Long-range oxidative damage in duplex DNA: the effect of bulged G in a G–C tract and tandem G/A mispairs

Abstract

Two series of duplex DNA oligomers were prepared having an anthraquinone derivative (AQ) covalently linked at a 5′-terminus. Irradiation of the AQ at 350 nm leads to injection of an electron hole (radical cation) into the DNA. The radical cation migrates through the DNA causing reaction primarily at G_n sequences. In one series, GA tandem mispairs are inserted between GG steps to assess the effect of the mispair on the transport of the radical cation, reaction (damage) caused by the radical cation at the mispair, and repair of the resulting damage by formamidopyrimidine DNA glycosylase (Fpg). In the second series, a bulged guanine in a G₃C₂ sequence is interposed between the GG steps. These experiments reveal that neither G/A tandem mispairs nor bulged guanines are significant barriers to long-range charge migration in DNA. The radical cation does not cause reaction at guanines in the G/A tandem mispair. Reaction does occur at the bulged guanine, but it is repaired by Fpg.

INTRODUCTION

Oxidative damage to DNA can cause disease or death and has been linked to as many as half of all human cancers (1). Since guanine has the lowest oxidation potential of the common DNA bases (2), oxidation (removal of an electron) from DNA to generate a radical cation (electron hole) often results in an irreversible reaction at a guanine; most commonly, at a 5′-G of a GG step and less often at the 5′-G of a GA sequence (3,4). Guanine radical cations in duplex DNA react with water to form, mainly, 8-oxoguanine (8-OxoG) (5,6). Unrepaired 8-OxoG lesions cause G→T transversions in prokaryotic and eukaryotic cells, which lead to harmful effects including mutations (7).

A collection of enzyme systems has evolved that recognize and repair DNA damage and help to maintain the integrity of the genome (8). In particular, the repair enzyme Fpg (formamidopyrimidine DNA glycosylase) which excises oxidized forms of guanine, is critically involved in the prevention of mutations in vivo. In fact, it seems reasonable to suppose that the adverse biological effects of oxidative damage to DNA will result only when Fpg or related repair systems fail. The sequence of bases and the local secondary structure surrounding an oxidative lesion affect the effectiveness of its repair by Fpg (9). For example, the excision of an 8-OxoG opposite an A in duplex DNA is poor (10), and similar relative reactivity has been observed for repair by the human glycosylase enzyme hOGH1 (11).

Recent experiments have demonstrated that radical cations introduced into DNA by an oxidative reaction can migrate >200 Å to a GG step where they react to form an oxidized guanine (4,12,13). The radical cation migrates by a hopping process during which it resides on bases between the site of its introduction and the GG step that ultimately reacts (14–17). The effect of base sequence on this process is not clearly understood (15,18–21). We considered the possibility that particular, unusual secondary structures in DNA might be especially susceptible to oxidative damage and simultaneously be relatively resistant to the action of enzymatic repair systems. Such structures would be especially prone to mutation.

There have been several investigations of the effect of atypical structures on the efficiency of long-distance radical cation transport in DNA. These include abasic sites (22), mis-matched base pairs (23–25), and unpaired bases (26–28). These studies reveal complex behavior that is strongly determined by details of structure. For example, a single G/A mispair does not present a barrier to charge transport, but a C/A mispair does (23). Furthermore, it has been reported that bulged bases inhibit radical cation migration and that this feature might protect cellular DNA from long-range oxidative damage (26). Significantly, the reactivity of a guanine in DNA is modulated by its environment. Only guanines in G/G mispairs, bulges or single-strand loops are oxidized by nickel complexes (29). Similarly, alkylation of guanine occurs selectively opposite a bulged base (30), and the solvent accessibility of a guanine involved in a mispair is enhanced compared with the standard Watson–Crick G/C pair (31). These findings suggest that unusual secondary structures in DNA might be specially involved in oxidative mutagenesis.

We report here results of an examination of radical cation transport and reaction in DNA oligomers constructed to contain atypical secondary structures. In one set of experiments, we examine the effect of G/A tandem mispairs on radical cations in DNA. These structures are intensively studied since they occur in highly conserved regions of single-stranded DNA viruses and in the peri-centromeric regions of human chromosomes (32). The detailed structure of the tandem mispair is determined by the nature of its flanking base pairs, and we examined mispairs in both the ‘sheared’ and ‘G_antiA_anti’ structures (33). Similarly, we have prepared and examined radical cation transport and reaction in DNA oligomers that contain a bulged guanine, since this structure has been identified as a mutational hot spot (34).

MATERIALS AND METHODS

General

Oligodeoxyribonucleotides were prepared by standard solid phase methods and purified by reverse-phase HPLC. The anthraquinone phosphoramidite was synthesized as previously described by Gasper and Schuster (22). The buffer used in all DNA experiments was 10 mM sodium phosphate at pH 7.0. The concentrations of the purified oligomers were determined by optical spectroscopy at 260 nm (22).

Melting temperatures were determined by the optical absorbance change at 260 nm for 2.5 µM samples in buffer solution in a sealed 1.0 cm quartz cell using a Cary 1E 3C spectrometer. The heating and cooling rates were 0.5°C/min with a temperature range of 15–99°C. The T_m value for each duplex was identified as the maximum of the first derivative plot of its absorbance versus the temperature. Circular dichroism (CD) spectra were obtained at room temperature on a JASCO (J-720) CD spectrophotometer in 10 mM sodium phosphate buffer solution. The results are an average of five scans. Images and quantification of images of gels or autoradiograms were obtained with a Fuji 2340 BAS-Images or a Scanalytics AMBIS 4000 Radioanalytic Imaging System; the results do not differ.

Preparation of radiolabeled DNA

³²P-5′-end-labeling was accomplished with T4 polynucleotide kinase and [γ-³²P]dATP (Amersham Pharmacia; used as received). DNA strands (5.0 µM) were incubated with 2.0 µl of [γ-³²P]dATP (6000 Ci/mmol) and 1.0 µl of T4 polynucleotide kinase in a total volume of 30 µl at 37°C for 45 min. After incubation, the labeled DNA was suspended in denaturing, loading buffer and purified on a 20% denaturing polyacrylamide gel. Labeled product was located by autoradiography. Bands corresponding to the DNA product were excised from the gel and eluted with 750 µl of elution buffer [0.5 M NH₄OAc, 10.0 mM Mg(OAc)₂, 1.0 mM EDTA and 0.1% SDS] at 37°C for 4 h. The supernatant from each sample was extracted and centrifuged at 12 000 g for 0.5 min. Precipitation proceeded by addition of 1.0 µl of glycogen and 700 µl of cold absolute ethanol. The mixtures were mixed on an agitator, placed on dry ice for 45 min to complete precipitation, and centrifuged for 30 min at 12 000 g in a Savant µSpeedFuge centrifugation. The supernatant was discarded. The resulting DNA pellets were washed twice at room temperature with 80% ethanol and dried with a Savant Speed Vac Plus for 45 min. The dried pellets were reconstituted in buffer solution, hybridized on a thermocycler at 90°C for 5.0 min and slowly cooled to room temperature over a 4-h period.

Photolysis and analysis

Samples for irradiation were prepared by incubating 5.0 µM of unlabeled strands with labeled complementary (5000 c.p.m.) strand in 70 µl of buffer solution. Samples were irradiated in 1.5 µl micro-centrifuge tubes using a Rayonet Photoreactor equipped with 350 nm lamps. The extent of reaction was kept below 10% in all cases. When the irradiation was complete, an aliquot was withdrawn for piperidine treatment and a second portion was removed for direct precipitation. Piperidine treatment consisted of adding 1 M (100 µl) of piperidine per sample. Each sample was mixed in an agitator for 15 s, heated at 90°C for 30 min and subjected for 5 s to a Savant Speed Vac Plus. The samples were dried for 1 h at medium heat. To ensure that all of the piperidine was removed, 20 µl of water was added to each sample and the drying process was repeated. This water-wash procedure was performed twice. Dried samples were dissolved in 5.0 µl of denaturing formamide-loading buffer. The photocleavage products were separated electrophoretically on a 20% polyacrylamide sequencing gel and detected by autoradiography after drying for 2.0 h on a Hoefer Drygel Sr. SE1160 Dryer.

Formamidopyrimidine DNA glycosylase (Fpg) digestion

The Fpg (Trevigen Corporation; used as received) reactions were performed in 20 µl of solution that contained 50 mM Tris–HCl (pH 7.5), 2.0 mM EDTA, 100 mM NaCl and 5.0 µM DNA at 37°C for 1 h. A 15.6 ng portion of Fpg was used in each experiment. The reaction was terminated by heating the sample to 90°C for 15 min after the digestion. The DNA was precipitated by the addition of ethanol and cooling to –20°C. Denaturing formamide loading buffer containing 0.1% bromphenol blue was added to the samples, which were then heated at 90°C for 10 min and quickly cooled to 4°C. The samples were electrophoresed on a 20% polyacrylamide gel and analyzed by phosphorimagery.

RESULTS

Design of oligonucleotides to test charge transport and reaction in duplexes containing G/A tandem mispairs and G bulges

Tandem G/A mispairs are recognized to be especially stable (35), to maintain well-defined B-form global structures (32), and to adopt a local structure that depends upon the identity of its neighboring bases (33). The G_antiA_anti base pairing scheme (Fig. 1) is observed for duplexes containing [5′-d(AGAT)-3′/3′-d(TAGA)-5′] mispairs, but ‘sheared’ base pairing is seen for duplexes containing the [5′-d(CGAG)-3′/3′-d(GAGC)-5′] sequence. The structure of the G_antiA_anti mispair causes a switch from intra-strand base stacking, usually seen in DNA, to a ‘crossover’ inter-strand adenine–adenine stack. In contrast, the sheared duplex has both strong intra-strand stacking with neighboring bases and excellent crossover stacking between the GA pairs (32). For this reason, we suspected that the efficiency of radical cation transport through a G/A mispair might be modulated by its neighboring bases.

Figure 1.

Structures of tandem GA mispairs in the G_antiG_anti and ‘sheared’ conformations. R represents connection to the deoxyribose backbone of the DNA strand.

A bulged guanine in a homopolymeric run of guanines is thought to play a role in frame-shift mutations at the hot spot of the λ C₁ gene (36). Woodson and Crothers assessed the structure of d(G_n/C_n–1) sequences in duplex DNA and found that the position of the bulge is averaged over the G_n sequence and that the flanking positions are bulged less often than the internal positions. They also found that the bulged guanine migrates with low activation energy on a millisecond time scale, which implies a fast base-pair opening rate and enhanced solvent access (37). For these reasons, we suspected that the d(G_n/C_n–1) sequence might be a site for preferred reaction of a radical cation migrating through DNA and that, because of the bulged structure, repair of these lesions by enzymes such as Fpg might by inhibited.

Figure 2 shows the series of DNA assemblies we constructed to examine radical cation transport, reaction and repair through tandem d(G/A) mispairs (Series I) and through a d(G₃/C₂) bulge sequence (Series II). Each DNA conjugate contains an anthraquinone group (AQ) linked covalently to a 5′-terminus and a variable sequence that is flanked by two GG steps: one that is proximal (closer to the AQ) and one distal. Irradiation of the AQ introduces a radical cation that migrates through the DNA causing reaction predominantly at GG and GGG sequences and secondarily at the G of 5′-GA-3′ sequences that is revealed as strand cleavage by subsequent treatment with piperidine (4). The conjugates designed to test the effect of tandem G/A mispairs contain the variable sequence (N₁–N₄/X₁–X₄), which allows us to examine both sheared and G_antiA_anti conformations as well as critical control compounds. The conjugates designed to test the effect of a bulged guanine have a three base pair variable sequence (N₁–N₃/X₁–X₃).

Figure 2.

Structures of modified DNA oligomers.

Characterization of DNA sequences containing tandem d(G/A) mispairs and a bulged guanine

The DNA conjugates used in this work were purified by HPLC and polyacrylamide gel electrophoresis (PAGE). The success of the synthesis was confirmed by Maxam–Gilbert sequence determination (38). The CD spectrum of each duplex shows that it is predominantly in a B-form DNA global structure. The CD spectra of DNA(1) and DNA(2) are shown in Figure 3. Even though they contain G/A mispairs in the G_antiA_anti and sheared configurations, respectively (see below), their spectra are similar to that of DNA(3), which contains only Watson–Crick base pairs. The melting temperatures (T_m) of the duplexes were measured to assess the effect of structure on stability. The data are summarized in Table 1. Comparison of the T_m value for DNA(1–2) with those of their fully complementary control compounds, DNA(3–6), shows that incorporation of the tandem G/A mispair sequence only slightly destabilizes the duplex. Similar behavior of tandem G/A mispairs has been reported previously (39).

Figure 3.

CD spectra of DNA(1–3), each sample is 2.5 µM in 10 mM sodium phosphate buffer solution at pH 7.0. Symbols represent the following: circles, [DNA(1)], the G_antiA_anti tandem mis-pair; squares, [DNA(2)], the sheared configuration tandem mis-pair; triangles, [DNA(3)], a fully complementary base pair sequence.

Table 1. Melting temperatures and reaction ratios for the AQ–DNA conjugates.

AQ–DNA	Tm (°C)	Gp/Gda	Gp/G1b	Gp/G2	Gp/G3
1	52.1	11.0
2	59.1	2.3
3	56.3	11.3
4	55.9	11.4
5	60.8	2.1
6	60.4	2.0
7	54.1	2.1 (2.1)	NDc	1.8 (1.2)	1.3 (1.7)
8	43.5	1.9 (2.2)	3.7 (4.3)	4.0 (3.3)	3.0 (5.0)
9	60.1	1.9 (1.9)	ND	2.8 (3.3)	2.8 (2.5)

^aRatio of strand cleavage observed at the 5′-G of the proximal (G_p) and distal (G_d) steps determined after treatment of the irradiated samples with piperidine or (in parenthesis) with Fpg. SE values from replicate runs are generally <10% of the reported values.

^bRatio of strand cleavage observed at the 5′-G of the proximal (G_p) step and the guanines of the G₃ sequence in DNA(7–9) determined after treatment of the irradiated samples with piperidine or (in parenthesis) with Fpg.

^cNo cleavage detectable above background level.

Irradiation of DNA sequences containing tandem d(G/A) mispairs

Both DNA(1) and DNA(2) contain a G/A tandem mispair. The mispair in DNA(1) is flanked on the 5′-side of the GA sequence by an A/T base pair and on its 3′-side by a T/A base pair. Tandem G/A mispairs in this general sequence (5′-pyrimidine-GA-purine-3′) have been shown to adopt the G_antiA_anti arrangement (39). In contrast, the G/A tandem mispair in DNA(2) is flanked on its 5′-side by a C/G base pair and on its 3′-side by a G/C base pair. Tandem G/A mispairs in this sequence (5′-purine-GA-pyrimidine-3′) adopt the sheared configuration (35). Thus, DNA(1) and DNA(2) allow the examination of the effect of tandem mispairs on transport and reactivity of radical cations in two distinct stacking configurations.

Careful control experiments are required in order to assess the effect of the G/A mispairs on the transport and reactivity of radical cations in duplex DNA. DNA(3) contains the same base pair sequence, X₁–X₄ (5′-TAGA-3′) on the complementary strand (i.e. the strand that is not linked to the AQ) as DNA(1), but its N₁–N₄ sequence (5′-ATCT-3′) removes the G/A tandem mispair and each base in DNA(3) is paired with its Watson–Crick complement. Similarly, DNA(4) is a control compound for DNA(1) that contains only Watson–Crick paired bases, but it differs from DNA(3) in that the X₁–X₄ sequence (5′-TCTA-3′) has been mutated; and, its N₁–N₄ sequence is the same as DNA(1). DNA(5,6) play the same role as control compounds to DNA(2) as DNA(3,4) do for DNA(1).

Samples of DNA(1–6) were radiolabeled at the 5′-termini of the complementary strand and irradiated for 15 min at 350 nm (only the AQ absorbs at this wavelength) in standard phosphate buffer solutions at pH 7.0. After the irradiation was complete, the samples were treated with hot piperidine to convert damaged bases to strand breaks. The treated samples were analyzed by PAGE and phosphorimagery. Figure 4 shows the results from the irradiation of DNA(1,2) in the form of autoradiograms of the gels. There is no strand cleavage (above background) before piperidine treatment, but strand cleavage is easily detected at the proximal and distal GG steps for both samples after treatment with piperidine. Reaction occurs primarily at the 5′-G of each of the GG steps, which is characteristic of the reaction of radical cations in duplex DNA. Importantly, no significant strand cleavage above the background level is seen at the tandem G/A mispair for either DNA(1) or DNA(2). The results of similar experiments carried out on control compounds DNA(3–6) are shown in Figure 5.

Figure 4.

PAGE results from the irradiation of DNA(1) and DNA(2) presented as autoradiograms. The 5′-terminus of the complementary strand is labeled with ³²P. Lanes 1 and 3 are ‘dark controls’, these samples were treated precisely as the experimental samples but were not irradiated. Experimental samples (5.0 µM DNA in 10.0 mM sodium phosphate buffer solution) were irradiated for 15 min at room temperature with 350 nm light and then treated with piperidine before PAGE. Lanes 2 and 4 are the results from irradiation of DNA(1) and DNA(2), respectively. Experiments were conducted under ‘single-hit conditions’, cleavage is revealed at the 5′-G of the proximal and distal GG steps in both experimental samples.

Figure 5.

PAGE results from the irradiation of DNA(3–6) presented as autoradiograms. The 5′-terminus of the complementary strand is labeled with ³²P. Lanes 1, 3 and 5 are ‘dark controls’. Experimental samples (5.0 µM DNA in 10.0 mM sodium phosphate buffer solution) were irradiated for 15 min at room temperature with 350 nm light and then treated with piperidine before PAGE. Lanes 2 and 4 are results from irradiation of DNA(3) and DNA(4), respectively; lanes 6 and 7 are results from irradiation of DNA(5) and DNA(6), respectively. Experiments were conducted under ‘single-hit conditions’, cleavage is revealed at the 5′-G of the proximal and distal GG steps in all experimental samples.

The efficiency of the radical cation-induced reactions at GG steps of DNA(1–6) was quantified by phosphorimagery. The data are reported in Table 1 in the form of ratios of strand cleavage observed at the proximal and distal GG steps. For DNA(1), which contains a tandem mispair in the G_antiA_anti configuration, the ratio of proximal to distal reactivity is 11, whereas for DNA(2), which contains a sheared G/A mispair, this ratio is 2.3. The sequence of bases between the two GG steps influences the relative reactivity. However, this is not due to the configuration of the G/A mispair since the same pattern reoccurs in the control compounds that contain only Watson–Crick base pairs. Clearly, the relative reaction efficiency at the GG steps is determined by factors beyond the structure of the G/A tandem mispair. Inspection of the structures of DNA(3,4) and DNA(5,6) shows that this remarkable effect must be due to conversion of N₁/X₁ from A/T to C/G and N₄/X₄ from T/A to G/C.

Irradiation of DNA sequences containing G-bulges

DNA(9) contains a GGG sequence between its proximal and distal GG steps. Molecular orbital calculations indicate that the ionization potential (IP) of GGG is significantly below that of G and GG. Consequently, GGG was expected to act as a ‘deep trap’ for radical cations; that is, one from which the radical cation cannot escape before it is consumed by reaction with water (40–43). Recent experiments have confirmed that the IP of GG and GGG sequences are below that of an ‘isolated’ G, but not so much that either GG or GGG behaves as a deep trap (17,44–46). DNA(9) serves as a control compound for DNA(7), which contains a bulged guanine in a d(G₃/C₂) sequence between the proximal and distal GG steps. The bulged guanine might function as a deep trap for a radical cation because it has a lowered IP (the hopping rate constant, k_h, is slowed), or it may become a trap because the exposed guanine radical cation reacts with water more rapidly than a guanine radical cation paired with a cytosine (k_t is accelerated).

Samples of DNA(7–9) were irradiated and analyzed as described previously: the results are shown in Table 1. These data reveal that a reaction leading to strand cleavage occurs at the GGG sequence for DNA(7–9). This sequence, whether it contains a bulged guanine or not, has little effect on the ratio of reaction at the proximal and distal GG steps (G_p/G_d), which indicates that in no case is it a deep trap. However, strand cleavage is detected in each case examined at guanines G₂ and G₃ of the 3′-G₁G₂G₃-5′ sequence, but there is no detectable cleavage at guanine G₁ except for DNA(8), which contains a bulged mispair.

The bulged guanine sequences might still be hot spots for mutation if the oxidized guanines in these structures were not repaired by enzymes such as Fpg. We assessed the ability of Fpg to recognize and cleave oxidized guanines in the d(G₃/C₂), or d(G₃/A₂) sequence. The results are shown in the form of an autoradiogram in Figure 6 and summarized in Table 1. Treatment of irradiated samples of DNA(7,8,9) with Fpg leads to strand cleavage which parallels that seen when the samples are treated with hot piperidine. Clearly, Fpg is competent to react with the oxidized guanines in these unusual secondary structures.

Figure 6.

PAGE results from the irradiation of DNA(7–9) presented as autoradiograms. The 5′-terminus of the complementary strand is labeled with ³²P. Lanes 1, 3 and 5 are ‘dark controls’. Experimental samples (5.0 µM DNA in 10.0 mM sodium phosphate buffer solution) were irradiated for 15 min at room temperature with 350 nm light and then treated with Fpg (15 ng, 37°C, 1 h). Lanes 2, 4 and 6 are results from irradiation of DNA(7), DNA(8) and DNA(9), respectively. Experiments were conducted under ‘single-hit conditions’, cleavage is revealed at the 5′-G of the proximal and distal GG steps in all experimental samples, reaction and at the central G_n sequence.

DISCUSSION

Effect of base stacking in tandem GA mispairs on radical cation transport

The canonical double helix of duplex DNA imparts an apparent order on its base pairs that has been claimed to facilitate π-stack-mediated electron transfer; a reaction whose distance dependence and rate is thought to be affected by the nuances of base stacking (47). For example, in certain instances the electron transfer reaction shows exquisite sensitivity to base mis-matches, and this property was suggested to be a feature that might be useful for detection of mispairs (48). In this regard, tandem GA mispairs provide well-defined structures with which to assess the affect of stacking on charge transport through DNA.

The tandem GA mispairs contained in a CGAG duplex sequence, as is found in DNA(2), adopt a sheared arrangement. Investigation of sheared tandem GA mispairs by 2D-NMR spectroscopy shows that they maintain an overall B-like structure in which the mispaired bases are over wound whereas the sequences flanking the mispair are under wound (49). This is one distortion among several that leads to cross-strand stacking in the mispair; that is, the G of one strand stacks upon the G of the complementary strand rather than on the A of its own strand. In contrast, for tandem mispairs contained in the sequence AGAT, as is found in DNA(1), the G and A bases of the mispair are fully stacked over one another as they are in B-DNA. Thus, variation of the identity of the base pairs flanking the GA mispair provides a means to manipulate the nature of the base stacking. If stacking strongly affects the efficiency of electron transfer reactions in DNA, then we expect DNA(1) to be similar to normal, fully complementary B-form DNA, with its dominant intrastrand stacking, and we anticipate that DNA(2) with its cross-strand stacking could exhibit anomalous charge transport characteristics. The results reported above reveal that the base sequence adjacent to the GA mispair, not the stacking mode, is the primary feature that controls the efficiency of charge transport.

Irradiation of the DNA oligomers in Series I leads to reaction at both proximal and distal GG steps that are located ∼17 and 61 Å from the site of charge injection (i.e. the AQ). This result demonstrates again that duplex DNA is capable of transporting charge (a radical cation in this case) over considerable distance. This phenomenon has been observed independently in several laboratories and the efficiency of this reaction generally shows an average exponential distance-dependent drop-off coefficient (γ) of –0.02 Å^–1 (4,12,15,50). The data reported in Table 1 reveal that γ for DNA(2,5,6) is –0.02 Å^–1, but that DNA(1,3,4) have γ = –0.06 Å^–1, which corresponds to a more pronounced distance dependence. The common feature distinguishing the structures of DNA(2,5,6) from that of DNA(1,3,4) is that for the former the GA mispair is flanked by guanine-containing base pairs and for the latter the flanking base pairs contain adenines.

The theory of charge transport in DNA is still being developed. It is clear now that long-distance charge transport occurs by a series of relatively short-range hops (4,13). Despite some initial confusion (51), it is now understood that the migrating positive charge can reside on either guanine- or adenine-containing base pairs (52), which is consistent with our earlier results (15,22).

The effect of base sequence on the efficiency of charge transport in duplex DNA is controlled by both thermodynamic factors, namely the difference in oxidation potential between G and A, and kinetic factors that lead to distortions in the DNA structure that occur as a consequence of positive charge localization (15). In the present case, it appears that both factors affect the efficiency of charge transport through the GA mispair variable sequence of DNA(1–6). Clearly, however, base pair stacking does not play a dominant role; consequently, the efficiency of charge transport is not a reliable indicator for the presence of mispairs in duplex DNA.

Effect of G₃/C_n sequences on radical cation transport

Selective reaction of radical cations with water at G_n sequences in DNA has been attributed to thermodynamic factors based on theoretical studies (40–43,53). These calculations, carried out with base pairs held in standard B-form with the sugar diphosphates removed and in the absence of water and counterions, indicate that the vertical IP of a G in a GG step is ∼0.5 eV below that of an isolated G and that the IP of G in a G₃ sequence is ∼0.7 eV below G. Recently, Ratner and coworkers (54) incorporated these estimates into an analysis of charge hopping in DNA that considers two dynamic regimes for relaxation of radical cations in G_n sequences. The results reported above show that the G₃ sequence is not an especially deep trap for the migrating radical cation, a finding that is consistent with recent work from other laboratories (17,44,45). More interestingly, the bulged guanine in the G₃/C₂ sequence of DNA(7) does not significantly affect the efficiency of reaction compared with the fully complementary G₃/C₃ sequence of DNA(9).

Irradiation of the DNA oligomers in Series II leads to reaction at the proximal and distal GG steps in addition to reaction at the intervening G₃/C_n sequence. Calculation of γ from these experiments yields a value of –0.018 Å^–1, which is within experimental error of that usually found for mixed sequence DNA (12,15). Clearly, neither the fully paired G₃/C₃ sequence nor the bulge-containing G₃/C₂ sequence provides a major kinetic or thermodynamic barrier for charge transport to the distal GG step. Surprisingly, reaction of the guanine radical cation at the bulged sequence is perturbed only slightly compared with its reaction at the fully complementary G₃/C₃ sequence. In particular, strand cleavage is observed only at G₃ and G₂ in both cases and the amount of reaction for the bulged sequence is only modestly enhanced. These findings are consistent with the earlier NMR data for the G₃/C₂ sequence that indicates a fluxional structure affecting all base pairs (37).

Mispairs and bulges—DNA damage and repair

It is clear from the findings reported above that loss of an electron from DNA to give a radical cation results in damage to G at locations far removed from the site of oxidation. The damage is located predominantly at G_n sequences with a small amount at the 5′-G of a GA step, but little at the isolated guanines (that not 5′ to a G or A) and, surprisingly, not at the G of tandem GA mispairs either in the G_antiG_anti or sheared configuration. Damage can become a mutation when it is not repaired or when its repair is slowed. The results reported in Table 1 indicate that Fpg, a component of excision repair, operates on damaged guanines when they are contained in G₃C₂ and G₃A₂ sequences. This may be a consequence of the ability of the bulged G in this case to migrate fluxionally (37) and present a base-paired structure to the enzyme.