Site-specific OH attack to the sugar moiety of DNA: a comparison of experimental data and computational simulation.

Little computational or experimental information is available on site-specific hydroxyl attack probabilities to DNA. In this study, an atomistic stochastic model of OH radical reactions with DNA was developed to compute relative OH attack probabilities at individual deoxyribose hydrogen atoms. A model of the self-complementary decamer duplex d(CCAACGTTGG) was created including Na(+) counter ions and the water molecules of the first hydration layer. Additionally, a method for accounting for steric hindrance from nonreacting atoms was implemented. The model was then used to calculate OH attack probabilities at the various C-H sites of the sugar moiety. Results from this computational model show that OH radicals exhibit preferential attack at different deoxyribose hydrogens, as suggested by their corresponding percentage solvent-accessible surface areas. The percentage OH attack probabilities for the deoxyribose hydrogens [1H(5')+2H(5'), H(4'), H(3'), 1H(2')+2H(2'), H(1')] were calculated as approximately 54.6%, 20.6%, 15.0%, 8.5% and 1.3%, respectively, averaged across the sequence. These results are in good agreement with the latest experimental site-specific DNA strand break data of Balasubramanian et al. [Proc. Natl. Acad. Sci. USA 95, 9738-9742 (1998)]. The data from this stochastic model suggest that steric hindrance from nonreacting atoms significantly influences site-specific hydroxyl radical attack probabilities in DNA. A number of previous DNA damage models have been based on the assumption that C(4') is the preferred site, or perhaps the only site, for OH-mediated DNA damage. However, the results of the present study are in good agreement the experimental results of Balasubramanian et al. in which OH radicals exhibit preferential initial attack at sugar hydrogen atoms in the order 1H(5')+2H(5') > H(4') > H(3') > 1H(2')+2H(2') > H(1').     

Radiolytic signature of Z-DNA.

Ionizing radiations induce various damages in DNA via the hydroxyl radical OH. generated by the radiolysis of water. We compare here the radiosensitivity of B- and Z-DNA, by using a Z-prone stretch included in a plasmid. In the supercoiled plasmid, the stretch is in the Z-form, whereas it is in the B-form when the plasmid is relaxed. Frank strand breaks (FSB) and alkali-revealed breaks (ARB) were located and quantified using sequencing gel electrophoresis. We show that B- and Z-DNA have the same mean sensitivity towards radiolytic attack, for both FSB and ARB. Nevertheless, the guanine sites are more sensitive, and the cytosine sites less sensitive in Z- than in B-DNA, leading to a characteristic signature of the Z-form. The comparison of experiments with the outcome of a Monte Carlo simulation of OH. radical attack suggests that transfer of initial damage from a guanine base to its attached sugar or the adjacent 3' cytosine is more important in Z-DNA than in B-DNA.     

RADACK, a stochastic simulation of hydroxyl radical attack to DNA

RADACK was conceived to simulate the radiation-induced attack to different DNA forms and complexes. It allows to separately calculate the probability of attack to each reactive atom of the sugar and of the base and takes into account the sequence-dependent structure of DNA as known from crystallographic or NMR studies or resulting from molecular modelling. The calculations are aimed to assess sequence-, structure- and ligand-dependent modulation of damages of sugar and bases, leading to single strand breaks (frank strand breaks, FSB) and alkali-labile base modifications (alkali-revealed breaks, ARB), respectively. The modelling procedure and the results of simulations for some representative structures (B, Z and quadruplex forms) are here described and discussed. The calculated relative probabilities of OH* radical attack to all reaction sites are compared to experimental FSB and ARB values. By a fitting procedure, the relative efficiencies of conversion of the C4' and C5'-centred radicals into FSB, epsilon (C4'): epsilon (C5'), and the relative efficiencies of base radicals- to- ARB conversion, epsilon(T) : epsilon(A) : epsilon(C) : epsilon(G), are then deduced for each DNA form. The ability of the model to account for the distribution of damages in DNA-ligand complexes is proven by its successful application to two DNA-protein systems : the lac repressor-lac operator complex and the nuclcosome core.     

Evidence against the generation of free hydroxyl radicals from the interaction of copper,zinc-superoxide dismutase and hydrogen peroxide

Prior spin trapping studies reported that H(2)O(2) is metabolized by copper,zinc-superoxide dismutase (SOD) to form (.)OH that is released from the enzyme, serving as a source of oxidative injury. Although this mechanism has been invoked in a number of diseases, controversy remains regarding whether the hydroxylation of spin traps by SOD is truly derived from free (.)OH or (.)OH scavenged off the Cu(2+) catalytic site. To distinguish whether (.)OH is released from the enzyme, a comprehensive EPR investigation of radical production and the kinetics of spin trapping was performed in the presence of a series of structurally different (.)OH scavengers including ethanol, formate, and azide. Although each of these have similar potency in scavenging (.)OH as the spin trap 5, 5-dimethyl-1-pyrroline-N-oxide and form secondary radical adducts, each exhibited very different potency in scavenging (.)OH from SOD. Ethanol was 1400-fold less potent than would be expected for reaction with free (.)OH. The anionic scavenger formate, which readily accesses the active site, was still 10-fold less effective than would be predicted for free (.)OH, whereas azide was almost 2-fold more potent than would be predicted. Analysis of initial rates of adduct formation indicated that these reactions did not involve free (.)OH. EPR studies of the copper center demonstrated that while high H(2)O(2) concentrations induce release of Cu(2+), the magnitude of spin adducts produced by free Cu(2+) was negligible compared with that from intact SOD. Further studies with a series of peroxidase substrates demonstrated that characteristic radicals formed by peroxidases were also efficiently generated by H(2)O(2) and SOD. Thus, SOD and H(2)O(2) oxidize and hydroxylate substrates and spin traps through a peroxidase reaction with bound (.)OH not release of (.)OH from the enzyme.     

Low-energy (5-40 eV) electron-stimulated desorption of anions from physisorbed DNA bases

We present the results of experiments on anion desorption from the physisorbed DNA bases adenine, thymine, guanine and cytosine induced by the impact of low-energy (5-40 eV) electrons. Electron bombardment of DNA base films induces ring fragmentation and desorption of H(-), O(-), OH(-), CN(-), OCN(- ) and CH(2)(-) anions through either single or complex multibond dissociation. We designate the variation of the yield of an anion with electron energy as the yield function. Below 15 eV incident electron energy, bond cleavage is controlled mainly by dissociative electron attachment. Above 15 eV, the portion of a yield function that increases linearly is attributed to nonresonant processes, such as dipolar dissociation. A resonant structure is superimposed on this signal around 20 eV in the anion yield functions. This structure implicates dissociative electron attachment and/or resonant decay of the transient anion into the dipolar dissociation channel, with a minimal contribution from multiple inelastic electron scattering. The yields of all desorbing anions clearly show that electron resonances contribute to the damage of all DNA bases bombarded with 5-40 eV electrons. Comparison of the ion yields indicates that adenine is the least sensitive base to slow electron attack. Electron-irradiated guanine films exhibit the largest yields of desorbed anions.     

Reduction in free-radical-induced DNA strand breaks and base damage through fast chemical repair by flavonoids

This paper provides evidence that dietary flavonoids can repair a range of oxidative radical damages on DNA, and thus give protection against radical-induced strand breaks and base alterations. We have irradiated dilute aqueous solutions of plasmid DNA in the absence and presence of flavonoids (F) in a "constant *OH radical scavenging environment", k of 1.5 x 10(7) s(-1) by decreasing the concentration of TRIS buffer in relation to the concentration of added flavonoids. We have shown that the flavonoids can reduce the incidence of single-strand breaks in double-stranded DNA as well as residual base damage (assayed as additional single-strand breaks upon post-irradiation incubation with endonucleases) with dose modification factors of up to 2.0+/-0.2 at [F] < 100 microM by a mechanism other than through direct scavenging of *OH radicals. Pulse radiolysis measurements support the mechanism of electron transfer or H* atom transfer from the flavonoids to free radical sites on DNA which result in the fast chemical repair of some of the oxidative damage on DNA resulting from *OH radical attack. These in vitro assays point to a possible additional role for antioxidants in reducing DNA damage.     

Pulse-radiolysis studies of 8-methoxypsoralen.

Both eaq- and .OH have been found to react with 8-methoxypsoralen (8-MOP), giving rate-constants of 1.1 X 10(10) M-1 s-1. Transient spectra of products from the reactions of eaq-, .OH with 8-MOP have been characterized. Rate-constants for the oxidation by 8-MOP of reduced and oxidized DNA bases have also been measured and found to lie in the range 3-6 X 10(9) M-1 s-1. Oxidation of reduced bases occurs by electron transfer with 100 per cent efficiency in all cases. However, for oxidized bases, only approximately 25 per cent of the intermediate yield produced by OH attack undergoes electron transfer; the balance of the oxidized base appears to form adducts with 8-MOP.     

Characterization of hydroxyl free radical mediated damage to plasmid pBR322 DNA

We have investigated hydroxyl free radical mediated damage to pBR322 DNA produced by ascorbate/iron and oxygen in a phosphate-buffered in vitro system. An observed lag phase in DNA nicking suggests a multi-target model of hydroxyl free radical attack on DNA. In the present report we further examine the model system and show that there is a "heat labile" component of the ascorbate/iron system which can be completely restored by the readdition of ascorbate. These observations have allowed us to rule out the possibility that intermediates build up in the reaction and act independently of ascorbate to increase the reaction rate. We have investigated the initial rate of OH production with two OH trapping agents, salicylate and deoxyguanosine, and find that the lag in DNA nicking is not due to a corresponding lag in the production of OH as assessed by formation of the products, dihydroxybenzoic acids and 8-hydroxydeoxyguanosine, respectively. We have found that the energy of activation for DNA supercoiled nicking is 13.9 kcal/mole and for OH trapping by salicylate is 21.1 kcal/nmole. These two activation energies are sufficiently different to suggest that the rate-limiting steps of these two reactions are different. Investigation of the rate of oxygen consumption during the ascorbate/iron-mediated DNA damage showed that oxygen was not a limiting component at any point in the reaction. The addition of catalase slowed down oxygen consumption by 31% and this data taken together with our previous observations on the model implicate hydrogen peroxide as a key intermediate in DNA damage caused by hydroxyl free radical.     

Dual effects of copper-zinc superoxide dismutase

Copper-zinc superoxide dismutase (CuZnSOD) specifically catalyzes the removal of superoxide radicals to protect cellular function against the generation of superoxide-dependent hydroxyl radicals ((.)OH). However, an unexpected observation reveals that denatured CuZnSOD (dCuZnSOD) itself induces (.)OH formation. This dCuZnSOD-dependent (.)OH generation was not inhibited by active CuZnSOD, suggesting that it is a superoxide-independent process. Sodium cyanide, histidine, and N,N'-diethyldithiocarbamate abolished (.)OH generation, implying that Cu may be responsible for dCuZnSOD-induced (.)OH formation. Catalase eliminated ()OH generation, suggesting that hydrogen peroxide may be involved in the mechanism of dCuZnSOD-mediated (.)OH production. Furthermore, nitric oxide ((.)NO) completely inhibited dCuZnSOD-induced (.)OH radical generation, indicating that (.)NO is an important (.)OH radical scavenger. Our results shed new light on the effect of dysfunctional CuZnSOD and suggest that structural disorder of the enzyme may be one of the endogenous pathways of toxic (.)OH formation in biological systems.     

Recent Advances in the Development of a Diagnostic Test for Irradiated Foodstuffs

Recent advances in the development and application of diagnostic tests for irradiated foodstuffs are reviewed. Exposure of water, the major chemical constituent of most foodstuffs to a source of ionising radiation initially generates the highly reactive radical species H', -OH and e^- (aq) which react very rapidly with a wide variety of biological molecules. The detection of foodstuffs subjected to irradiation processing requires the identification and/or quantification of 'unnatural' chemical species (i.e. those not usually formed by normal metabolic processes) produced by the attack of ·OH radical or e^-(aq) on suitable 'target' molecules. Modern methods for the analysis of a series of these 'unnatural' products arising from the interaction of radiolytically-generated ·OH radical or e^-(aq) with polyunsaturated fatty acids, DNA, aromatic compounds and other biologically important scavenger molecules are examined. It is concluded that the analytical test to be conducted is highly dependent on the nature of the foodstuff to be tested.     

Calculation on spectrum of direct DNA damage induced by low-energy electrons including dissociative electron attachment

In this work, direct DNA damage induced by low-energy electrons (sub-keV) is simulated using a Monte Carlo method. The characteristics of the present simulation are to consider the new mechanism of DNA damage due to dissociative electron attachment (DEA) and to allow determining damage to specific bases (i.e., adenine, thymine, guanine, or cytosine). The electron track structure in liquid water is generated, based on the dielectric response model for describing electron inelastic scattering and on a free-parameter theoretical model and the NIST database for calculating electron elastic scattering. Ionization cross sections of DNA bases are used to generate base radicals, and available DEA cross sections of DNA components are applied for determining DNA-strand breaks and base damage induced by sub-ionization electrons. The electron elastic scattering from DNA components is simulated using cross sections from different theoretical calculations. The resulting yields of various strand breaks and base damage in cellular environment are given. Especially, the contributions of sub-ionization electrons to various strand breaks and base damage are quantitatively presented, and the correlation between complex clustered DNA damage and the corresponding damaged bases is explored. This work shows that the contribution of sub-ionization electrons to strand breaks is substantial, up to about 40–70%, and this contribution is mainly focused on single-strand break. In addition, the base damage induced by sub-ionization electrons contributes to about 20–40% of the total base damage, and there is an evident correlation between single-strand break and damaged base pair A–T.     

Reaction of substituted pyrimidines with photochemically generated t-BuO* radicals

Humans are exposed to various organic peroxides through chemical, pharmaceutical and cosmetic products. On photolysis, these peroxides produce alkoxyl radicals and hydroxyl radicals. The reaction of *OH radicals with DNA and its constituents have been extensively studied, but very little is known about the reactions of alkoxyl radicals with DNA and its constituents. In view of this, the oxidation of pyrimidine bases viz., thymine, uracil, cytosine, 5-bromouracil, 6-methyluracil and 1,3-dimethyluracil by t-BuO* radicals in aqueous solution at pH 7.5 has been carried out. The reaction between pyrimidine and t-BuO* is followed by measuring the absorbance of pyrimidine at the respective lambdamax. The rates of oxidation of pyrimidines are calculated from the plot of absorbance vs time. The rates of oxidation of pyrimidines have been found to increase with increase in [t-BuOOH], [pyrimidine] and light intensity. The quantum yields are calculated from the initial rates of oxidation of pyrimidine and the measured light intensity at 254 nm the wavelength at which t-BuOOH is activated to give radicals. The quantum yields are found to depend on [pyrimidine] as well as on [t-BuOOH] while they are independent of light intensity. The product analysis was carried out on HPLC with UV-visible detector. The corresponding 5,6-dihydroxypyrimidine and isobarbituric acid have been identified by comparing the retention times of the authentic samples. On the basis of experimental results and product analysis, it is suggested that t-BuOOH on photolysis gives t-BuO* radical, which initiates the reaction by adding to C (5) or C (6) position of pyrimidine base, leading to the formation of pyrimidine base radical via hydrolysis. The pyrimidine radical further reacts with t-BuO* radical to give the final product. This study predicts the probable transient pyrimidine radicals.     

Computational mapping reveals dramatic effect of Hoogsteen breathing on duplex DNA reactivity with formaldehyde

Formaldehyde has long been recognized as a hazardous environmental agent highly reactive with DNA. Recently, it has been realized that due to the activity of histone demethylation enzymes within the cell nucleus, formaldehyde is produced endogenously, in direct vicinity of genomic DNA. Should it lead to extensive DNA damage? We address this question with the aid of a computational mapping method, analogous to X-ray and nuclear magnetic resonance techniques for observing weakly specific interactions of small organic compounds with a macromolecule in order to establish important functional sites. We concentrate on the leading reaction of formaldehyde with free bases: hydroxymethylation of cytosine amino groups. Our results show that in B-DNA, cytosine amino groups are totally inaccessible for the formaldehyde attack. Then, we explore the effect of recently discovered transient flipping of Watson–Crick (WC) pairs into Hoogsteen (HG) pairs (HG breathing). Our results show that the HG base pair formation dramatically affects the accessibility for formaldehyde of cytosine amino nitrogens within WC base pairs adjacent to HG base pairs. The extensive literature on DNA interaction with formaldehyde is analyzed in light of the new findings. The obtained data emphasize the significance of DNA HG breathing.     

Radiation damage to DNA in DNA-protein complexes

The most aggressive product of water radiolysis, the hydroxyl (OH) radical, is responsible for the indirect effect of ionizing radiations on DNA in solution and aerobic conditions. According to radiolytic footprinting experiments, the resulting strand breaks and base modifications are inhomogeneously distributed along the DNA molecule irradiated free or bound to ligands (polyamines, thiols, proteins). A Monte-Carlo based model of simulation of the reaction of OH radicals with the macromolecules, called RADACK, allows calculating the relative probability of damage of each nucleotide of DNA irradiated alone or in complexes with proteins. RADACK calculations require the knowledge of the three dimensional structure of DNA and its complexes (determined by X-ray crystallography, NMR spectroscopy or molecular modeling). The confrontation of the calculated values with the results of the radiolytic footprinting experiments together with molecular modeling calculations show that: (1) the extent and location of the lesions are strongly dependent on the structure of DNA, which in turns is modulated by the base sequence and by the binding of proteins and (2) the regions in contact with the protein can be protected against the attack by the hydroxyl radicals via masking of the binding site and by scavenging of the radicals.     

Recent Advances in the Development of a Diagnostic Test for Irradiated Foodstuffs

Recent advances in the development and application of diagnostic tests for irradiated foodstuffs are reviewed. Exposure of water, the major chemical constituent of most foodstuffs to a source of ionising radiation initially generates the highly reactive radical species H', -OH and e^- (aq) which react very rapidly with a wide variety of biological molecules. The detection of foodstuffs subjected to irradiation processing requires the identification and/or quantification of ‘unnatural’ chemical species (i.e. those not usually formed by normal metabolic processes) produced by the attack of ·OH radical or e^-(aq) on suitable ‘target’ molecules. Modern methods for the analysis of a series of these ‘unnatural’ products arising from the interaction of radiolytically-generated ·OH radical or e^-(aq) with polyunsaturated fatty acids, DNA, aromatic compounds and other biologically important scavenger molecules are examined. It is concluded that the analytical test to be conducted is highly dependent on the nature of the foodstuff to be tested.     

Reduction in free-radical-induced DNA strand breaks and base damage through fast chemical repair by flavonoids

This paper provides evidence that dietary flavonoids can repair a range of oxidative radical damages on DNA, and thus give protection against radical-induced strand breaks and base alterations. We have irradiated dilute aqueous solutions of plasmid DNA in the absence and presence of flavonoids (F) in a “constant ^·OH radical scavenging environment”, k of 1.5 × 10⁷ s^-1 by decreasing the concentration of TRIS buffer in relation to the concentration of added flavonoids. We have shown that the flavonoids can reduce the incidence of single-strand breaks in double-stranded DNA as well as residual base damage (assayed as additional single-strand breaks upon post-irradiation incubation with endonucleases) with dose modification factors of up to 2.0 ± 0.2 at [F] < 100 μM by a mechanism other than through direct scavenging of ^·OH radicals. Pulse radiolysis measurements support the mechanism of electron transfer or H^· atom transfer from the flavonoids to free radical sites on DNA which result in the fast chemical repair of some of the oxidative damage on DNA resulting from ^·OH radical attack. These in vitro assays point to a possible additional role for antioxidants in reducing DNA damage.     

Yield of DNA strand breaks after base oxidation of plasmid DNA

We have irradiated aerobic aqueous solutions of plasmid DNA with 137Cs gamma rays in the presence of inorganic radical scavengers including nitrite, iodide, azide, thiocyanate and bromide. These scavengers react with the strongly oxidizing hydroxyl radical (*OH) to produce less powerful oxidants. Of these scavengers, only thiocyanate and bromide result in the formation of oxidizing species [(SCN)2*- and Br2*-, respectively] which are capable of reacting with the bases in DNA. The oxidized bases were detected after incubation of the irradiated plasmid with the two E. coli DNA base excision repair endonucleases, formamidopyrimidine-DNA N-glycosylase and endonuclease III. Depending on the experimental conditions, the intermediate base radicals may ultimately form stable oxidized bases in very high yields (within an order of magnitude of the *OH yield), and possibly also single-strand breaks (SSBs) in much lower yield (between 0.1 and 1% of the total yield of base damage). By competing for (SCN)2*- with an additional species (nitrite), it was possible to estimate the second-order rate constant for the reaction of (SCN)2*- with DNA as 1.6 x 10(4) dm3 mol(-1) s(-1), and also to demonstrate a correlation between the large yield of damaged bases and the much smaller increase in the yield of SSBs over background levels due to *OH. The efficiency of transfer of damage from oxidized base to sugar is estimated as about 0.5% or 5%, depending on whether purine or pyrimidine base radicals are responsible for the base to sugar damage transfer.     

Studies of electron migration in DNA in aqueous solution using intercalating dyes.

The reactions of the hydrated electron (e-aq) and of the hydroxyl radical (OH) with double-stranded DNA in aqueous solution at room temperature have been studied through the use of the intercalating dyes, proflavine and ethidium. These dyes react with e-aq with rate constants of (2.5 +/- 0.2) - 10(10) M-1 - s-1 and (3.0 +/- 0.3) - 10(10) M-1 - s-1, respectively; the rate constant for the reaction of OH with proflavine is (1.0 +/- 0.2) - 10(10) M-1 - s-1. When these molecules are bound within the DNA structure both the yields and the rate constants of reaction with e-aq are reduced in a manner entirely consistent with a simple competition between the DNA bases and restricted dye molecules reacting with a bimolecular rate constant of about 2 - 10(9) M-1 - s-1. No evidence of free electron migration in the DNA was obtained, and an upper limit of five base pairs for the range of such migration was derived. Reactions of the hydroxyl radical with DNA-bound proflavine also lead to a rate constant of about 2 - 10(9) M-1 - s-1. These rate constants are in good agreement with rate predictions (per base unit) for a diffusion-controlled reaction with the DNA structure.     

Radiolysis of spin-labeled DNA: an electron spin resonance investigation

The reactions of free and DNA-bound 2,2,5,5-tetramethylpyrrolidine-N-oxyl (PROXYL) probes with radicals generated during radiolysis of dilute aqueous solutions of DNA were examined. For the free PROXYL probe in deaerated solution with each of the four nucleotides (dAMP, dCMP, dGMP, and TMP) it was found that the pyrimidine radicals were more reactive toward the probe than were the purine radicals. Reactions of the electron adduct of TMP and the hydroxyl radical adducts of dAMP, dGMP, and TMP with the probe resulted in little or no reduction of the probe. For TMP these results are consistent with the fact that both the protonated electron and hydroxyl radical adducts of TMP will covalently bind to the nitroxide function of the probe. Reduction of the PROXYL probe was observed in reactions with the hydroxyl radical adduct of dCMP and with the electron adducts of dAMP, dCMP, and dGMP. Results of the radiolysis of the free PROXYL probe in deaerated dilute solution of DNA suggest that the PROXYL probe protects the DNA from water radical attack as the ratio of DNA bases to PROXYL probe increases above 50:1. Reactions of DNA-bound probes are dependent on the depth of the nitroxide function in relation to the major groove of the DNA helix. Two probes with tether lengths which are less than the depth of the major groove show an expected increase in reactions with DNA base radicals as compared to a probe with a tether that extends beyond the groove. The longer probe is involved largely in reactions with sugar and water radicals along the periphery of the DNA helix. In the presence of oxygen, there is a dramatic decrease in the loss of both the free and DNA-bound probes due to the lack of reaction of these probes with peroxyl radicals formed by the addition of molecular oxygen to DNA radicals.     

The formation of DNA sugar radicals from photoexcitation of guanine cation radicals

In this investigation of radical formation and reaction in gamma- irradiated DNA and model compounds, we report the conversion of the guanine cation radical (one-electron oxidized guanine, G(.+)) to the C1' sugar radical and another sugar radical at the C3' or C4' position (designated C3'(.)/C4'(.)) by visible and UV photolysis. Electron spin resonance (ESR) spectroscopic investigations were performed on salmon testes DNA as well as 5'-dGMP, 3'-dGMP, 2'-deoxyguanosine and other nucleosides/nucleotides as model systems. DNA samples (25- 150 mg/ml D(2)O) were prepared with Tl(3+) or Fe(CN)(3-)(6) as electron scavengers. Upon gamma irradiation of such samples at 77 K, the electron-gain path in the DNA is strongly suppressed and predominantly G(.+) is found; after UV or visible photolysis, the fraction of the C1' sugar radical increases with a concomitant reduction in the fraction of G(.+). In model systems, 3'- dGMP(+.) and 5'-dGMP(+.) were produced by attack of Cl(.-)(2) on the parent nucleotide in 7 M LiCl glass. Subsequent visible photolysis of the 3'-dGMP(+.) (77 K) results predominantly in formation of C1'(.) whereas photolysis of 5'-dGMP(+.) results predominantly in formation of C3'(.)/C4'(.). We propose that sugar radical formation is a result of delocalization of the hole in the electronically excited base cation radical into the sugar ring, followed by deprotonation at specific sites on the sugar.