyde, or B, the malondialdehyde-like product of DNA degradation by oxygenated Fe(II).bleomycin. The baselines have been arbitrarily offset. The optical spectra were obtained from: A, an unproductive 200-pl reaction mixture (0.1 m M Fe(I1) added to 0.1 mM bleomycin, followed by 1 m M DNA), supplemented with 2.2 nmol of authentic malondialdehyde, or B, a productive reaction mixture. Each was treated with 2-thiobarbituric acid. The shoulder at 453 nm is due to a reaction of 2-thiobarbituric acid with the iron present. The NMR spectra were obtained from the crystallized products of 2-tbiobarbituric acid refluxed with malondialdehyde bis(dimethy1 acetal) ( A ) , or with the distillate of a bleomycin/DNA reaction mixture ( B ) , as described under "Experimental Procedures." NMR sample A contained 10 mg of adduct/ml, and sample E contained 0.1 mg/ml. The indicated chemical shifts were determined with respect to tetramethyl silane. The structure proposed for the malondialdehyde adduct (18) is shown with our NMR assignments. The exchangeable protons give a broad resonance at 9.06 ppm (not shown). Doublet and triplet coupling constants are 14Hz.
A minor but consistent difference is also seen in the effect of ethanol on the yield of adduct. Addition of 1 to 5 M ethanol to the 2-thiobarbituric acid assay mixture has no effect on color development with malondialdehyde, but enhances the yield from the DNA products by 8%.
These results are consistent with the hypothesis that reaction mixtures contain a product that may be converted to malondialdehyde by hydrolysis. The product is less stable, and reacts faster than does malondialdehyde in forming the 2-thiobarbituric acid adduct.
Fractionation of 2-Thiobarbituric Acid Chromogens-Malondialdehyde does not co-precipitate with DNA in cold ethanol, but when bleomycin-treated DNA is ethanol-precipitated, as described in the legend to Table I, as much as 88% of its chromogenic product is recoverable in the precipitate. Such completeness of precipitation is lost if reactions are performed at higher temperatures, for longer times, or contain a smaller ratio of DNA to bleomycin.
A more detailed analysis of the DNA degradation products was obtained by fractionating reaction mixtures on a Sephadex G-10 column (Fig. 2). Four chromogenic fractions were resolved, each present in an amount depending on the conditions of the reaction and of postreaction treatment.
When a 0°C reaction mixture is applied directly to the column (Fig. 2b) A rechromatography experiment suggests that the chrom- ogen appearing in the region between peaks 1 and 3 (Fig. 2, b, e, and h) is due to the release of chromogen from DNA during chromatography. When the central Peak 1 fractions of a reaction mixture were applied within 1 h of collection to an identical column, the chromogenic material eluted mainly as Peak 1, but 15% of it eluted as Peak 3.
When DNA cleavage reactions are run at room temperature or warmed after their completion, the transfer of chromogen from Peak 1 to Peaks 3 and 4 is enhanced. It is almost complete in mixtures heated to 50°C (at pH 7) for 10 min (Fig.  2c). Chromogenic material eluting like authentic malondialdehyde is now manifest in a minor shoulder preceding Peak 3; no trace of color is obtained between the vestige of Peak 1 and this shoulder.
When reaction mixtures are treated with 0.1 N NaOH or HC1 at 92°C for 10 min, then chilled, neutralized, and chromatographed (Fig. 2d), all chromogenic material elutes like malondialdehyde, in Peak 2. The rate of color development in the 2-thiobarbituric acid assay of these fractions is also characteristic of authentic malondialdehyde and is unlike that of the chromogens eluting elsewhere (Peaks 1,3, and 4).
Fate of Thymine Radioactivity-The origin of the materials eluting in Peaks 2, 3, and 4 was investigated by fractionating reaction products of radioactive DNA. A reaction run at -5"C, stopped at <75% completion, and ethanol-precipitated released 0.13 eq of ethanol-soluble [6-3H]thymine radioactivity/ mol of 2-thiobarbituric acid chromogen formed (Table I). This amount of radioactivity is one-third of that expected if chromogen production were contingent on base release, since thymine normally accounts for about half of the base released by bleomycin from typical DNA's (5-7, 15). At slightly higher temperatures (6°C; Fig. 2, b and h), 0.43 eq of [€k3H]thyrnine are released per mol of chromogen formed. Other experiments (below) indicate that the deoxyribose-base linkage also remains intact in material cleaved from DNA by bleomycin reactions analyzed at 6°C.
Intact DNA labeled with [6-3H]-or [methyZ-'4C]thymidine elutes from Sephadex G-10 columns entirely in Peak 1. After incubation with Fe(I1). bleomycin at 0°C and chromatography a t 6°C (Fig. 2h), about 10% of the label is eluted subsequent to Peak 1, just like the chromogens described above. Heating to 50°C (Fig. Xi) doubles the amount of label transferred from Peak 1 to Peak 3, while reducing the amount eluting between them. Hydrolysis with base (Fig. 2j)  Sephadex G-10 column fractionation of Fe(I1)-bleomycin-degraded DNA products. The DNA cleavage reactions to be fractionated were incubated at O'C, and the reaction mixtures were applied to the column directly or after the indicated treatments. Columns were run at 6°C and pH 7.0, as described under "Experimental Procedures." a, the elution of the indicated reference compounds (arbitrary ordinate); b to d, the elution of chromogenic incubation products; e to g, the elution of radioactivity from DNA with [U-'4C]thymidine; h toj, the elution of radioactivity from DNA with [6-3H]thymidine. The ordinates indicate the radioactivity per fraction and the A 6 3 2 developed in assaying a 0.3-ml aliquot with 2thiobarbituric acid. The arrows indicate the elution position of malondialdehyde (MDA). The peaks are numbered for reference to the text. Reference compounds tested but not shown here are: formate, thymidine, and cytosine, which elute with peaks centered at 10, 13, and 15.5 ml, respectively.
chromogenic Peak 3 product is not demonstrable using Sephadex G-10 columns, since thymine itself elutes in Peak 3.
Thin layer chromatography of Fe(I1). bleomycin/DNA reaction mixtures reveals that much of the thymine label first released from DNA partitions not as free thymine, but as a separable species that is susceptible to hydrolysis, and then yields a product with the mobility of the free base. When a digest of [6-3H]thymidine-labeled DNA is fractionated by reversed-phase thin layer chromatography (Fig. 3 ) , most of the radioactivity remains associated with oligonucleotides near the origin, but the remainder is found in two mobile fractions. One (RF = 0.73) co-migrates with an authentic [14C]thymine internal marker; the other is less mobile (RF = 0.56). Unlike thymidine (RF = 0.75; not shown), this less mobile fraction is susceptible to hydrolysis in 0.1 N NaOH at 92°C for 10 min. When a reaction aliquot is hydrolyzed before chromatography, the fraction having RF = 0.56 is absent, but the radioactivity found to co-migrate with thymine is enhanced by the amount otherwise found in the missing fraction. The ratio of these two products is variable and depends on the separation system used. We obtained results similar to those of Povirk et al. ( 7 ) when we used a cellulose thin layer, but found that the recovery of the non-thymine (RF = 0.56) product is much enhanced with reversed-phase chromatography.

Fate of Deoxyribose Radioactivity-On Sephadex G-10
fractionation, all the radioactivity in digests of [5'-3H]thymidine-labeled DNA eluted with the oligonucleotide fraction, as in Fig. 2, Peak 1, unless the completed reaction had been hydrolyzed prior to fractionation. The radioactivity then released (5%) eluted as 3H20 and probably derives from tritium exchange with the solvent, No radioactive formate was detected in any of our reaction mixtures.
When DNA containing [U-'4C]thymidine is incubated with bleomycin but not hydrolyzed (Fig. 2, e and f ), the distribution of label in the Sephadex G-10 effluent is qualitatively like that of thymine-labeled DNA (Fig. 2, h to j ) . When the reaction mixture was first hydrolyzed, radioactivity was also found in the Peak 2 eluant (Fig, 2g) Fig. 2, but the detection of DNA tritium at the sample origin is impaired by quenching.
The release of deoxyribose fragments was studied quantitatively using DNA doubly labeled with [U-'4C]thymidine and [6-3H]thymine. Fe(I1) ebleomycin digests of this DNA which were otherwise untreated (O'C), warmed to 50"C, or base-hydrolyzed were fractionated on Sephadex G-10 columns, and the ratios of isotopes released permitted calculation of the fraction of thymidine carbons appearing in Peak 2 and 3 eluants. Thus, for example, if only the thymine base moiety were released in a particular reaction, the fraction of [6-"Hlthymine label released would be twice that of the [U-l4C]t,hyrnidine label released, since only half of the thymidine carbon atoms are in the base. Such calculations are interpreted cautiously: they express averages that reflect a possible mixture of products.
The results of these experiments are summarized in Table  11. When a 0°C reaction mixture is applied directly to the column, the material eluting after Peak 1 contains 8 (of 10) thymidine I4C carbons for every [6-3H]thymine equivalent released. The 5'-3H (and, presumably, the 5"carbon) was seen to remain associated with the oligonucleotide fraction (Peak I ) , and no thymidine was detected.
When reactions run at 0°C are heated to 50°C before column chromatography, our calculations indicate that 92% of the increase in released 14C radioactivity derives from thymine and 8% derives from deoxyribose products. The number of thymidine carbons found per eq of [6-3H]thymine, either in Peak 3 or in all fractions subsequent to Peak 1, is now 6. This would result if, a t 50"C, for every new fragment containing 8 thymidine carbons, four fragments now appeared, containing only 5 carbons. Although the release of thymine label more than doubles on heating to 50"C, no significant increase in total chromogen is seen.
When the same reaction products are hydrolyzed before column chromatography, little additional radioactivity is released, but the products are altered so that only Peak 2 is chromogenic. It now includes some deoxyribose radioactivity that would otherwise have appeared in Peak 3 (Fig. 2, f and   g). The overall I4C:'H fractional release ratio remains 0.6, but when Peak 3 alone is considered, the ratio is 0.5, indicating that Peak 3 now contains only thymine. The deoxyribose carbons now elute as malondialdehyde in Peak 2.
A parallel experiment was done using DNA labeled with [1',2',n~ethyZ-~H]-and [methyZ-'4C]thymidine (Fig. 4). DNA cleaved a t 0°C released equal fractions of both labels, which is consistent with a continuing association of base with deoxyribose carbons 1'-3'. However, in this experiment, not all the released 3H appears in Peak 3 about 20% of the released "H elutes as 3H20. An exchange of 2'-3H with solvent could result from enolization of a 3'-aldehyde, which is a possible structure for the base-sugar fragment. As expected, heating the reaction mixture to 50°C at pH 7 releases additional radioactivity, with C predominating, but no additional 3Hz0 appears. The 'H20 detected exceeds by 5-fold that found in similarly treated controls containing undigested DNA. Base hydrolysis releases 9% of DNA tritium as 'H20 after incubation with bleomycin, but releases very little from untreated DNA. These experiments are refractory to more complete interpretation, since the quantitative distribution of tritium in this thymidine is not precisely known.
The release at 50°C of chromogen from DNA (Fig. 2c) without the release of equivalent thymidine deoxyribose label (Table 11) requires comment. It appears that the chromogen formed at 0°C but released from DNA at 50°C must derive mainly from nucleosides other than thymidine. Conversely, the chromogen released at 6°C might be expected to derive mainly from thymidine, and the ratio of chromogen recovered (Fig. 221) to the thymine and deoxyribose radioactivity released (Fig. 2, e and h) indicates that this is so. Thus, it appears that release from DNA of different nucleoside degradation products is differentially affected by incubation temperature.

DISCUSSION
The fist observed effects of bleomycin on DNA were a reduction in melting temperature and sedimentation velocity, reflecting DNA polymer cleavage in vivo and in vitro (26).
Muller et al. (4) observed the formation of aldehyde groups and proposed this to be a probable consequence of the liberation of free thymine. They titrated 0.58 aldehyde eq per thymine released. The aldehydic species was characterized by Kuo and Haide (8) as malondialdehyde-like in forming the characteristic 2-thiobarbituric acid adduct. They noted the similarity of products of DNA damaged by x-rays and by [VI b'c. SCHEME 1. Steps in DNA degradation following ferrous bleomycin treatment. Two modes of DNA disintegration have been resolved, one releasing a chromogenic compound comprising the carbon atoms of the base and a %carbon deoxyribose fragment, and the other releasing free base. The early DNA products, [XI and [yl, are ethanol-insoluble, but upon warming, release the fragments depicted. The products enclosed in brackets have not been definitively characterized, but the carbon linkages depicted in the 8-carbon nucleoside fragment seem probable in light of its precursor and products.
Haidle et al. (3) and subsequent workers (5-7) observed liberation of all four bases. Haidle et al. (3) proposed that the bleomycin acted primarily by removing bases, as an alkylating agent. Such lesions would then render the DNA polymer unstable. Closer examination of the products of the drugdegraded DNA by Povirk et al. (7,27) revealed further complications: the DNA contained alkali-labile sites in addition to breaks, and the base-like products included species that were distinguishable from authentic free base. The alkali-labile sites were attributed to base-free deoxyribose residues (27).
The possibility that the breaks occurring without alkali treatment give rise to a derivatized base species is suggested by the discovery (14) among the products of irradiated deoxynucleotides, of compounds comprising base and deoxyribose fragments, as well as the malondialdehyde-like chromogen. Products originating in DNA degradation and in lipid oxidation resemble malondialdehyde in their product with 2thiobarbituric acid but appear by other criteria to be different from, though possibly precursors of, malondialdehyde (10,28). Scheme 1 summarizes our interpretation of the reactions initiated by bleomycin. The early products of DNA degradation, X and Y, are relatively stable at low temperatures and are macromolecular. They must contain lesions, however, which predispose them to two modes of disintegration. Even a t low temperatures, Y slowly releases a compound containing the carbon atoms of the nucleic base and 3 of the 5 deoxyribose carbons. This compound is relatively stable in its chromogenic properties, and a hydrolysis procedure is necessary to cleave the deoxyribose fragment, as malondialdehyde, from the nucleic base moiety. (At room temperature and pH 7, the stoichiometry of chromogen formed per DNA cleavage is about 1.)' Following cleavage of the deoxyribose 3'-4' bond, carbons 5' and, probably, 4' remain associated with the degraded oligomer. A mechanism for such a cleavage has been proposed (29).
The other mode of disintegration takes effect when reaction ' R. M. Burger and S. B. Horwitz, manuscript in preparation. The 8-carbon compound is released at low temperatures (-6°C); the free base is released at higher temperatures. The former may be hydrolyzed, yielding base and malondialdehyde as shown. The C-3'aldehyde and C-4'-hydroxyl functions shown are conjectured, to account for patterns of tritium exchange from carbon atoms 2' and 5'. Hydrogens bonded to carbon or nitrogen have been omitted from these drawings. The released bases are shown as thymine, the one most often released; the polymer cleavage event is shown occurring adjacent to a 3"guanidylphosphate, the highly preferred site.
mixtures are warmed, releasing free thymine and, presumably, other bases. This lesion yields no malondialdehyde-like chromogen, but the removal of nucleic base renders the residual phosphodeoxyribose oligomer susceptible to cleavage (27) in moderate alkali. Thus, under appropriate conditions, DNA incubated with oxygenated ferrous bleomycin may release malondialdehyde and free nucleic bases, a compound comprising the nucleic base and three deoxyribose carbon atoms, a mixture of the latter two, or nothing. The hypothesis that DNA cleavage results as a consequence of free base release is only partly true, since the released compound combining base and deoxyribose carbons 1'-3' preserves the glycosidic bond. The hypothesis that the free base detected is a breakdown product of this compound is, likewise, only partly true, since base is also released independently.
The events preceding the cleavage reactions were elucidated by Sausville et al. (30,31), who demonstrated the necessary participation of both Fe(I1) and 0 2 in the bleomycin-catalyzed reaction. They appreciated the radiomimetic aspects of bleomycin activity and proposed that such free radicals as .OH and -Onmight be formed as a consequence of Fe(I1). bleomycin oxidation, and that these might attack DNA. Indeed, the detection of free radicals using spin traps in aerobic Fe(I1). bleomycin mixtures (32,33) has been interpreted as c o n f i iing the proposal that .OH or . 0 2 -accumulate and damage DNA in a way analogous to that resulting from irradiation or from treatment with aerobic Fe(I1) solutions (34-36). Although this proposal is attractive, there is no compulsion to assume that radicals formed from 02-Fe(II).bleomycin autooxidation are the species responsible for the specific DNA cleavage outlined in this paper. The oxygenated complex itself (2) may be the active species or may give rise to one that is not necessarily a free radical.
Oxidation reactions are well known for the heterogeneity of their pathways and products, so the observed release of base, both with and without the chromogenic deoxyribose fragment, does not in itself require that bleomycin inflict upon DNA more than one kind of primary lesion. However, it seems likely that the early intermediates in Scheme 1, X and Y, are Malondialdehyde from DNA Degraded by Fe(II) -Bleomycin different, since the prior formation of one final product does not seem to prejudice the yield of the other.