Direct observation of ultrafast-electron-transfer reactions unravels high effectiveness of reductive DNA damage

Abstract

Both water and electron-transfer reactions play important roles in chemistry, physics, biology, and the environment. Oxidative DNA damage is a well-known mechanism, whereas the relative role of reductive DNA damage is unknown. The prehydrated electron (), a novel species of electrons in water, is a fascinating species due to its fundamental importance in chemistry, biology, and the environment. is an ideal agent to observe reductive DNA damage. Here, we report both the first in situ femtosecond time-resolved laser spectroscopy measurements of ultrafast-electron-transfer (UET) reactions of with various scavengers (KNO₃, isopropanol, and dimethyl sulfoxide) and the first gel electrophoresis measurements of DNA strand breaks induced by and OH^• radicals co-produced by two-UV-photon photolysis of water. We strikingly found that the yield of reductive DNA strand breaks induced by each is twice the yield of oxidative DNA strand breaks induced by each OH^• radical. Our results not only unravel the long-standing mystery about the relative role of radicals in inducing DNA damage under ionizing radiation, but also challenge the conventional notion that oxidative damage is the main pathway for DNA damage. The results also show the potential of femtomedicine as a new transdisciplinary frontier and the broad significance of UET reactions of in many processes in chemistry, physics, biology, and the environment.

Water plays a crucial role in biophysics, biochemistry, cell biology, and Earth’s environment (1–4). In particular, water forms between 70 and 95% of the mass of the cell. It is a simple molecule yet essential to life, shaping biomolecules and controlling their functions. Electron-transfer reactions play an important role in many processes in chemistry, physics, biology, and the environment (3–12). DNA damage is a central mechanism in the pathogenesis and treatment of human diseases, notably cancer (13). Oxidative DNA damage is a well-known pathway, whereas the relative role of reductive DNA damage is unknown (12).

There has been continued interest in studying the solvation dynamics and reactivity of excess electrons in liquid water (14–36). First discovered in the 1960s (14), the hydrated electron () is a well-known species produced by radiolysis of water under ionizing radiation and its reactivity has been well determined (21, 22). Since the advent of femtosecond (fs) (1 fs = 10^-15 s) time-resolved laser spectroscopy (fs-TRLS), the precursor to (henceforth denoted ) has been directly observed (15, 17–20). The high efficiencies of attachment of to some molecules including conventional electron scavengers have been revealed (23–29, 36). In a series of fs-TRLS experiments, we have resolved the long controversies about the lifetimes and physical nature of states (20) and observed -induced bond-breaking reactions of environmental and biological molecules in water via a dissociative ultrafast-electron-transfer (UET) mechanism (4, 12, 20, 30–35). Particularly, we have directly observed chemical bond breaking of nucleotides, most effective at the guanine (G) base, induced by the UET reaction of (12). Similar to the attack by an oxidizing species, such a bond break at G may cause severe structural deformation and potential lesions to the DNA, as theoretically shown by Bera and Schaefer (37). In water, a free electron is rapidly solvated by surrounding H₂O molecules to form a weakly bound at around −1.5 eV that has lifetimes of approximately 500 fs (16–20), about two orders of magnitude higher than the residence times of free electrons and quasi-free electrons in nonpolar media, and UET reactions of with foreign molecules are highly effective, as reviewed recently (3, 4). Hence, the UET reaction of may play a key role in causing damage to aqueous DNA, and is a fascinating species due to its fundamental importance in chemistry, biology, and the environment.

Radiotherapy is still the primary method for curative cancer therapy (38–41). Exposure of living cells to ionizing radiation, such as hard X-rays and γ-rays, leads to biological damage by both direct and indirect interactions with the cell components (mainly water). It is known that indirect damage arising from radicals formed by radiolysis of water surrounding the DNA far exceeds direct damage due to energy deposited at DNA itself (39–41). Indeed, Ito et al. (42) have observed that the yields of γ-ray-induced single-strand breaks (SSBs) and double-strand breaks (DSBs) of DNA are three orders of magnitude higher in aqueous solution than in dry conditions. Hence, water clearly plays the dominant role in radiation-induced DNA damage.

However, the key question is how the radiolysis of water leads to DNA damage. The H₂O radiolysis is known to generate an oxidizing OH^• (H^•) radical and a free electron. The conventional understanding is that the major radicals are OH^• and and their reaction efficiencies with various molecules are well-known (21, 22). However, is ineffective at inducing biological damage; the oxidizing OH^• radical is thought to be the main species for indirect damage to DNA (39–41). But it was also found that cell killing or DNA damage, especially DSBs, cannot be completely suppressed even with high concentrations (up to 2 M) of OH^• scavengers such as DMSO; there is approximately 35% cell killing or DNA damage “nonscavengable” by OH^• scavengers (40, 41). In the 2000s, Sanche and co-workers (43, 44) found that anion resonances formed via dissociative electron attachment (DEA) of secondary free electrons in energy ranges of 3–20 eV and 0–4 eV, respectively, caused SSBs plus DSBs and SSBs only of DNA in dry (vacuum) conditions. In water, however, a free electron rapidly (in the order of fs) becomes an and finally an (15–20, 29–36). Moreover, it has also been observed that a polar molecular environment (e.g., on H₂O/NH₃ ice) led to a nearly complete quenching of DEA resonances of molecules at electron energies above 1.0 eV (45, 46), whereas DEA resonances for electron energies near 0 eV were significantly enhanced (28, 46). The latter was attributed to an UET mechanism: The nearly 0-eV electron instantly becomes an in H₂O/NH₃ ice, which is then transferred to the molecule (3, 9, 28, 46). In fact, if the 1/3 “nonscavengable” damage were attributed to direct damage to DNA by primary particles or secondary free electrons without involving water, then it would not align with the observed enhancements by orders of magnitude of DNA strand breaks induced by ionizing radiation in aqueous solutions (42). Thus, there is a long-standing mystery how radicals produced from water radiolysis precisely induce DNA SSBs and DSBs (39–41). The DSB yield is typically only a few percent of the SSB yield, but unrepaired DSBs are the most lethal form of DNA damage and directly relate to final cell kill (39).

Owing to its being weakly bound at around −1.5 eV, finite lifetimes of approximately 500 fs and readily co-production with OH^• radical from water photolysis, is an ideal species to observe the efficiency of reductive damage versus oxidative damage of DNA. So far, however, no measurement of -induced DNA strand breaks has been reported. Here, we present in situ studies of both real-time fs-TRLS observation of ultrafast scavenging reactions of and resultant DNA strand breaks. As schematically shown in Fig. 1, and OH^• were co-produced via two-photon photolysis of water and respectively scavenged by various scavengers in this well-controlled experiment. Strikingly, we show that is the major species to cause damage to aqueous DNA under ionizing radiation and that reductive damage is much more effective than oxidative damage.

Fig. 1.

Two-photon photolysis of water. Two-photon excitation of water readily generates the main radicals ( and OH^•) of radiolysis of water within 10^-14 s after irradiation at a UV wavelength of 266–380 nm. The attacks of radicals can then cause single-strand and double-strand breaks of the DNA.

Results

Direct Observation of UET Reactions of with Various Scavengers.

In analogy to the radiolysis of water using X-ray or γ-ray, a finely controlled laser photolysis experiment can readily generate the radicals (, , OH^•, and H^•) through two-UV-photon excitation of a H₂O molecule at wavelengths of 266–380 nm, equivalent to single-photon radiation at 9.3–6.5 eV (15, 17–20, 47–49). The laser excitation can lead to ionization of H₂O:

[1]

followed by

[2]

[3]

The excited water molecule can also dissociate:

[4]

The advantage of photolysis over radiolysis using X-ray or γ-ray is that the yields of the major radicals ( and OH^•) can be well determined and their reactions can be monitored in real time by in situ fs-TRLS measurements (12, 20, 30–35). The relative yield ratio r = [OH^•]/[e^-] generally increases with decreasing photon excitation energy: At high energy (similar to the case in radiolysis), ionization (formula 1) is the dominant process, whereas dissociation (formula 4) becomes dominant at low energy. It was observed that r = 1.1 ± 0.3 at 200 nm (48), 1.9 (47) or 1.7 ± 0.9 (48) at 266 nm, and 3.3 ± 1.0 or > 8 at 299 nm (48). In particular, UV irradiation at ≥300 nm can avoid not only the production of conduction-band or free electrons (49), leading to a neat production of the above radicals, but the DNA damage from direct photoexcitation of DNA (absorption by DNA at ≥300 nm is negligible). Furthermore, to minimize the absorption by scavengers, we used an excitation wavelength of 330 nm, at which all isopropanol, DMSO, and KNO₃ have no or negligible absorption (as determined from detailed absorption measurements) and hence nearly all the energy is absorbed by water.

Our fs-TRLS measurements, as described previously (12, 20, 30–35) and in Methods, were performed to directly observe the UET reactions between and various electron and OH^• scavengers: . As previously demonstrated in detail (20), the scavenging reaction can be directly observed by monitoring the change in decay kinetics of observable at near infrared wavelengths with the introduction of a scavenger or by alternatively monitoring the formation kinetics of observable at visible wavelengths centered at 720 nm in the first picoseconds after the excitation of H₂O. This is because the decay kinetics of corresponds to the formation (rising) kinetics of (15, 17–20, 29). Because the intrinsic lifetime of is within 1.0 ps, the yield of surviving from the scavenging is given by the initial yield of at approximately 1.0 ps. For readiness of our experiments, we measured the formation kinetics to observe the UET reactions of with KNO₃, isopropanol and DMSO, respectively.

As shown in Fig. 2A, the initial yield of at approximately 1.0 ps decreases with increasing concentrations of KNO₃, showing the reaction of the nitrate ion with . The scavenging reaction can be expressed as

[5]

This scavenging reaction competes with the hydration of dynamically and can be approximated as the first-order kinetic reaction. Then, it can be deduced that the reciprocal of the surviving percentage Y of the yield corresponding to the initial yield at approximately 1.0 ps has an approximate linear relationship with scavenger concentration [S]:

[6]

where τ_pre is the lifetime of at [S] = 0 and k_pre is the reaction rate constant of the scavenger with . The measured 1/Y versus KNO₃ concentration is shown in Fig. 2B; the slope of the linear line is thus the product of the reaction rate constant k_pre and the lifetime of . Recently, we have determined τ_pre to be 180 ± 30 and 545 ± 30 fs, after a spike effect had been identified and removed from pump–probe kinetic trace measurements (20). These values are consistent with the theoretical prediction by Rossky and Schnitker (16) and measurements by the groups of Eisenthal (17) and Laubereau (18). Taking τ_pre = 540 fs, the reaction rate constant of with KNO₃ is determined to be k_pre = (1.2 ± 0.5) × 10¹³ M^-1 s^-1. This value is three orders of magnitude larger than the reaction rate constant of with (22), which confirms the high reactivity of . On the other hand, it should also be noted that the scavenging reaction of with KNO₃ will lead to the formation of the OH radical, as seen from formula 5. This is similar to the reaction of the well-known electron scavenger N₂O with , which also leads to the formation of the OH^• radical (41, 50). Thus, special caution must be paid to explain the results about DNA damage affected by these molecules used as electron scavengers.

Fig. 2.

Real-time observation of UET reactions of with various scavengers. Femtosecond transient absorption kinetic traces of with the presence of KNO₃/isopropanol/DMSO. (A, C, and E) Data were corrected after eliminating the coherence spike at time zero (20). (B, D, and F) Plot of the reciprocal of the surviving yield Y of the versus scavenger concentration, where Y is determined by the yield at 1.0 ps and normalized to the value without the scavenger.

Similar measurements were made for reactions of with two well-known OH^• scavengers, isopropanol and DMSO, in buffer solutions. Their slight difference is that DMSO is a well-known scavenger of the OH^• radical, whereas isopropanol can effectively scavenge both OH^• and H^• radicals (50). The results for UET reactions of with isopropanol and DMSO are shown in Fig. 2 C–F. Strikingly, we observe that these often-used OH^• scavengers also show strong reactions with . At a high scavenger concentration of 2 M, only 75–80% and 50–54% of survive from the reactions with isopropanol and DMSO, respectively. The reaction rate constants determined from Eq. 6 are k_pre = (2.3 ± 0.5) × 10¹¹ and (8.1 ± 0.5) × 10¹¹ M^-1 s^-1 for the reactions with isopropanol and DMSO, respectively. These k_pre values are about two orders of magnitude higher than the reaction rate constants k_OH for the reactions of OH^• with isopropanol and DMSO, which were well determined to be k_OH = 2 × 10⁹ and 7 × 10⁹ M^-1 s^-1, respectively (22). Thus, the present results clearly demonstrate that these molecules are not only effective OH^• scavengers but highly effective scavengers.

DNA Strand Break Measurements.

Using exactly the same setup for the above fs-TRLS measurements of scavenging reactions, we irradiated DNA samples, which include DNA in buffer solutions with/without the presence of 2 M KNO₃/isopropanol/DMSO. We found that a PBS buffer (20 mM phosphate buffer, 0.15 M NaCl, pH 7.40) was required to stabilize DNA and to minimize the effects of scavenging molecules on DNA. Prior to irradiation, the samples were bubbled in N₂ for 20–30 min to remove oxygen. As described in Methods and as shown in Fig. S1, the DNA damage was analyzed by agarose gel electrophoresis and characterized as the undamaged supercoiled (SC) conformation, circular relaxed (C) conformation caused by SSBs, and the linear (L) conformation induced by DSBs (42, 51). Gel densitograms for the four different DNA solutions with various irradiation times are presented in Fig. 3 A–D. The percentages of SSBs and DSBs as a function of irradiation time (dose), determined from the fractions of detected SC, C, and L form of the DNA, are shown in Fig. 4 A and B. The control DNA sample (without irradiation) corresponded to approximately 95% undamaged (SC) DNA, ≤ 4.5% SSBs, and < 0.5% DSBs, which were typically caused by the DNA extraction and purification process.

Fig. 3.

Strand-break measurements of irradiated DNA. Agarose gel densitograms for irradiated DNA in buffer solutions with/without 2M KNO₃/DMSO/isopropanol with various irradiation times, where the bands for SC, open-C, and L forms of DNA are visible. The baseline (background) has been subtracted. The band intensities for different samples were normalized to that of the control sample without irradiation.

Fig. 4.

Percentages of SSBs and DSBs in irradiated DNA. These quantitative data were obtained from the gel densitograms shown in Fig. 3, by integrating the peak areas.

At first glance, the results in Figs. 3 and 4 show that the introduction of OH^• scavengers (isopropanol and DMSO) leads to a more significant reduction in SSBs and DSBs, compared with the introduction of the scavenger (KNO₃). Indeed, only 36 ± 4% of the SSBs and approximately 43 ± 4% and approximately 27 ± 3% of the DSBs survive from the presence of isopropanol and DMSO, respectively. These results are generally consistent with the results reviewed in the literature (40, 41): About 1/3 DNA damage or cell killing is nonscavengable by OH^• scavengers. Those previous studies, however, assumed that OH^• scavengers would not kill but OH^• radical solely. Based on our in situ, direct fs-TRLS measurements (Fig. 2 C–F), one can now see that only 75–80% and 50–54% of could survive at 2 M isopropanol and DMSO, respectively. Given the well-known reaction rate constants of OH^• and H^• with isopropanol (22), one can get that is the only survival radical that could cause the DNA damage with the presence of a high concentration (2 M) of isopropanol (50). Thus, the measured survival percentages of SSBs and DSBs (36 ± 4% and 43 ± 4%, respectively) show reliable evidence that accounts for approximately 45 ± 4% of the SSBs and 54 ± 4% of the DSBs produced by the two-photon excitation of water at 330 nm. That is, only approximately 50% of the DNA damage is due to the OH radical.

The data in Fig. 4 also show that the fraction of SSBs scavenged by DMSO is close to that by isopropanol, whereas the fraction of survival DSBs from DMSO is considerably smaller than that from isopropanol (27% vs. 43%). With the observed fact that about half of the yield is scavenged by DMSO (Fig. 2 E and F), this also indicates that about 54 ± 3% of DNA DSBs is caused by with the absence of DMSO. This result is in excellent agreement with that for the scavenging by isopropanol. The nearly identical survival percentage of SSBs for both scavengers indicates that H^• radical also contributes partially to SSBs but no DSBs. For the presence of DMSO, the survival H^• radical compensates the larger loss of in inducing SSBs.

It is also interesting to note that 2 M KNO₃, which scavenges solely, reduces the percentages of survival SSBs and DSBs to 72 ± 5% and 70 ± 5%, respectively. The presence of 2 M KNO₃ leads to a nearly complete killing of (Fig. 2). This scavenging reaction, however, also results in an increase of the OH^• yield (formula 5). From previous experimental observations (47, 48), one can make a fairly reliable estimate that the yield ratio of is ≥2∶1 with the two-photon excitation wavelength at 330 nm. Taking the yield ratio , the reaction causing a complete scavenging of by KNO₃ can lead to an increase of the OH^• yield by 50% (see formula 5), i.e., the OH^• yield can be enhanced by a factor of 1.5. Thus, the OH^• radical produced by two-photon excitation of water accounts for 48 ± 5% of SSBs and 47 ± 5% of DSBs of the DNA when there is no scavenger. That is, each species of and OH^• contributes to approximately 50% of the SSBs and DSBs of DNA under the yield ratio produced by the two-photon photolysis of water at 330 nm.

From the results described above, one can see that three independent experiments with different scavengers (isopropanol, DMSO, and KNO₃) lead to a consistent observation. These results lead to a surprising conclusion that on the basis of the SSB and DSB yields per radical, nearly doubles the effectiveness of causing DNA damage, compared with the OH^• radical.

Discussion

Oxidative DNA damage is a relatively well-known process both in mutagenesis during oncogenesis and cancer cell killing during radiotherapy, and the conventional notion in radiobiology is that damage to the genome by ionizing radiation is mainly induced by the oxidizing OH^• radical (39–41). However, little was known about reductive DNA damage. Surprisingly, here we found that reductive DNA damage induced per are doubly effective than oxidative DNA damage induced per OH^• radical.

For conventional ionizing radiation using X-rays and γ-rays in radiotherapy (more discussion in SI Text), it is well-known that unlike the UV photolysis of water, the ionization of the excited water becomes the dominant process; that is, the dissociation of water becomes a negligible pathway. This gives rise to the yield ratio [OH^•]∶[e^-] ≈ 1∶1, which can be inferred from the reaction (formulas 1–3). Note that the quantum yield (the G value) of the OH^• radical at 10^-6 s after irradiation is 2.4–2.8, nearly identical to that of (22, 40, 41). The latter should be about the half of the initial yield of (4, 18, 22). Nevertheless, the initial yield of the OH^• radical was estimated to be about 4.8 at 10^-12 s after irradiation (22), which also doubles its yield at 10^-6 s. Like the contrast of the ultrashort-lived highly reactive to the long-lived, inert , the “hot” OH^• radical at the 10^-14–10^-12 s scale should be much more reactive and plays a more significant role in DNA damage than the “cold” OH radical detected on the microsecond or sub-micro-second scale. Again, this also indicates that the yield ratio of produced in conventional ionizing irradiation should approximately be 1∶1. The present observation therefore indicates that in conventional ionizing irradiation, the yield of DNA strand breaks induced by is nearly twice that induced by the OH radical. That is, only about one-third of the DNA damage is caused by the OH^• radical, whereas about two-thirds of the damage results from attacks by . Based on direct observation of the UET reactions shown in Fig. 2, the DNA damage induced by not only all OH^• but approximately 50% was killed by high concentrations of OH^• scavengers like DMSO. That is, only approximately 50% of the -induced DNA damage survived from OH^• scavengers. This understanding can now directly unravel the long-standing mystery why about 1/3 of DNA damage or cell killing is nonscavengable by those OH^• scavengers.

Conclusion

In summary, we report the direct observation of highly effective UET reactions of with both the scavenger and the so-called OH^• scavengers. We show that the unique with lifetimes of approximately 5 × 10^-13 s and energies at approximately -1.5 eV is the major species inducing SSBs and DSBs of aqueous DNA under ionizing radiation. Moreover, we found that on the basis of the SSB and DSB yields per radical, the effectiveness of reductive DNA damage induced by is nearly double that of oxidative DNA damage by OH^• radical. These results not only challenge the conventional belief that damage to the genome by ionizing radiation is mainly induced by the oxidizing OH^• radical, but more importantly, provide a fundamental mechanistic understanding of reductive DNA damage on the molecular level.

Our finding of highly effective DNA strand breaks induced by has great significance. (i) The knowledge of how individual radicals (both and OH^•) interact with cells, their DNA, and other molecular constituents is required to evaluate the health effects of humans exposed to ionizing radiation. (ii) Our findings can be applied to develop better strategies for more effective radiotherapies of diseases such as cancer. The dissociative UET reaction pathway with can be strategically utilized to design previously undescribed drugs that can effectively cause DNA damage and tumor cell kill. (iii) Because there are numerous sources of weakly bound electrons in biological systems, the direct observation of reductive DNA damage by the weakly bound has particular significance. If the DNA strand breaks, especially DSBs, are not repaired quickly and efficiently, genetic mutations can arise, causing serious diseases such as cancer. The oxidative damage at the G base and its relation to human diseases are well-known. Our present finding of the high effectiveness of -induced DNA strand breaks indicates that reductive DNA damage could also play a vital role in human diseases. Our results may therefore have far-reaching significance for a deeper understanding of DNA damage and repair processes in biological systems, which is crucial to the development of effective therapies for various diseases such as cancer and stroke. (iv) These studies show promise of the emerging transdisciplinary frontier, femtomedicine (4, 35), to provide an unprecedented understanding of important biological processes with close relevance to diseases and their treatments. (v) The UET reactions of have also great implications beyond biology and medicine, e.g., for breakups of environmentally important halogenated molecules in aqueous solutions or on water ice, which plays an important role in forming the polar ozone hole (3). Thus, our results have broad implications for understanding of the role of water in electron-initiated reactions in many chemical, biological, and environmental systems (3, 4, 9–12).

Methods

Materials.

Ultrapure water for life science with a resistivity of > 18.2 MΩ/cm and TOC < 1 ppm obtained freshly from a Barnstead Nanopure water system was used. Chemicals (isopropanol, dimethylsulfoxide, and KNO₃) and buffer compositions were obtained from Sigma-Aldrich. Plasmid DNA [pGEM 3Zf(-), 3197 kbp] was extracted from Escherichia coli JM 109 and purified using QIAprep Kit (Qiagen). An ultrapure N₂ gas was used to purge the DNA solutions.

Femtosecond Time-Resolved Laser Spectroscopic Measurements.

The standard methodology for fs pump–probe transient absorption measurements of UET reactions of with scavenging molecules (KNO₃, dimethyl sulfoxide, and isopropanol) in PBS buffer solutions (20 mM phosphate buffer, 0.15 M NaCl, pH 7.40) has been described previously (12, 20, 30–35).

Agarose Gel Electrophoresis.

The pump beam at 330 nm, identical to that for fs-TRLS measurements, was used to produce radicals via two-photon excitation of water in DNA solutions. The laser beam was focused into a quartz cell containing 3.0 μg of DNA in 200 μL of buffer solutions with/without the presence of 2 M KNO₃/isopropanol/DMSO. The solutions were stirred during irradiation to produce uniform DNA damage throughout the samples. Aliquots equivalent to 96 ng DNA were removed from the cell in the irradiation times of 0, 15, 30, 60, 90, 120, and 150 min for a total of seven samples for gel electrophoresis. All aliquots were analyzed with a standard agarose gel electrophoresis method, namely, on 1% neutral TAE agarose gel in TAE running buffer. The gel was prestained with 0.5 μg/mL ethidium bromide. The image of the gel was taken on a FluorChem imaging station (Alpha Innotech) and exhibits various DNA topological forms, including closed-circular SC (undamaged DNA), open C (SSB), and L form (DSB). They were quantified with an AlphaEase FC software. To account for the lower binding constant of EtBr to supercoiled DNA compared to linear and open-circular forms, a correction factor of 1.4 was used for the supercoiled DNA form (51).