A Comprehensive Study of Sugar Radicals in Irradiated DNA - The

Gauthier C. A. M. Vanhaelewyn, Ewald Pauwels, Freddy J. Callens, Michel Waroquier, Einar Sagstuen, and Paul F. A. E. Matthys. The Journal of Physical ...
0 downloads 0 Views 135KB Size
7674

J. Phys. Chem. B 1998, 102, 7674-7686

A Comprehensive Study of Sugar Radicals in Irradiated DNA Stacey D. Wetmore and Russell J. Boyd* Department of Chemistry, Dalhousie UniVersity, Halifax, NoVa Scotia, Canada B3H 4J6

Leif A. Eriksson Department of Physics, UniVersity of Stockholm, Box 6370, 113 85, Stockholm, Sweden, ReceiVed: June 2, 1998; In Final Form: July 9, 1998

Density functional theory is used to study the energetics, geometries, and hyperfine couplings in sugar radicals which are generated through irradiation of DNA. The C4′-S and the C3′-S radicals are determined to be the lowest lying species of the radicals formed through abstraction of a hydrogen or a hydroxyl group from a model of the sugar present in DNA, respectively. The C2′ radical has the highest energy and the smallest degree of ring puckering of all possible carbon-centered radicals formed via hydrogen abstraction. In addition to the possible dehydrogenated and dehydroxylated products, various radicals which lead to substantial ring alterations, such as ring breaks or flattening of the ring, are also studied. In most cases, the calculated hyperfine coupling constants directly support the assignment of the experimentally observed couplings to the specific radicals. The effects of rotation about the C5′C4′ bond on the HFCCs in the C5′ and O5′ radicals are examined in order to compare the experimental and theoretical results. In cases where experiment and theory differ, the calculated results facilitate the assignment of the experimental couplings to alternative radicals.

Introduction Many recent papers in the literature are devoted to the investigation of the various radicals formed upon irradiation of DNA.1,2 Interest in this area has arisen for a number of reasons, including the depletion of the stratospheric ozone. The effects of radiation on DNA include alterations to the base and sugar molecules1 which subsequently lead to strand breaks3 and DNA-protein cross-links.4-6 In particular, the formation of sugar radicals is of great interest since it is now widely accepted that single-strand breaks in DNA occur via these intermediates.7,8 Sugar radicals can be formed through either direct or indirect mechanisms. Direct formation occurs via alkoxyl or base radicals, whereas indirect formation occurs through the attack of hydrogen or hydroxyl radicals generated from water radiolysis. In an important study of D-glucose, Schuchmann and von Sonntag9 concluded that the six carbon atoms in the sugar are attacked by hydroxyl radicals to an equal extent. However, electron spin resonance (ESR) techniques proved to be unable to reliably detect sugar radicals in irradiated DNA. In fact, Hole et al.10 were the first to observe a large variety of sugar radicals in their study of 2′-deoxyguanosine 5′-monophosphate, although numerous sugar radicals were previously observed in studies of different nucleotides and nucleosides. In their study, Hole et al. characterized nine sugar radicals (which were not easily detected by ESR techniques), indicating that almost every carbon site in the sugar is affected by the radiation. A subsequent ENDOR study of single crystals of deoxyadenosine11 supported the hypothesis of the formation of sugar radicals upon application of small radiation doses at low temperatures. These studies indicate that the sugar radicals, in addition to the various bases, may be a site of significant radiation damage in DNA, although detection of these radicals in full DNA appears to be difficult.12 A few theoretical investigations of the possible carboncentered sugar radicals generated in irradiated DNA have

appeared in the literature.13,14 In these studies, geometries, relative energies, spin density distributions, and hyperfine coupling constants were calculated at the HF level. Both studies were very complete and carefully performed at the level of theory chosen. However, it is well-known that electron correlation can have considerable effects on molecular properties and, thus, should be taken into account. In particular, it is wellknown that Hartree-Fock level calculations overestimate the hyperfine coupling constants considerably.15 It is of interest to calculate the hyperfine coupling constants (HFCCs) of possible radicals in the DNA sugar moiety since even with the detailed ENDOR techniques the spectra of the radicals are complicated. This is because the spectra of different radicals often overlap and frequently assumptions are required for a full analysis. Thus, through the comparison of accurate theoretical calculations and experimental hyperfine coupling constants, assignment of the spectra to particular radicals can be made with greater confidence. Once the radicals which are formed in single crystals are fully characterized, experimentalists will have a better understanding of how to recognize these radicals in full DNA and be able to answer an important question, namely, whether sugar radicals are formed in DNA.12 Due to the difficulties mentioned above with HF level HFCCs, an alternate theoretical technique must be found which yields accurate coupling constants. This method must also be computationally efficient due to the number of atoms involved in the sugar moiety. Gradient-corrected or hybrid density functional theory (DFT) techniques fit both of these requirements. In fact, recently DFT has been used to analyze the radicals formed in the DNA bases cytosine16 and thymine17 with a great deal of success. In this paper, we employ DFT in order to examine the geometry and HFCCs of various possible sugar radicals formed upon irradiation of DNA. The structure and standard atomic numbering of the unsubstituted sugar which appears in DNA,

S1089-5647(98)02440-7 CCC: $15.00 © 1998 American Chemical Society Published on Web 09/10/1998

Sugar Radicals in Irradiated DNA

J. Phys. Chem. B, Vol. 102, No. 39, 1998 7675 TABLE 1: Relative Energies (kcal/mol) of the Sugar Radicals Calculated at the B3LYP/6-311G(2df,p)//B3LYP/ 6-31G(d,p) Level radical

Figure 1. Structure and numbering of 2′-deoxyribose (I) and the model system used in the present study (II).

2′-deoxyribose, are presented in Figure 1, structure I. Nucleosides are formed by joining the C1′ position of the sugar to the N1 position of one of the four DNA bases. Nucleotides, which form the molecular building blocks of DNA, are nucleosides with phosphates esterfied at the C3′ and C5′ positions. Due to the number of atoms involved in the nucleotides, a model system was used where the phosphate groups are represented by hydroxyl groups and the DNA base has been represented by an amine group (Figure 1, structure II). Geometrical effects generated by substituting the DNA base at the C1′ position with an amino group have been shown to be small.14 The sugar radicals examined in this study include hydrogen abstraction radicals formed by removal of hydrogen from all carbon and oxygen atoms, radicals formed via removal of either of the hydroxyl groups in the model system, and a variety of radicals which lead to significant sugar ring alterations. Theoretical Details All geometries were optimized using Becke’s three-parameter exchange functional (B3)18 in combination with Lee, Yang, and Parr’s correlation functional (LYP)19 and Pople’s 6-31G(d,p) basis set.20 Frequency analyses were performed to ensure that stationary points were local minima. Subsequent single-point calculations were performed at the B3LYP level with Pople’s 6-311G(2df,p) basis set to obtain relative energies, spin densities, and dipole moments of the global minima. These calculations were accomplished using Gaussian 94.21 The hyperfine coupling constants were obtained with Perdew and Wang’s nonlocal exchange (PW),22 Perdew’s nonlocal correlation functional (P86),23 and Pople’s 6-311G(2d,p) basis through the deMon program.24 The units of gauss (1 G ) 2.8025 MHz) are used throughout this work for the HFCCs. Many good reviews on the accurate calculation of HFCCs exist in the literature,25,26 and thus, the details of theoretical requirements and methodology will not be reviewed here. However, to provide a fair and accurate interpretation of the results presented within, a few points should be addressed. Accurate isotropic HFCCs require both a good description of electron correlation and a well-defined basis set, whereas satisfactory anisotropic HFCCs can be obtained with almost any theoretical method and basis set provided the structure is qualitatively correct. Thus, comparison of anisotropic hyperfine tensors can be used as an accurate guide to identify radical sites even when less satisfactory agreement is obtained for the isotropic component. Results and Discussion Energetics and Geometrical Parameters. Two different puckering modes were examined for each possible radical corresponding to north (N) and south (S) types, which are defined according to where the radical is located on the pseudorotation cycle.27 It is convenient to analyze the puckering amplitudes in the sugar molecules through the use of the

C4′-S C5′-N C3′-S C1′-S C1′-N C3′-N C5′-S C4′-N C2′-N C2′-S O3′-S O3′-N O5′-N O5′-S C3′-S C3′-N C5′-S C5′-N

rel energy + ZPE Hydrogen Removal 0 1.4 1.8 2.3 2.3 2.5 3.0 3.3 8.3 8.7 10.2 12.9 12.9 Hydroxyl Removal 0 0.7 3.2 3.7

rel energy 0 1.6 2.3 3.1 3.2 2.8 3.2 4.1 9.5 9.7 10.4 12.2 13.9 14.1 0 0.5 4.1 4.5

pseudorotational phase angle27 defined as

tan P )

(ν4 + ν1) - (ν3 + ν0) 2ν2(sin 36° + sin 72°)

(1)

where the νj are the ring dihedral angles: ν0 ) C4′O1′C1′C2′, ν1 ) O1′C1′C2′C3′, ν2 ) C1′C2′C3′C4′, ν3 ) C2′C3′C4′O1′, and ν4 ) C3′C4′O1′C1′. The puckering amplitude,13 defined as

τm2 )

2

4

∑νj2

5 j)0

(2)

is also useful for discussing the puckering of the sugar ring. A diagram depicting the pseudorotation cycle and the relation of P to this cycle has been given in previous papers on the sugar molecule and will not be repeated here.13,14 The puckering in the sugar molecules can be considered to be either an envelope (E) form where four atoms are located in a plane and the fifth atom is located out of the plane or a twist (T) form where three atoms are in a plane and the other two atoms are displaced on opposite sides of the plane. The displaced atoms are categorized as endo or exo according to whether they are displaced to the same side or the opposite side of C5′, respectively. A superscript (subscript) on the left side of the puckering symbol is used to represent endo (exo) puckering. For example, the C3′-endo and C2′-endo forms which are observed in the nonradical sugar molecules in A-DNA and B-DNA, respectively, are both envelope conformations and can be represented as 3E and 2E. Radical Energetics. The relative energies of the various sugar radicals formed by hydrogen abstraction are displayed in Table 1. These energies were obtained from single-point calculations at the B3LYP/6-311G(2df,p) level, and the zeropoint vibrational energy has been accounted for through the use of a scale factor of 0.9804.28 From the results, it can be seen that the C4′-S radical is the lowest in energy of all the radicals formed by hydrogen abstraction. The north- and south-type conformers for each of the C1′ and C2′ radical pairs are very close in energy, with a separation on average of only 0.2 kcal/ mol. The C3′ radicals are also close in energy, with a separation of only 0.7 kcal/mol. There is a much larger difference for the

7676 J. Phys. Chem. B, Vol. 102, No. 39, 1998

Wetmore et al.

TABLE 2: Structural Parameters for the Sugar Radicals Calculated at the B3LYP/6-31G(d,p) Level (deg)

ν0 ν1 ν2 ν3 ν4 P τm µ, D

C1′

C2′

C3′

4E

3 2T

2E

-23.6 2.0 18.4 -32.6 35.1 58.7 35.7 2.29 2E

ν0 ν1 ν2 ν3 ν4 P τm µ, D a

-17.6 29.2 -28.6 19.5 -1.9 163.9 30.8 2.79

-27.3 9.5 9.9 -26.3 34.4 -1.4 33.5 1.58 3E

3.9 11.6 -21.0 22.7 -17.1 206.4 23.7 2.85

24.8 -30.2 26.2 -12.1 -8.3 -32.4 31.1 0.73 0E

-34.0 24.2 -7.4 -12.2 28.8 103.0 33.3 3.00

C4′

C5′

North Radicals 1 0T 14.4 35.8 -25.2 -31.1 26.0 15.5 -18.3 5.8 2.7 -26.4 -13.1 -64.3 27.3 35.9 2.34 0.91 2E

South Radicals 2 3T 27.1 -15.2 -30.4 32.7 22.6 -36.9 -6.7 28.3 -18.0 -8.4 135.3 174.4 31.9 37.6 1.99 2.11 2 1T

O5′ 3 4T

-15.4 -9.3 27.9 -37.7 33.7 42.1 38.3 1.96 2 1T

-18.0 38.6 -27.4 6.7 18.4 141.0 34.3 3.37

O3′ 3Ea

-7.8 -18.8 36.3 -41.6 31.7 29.1 42.3 2.60 2E

-27.6 37.7 -33.2 17.4 6.2 152.1 38.1 2.24

C3′ (OH) 1 2T

25.9 -26.8 19.4 -4.6 -13.7 -46.0 28.1 1.81 1E

-29.8 26.9 -15.7 -1.5 19.8 113.5 30.0 1.69

C5′ (OH) 3 4T

-18.8 -7.4 27.9 -39.3 36.7 45.5 41.4 1.55 2E

-27.5 38.7 -34.9 19.3 5.1 154.2 39.3 1.77

MP2 geometry (see text for further details).

north- and south-type C4′ and C5′ radicals. The C4′ north conformer is 3.3 kcal/mol higher in energy than its south-type counterpart, while the C5′-N radical is 1.6 kcal/mol lower in energy than the C5′-S conformer. The C2′ radicals are the highest energy radicals formed through the abstraction of a hydrogen from one of the sugar carbons, lying on average 8.5 kcal/mol higher in energy with respect to the lowest energy C4′-S radical. The north and south radicals formed via hydrogen removal from a hydroxyl group in the model sugar are close in energy for both the O3′- and O5′-centered radicals. These alkoxyl radicals are very high in energy, being on average 10.2 and 12.9 kcal/mol above the C4′-S radical for the O3′ and O5′ hydrogen abstraction radicals, respectively. The results presented in Table 1 differ somewhat from those obtained from Miaskiewicz and Osman’s MP2/6-31G(d) singlepoint calculations.13 In their study, it was determined that the C3′-S radical is the lowest energy radical, with C1′-S, C1′N, C3′-N, and C4′-S lying at most 1.4 kcal/mol higher in energy. The ROHF/3-21G results of Colson and Sevilla14 for the S-type radicals also differ in that the C1′ radical was determined to be the lowest in energy, with the C4′ radical close behind. All studies agree that of the carbon-centered radicals, the C2′ radicals are highest in energy, although the 8.5 kcal/ mol difference from the lowest energy radical obtained in this study is approximately twice the difference observed in the other two studies (approximately 5 kcal/mol). Due to the relatively small stabilization energy of one radical over another, it is not surprising that differences arise in the results once electron correlation is included in the geometry optimizations. In addition, in the previous studies, it is not clear whether corrections were made for the zero-point energy. When the ZPE is not taken into account, the magnitude of the energy difference between radicals increases (Table 1). In addition, when ZPE is not corrected for, the relative order of the radicals may change. For example, without ZPE, the C3′-S radical is lower in energy than the two C1′ radicals, but inclusion of the ZPE indicates that the C1′ radicals are lower in energy. The model systems used in the previous studies differ from the model used in the current study. Miaskiewicz and Osman13 is used a system very similar to ours, differing only by the absence of the hydroxyl group at C5′. Colson and Sevilla,14

on the other hand, used three different models: the same model as Miaskiewicz and Osman and two model systems with phosphate groups at C3′ and C5′, one with an amine group present at C1′ and one without. The energetic differences for the three model systems of Colson and Sevilla were small except for the fact that the C3′ radical exhibited destabilization due to the phosphate groups. It was indicated in a previous study that solvation effects might be important when transferring results to full DNA.13 It was determined that S-type conformers have larger dipole moments and, thus, would be affected to a greater extent than the corresponding N-type radicals. Our results for the dipole moments are displayed in Table 2, and the results also indicate that for most cases, the S-type conformers have a larger dipole moment. However, the two studies exhibited differences in some instances. For example, the dipole moments determined in this study for C3′-N and C3′-S are 0.73 and 3.00 D, respectively, whereas Miaskiewicz and Osman13 determined that these conformers possess very similar dipole moments. The large difference observed in the present study for the two conformers of the C3′ radical will lead to a much greater stabilization of the S-type radical relative to the N-type radical in solution. This stabilization would cause a greater variation in their relative energies, which without consideration of solvation effects is very small. The relative energies for the radicals formed through net hydroxyl radical removal in the model system also appear in Table 1. These radicals correspond to a breakage of a phosphoester bond in DNA where the hydroxyl groups are replaced by phosphates. For both radical sets, the N- and S-type radicals are very close in energy. The radicals formed via abstraction of a hydroxyl group from the C3′ position are approximately 4 kcal/mol lower in energy than the corresponding C5′ radicals, identifying this as a possible site for singlestrand breakage in DNA. The dipole moments for both conformers of each type of radical are highly similar (Table 2), and thus, the solvation effect on the relative stability of these radicals will be negligible. These radicals were not examined in the previous studies, and thus, no comparison can be made. Geometrical Parameters. Table 2 displays the various geometrical parameters for the dehydrogenated radicals. The

Sugar Radicals in Irradiated DNA

J. Phys. Chem. B, Vol. 102, No. 39, 1998 7677

TABLE 3: Bond Length (Å) and Bond Angles (deg) for the Sugar Ring in the Sugar Radicals Calculated at the B3LYP/ 6-31G(d,p) Level nonradial

C1′

C2′

C3′

C4′

C1′C2′ C2′C3′ C3′C4′ C4′O1′ O1′C1′ C4′O1′C1′ O1′C1′C2′ C1′C2′C3′ C2′C3′C4′ C3′C4′O1′

1.548 1.546 1.544 1.419 1.450 107.3 105.1 105.0 102.6 104.3

1.505 1.539 1.544 1.438 1.408 107.1 109.8 103.8 102.7 104.3

1.496 1.488 1.551 1.431 1.459 108.4 103.3 110.9 101.3 105.4

1.536 1.494 1.506 1.456 1.462 110.8 104.6 101.6 108.8 104.8

North Radicals 1.534 1.537 1.501 1.376 1.475 110.1 104.2 105.0 101.7 111.7

C1′C2′ C2′C3′ C3′C4′ C4′O1′ O1′C1′ C4′O1′C1′ O1′C1′C2′ C1′C2′C3′ C2′C3′C4′ C3′C4′O1′

1.530 1.534 1.533 1.428 1.445 108.9 102.8 102.8 103.1 108.9

1.496 1.540 1.540 1.447 1.403 109.7 108.8 102.6 103.0 106.8

1.493 1.489 1.535 1.443 1.456 111.3 104.2 110.6 101.9 106.7

1.545 1.503 1.504 1.435 1.452 108.9 104.6 102.5 109.5 104.0

South Radicals 1.532 1.551 1.511 1.370 1.474 108.9 103.5 104.5 101.7 112.2

a

C5′

O3′

O5′

C3′(OH)

C5′(OH)

1.533 1.539 1.554 1.442 1.460 108.1 103.3 105.7 103.4 107.3

1.537a 1.544 1.528 1.433 1.454 109.6 105.5 103.3 100.1 103.8

1.547 1.538 1.537 1.418 1.460 109.7 105.5 104.5 104.4 109.7

1.542 1.491 1.491 1.442 1.453 110.5 104.6 102.6 109.6 105.2

1.547 1.528 1.560 1.428 1.455 109.2 105.7 104.7 101.5 102.9

1.530 1.531 1.537 1.443 1.466 110.5 104.2 103.2 101.7 106.3

1.526 1.538 1.590 1.421 1.465 110.8 103.5 103.4 101.5 106.6

1.530 1.534 1.553 1.426 1.449 109.0 102.7 103.0 103.1 107.0

1.541 1.489 1.495 1.437 1.453 109.7 104.5 102.6 109.7 105.0

1.529 1.528 1.563 1.434 1.449 110.4 103.4 102.5 102.1 106.2

MP2 geometry (see text for further details).

complete geometries for all of the radicals discussed within are available as supporting information. The sugar puckering modes for C2′-N, C2′-S, and C4′-S differ from those obtained by Miaskiewicz and Osman.13 The other puckering modes for C1′, C3′, and C4′ are in agreement with the previous results. In addition, the magnitude of the pseudorotation phase angle is in good agreement. In their study, Miaskiewicz and Osman assigned the C3′ radical with P ) 84.4° to a south-type 0E envelope form. From the results in Table 2, it can be seen that this is indeed an 0E form with P ) 103.3°. Our results for the ring puckering are also different from Colson and Sevilla’s,14 but the reader is reminded of differences in both model systems and the level of theory implemented. In particular, this study includes electron correlation which may have a large effect on the ring puckering. Colson and Sevilla determined that there exists a dependence of the ring puckering on whether an amine group is present at the C1′ position. Using hydroxyl groups rather than phosphate groups in the chemical model may also lead to some geometrical differences. It is furthermore noted that although replacement of hydrogen at C1′ with amine led to puckering differences, replacement of the amine group with cytosine, the smallest DNA base unit, led to negligible differences. Along with the endocyclic dihedral angles and the pseudorotation phase angle, Table 2 also displays the puckering amplitude. This parameter will yield an indication of the degree of puckering in the sugar ring, where a low value of νm indicates a relatively flat sugar ring. In their study, Miaskiewicz and Osman13 determined that the C2′-S radical has the smallest puckering, while the C4′-N and C1′-N radicals have the largest values of νm. The results displayed in Table 2 indicate that when electron correlation is accounted for, a similar situation arises where, of the possible carbon-centered radicals, C2′-S has the lowest νm value, while C5′-S has the largest puckering amplitude. These results are also in agreement with Colson and Sevilla’s observations which indicate that only small effects are generated by using hydroxyl groups in the model system. It was suggested by Miaskiewicz and Osman that the relatively flat structure for the C2′ radicals occurs due to the absence of an oxygen next to the radical center. They proposed that in

the other carbon-centered radicals, the oxygen lone pairs and the unpaired electron interact to form a more pyramidal-like structure. Table 3 compares the bond lengths and bond angles for the various sugar radicals to those in the parent molecule. For the C1′, C2′, C3′, and C4′ radicals, the major geometrical alterations are the contraction of the bonds between the radical center and surrounding atoms. The bond lengths are shortened between 0.04 and 0.06 Å. The bond angle in which the radical center is the central atom changes by 3-8°, more than any other angle in the radicals. The remainder of the sugar ring geometry is relatively unaffected. The geometry of the sugar ring in the C5′ radical is altered only slightly upon radical formation. For both of the N- and S-type radicals, the major alterations occur at C5′, where the O5′C5′ bond length is decreased by 0.06 Å and bond angles are altered by 2°. Similarly, the major alterations upon radical formation in the O5′ radical occur at C5′. The major differences in geometry from the nonradical forms include changes in the O5′C5′ bond length of 0.07 Å and slight alterations in bond angles. Thus, for all of the dehydrogenated radicals, alterations in the geometry relative to the nonradical forms are limited to the radical site. This is in agreement with the observation obtained at the HF level by Miaskiewicz and Osman.13 Table 2 also displays the various structural parameters for the radicals formed by abstraction of a hydroxyl group in the model system. The pseudorotation angles and, thus, the puckering modes for the two sets of radicals differ from one another. The radicals formed via abstraction of the hydroxyl group from the C3′ center have significantly lower puckering amplitudes relative to the similar C5′ radicals. In fact, the C3′-N radical is nearly planar. The effects of radical formation on the sugar ring geometry are displayed in Table 3. The C3′ radicals formed by OH removal have similar geometries to those formed by H removal. The geometry differs from the nonradical forms by the shortening of the C2′C3′ and C3′C4′ bonds by approximately 0.06 Å and the increase in the C2′C3′C4′ angle by 7°. The remainder of the sugar ring is relatively unaffected. The geometry changes in the sugar ring for the C5′ radicals formed via OH removal

7678 J. Phys. Chem. B, Vol. 102, No. 39, 1998 are all relatively small, as determined for the corresponding dehydrogenated radicals. The major geometric alterations for this radical occur outside the sugar ring where the C4′C5′ bond length is decreased by 0.03 Å and the H′5C5′H′5 angle is increased by approximately 11°, as a result of the altered hybridization of the C5 carbon. Hyperfine Coupling Constants. The results for the hyperfine coupling constants of many different sugar radicals obtained from a variety of experimental studies10,11,29-36,38,39 are displayed in Table 4. The results obtained from the present study for the various possible radicals formed via removal of a hydrogen atom or breakage of a phosphoester bond are displayed in Tables 5 and 6, respectively, for both the north- and south-type radicals. Table 7 displays the location of the majority of the spin density in these radicals obtained from the Mulliken population analysis. The HFCCs for the ring-altering radicals are listed in Table 8. Dehydrogenated Carbon-Centered Radicals. It has been argued in the past that sugar radicals are all generated from alkoxyl radicals and, thus, no sugar radicals should be formed in DNA due to the lack of hydroxyl groups.37 However, due to the number of sugar radicals generated in 2′-deoxyguanosine 5′-monophosphate,10 which contains only one hydroxyl group, it was suggested that other mechanisms for the formation of these radicals must be considered. For example, carbon-centered radicals have been proposed to be formed via H abstraction by hydroxyl or hydrogen radicals. In addition, it has been suggested that carbon-centered radicals can possibly be formed through (super)excitation followed by homolytic cleavage of CH bonds. The couplings present in the spectra of a number of irradiated DNA molecules have been assigned to the C1′ radical.10,11,29,30 This radical could be formed via hydrogen abstraction or (super)excitation, as previously mentioned. Another possible mechanism for formation is through deprotonation of a parent sugar radical cation at the C1′ position.36 This mechanism has been previously suggested for the radical formed by net hydrogen removal at the C4′ position. It was suggested that if the electron vacancy is primarily located on O1′, the positively charged radical is most likely stabilized through deprotonation at positions close to O1′ (i.e., C4′ or C1′). Hole et al.10 determined that the π-spin density at C1′ is 0.64. Our results indicate that the spin density at the C1′ position is greater for this radical (0.75). Comparison of all of the experimental results in Table 4 with the values obtained in this study for the HFCCs in the C1′ radical indicates that the present results support the experimental assignment. In particular, from the results displayed in Table 5, it can be seen that the experimental results agree more closely with those calculated for the N-type radical. The S-type C1′ radical possesses a significantly lower C2-H coupling (9.1 G) relative to the experimental results. This is a nice example of the effects of the puckering amplitude on the HFCCs. Note that although the isotropic components differ between the values calculated for the C2-H couplings in the N- and S-type C1′ radicals, the anisotropic values are almost identical. Energetically, both systems are equal in stability. The C2′ hydrogen abstraction radical has been observed by Hole et al.,10 where it appears only as a minor radical at 10 K. This is not surprising since the energetics observed for this radical indicate that it is higher in energy than any of the other carbon hydrogen abstraction radicals. From the experimental results, Hole et al. suggested an sp2 configuration at the C2′ center, which would involve a rehybridization of this center upon radical formation. An sp2 hybridization would help to explain the relatively flat structure for this radical. The spin density at

Wetmore et al. C2′ was suggested to be 0.89, which is again slightly smaller than the calculated value (0.96). The experimental HFCCs displayed in Table 4 for the C2′ radical are in qualitative agreement with the calculated values. The C2′-H HFCC exhibits a great deal of anisotropy in the experimental results, and the calculated results support this finding. As previously mentioned, both C2′ radicals are very close in energy. However, from Table 5, it can be seen that the HFCCs differ in the N- and S-type radicals. This difference arises mainly in the C3′-H coupling where a larger isotropic HFCC was obtained for the north conformation (38.2 G) relative to the south conformation (7.9 G). Experimentally, no C3′-H coupling was observed, and thus, the radical observed in the experiment is likely to be the south-type conformer. The C3′ dehydrogenated radical was observed in an investigation of irradiated 2′-deoxyguanosine 5′-monophosphate.10 The experimental results indicated that rehybridization at C3′ does not take place as in the C2′ radical; e.g., the C3′ remains sp3 hybridized. The calculated HFCCs for the N- and S-type C3′ radicals differ in the absence of a C4′-H coupling for the south conformation. The magnitudes of the remaining couplings also differ between the two conformations. The experimental results are in fair agreement with the N-type C3′ radical. The calculated values indicate that the O3′ hydrogen has a small isotropic coupling and a relatively large anisotropic contribution which were not detected in the experimental study. However, experimentally, there was another coupling observed for which only the full tensor components were resolved and assignment to a particular atom was not made. The unassigned couplings are not unlike the couplings of the C2′ hydrogens and could possibly be due to the C2′-H in another conformation. The difference between the experimental and the calculated isotropic hyperfine coupling constants in this radical could be due to the presence of a phosphate group at the C5′ position in the experimental study. Colson and Sevilla14 previously determined that the presence of the phosphate groups affects the HFCCs in the C3′ radical. Radicals formed through hydrogen abstraction from the C4′ position have been observed in three different crystals: uridine 5′-phosphate,36 inosine,38 and adenosine:5-bromouridine.39 Formation of the C4′ hydrogen abstraction radical has been speculated to be due to one of two mechanisms. First, it could arise from a hydroxyl alkyl radical, formed by abstraction of a hydrogen at C5′, followed by a 4′,5′-hydrogen atom transfer. Second, it was speculated to be formed, as previously discussed for the C1′ hydrogen abstraction radical, through deprotonation of the parent cation at C4′. The HFCCs of the north and south conformers of this radical are quite different from one another. The N-type radical exhibits two C5′-H couplings, one of substantial magnitude (27.9 G) and no O5′-H coupling. The S-type conformer, on the other hand, has a significant O5′-H coupling (5.6 G) and only one small C5′-H coupling (6.2 G). Experimentally, Sagstuen observed three substantial couplings of 36, 25, and 24 G in the crystals of uridine 5′-phosphate.36 In addition, two small couplings were observed at only certain orientations, and therefore, accurate HFCCs could not be abstracted. In inosine,38 large C3′-H and C5′-H couplings of 34.7 and 33.4 G, respectively, and a smaller C5′-H coupling of 3.4 G were obtained. In adenosine:5′-bromouridine, two couplings were resolved which correspond to C3′-H and C5′-H couplings of 21.0 and 10.0 G, respectively. Overall, poor agreement between the theoretical and experimental HFCCs for this radical indicates that either better data must be obtained or other radical

Sugar Radicals in Irradiated DNA

J. Phys. Chem. B, Vol. 102, No. 39, 1998 7679

TABLE 4: Experimental HFCCs for Sugar Radicals (G) radical C1′-H abstraction

molecule 2′-deoxyguanosine 5′-monophosphate10 adenosine-HCl29 deoxyadenosine monohydrate11 cytosine 5′-monophosphate30

C2′-H abstraction

2′-deoxyguanosine 5′-monophosphate10

C3′-H abstraction

2′-deoxyguanosine 5′-monophosphate10

C4′-H abstraction

uridine 5′-phosphate36 inosine38 adenosine:5-bromouridine39

C5′-H abstraction

2′-deoxyguanosine 5′-monophosphate10

deoxyadenosine monohydrate31 deoxyadenosine monohydrate11 cytosine 3′-monophosphate30 5-chloro- and 5-bromodeoxyuridine30 breakage of C5′-OPO32-

deoxycytidine 5′-monophosphate30 2′-deoxyguanosine 5′-monophosphate10

O5′ alkoxyl

thymidine30 uracil-β-D-arabinofuroside40 adenosine-HCl32 5-chlorodeoxyuridine32 5-bromodeoxyuridine32 cytosine 3′-monophosphate32 deoxyribose adenosine-H2O32

O3′ alkoxyl C4′ ring opened

2′-deoxyguanosine 5′-monophosphate10,34 2′-deoxyguanosine 5′-monophosphate33 uridine 5′-phosphate35

ring breaking

atom C2′-H C2′-H C2′-H C2′-H C2′-H C2′-H C2′-H C2′-H C1′-H C2′-H C2′-H C4′-H unassigneda β-H β-H β-H C3′-H C5′-H C5′-H C3′-H C5′-H C5′-H C5′-H C5′-H C5′-H C4′-H C5′-OHa C5′-H C4′-H C5′-H C5′-H C4′-H C5′-OH C5′-H C5′-OH C4′-H C5′-H (2) C4′-H C5′-H C5′-H C4′-H C5′-H C4′-H C5′-H C5′-H C5′-H C5′-H C5′-H C5′-H C5′-H C5′-H C5′-H C5′-H C5′-H C5′-H C5′-H C5′-H C3′-H C5′-H C5′-H C4′-H C4′-H C5′-H C5′-H

2′-deoxyguanosine 5′-monophosphate10 2′-deoxyguanosine 5′-monophosphate34

H2O + H removal

a

Full tensor components.

2′-deoxyguanosine 5′-monophosphate34

R-H β-H β-H β-H

Aiso 28.0 16.1 25.8 17.2 25.4 19.6 19.6 -23.1 32.4 16.7 38.1 27.5 36 ( 2 25 ( 3 24 ( 3 34.7 33.4 3.4 21.0 10.0 -22.2 -20.9 -20.8 -19.6 2.5 -14.7 7.0 ( 1 -17.5 -22.7 4.5 20.8 -20.7 8.6 18.9 -21.4 36.0 -15.5 -17.3 6 -18.0 6 80.4 71.3 90.0 47.3 93.5 48.0 83.3 85.6 58.5 57.3 ∼82 ∼59 ∼100 ∼53 20.3 27 37 32.2 -18.8 48 13 -15.5 4.0 -16.9 3.8 -16.0 25.0 23.0 12.9

TXX

TYY

TZZ

-2.1

-1.8

3.9

-1.8 -1.9 -1.7 -1.4 -2.0 -12.2 -2.1 -1.6 -2.1 -1.7 (16)

-0.5 -1.7 -0.7 -0.8 -1.4 1.0 -0.3 -0.9 -1.4 -0.9 (∼22)

2.3 3.6 2.3 2.2 3.3 11.1 2.4 2.5 3.4 2.7 (29)

-8.7 -8.6 -8.8 -8.7

0.8 0.6 0.8 0.5

7.9 8.0 8.0 8.2

(16.3) -7.9

(20.2) -1.7

(28.1) 9.7

-11.8 -9.3 -3.0 -4.3 -12.5 -3.1 -1.6 -8.3

0.8 0.7 0.1 -1.5 2.9 -1.0 -0.2 1.0

11.0 8.5 3.0 5.8 9.6 4.2 1.9 7.4

-8.9 -9.7

-1.2 -0.2

10.2 9.9

-9.8

-0.4

10.2

-3.1 -3.1 -1.0 -1.3 -3.3 -2.7 -4.1 -3.5 -3.3 -2.6

-1.7 -1.7 -1.0 -0.3 -2.0 -1.9 -1.9 -0.9 -0.3 -2.1

4.8 4.8 2.0 1.7 5.3 4.5 5.9 4.3 3.7 4.6

-4.7

-0.5

5.1

-9.8

-0.2

10.0

-11.0 -1.4 -8.0 -1.6 -11.7 -3.8 -6.7 -3.3

1.1 -0.2 0.2 -0.2 1.9 0.8 -0.2 -0.7

9.8 1.5 7.8 1.9 9.8 3.0 7.0 4.1

7680 J. Phys. Chem. B, Vol. 102, No. 39, 1998

Wetmore et al.

TABLE 5: Calculated HFCCs for Dehydrogenated Sugar Radicals (G) north radical C1′ C2′

C3′

C4′

C5′ O5′ O3′ a

south

atom

Aiso

TXX

TYY

TZZ

Aiso

TXX

TYY

TZZ

C2′-H C2′-H C2′-H C1′-H C3′-H O3′-H C2′-H C2′-H O3′-H C4′-H C5′-H C5′-H C4′-H O5′-H C5′-H C4′-H O5′-H C5′-H C5′-H C1′-H C3′-H C1′-H

18.5 22.7 -21.3 30.9 38.2 1.9 18.9 34.0 -2.8 22.5 27.9 2.8 22.1

-1.4 -1.9 -13.1 -2.0 -2.3 -1.4 -1.7 -1.5 -4.3 -1.6 -1.8 -2.1 -1.8

-1.0 -1.6 -0.2 -1.0 -0.9 -0.6 -1.5 -1.1 -3.1 -1.3 -1.1 -1.4 -1.1

2.4 3.4 13.4 3.0 3.2 2.0 3.2 2.6 7.4 2.8 2.8 3.5 2.8

29.3 9.1 -20.7 31.6 7.9 -2.4 12.4 31.2 -2.3

-1.4 -1.9 -13.0 -1.9 -2.2 -3.5 -1.6 -1.9 -4.5

-1.1 -1.4 -0.2 -0.9 -1.0 -0.5 -1.1 -1.5 -3.3

2.5 3.3 13.3 2.9 3.2 4.0 2.7 3.5 7.7

-11.2 -1.7 -5.1 -2.6 -3.2 -0.3 -2.1 -0.7

-0.8 -0.8 -3.4 -1.6 -1.4 -0.2 -1.5 -0.3

12.0 2.5 8.5 4.1 4.6 0.5 3.6 1.0

6.2 31.4 5.6 -10.4 35.3 -3.9 17.5 90.3 2.1 12.5 2.7

-2.4 -1.9 -1.6 -10.9 -1.6 -5.0 -2.5 -3.1 -0.3 -2.0 -1.0

-1.1 -1.2 -0.8 -0.7 -0.8 -3.3 -1.5 -1.5 -0.3 -1.7 -0.7

3.5 3.2 2.5 11.7 2.4 8.3 4.0 4.5 0.6 3.7 1.7

-9.4 33.6 -4.3 20.4 92.5 4.8 3.2a 5.9

MP2 geometry used in single-point calculations (see text for further details).

TABLE 6: Calculated HFCCs for Sugar Radicals Resulting from a Breakage of a Phosphoester Bond (G) north radical C5′ C3′

south

atom

Aiso

TXX

TYY

TZZ

Aiso

TXX

TYY

TZZ

C5′-H C5′-H C4′-H C4′-H C3′-H C2′-H C2′-H

-21.7 -21.6 12.8 36.3 -20.0 23.5 49.0

-13.4 -13.6 -2.0 -1.7 -13.1 -1.7 -1.8

-0.1 0.0 -1.4 -1.1 0.0 -1.3 -1.1

13.5 13.6 3.4 2.8 13.2 3.0 2.9

-22.6 -21.4 31.8 33.9 -21.4 23.5 47.4

-13.9 -13.8 -1.8 -1.8 -13.4 -1.6 -1.8

0.1 -0.2 -1.4 -1.4 0.0 -1.6 -1.8

13.8 14.0 3.3 3.1 13.4 2.9 3.0

TABLE 7: Spin Density Distribution in Sugar Radicals radical C1′ C2′ C3′ C4′ C5′ O5′ O3′

C5′ C3′

atom

north

south

Hydrogen Removal C1′ 0.75 O1′ 0.10 C2′ 0.96 C3′ 0.78 O3′ 0.12 C4′ 0.76 O1′ 0.14 C5′ 0.81 O5′ 0.16 C5′ 0.11 O5′ 0.89 C4′ O3′ 0.86a

0.75 0.11 0.94 0.79 0.13 0.75 0.16 0.79 0.16 0.12 0.89 0.11 0.84

Hydroxyl Removal C5′ 1.0 C3′ 0.95

1.0 0.96

TABLE 8: Calculated HFCCs (G) for the Ring-Altering Radicals Illustrated in Figure 4 radical

a MP2 geometry used in single point calculations (see text for further details).

possibilities must be considered for the experimentally observed radical. It should be noted that the calculated couplings are subject to alterations if the hydroxyl groups are replaced by phosphates, as discussed by Colson and Sevilla, and this could be leading to differences between experiment and theory.14 The C5′ hydrogen abstraction radical has been thought to be detected in several studies of various DNA constituents,10,11,30,31 and has appeared in discussions in the literature throughout the years. In some cases, the possibility of the formation of this radical was eliminated and various other ring-breaking or ring-

C4′ ring opened (structure I)

atom

C2′-H C3′-H C4′-H C5′-H C5′-H O5′-H C1′ ring opened (structure II) C1′-H C2′-H C2′-H ring breaking (structure V) C4′-H C5′-H O5′-H ring breaking with phosphorus C5′-H (structure VII) O5′-P H2O + H removal (structure VIII) C2′-H C1′-H C5′-H

Aiso

TXX

TYY

TZZ

2.2 -0.9 -0.7 1.6 32.4 -1.9 -1.0 3.0 -21.3 -13.0 0.0 13.0 32.8 -2.3 -0.8 3.1 3.8 -2.1 -1.5 3.5 3.4 -1.5 -0.7 2.1 -11.9 -11.4 -0.4 11.8 12.5 -1.7 -1.4 3.1 33.0 -1.5 -1.1 2.6 -2.8 -1.8 -1.3 3.0 -14.1 -7.6 -1.1 8.8 -4.2 -4.4 -3.0 7.4 0.0 -0.8 -0.7 1.5 -21.1 -2.2 1.0 1.2 -12.7 -7.8 0.0 7.8 27.3 -1.0 -0.4 -1.5 11.2 -1.3 -0.7 2.0

opened radicals were thought to arise.33 Close et al.11 observed evidence of the formation of this radical in irradiated deoxyadenosine monohydrate at 10 K. They recovered a large coupling (Aiso ) -17.5 G) which was thought to be due to C5′H. This coupling is much larger in magnitude than the calculated results obtained in this study for either the N- or S-type radicals (Table 5). However, the experimentally observed coupling exhibits considerable anisotropy (TXX ) -11.8, TYY ) 0.8, TZZ ) 11.0 G) which is not unlike the calculated C5′-H values displayed in Table 5 (TXX ) -11, TYY ) -1, TZZ ) 12 G). The reader is reminded that it is possible to calculate the anisotropic HFCC to a greater degree of accuracy

Sugar Radicals in Irradiated DNA

J. Phys. Chem. B, Vol. 102, No. 39, 1998 7681

Figure 2. C4′, C5′, and O5′ hydrogen HFCCs (G) versus the rotation angle (deg) about the C5′C4′ bond for the C5′-N radical.

than the isotropic component, and thus, the calculated results support the experimental assignment. In their investigation, two weak couplings were also observed, but they could not be completely resolved. From the calculated values, it can be seen that these weak couplings probably arise due to the C4′-H and the O5′-H couplings. Close et al. estimated that 88% of the spin density is located on the C5′ center, which is in agreement with the value obtained in the present study (81%). Hole et al.10 previously observed four parallel lines in the spectrum of irradiated 2′-deoxyguanosine 5′-monophosphate which were assigned to the C5′-H couplings of different conformers of the C5′ hydrogen abstraction radical. From the results accumulated in Table 4, it can be seen that all of these couplings are highly similar with considerable anisotropy and their conclusion appears to be legitimate. However, the values for the anisotropic tensor obtained by Hole and co-workers are not as large as those observed in deoxyadenosine monohydrate11 or those obtained in the present study. They proposed that the spin density on C5′ is between 75 and 86%. In addition to the large couplings, a small isotropic coupling was also observed for each conformer of approximately 2.5 G, which was suggested to arise due to C4′-H. The results in Table 5 indicate that the C4′-H coupling is expected to be larger than this value. The small experimental coupling is closer to the calculated O5′-H coupling, although this assignment is unlikely since it would imply that the C4′-H coupling was not observed even though the calculations indicate this coupling to be much larger in magnitude. The system investigated by Hole et al.10 and discussed above differs from the model radical used in the present study in that a phosphate group was present at the C5′ position. However, it was suggested that a similar radical with the phosphate group replaced by a hydroxyl group was also observed in the experimental spectrum. Principal values were obtained which were typical of β(OH) couplings (AXX ) 16.3, AYY ) 20.2, AZZ ) 28.1 G). The calculated values for this coupling are, however, much smaller in magnitude. Alexander and Franklin observed a radical upon irradiation of deoxyadenosine which was debated to be either the C5′ radical or a radical which is formed upon breakage of the C4′O1′

bond within the sugar ring.31 The radical was concluded to be the C5′ radical, and a considerable degree of anisotropy, as seen in other studies, was observed. However, comparison of their isotropic results with those obtained in other studies (Table 4) and the presently calculated results indicate that perhaps reexamination of this radical is necessary. The results obtained by Alexander and Franklin will be discussed further below. Table 4 also displays the results obtained for the C5′ radical upon irradiation of cytosine 3′-monophosphate, 5-chlorodeoxyuridine, and 5-bromodeoxyuridine.30 The C5′-H isotropic values are not unlike those obtained by Hole et al.,10 while the anisotropic values are highly similar to those obtained by the same group in a later study.11 In addition to the C5′-H HFCCs, two other significant couplings were observed in all samples upon irradiation which were assigned to the O5′ and C4′ hydrogens. However, the magnitude of the couplings is quite different in the two sets of data, neither of which match the results obtained in the present study. Due to the discrepancies between the experimental and theoretical results, a more in-depth look at the couplings assigned to the radical formed via abstraction of the C5′ hydrogen is required. Due to the significant effects of the geometry on the HFCCs, an investigation of the dependence of the HFCCs in the C5′ radical on rotation about the C5′C4′ bond was undertaken in which the XC5′C4′C3′, X ) O5′ or H5′, dihedral angles were varied by increments of 15° starting from the optimized geometry. The dihedral angles in the optimized geometry were 289.3 and 14.4 4° for X ) H5′ and O5′, respectively. The results for the variation in the C4′-H, C5′H, and O5′-H HFCCs as a function of rotation angle are displayed in Figure 2. It is interesting to note that upon rigid rotation, the isotropic component of the HFCCs changes considerably, whereas the anisotropic components (not shown) do not differ more than 20% from the values displayed in Table 4. On average, the rotation barrier about the C4′C5′ bond is 8.6 kcal/mol, with maximum and minimum values occurring at 90° (14.4 kcal/mol) and 15° (1.4 kcal/mol) rotations, respectively. The results in Figure 2 shed some light on the dependence of the HFCCs on the rotation about the C5′C4′ bond. The

7682 J. Phys. Chem. B, Vol. 102, No. 39, 1998 calculated C5′-H isotropic HFCC does not reach the experimental value of -22 G obtained in 2′-deoxyguanosine 5′monophosphate but comes close to the value obtained in deoxyadenosine monohydrate (-17 G) upon a 300° rotation (-16.7 G). The variation between the O5′-H and C4′-H results obtained for cytosine 3′-monophosphate and 5-chloroor 5-bromodeoxyuridine can be understood from these calculations. For the cytosine results, the best calculated values which satisfy both the C4′ and O5′ couplings occur at a 130° or 240° rotation, where Aiso(O5′) ) 22.6 G and Aiso(C4′) ) 8.1 G (experimental values are 20.8 and 4.5 G, respectively). It should be noted that the best C5′-H HFCCs are obtained at a 130° rotation. Through similar analysis, the results in best agreement with the 5-chloro- or 5-bromodeoxyuridine HFCCs occur upon a 150° rotation, where Aiso(C4′) ) 17.7 G and Aiso(O5′) ) 10.3 G (experimental values are 18.9 and 8.6 G, respectively). Hence, the calculated results agree very well with the HFCCs which were obtained experimentally in these studies. At 130° rotation, the results obtained by Alexander and Franklin in deoxyadenosine for the C5′-H and C4′-H HFCCs (-14 and 7 G, respectively) are also in good agreement with the calculated values (-16.7 and 8.4 G). However, at this degree of rotation, the calculated results indicate a large O5′-H HFCC as well (30.5 G) which was not detected in the experiment. Thus, the results of Alexander and Franklin31 cannot be understood through this rotation analysis, and other radical possibilities must be considered. Radicals Formed through Breakage of a Phosphoester Bond. Two radicals formed through breakage of a phosphoester bond at the C3′ and C5′ positions have been observed experimentally. In our model system, these radicals would be formed through net removal of a hydroxyl radical at the C5′ and C3′ positions. Hole et al.10 observed the radical formed through breakage at the C5′ position in 2′-deoxyguanosine 5′monophosphate at temperatures below 10 K. The formation of this radical at such low temperatures led the authors to conclude that it is unlikely to arise from a base radical but is probably formed via direct reduction or (super)excitation. No experimental data for the similar C3′-centered radical were found. The energetics, displayed in Table 1 under hydroxyl removal, indicate that the C3′-centered radical is approximately 3 kcal/mol lower in energy than the C5′-centered radical. Hence, some other factor must be responsible for the stability of the C5′-centered radical over the similar C3′-centered radical. Formation of both radicals would lead to single-strand breaks in DNA. The HFCCs calculated for the net hydroxyl removal (from the model system) radicals are displayed in Table 6. The calculated HFCCs for the N- and S-type C5′-centered radicals have similar characteristics, each containing two large isotropic C5′-H couplings of equal magnitude with a high degree of anisotropy. However, the HFCCs in the two conformers of this radical differ through the C4′-H coupling, which is much larger for the south radical (31.8 G) than its north counterpart (12.8 G). Experimentally, this radical has been observed in deoxycytidine 5′-monophosphate (Table 4).30 The calculated C5′-H couplings are in good agreement with experimental values, although the calculated anisotropic values are greater in magnitude (largest component for the tensors are -8.3 and -14 G for experiment and theory, respectively). The large isotropic C4′-H coupling (36.0 G) obtained experimentally indicates that the observed radical is most likely in a south conformation (theory: 31.8 G). This C5′ radical was also suggested to be observed in 2′-deoxyguanosine 5′-monophosphate in two dif-

Wetmore et al. ferent conformers.10 The experimental C5′-H couplings are slightly smaller in magnitude than the calculated values (∼ -17 versus -22 G). In addition, a small coupling of approximately 6 G was observed for both conformers and assigned to C4′-H. This value, although slightly smaller than that calculated (13 G), indicates that the observed radical was probably a north-type conformer. The HFCCs in the radical formed by net abstraction of a hydroxyl group from C3′ in our model structure are also displayed in Table 6. The couplings in both the N- and S-type radicals are very similar, consisting of four large couplings: a C4′-H coupling of approximately 35 G, a C3′-H coupling of -21 G, and two C2′-H couplings of 24 and 48 G. In addition, the C3′-H coupling exhibits a large degree of anisotropy (the largest component of the tensor is 13 G). These values cannot be compared to experiment due to lack of data. Alkoxyl Radicals. There are two possible alkoxyl radicals that can be formed through abstraction of a hydrogen from the oxygens at the C5′ and C3′ positions. Both forms have been assigned to various observed spectra on different occasions.10,30,32,34 In DNA, these radicals would represent breakage of a bond within the phosphate group and, hence, would also lead to strand breaks. Bernhard and co-workers performed a detailed examination of the O5′ alkoxyl radicals formed in a variety of compounds.32 In their work, it was noted that alkoxyl radicals are relatively unstable, decaying at 4.2 K in serine and in other crystals between 77 and 120 K. The results obtained by Bernhard et al. are displayed in Table 4, and from here, it can be seen that there is a variation in the magnitude of the two C5′-H couplings when different compounds are considered. It was pointed out in their work, however, that the sum of the two couplings varies over a small range between 134 and 145 G. The calculated results are displayed in Table 5. The couplings for both the north- and south-type O5′ radicals are very similar, which is not surprising since the radical center is outside the sugar ring and, hence, puckering effects on the HFCCs are expected to be small. The HFCCs of both conformers consist of two large C5′-H couplings of 91 and 19 G on average and a smaller C1′ coupling of approximately 3 G. Due to the difference in the magnitude of the couplings presented in Table 4 and the calculated results in Table 5, a further examination of the couplings in this radical was undertaken. As for the C5′ hydrogen abstraction radical, the effects of rotation about the C5′-C4′ bond on the HFCCs were examined. The results are displayed in Figure 3 as a function of an increase in the O5′C5′C4′C3′ dihedral angle by increments of 15° starting from the optimized geometry. The two C5′-H and the O5′-optimized dihedral angles with respect to C3′ are 64.2, 309.4, and 193.2°, respectively. The average rotation barrier about the C4′C5′ bond (2.7 kcal/mol) is much smaller than the barrier for the C5′ radical. The maximum (7.0 kcal/ mol) and minimum (0.2 kcal/mol) barriers occur at 315 and 210° rotations, respectively. From the rotation investigation, it can be seen that the results vary greatly, although not in the smooth manner observed for the C5′ radical. In some instances, the rotation study clarifies the discrepancies between experiment and theory. For example, the C5′-H experimental couplings observed in uracil-β-Darabinofuroside30 (90.0 and 47.3 G) and adenosine-HCl32 (93.5 and 48.0 G) are in better agreement with the results obtained upon 45° rotation (88.2 and 48.6 G) than the calculated values obtained at the optimized geometry (Table 5). In other instances, the rotation study does not explain the experimental

Sugar Radicals in Irradiated DNA

J. Phys. Chem. B, Vol. 102, No. 39, 1998 7683

Figure 3. C5′ hydrogen HFCCs versus the rotation angle (deg) about the C5′C4′ bond and the sum of these couplings for the O5′-N radical.

results. For example, the two C5′-H couplings in 5-chlorodeoxyuridine32 (83.3 and 85.6 G) and those obtained in 5-bromodeoxyuridine (58.5 and 57.3 G) are equal in magnitude. However, although the calculated C5′-H couplings come close in value at a 105° rotation (66.0 and 71.0 G), they do not attain either of the above values at the same time. It should be noted that there exists an extensive hydrogen-bonding scheme with respect to O5′ in these molecular systems which was not accounted for in the calculations, and thus, theory and experiment are in remarkable agreement. Figure 3 also displays the values obtained for the sum of the two C5′-H couplings. Although the calculated values for the sum are on average slightly smaller than those obtained by Bernhard et al.,32 it can be seen that the calculated sum varies over only a small range of approximately 25 G, close to the experimental range of 22 G. In addition, the ratios of the two couplings obtained from the calculations vary from 1 to 5, which are in good agreement with the experimental ratios of 1 to 6. Taken together, all of the above information supports the assignment of these couplings to the O5′ alkoxyl radical. The O3′ alkoxyl radical was observed by Hole and coworkers10,34 upon irradiation of 2′-deoxyguanosine 5′-monophosphate. The values displayed in Table 4 indicate that significant isotropic (20.3 G) and anisotropic components (-4.7, -0.5, 5.1 G) were exhibited for this radical. The results obtained for the south conformer are in poor agreement with the experimental values. The isotropic component is much too small in magnitude (12.5 G), and a relatively small degree of anisotropy is exhibited (-2.0, -1.7, 3.7 G). Unfortunately, the north conformer has not been detected upon optimization at the DFT level. This radical has been isolated at the HF and MP2 levels with the 6-31G(d,p) basis set, but upon optimization with the same basis set and the B3LYP functional, the south-type radical previously studied is obtained. Since MP2 and DFT geometries are comparable, the MP2-optimized geometry (Tables 2 and 3) was used for the DFT single-point calculations in order to obtain the HFCCs (Table 5) and the spin density

Figure 4. Model systems for various ring-altering sugar radicals: C4′centered ring-opened radical (I), C1′-centered ring-opened radical (II), ring-breaking radicals observed experimentally (III and IV), model ringbreaking radical (V), ring-breaking C5′-centered radical (VI), model ring-breaking radical with phosphate group (VII), and radical formed via H2O elimination from products formed by hydrogen abstraction at C2′ or C4′ (VIII).

distribution (Table 7) in the O3′-N radical. The C3′-H isotropic HFCC calculated for the N-type radical is much smaller in magnitude (3.2 G) than that obtained for the south conformer (12.5 G) or the experimental radical (20.3 G). In addition, the anisotropic couplings are nearly identical to those obtained for the south conformer, which are in poor agreement with the experimental couplings. The reason for such poor agreement between experiment and theory is not available at this time. Ring-Breaking Radicals. On numerous occasions in the literature, various radicals (Figure 4) have been postulated which involve some kind of sugar ring breakage or more extensive damage to the ring than sole removal of a hydrogen atom or breakage of a phosphoester bond.10,31,33-35 The first such radical to be discussed involves breaking of the sugar ring through the

7684 J. Phys. Chem. B, Vol. 102, No. 39, 1998 generation of a radical at the C4′ center. This radical is illustrated in Figure 4, structure I. Although this radical has been proposed in several instances,10,33,35 the idea of its formation has often been rejected33 and the HFCCs for this radical have only been isolated in two cases.10,35 The experimental results for this radical are displayed in Table 4 (“C4′ ring-opened radical”). The two sets of experimental results exhibit great differences in the magnitude of the couplings and in the centers to which the couplings are assigned. For example, the C5′-H couplings in the two studies differ greatly. However, it is interesting to note that the sums of the couplings are very similar (61 versus 64 G), indicating that alternative conformers may be responsible for the differences. The 2′-deoxyguanosine 5′-monophosphate spectrum33 contains a large coupling of 32.2 G that was assigned to C4′-H. On the other hand, a different C4′-H isotropic coupling was observed in uridine 5′-phosphate35 (-18.8 G). The calculated results for this radical are listed in Table 8. The calculations indicate that 95% of the spin density resides on C4′. From the results, it can be seen that a large isotropic coupling is obtained at the C4′ hydrogen (-21.3 G) which has significant anisotropy (-13.0, 0.0, 13.0 G). This is not unlike the situation observed in uridine 5′-phosphate.35 Two additional couplings of substantial magnitude were also obtained from the calculations for the C3′ (32.4 G) and the C5′ (32.8 G) hydrogens. However, these couplings do not correspond to those observed in uridine 5′-phosphate or those obtained by Hole et al. for 2′deoxyguanosine 5′-monophosphate.10 In addition, the large isotropic HFCC calculated for C3′-H (32.4 G) is not unlike the large coupling assigned to C5′-H (27 G) in 2′-deoxyguanosine 5′-monophosphate. Unfortunately, due to the discrepancies between experiment and theory for this proposed radical, further information cannot be obtained at this time. Experimental studies of this radical with partially deuterated samples would aid in the determination of the entire coupling tensor to a greater degree of accuracy, whereby the deviations in experimental assignment possibly could be unveiled. It should be noted, however, that these studies are unavailable due to the fact that the C3′, C4′, and C5′ hydrogens are not easily replaced. The sugar ring-opened radical was also suggested to have been observed in deoxyadenosine monohydrate by Alexander and Franklin.31 It was determined in their study that this radical was probably not responsible for the observed couplings, and the spectrum was eventually assigned to the C5′ radical. However, as previously discussed, the C5′ radical is probably not responsible for the observed experimental couplings in this case. A comparison of their results (under C5′ radical in Table 4) with the previous experimental and calculated values for the ring-opened radical furthermore suggests that the present radical is unlikely to give rise to the couplings observed by Alexander and Franklin. A similar ring-opened radical can also be considered in which C1′ is the radical center (Figure 4, structure II). Formation of this radical involves the breakage of the O1′C1′ bond in the sugar ring. This ring-opened radical has, to the best of our knowledge, not been suggested in experimental studies to be formed upon irradiation. Single-point calculations at the B3LYP/6-311G(2df,p) level indicate that the C4′-centered radical lies 16.2 kcal/mol lower in energy than the corresponding C1′ radical. This great energy difference could be the reason that the C1′-centered radical has not been detected. The calculated HFCCs for this radical appear in Table 8. From the results, it can be seen that C1′-H has a large isotropic coupling (-11.9 G) which has considerable anisotropy. In addition, both

Wetmore et al. C2′ hydrogens have large isotropic HFCCs. The spin density distribution in this radical is mostly located on C1′ (0.81). However, some of the spin density is delocalized to N1 (0.17). This indicates that if a base is attached at this position, rather than our model amine group, some of the spin density would be distributed throughout the base. The second series of radicals leading to sugar ring breaks is depicted in Figure 4, structure III. This radical has been observed in 2′-deoxyguanosine 5′-monophosphate33 and later examined by Hole and Sagstuen.34 This radical has been proposed to be formed in nucleotides by abstraction of a hydrogen atom from the C5′ position by a base, followed by breakage of the sugar ring and reorientation of O1′.33 A very similar radical appears in Figure 4, structure IV, where this radical was observed only after irradiation at room temperature.34 The coupling constants in these radicals were calculated using a model system displayed in Figure 4, structure V, where the hydroxyl group represents either a phosphate group33 or a carbon group.34 The experimental couplings are displayed in Table 4 (“ringbreaking radical”). The two sets of experimental results are in good agreement with a large isotropic coupling of approximately -17 G (C5′-H) which exhibits considerable anisotropy and a smaller nearly isotropic coupling of approximately 4 G (C4′H). The major difference in the two sets of results is that the magnitude of the largest component of the anisotropic tensor is much greater for the results obtained in an earlier study (11.0 G) relative to the values obtained by Hole and co-workers more recently (8.0 G). The calculated results for this radical are displayed in Table 8, and good agreement is observed with both sets of experimental results. In particular, the calculated anisotropic results agree more closely with the results of Hole et al.34 In addition to the couplings observed for the two hydrogens attached to the carbons, the hydrogen in the hydroxyl group also exhibits a notable coupling. It is pointed out, however, that in the experimental setting, this coupling is not possible since the hydroxyl group was chosen to model the much larger phosphate or carbon groups. Hole et al.34 also suggested that the couplings observed in their study could arise from the radical displayed in Figure 4, structure VI, which also displays many similarities to structure II. They claimed that the experimental coupling of -17 G could arise from the phosphate group. This possibility was explored in the present study through the use of the model system of structure VII. The results obtained from the calculations (Table 8, “ring breaking with phosphorus”) indicate that the phosphorus does give rise to a coupling (-21 G) very similar to that observed experimentally (-17 G). However, the calculated phosphorus coupling does not exhibit the same anisotropy as the experimental HFCC. In addition, the C5′-H in this model system does not exhibit a coupling of 3.8 G which was observed experimentally. Thus, due to the good agreement obtained for the other ring-breaking radical modeled by structure V and the lack of the anisotropy in the phosphorus coupling, as well as the small C5′-H coupling, in the model illustrated in structure VII, it can be concluded that the most likely structure for the radical observed in their study is that displayed in structure III. A mechanism for the formation of the radical depicted in structure III has been discussed previously by Rakvin and Herak.33 In addition to the good agreement with experiment for the calculated couplings of the ring-breaking radical, these values can lead to some valuable insight into the problem of the poor

Sugar Radicals in Irradiated DNA agreement of Alexander and Franklin’s results with the values calculated using their proposed radical structures. Recall that their couplings displayed in Table 4 for the C5′ radical were in poor agreement with the calculated and other experimental results for this radical. In addition, their suggestion that the couplings could be due to the ring-opened radical (structure I) was determined to be unlikely, as concluded in their paper. Comparison of the results obtained in their study (under deoxyadenosine monohydrate C5′ radical in Table 4) and the calculated results for the ring-breaking radical discussed in this section leads to the conclusion that a radical similar to that depicted in Figure 4, structure III, is most likely responsible for the couplings observed by Alexander and Franklin. In addition, the other C5′-H coupling listed in Table 4 for deoxyadenosine monohydrate (under the C5′ radical) is more similar to the values obtained for the ring-breaking radical than those obtained for the C5′ radical. The final radical to be discussed is structure VIII in Figure 4. This radical can be formed either through abstraction of a hydrogen from C2′ followed by removal of water (C3′-OH and C4′-H) or through abstraction of a hydrogen from C4′ followed by removal of water (C3′-OH and C2′-H). Since the formation of this radical involves breakage of the C3′O bond, this would lead to single-strand breaks in DNA when the hydroxyl groups are replaced by phosphates. This radical has been proposed to give rise to couplings observed in 2′deoxyguanosine 5′-monophosphate,34 and the experimental results are displayed in Table 4 under the H2O + H removal radical. Four large couplings were observed experimentally for this proposed radical, two of which have considerable anisotropy. The optimized geometry of this radical is planar, and the calculated couplings, displayed in Table 8, support the planarity in that this system is a π-radical with the spin density distributed throughout the molecule. Only three large couplings were obtained from the calculations, and the degree of anisotropy was not as high as that observed in the experimental results. The experimentally assigned R coupling (-16.0 G) is not unlike the calculated coupling for C2′-H (-12.7 G). However, although both couplings exhibit a large anisotropy, there is a difference in its magnitude. Two of the experimentally assigned R couplings (12.9 and 25.0 G) are similar in magnitude to the C5′-H and the C1′-H HFCCs, respectively (11.2 and 27.3 G). The third large coupling exhibited in the experiments (23.0 G) cannot be accounted for in the calculations. It is possible that DFT has incorrectly predicted this radical to be planar as observed in the studies of thymine17 and cytosine16 where an inadequate description of the ring puckering was noted in some cases. Another possible explanation could be that the third large R coupling arises from the other C5′-H and that this coupling is not observed in the calculations due to the fixed orientation of the methoxyl group, but experimentally, a rotation of this group is observed (Figures 2 and 3 and discussion above). Further insight into the presence of an additional large coupling in the experimental results and the corresponding absence of this coupling in the theoretical results is not available at this time. Conclusions In this study, possible sugar radicals formed upon irradiation of DNA were examined through the use of DFT. The types of radicals examined included various hydrogen abstraction radicals, radicals formed via breakage of a phosphoester bond, and different radicals arising from significant alterations of the sugar ring. The results calculated for the hyperfine coupling constants

J. Phys. Chem. B, Vol. 102, No. 39, 1998 7685 in these radicals were compared to experimental values which have appeared in the literature. The energetics indicate that the C4′ south-type radical and the C3′ south-type radical are the lowest lying species for radicals formed via hydrogen abstraction and removal of a hydroxyl group, respectively. The C2′-centered radical is higher in energy than any other carbon-centered radical and has a relatively small ring puckering. Alterations in the sugar ring geometry were found to occur predominantly at the radical center for all radicals. The dipole moments are larger for the south-type radicals relative to their north counterparts, which indicates a greater stabilization for the S-type radicals upon inclusion of solvation effects. The calculated hyperfine couplings in the dehydrogenated radicals support the experimental assignments of the various radical forms in most cases. The agreement between experiment and theory is extremely good in spite of the fact that differences in radical geometries may arise due to crystal interactions in the experimental studies which were not accounted for in the theoretical model. Differences in the couplings of north- and south-type radicals, arising from differences in the puckering amplitudes, lead to some speculations of which forms were observed in the experimental setting. Studies of couplings versus rotation about the C5′C4′ bond were required for the C5′ and O5′ hydrogen abstraction radicals in order to confidently support the experimental assignments of these radicals. In addition to the radicals formed through removal of a hydrogen or a hydroxyl group in the model system, different ring-altering radicals were examined and attempts to clarify experimental discrepancies were made. Acknowledgment. We gratefully acknowledge the Natural Sciences and Engineering Research Council of Canada (NSERC), the Swedish Natural Science Research Council (NFR), and the Killam Trust for financial support. We also thank the Computing and Network Services at the University of Alberta and the Center for Parallel Computing at the Institute of Technology, Stockholm, for grants of computer time. Supporting Information Available: Complete geometries for all radicals (6 pages). Ordering information is given on any current masthead page. References and Notes (1) Close, D. M. Rad. Res. 1995, 1, 135 and references therein. (2) Colson, A.-O.; Sevilla, M. D. Int. J. Radiat. Bio. 1995, 67, 627. (3) Prakash Rao, P. J.; Bothe, E.; Schulte-Frohlinde, D. Int. J. Radiat. Bio. 1992, 61, 577. (4) Dizdaroglu, M.; Gajewski, E.; Reddy, P.; Margolis, S. A. Biochemistry 1989, 28, 3625. (5) Olinski, R.; Briggs, R. C.; Hnilica, L. S.; Stein, J.; Stein, G. Radiat. Res. 1981, 86, 102. (6) Gajewski, E.; Dizdaroglu, M. Biochemistry 1990, 29, 977. (7) von Sonntag, C. The Chemical Basis of Radiation Biology; Taylor and Francis: New York, 1987. (8) Becker, D.; Sevilla, M. D. AdVances in Radiation Biology; Academic: New York, 1993; p 121. (9) Schuchmann, M. N.; von Sonntag, C. J. Chem. Soc., Perkin Trans. 1977, 2, 1958. (10) Hole, E. O.; Nelson, W. H.; Sagstuen, E.; Close, D. M. Rad. Res. 1992, 129, 119. (11) Close, D. M.; Nelson, W. H.; Sagstuen, E.; Hole, E. O. Rad. Res. 1994, 137, 300. (12) Close, D. M. Rad. Res. 1997, 147, 663. (13) Miaskiewicz, K.; Osman, R. J. Am. Chem. Soc. 1994, 116, 232. (14) Colson, A.-O.; Sevilla, M. D. J. Phys. Chem. 1995, 99, 3867. (15) Feller, D.; Glendening, E. D.; McCullough, E. A., Jr.; Miller, R. J. J. Chem. Phys. 1993, 99, 2829. (16) Wetmore, S. D.; Himo, F.; Boyd, R. J.; Eriksson, L. A. J. Phys. Chem. B, in press.

7686 J. Phys. Chem. B, Vol. 102, No. 39, 1998 (17) Wetmore, S. D.; Boyd, R. J.; Eriksson, L. A. J. Phys. Chem. B 1998, 102, 5369. (18) The original three-parameter hybrid suggested by Becke can be found in: Becke, A. D. J. Chem. Phys. 1993, 98, 1372. A slightly modified form implemented in the Gaussian programs can be found in: Stephens, P. J.; Devlin, F. J.; Chablowski, C. F.; Frisch, M. J. J. Phys. Chem. 1994, 98, 11623. (19) Lee, C.; Yang, W.; Parr, R. G. Phys. ReV. B 1988, 37, 785. (20) Ditchfield, R.; Hehre, W. J.; Pople, J. A. J. Chem. Phys. 1971, 54, 724. Hehre, W. J.; Ditchfield, R.; Pople, J. A. J. Chem. Phys. 1972, 56, 2257. Hariharan, P. C.; Pople, J. A. Mol. Phys. 1974, 27, 209. Gordon, M. S. Chem. Phys. Lett. 1980, 76, 163. Hariharan, P. C.; Pople, J. A. Theor. Chim. Acta 1973, 28, 213. McLean, A. D.; Chandler, G. S. J. Chem. Phys. 1980, 72, 5639. Krishnan, R.; Binkley, J. S.; Seeger, R.; Pople, J. A. J. Chem. Phys. 1980, 72, 650. Clark, T.; Chandrasekhar, J.; Spitznagel, G. W.; Schleyer, P. v. R. J. Comput. Chem. 1983, 4, 294. Frisch, M. J.; Pople, J. A.; Binkley, J. S. J. Chem. Phys. 1984, 80, 3265. (21) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Gill, P. M. W.; Johnson, B. G.; Robb, M. A.; Cheeseman, J. R.; Keith, T. A.; Petersson, G. A.; Montgomery, J. A.; Raghavachari, K.; Al-Laham, M. A.; Zakrzewske, V. G.; Ortiz, J. V.; Foresman, J. B.; Cioslowski, J.; Stefanov, B. B.; Nanayakkara, A.; Challacombe, M.; Peng, C. Y.; Ayala, P. Y.; Chen, W.; Wong, M. W.; Andres, J. L.; Replogle, E. S.; Gomperts, R.; Martin, R. L.; Fox, D. J.; Binkley, J. S.; Defrees, D. J.; Baker, J.; Stewart, J. P.; HeadGordon, M.; Gonzalez, C.; Pople, J. A. Gaussian 94 (Revision B.2); Gaussian, Inc.: Pittsburgh, PA, 1995. (22) Perdew, J. P.; Wang, Y. Phys. ReV. B 1986, 33, 8800. (23) (a) Perdew, J. P. Phys. ReV. B 1986, 33, 8822. (b) Perdew, J. P. Phys. ReV. B 1986, 34, 7406.

Wetmore et al. (24) St-Amant, A.; Salahub, D. R. Chem. Phys. Lett. 1990, 169, 387. St-Amant, A. Ph.D. Thesis, Universite´ de Montre´al, 1991. Salahub, D. R.; Fournier, R.; Mlynarski, P.; Papai, I.; St-Amant, A.; Ushio, J. In Density Functional Methods in Chemistry; Labanowski, J., Andzelm, J., Eds.; Springer: New York, 1991. (25) Malkin, V. G.; Malkina, O. L.; Eriksson, L. A.; Salahub, D. R. In Modern Density Functional Theory, A Tool for Chemistry; Seminario, J. M., Politzer, P., Eds.; Elsevier: New York, 1995. (26) Engels, B.; Eriksson, L. A.; Lunell, S. AdV. Quan. Chem. 1997, 27, 297. (27) Saenger, W. Principles of Nucleic Acid Structure; SpringerVerlag: New York, 1984. (28) Bauschlicher, C. W., Jr.; Partridge, H. J. Chem. Phys. 1995, 103, 1788. (29) Nelson, W. H.; Sagstuen, E.; Hole, E. O.; Close, D. M. Rad. Res. 1992, 131, 272. (30) . Effects of Ionizing Radiation on DNA; Hu¨ttermann, J., Ko¨hnleif, W., Teoule, R., Bertinchamps, A. J., Eds.; Springer: Heidelberg, 1978. (31) Alexander, C., Jr.; Franklin, C. E. J. Chem. Phys. 1971, 54, 1909. (32) Bernhard, W. A.; Close, D. M.; Hu¨ttermann, J.; Zehner, H. J. Chem. Phys. 1977, 67, 1211. (33) Rakvin, B.; Herak, J. N. Rad. Res. 1981, 88, 240. (34) Hole, E. O.; Sagstuen, E. Rad. Res. 1987, 109, 190. (35) Sagstuen, E. Rad. Res. 1980, 84, 164. (36) Sagstuen, E. J. Magn. Reson. 1981, 44, 518. (37) Hu¨ttermann, J. Ultramicroscopy 1982, 10, 25. (38) Hole, E. O.; Nelson, W. H.; Sagstuen, E.; Close, D. M. Rad. Res. 1992, 130, 148. (39) Kar, L.; Bernhard, W. A. Rad. Res. 1983, 93, 232.