**8.2. Transient anions in DNA bound to platinum chemotherapeutic agents**

tization properties of halogen compounds are more complicated than previously anticipated [168]. Within the 0–7 eV energy range, resonant electron scattering mechanisms with halour‐ acils lead to more complex molecular fragmentation than that occurs with thymine, which produces a different range of anionic and neutral radical fragments. When formed within DNA, such fragments could react with local subunits, and thus lead to lethal clustered damage, further to that already occurring in unsensitized DNA. The most striking evidence of a huge enhancement of LEE damage obtained upon Br substitution in thymine is seen in the early results of Klyachko et al. [160], who found that, in the presence of water, DEA to bromouracil could be enhanced by orders of magnitude compared to the dry compound. Differences between wet and dry TMA states of halogenated pyrimidines have recently been investigated by Cheng et al. [172]. They applied Koopman's theorem in the framework of long‐range corrected density functional theory for calculation of the TMA states and self‐consistent reaction field methods in a polarized continuum to account for the solvent. Their results indicate that the TMAs of these molecules are more stable in water, but to differing degrees.

The radiosensitization properties of halouracils depend not only on hydrated electrons, but also on LEEs and on DEA. However, the high propensity of LEEs of very low energies (i.e., <1 eV) to fragment bromouracil and deoxybromouridine (BrUdR) may, according to the theory, exist only in single‐stranded DNA [165]. This important prediction was confirmed by Cecchini et al. [173] for the case of solvated electrons and was commented upon by Sevilla [174]. Solutions of single‐ and double‐stranded oligonucleotides, and of double‐stranded oligos containing mismatched bubble regions, were irradiated with *γ*‐rays, and the concentrations of various reactive species produced, including solvated electrons, were controlled with scav‐ engers. When in the absence of oxygen, OH radicals were scavenged, BrUdR was shown to sensitize single‐stranded DNA, but could not sensitize complementary double‐stranded DNA. However, when BrUdR was incorporated in one strand within a mismatch bubble, the nonbase‐paired nucleotides adjacent to the BrUdR, as well as several unpaired sites on the opposite unsubstituted strand, were highly sensitive to *γ*‐irradiation. Since LEEs and solvated electrons fragment BrUdR by the same DEA mechanism [162–165, 168], these results imply that the strong sensitizing action of BrUdR to electron‐induced damage is limited to single‐ stranded DNA, which can be found in transcription bubbles, replication forks, DNA bulges, and the loop region of telomeres. These results are clinically relevant since they suggest that BrUdR sensitization should be greatest for rapidly proliferating cells [173, 174]. When injected into a patient being treated for cancer, BrUdR quickly replaces a portion of the thymidine in the DNA of the fast‐growing malignant cells, but radiosensitization occurs only when DNA is in a single‐stranded configuration (e.g., at the replication forks during irradiation). From this conclusion, it appears advantageous to administer to patients receiving BrUdR, another approved drug, such as hydroxyurea, to increase the duration of the S‐phase of cancer cells (i.e., the replication cycle). This addition would increase the probability that SEs would interact with bromouracil while bound to DNA in its single‐strand form. Such a modality provides an example of how our understanding the mechanisms of LEE‐induced damage can help to

improve radiotherapy [174].

202 Radiation Effects in Materials

Considering that it can often take years, if not decades, before potential new radiosensitizers arrive in the clinic, Zheng et al. [175] hypothesized that present clinical protocols involving high‐energy radiation and platinum (Pt) chemotherapeutic agents could be improved by considering the fundamental principles of energy disposition, including the results of LEE experiments. Their initial goal was to explain the superadditive effect occurring in tumor treatments, when cisplatin and radiation were administered in concomitance [176, 177]. Zheng et al. [175] found that, with cisplatin bound to DNA as in the cancer cells, damage to the molecule increases by factors varying from 1.3 for high‐energy electrons to 4.4 at 10 eV. Considering the much higher enhancement factor (EFs) at 10 eV, the increase in bond disso‐ ciation was interpreted as being triggered by an increase in DNA damage induced by LEEs.

In their experiments, Zheng et al. [175] deposited lyophilized films of pure plasmid and plasmid–cisplatin complexes on a clean tantalum foil. The films were bombarded under UHV with electrons of 1–60 keV. Under these conditions, 90% of the plasmid–cisplatin complexes consisted of a cisplatin molecule chemically bound to DNA, preferentially at the N7 atom of two guanines producing an interstrand adduct. The films had the necessary thickness to absorb most of the energy of the electrons. The different forms of DNA corresponding to SSBs and DSBs were separated by gel electrophoresis, and the percentage of each form quantified by fluorescence. Exposure response curves were obtained for several incident electron energies for cisplatin bound or not to plasmid DNA. **Table 1** gives the results for exposure to 1, 10, 100, and 60,000 eV electrons of films of pure DNA and cisplatin/plasmid complexes with a ratio (*R*) of 2:1 and 8:1. For both *R* values, cisplatin binding to DNA increases the production of SSBs and DSBs, but in quite different proportions depending on electron energy. Considering that it takes about 5 eV to produce a DSB with electrons [90], the most striking result of **Table 1** is clearly the production of DSBs by 1 eV electrons. Later, Rezaee et al. [178] demonstrated that even 0.5 eV electrons could induce DSBs in DNA containing Pt adducts in similar proportions and more efficiently than other types of radiation, including X‐rays and high‐energy electrons. The formation of DSBs by 0.5 eV electrons resulted from a single‐hit process. Gamma radiolysis experiments with plasmid DNA dissolved in water, further demonstrated that even solvated electrons could react with cisplatin–DNA complexes to induce DSBs [179]. The results of Zheng et al. [175] at higher energy were later confirmed by those of Rezaee et al. [180], who showed that increased damage via the formation of TMA could explain, at least partially, the concom‐ itance effect in chemoradiation therapy for cisplatin, as well as for the other platinated chemotherapeutic drugs such as oxaliplatin and carboplatin.

This type of radiosensitization was investigated in more detail by irradiating with a *γ* source the oligonucleotide TTTTTGTTGTTT with or without cisplatin bound to the guanines [181]. Using scavengers and by eliminating oxygen, the oligonucleotide was shown to react with hydrated electrons. Prior to irradiation, the structure of the initial cisplatin adduct was identified by mass spectrometry as G‐cisplatin‐G. Radiation damage to DNA bases was quantified by HPLC, after enzymatic digestion of the TTTTTGTGTTT–cisplatin complex to deoxyribonucleosides. Platinum adducts were following digestion and separation by HPLC, quantified by mass spectrometry. The results demonstrated that hydrated electrons induce damage to thymines as well as detachment of the cisplatin moiety from both guanines in the oligonucleotide. The amount of free cisplatin (i.e., the cleavage of two Pt–G bonds) was found to be much larger than that of the products resulting from the cleavage of a single bond [181,182].


ND, Not detected.

The errors represent the deviation of three identical measurements.

**Table 1.** Yields (in 10‐15 electron‐1 molecule‐1) for the formation of SSB and DSB induced by 1, 10, and 100 eV electron impact on 5 ML DNA films and 60 keV electron impact on 2900 nm DNA films deposited on a tantalum substrate.

These results suggest two major pathways by which hydrated electrons interact destructively with TTTTTGTGTTT–cisplatin [181, 182]. First, the hydrated electron is captured initially on a thymine base and is transferred to the guanine site by base to base electron hopping, where DEA detaches the cisplatin moiety from the oligonucleotide. Alternatively, the hydrated electron interacts directly with the platinum–guanine adduct, and cisplatin is detached via DEA. These hypotheses were consistent with those proposed by Rezaee et al. [178] for LEE‐ induced damage to plasmid DNA. Additionally, Rezaee et al. suggested that in the double‐ stranded configuration, the cisplatin molecule weakens many of the DNA chemical bonds and changes the topology of the molecule; these modifications render DNA much more sensitive to damage over large distances [180]. Of course, under high‐energy irradiation conditions, the increase in ionization cross section, due to the presence of the Pt atom, also increases the quantity of LEEs near cisplatin and therefore may indirectly contribute to the increase in damage.

More recently, the energy dependence of conformational damage induced to pure plasmid DNA [183] and cisplatin–plasmid DNA complexes [184] was investigated in the range 2–20 eV. In addition to the strong resonances (i.e., TMAs) in pure DNA around 5 and 10 eV, further TMA specific to cisplatin‐modified DNA were observed in the yield function of SSBs at 13.6 and 17.6 eV. Moreover, the presence of cisplatin lowered the threshold energy for the formation of DSBs to 1.6 eV, considerably below that observed with electrons in pure DNA films. In all cases, the measured yields were larger than those measured with nonmodified DNA. To reconcile all existing results starting from those obtained with hydrated electrons to those generated up to 20 eV, Bao et al. [184] suggested a single mechanism that could apply to shape and core‐excited resonances, depending or not if electronic excitation of the Pt or guanines was involved in TMA formation. This mechanism, previously proposed for shape resonances by Rezaee et al. [178], can be explained with reference to **Figure 6**. When the TMA is formed on the Pt adduct, the extra electron is delocalized and occupies simultaneously, with identical wave functions, the two bonds linking the Pt atom to guanine bases on opposite strands. Occupancy of the dissociative *σ*\* orbitals induces equal repulsive impulses on the two bonds between platinum and guanines (Pt–G), due to the symmetrical delocalization of the excess electron. If the extra electron autodetaches when the gained kinetic energy is larger than the energy barrier to dissociate the Pt–G bonds, both bonds can be simultaneously broken. The extra energy for dissociation is supplied to the complex by autodetachment from the *σ*\* bond, leaving the additional electron stabilized at the bottom of the potential well of the Pt. The simultaneous cleavage of two Pt–G bonds and formation of two guanine radicals are followed

damage to thymines as well as detachment of the cisplatin moiety from both guanines in the oligonucleotide. The amount of free cisplatin (i.e., the cleavage of two Pt–G bonds) was found to be much larger than that of the products resulting from the cleavage of a single bond

**Energy (eV)** 1 10 100 60,000 1 10 100 60,000 **Thickness** 5 ML 2900 nm 5 ML 2900 nm **DNA** 27± 3 33 ± 3 57 ± 5.5 1.2 ±0.1 ND 10 ± 1 13 ± 2 0.4 ± 0.2 **Cisplatin:DNA = 2:1** 38 ± 3 120 ± 11 150 ± 15 2.4 ±0.3 5 ±1 17 ± 1 36 ± 4 0.5 ± 0.2 **Cisplatin:DNA = 8:1** 52 ± 5 143 ± 14 199 ± 18 3.0 ±0.4 5 ±2 29 ± 2 44 ± 4 0.7 ± 0.1

**Table 1.** Yields (in 10‐15 electron‐1 molecule‐1) for the formation of SSB and DSB induced by 1, 10, and 100 eV electron impact on 5 ML DNA films and 60 keV electron impact on 2900 nm DNA films deposited on a tantalum substrate.

These results suggest two major pathways by which hydrated electrons interact destructively with TTTTTGTGTTT–cisplatin [181, 182]. First, the hydrated electron is captured initially on a thymine base and is transferred to the guanine site by base to base electron hopping, where DEA detaches the cisplatin moiety from the oligonucleotide. Alternatively, the hydrated electron interacts directly with the platinum–guanine adduct, and cisplatin is detached via DEA. These hypotheses were consistent with those proposed by Rezaee et al. [178] for LEE‐ induced damage to plasmid DNA. Additionally, Rezaee et al. suggested that in the double‐ stranded configuration, the cisplatin molecule weakens many of the DNA chemical bonds and changes the topology of the molecule; these modifications render DNA much more sensitive to damage over large distances [180]. Of course, under high‐energy irradiation conditions, the increase in ionization cross section, due to the presence of the Pt atom, also increases the quantity of LEEs near cisplatin and therefore may indirectly contribute to the increase in

More recently, the energy dependence of conformational damage induced to pure plasmid DNA [183] and cisplatin–plasmid DNA complexes [184] was investigated in the range 2–20 eV. In addition to the strong resonances (i.e., TMAs) in pure DNA around 5 and 10 eV, further TMA specific to cisplatin‐modified DNA were observed in the yield function of SSBs at 13.6 and 17.6 eV. Moreover, the presence of cisplatin lowered the threshold energy for the formation of DSBs to 1.6 eV, considerably below that observed with electrons in pure DNA films. In all cases, the measured yields were larger than those measured with nonmodified DNA. To reconcile all existing results starting from those obtained with hydrated electrons to those generated up to 20 eV, Bao et al. [184] suggested a single mechanism that could apply to shape and core‐excited resonances, depending or not if electronic excitation of the Pt or guanines was involved in TMA formation. This mechanism, previously proposed for shape resonances by

**Form of damage SSB DSB**

The errors represent the deviation of three identical measurements.

[181,182].

204 Radiation Effects in Materials

ND, Not detected.

damage.

**Figure 6.** Possible mechanism for the formation of a DSB by a single electron, when cisplatin links two guanine (G) bases on opposite strands. (a) Electron capture into two identical dissociative orbitals between Pt and two Gs. (b) The transient anion thus formed dissociates, leaving the electron on the (NH3)2Pt moiety and causing simultaneous cleav‐ age of the two symmetrical Pt–G bonds. The resulting two guanine radicals (G●) abstract hydrogen from the back‐ bones, causing cleavage of phosphodiester bonds on opposite strands. (c) Resulting DSB [178].

by hydrogen abstraction from the backbone. This abstraction cleaves the phosphodiester bonds in opposite strands, forming a DSB. Considering the results obtained with carboplatin and oxaliplatin [180], which are similar to those obtained with cisplatin, the mechanism depicted in the diagram of **Figure 6** is likely to apply also to these chemotherapeutic drugs. Since these latter behave as cisplatin and bind similarly to DNA, we can replace cisplatin by carboplatin in **Figure 6**; to represent oxaliplatin in the figure, NH3 has to be replaced by C6H10(NH2)2.

The LEE enhancement mechanism of damage in DNA–Pt drug complexes acts on a femtosec‐ ond timescale, which quite unlike other biological mechanisms of radiosensitization, act over macroscopic times that can range from hours to days. These considerations imply that the mechanism (e.g., physicochemical vs biological) of radiosensitization by Pt agents in concom‐ itant chemoradiation therapy may be sensitive to the timing between the injection of the drug to the patient and the irradiation. Thus, if TMA formation in DNA plays a major role in radiosentization by Pt drugs, maximal cancer cell killing should be achieved, if these cells are irradiated when the maximum amount of Pt is bound to their nuclear DNA.

Led by this hypothesis, Tippayamontri et al. [185, 186] determined the optimal conditions for concomitant chemoradiation treatment of colorectal cancer with cisplatin, oxaliplatin, and their liposomal formulations Lipoplatin and Lipoxal [187, 188]. Using an animal model of human colorectal cancer, they determined the time window for maximum radiosensitization and synergy with irradiation, by studying the pharmacokinetics and time‐dependent intra‐ cellular distribution of the Pt drugs. This, in turn, is determined by the reaction kinetics of the drug with DNA and the DNA repair kinetics.

In nude mice bearing HCT116 colorectal carcinoma, treated with the Pt drugs, they measured by inductively coupled plasma mass spectrometry, the platinum accumulation in blood, serum, different normal tissues, tumor, and different tumor cell compartments, including the amount of Pt bound to nuclear DNA [185, 186] **Figure 7a** indicates the positions of binding of cispelatin to DNA. Examples of the amount of cisplatin and Lipoplatin binding to the DNA of HCT116 colorectal cancer cells in mice are shown in **Figure 7b** as a function of time after injection of the drug. Radiation treatment (15 Gy) was given 4, 24, and 48 h after drug admin‐ istration. The resulting tumor growth delay was reported and correlated with apoptosis analyses. Optimal survival of the mice and highest apoptosis were observed when radiation was given at 4 or 48 h after drug injection. These times corresponded to the times of maximal platinum binding to tumor DNA, as shown in **Figure 7b** for cisplatin and Lipoplatin. When tumor irradiation was performed at 48 h, the ratio of tumor growth delay for the group having the combined treatment compared to delay for the group treated with chemotherapy alone varied from 4.09 to 13.00, depending on the drug. The most efficient combination treatment was observed when the amount of Pt drug binding to DNA was highest, as predicted from fundamental considerations [178–182]. Such results testify our fundamental understanding of the mechanisms of platinum‐induced radiosensitization and should have significant impact on the design of more efficient treatment protocols.

Transient Anions in Radiobiology and Radiotherapy: From Gaseous Biomolecules to Condensed Organic and Biomolecular Solids http://dx.doi.org/10.5772/63293 207

by hydrogen abstraction from the backbone. This abstraction cleaves the phosphodiester bonds in opposite strands, forming a DSB. Considering the results obtained with carboplatin and oxaliplatin [180], which are similar to those obtained with cisplatin, the mechanism depicted in the diagram of **Figure 6** is likely to apply also to these chemotherapeutic drugs. Since these latter behave as cisplatin and bind similarly to DNA, we can replace cisplatin by carboplatin in **Figure 6**; to represent oxaliplatin in the figure, NH3 has to be replaced by C6H10(NH2)2.

The LEE enhancement mechanism of damage in DNA–Pt drug complexes acts on a femtosec‐ ond timescale, which quite unlike other biological mechanisms of radiosensitization, act over macroscopic times that can range from hours to days. These considerations imply that the mechanism (e.g., physicochemical vs biological) of radiosensitization by Pt agents in concom‐ itant chemoradiation therapy may be sensitive to the timing between the injection of the drug to the patient and the irradiation. Thus, if TMA formation in DNA plays a major role in radiosentization by Pt drugs, maximal cancer cell killing should be achieved, if these cells are

Led by this hypothesis, Tippayamontri et al. [185, 186] determined the optimal conditions for concomitant chemoradiation treatment of colorectal cancer with cisplatin, oxaliplatin, and their liposomal formulations Lipoplatin and Lipoxal [187, 188]. Using an animal model of human colorectal cancer, they determined the time window for maximum radiosensitization and synergy with irradiation, by studying the pharmacokinetics and time‐dependent intra‐ cellular distribution of the Pt drugs. This, in turn, is determined by the reaction kinetics of the

In nude mice bearing HCT116 colorectal carcinoma, treated with the Pt drugs, they measured by inductively coupled plasma mass spectrometry, the platinum accumulation in blood, serum, different normal tissues, tumor, and different tumor cell compartments, including the amount of Pt bound to nuclear DNA [185, 186] **Figure 7a** indicates the positions of binding of cispelatin to DNA. Examples of the amount of cisplatin and Lipoplatin binding to the DNA of HCT116 colorectal cancer cells in mice are shown in **Figure 7b** as a function of time after injection of the drug. Radiation treatment (15 Gy) was given 4, 24, and 48 h after drug admin‐ istration. The resulting tumor growth delay was reported and correlated with apoptosis analyses. Optimal survival of the mice and highest apoptosis were observed when radiation was given at 4 or 48 h after drug injection. These times corresponded to the times of maximal platinum binding to tumor DNA, as shown in **Figure 7b** for cisplatin and Lipoplatin. When tumor irradiation was performed at 48 h, the ratio of tumor growth delay for the group having the combined treatment compared to delay for the group treated with chemotherapy alone varied from 4.09 to 13.00, depending on the drug. The most efficient combination treatment was observed when the amount of Pt drug binding to DNA was highest, as predicted from fundamental considerations [178–182]. Such results testify our fundamental understanding of the mechanisms of platinum‐induced radiosensitization and should have significant impact

irradiated when the maximum amount of Pt is bound to their nuclear DNA.

drug with DNA and the DNA repair kinetics.

206 Radiation Effects in Materials

on the design of more efficient treatment protocols.

**Figure 7.** (a) Diverse sites of intrastrand and interstrand binding of cisplatin to cellular DNA. (b) Concentration of Pt– DNA adducts in the nucleus of human colorectal cancer cells of mice bearing HCT116 xenografts, as a function of time after administration of cisplatin and LipoplatinTM. The mice were irradiated at 4, 24, and 48 h after injection of the che‐ motherapeutic agents.

#### **8.3. Interaction of LEEs with DNA bound to gold nanoparticles**

So far in this section, we have shown that cancer cells can be made more sensitive to high‐ energy radiation by chemically modifying their nuclear DNA with small molecules. These latter provide at least some of their radiosensitizing action, by increasing the interaction of LEEs with DNA, the products of DEA, and the resulting induced damage. Another approach consists of simply increasing the numbers of LEEs near the DNA of cancer cells. The best examples of this type of radiosensitization have been provided by the numerous fundamental, in vitro, and in vivo investigations of enhanced radiation absorption by gold nanoparticles (GNPs).

Both in vitro and in vivo experiments [189–204] have shown radiation enhancement effects due to the presence of GNPs. Several models have been developed to account for dose enhancement in cells by considering the increase in radiation energy deposition [205–211], due to additional energy absorption by the GNPs, as a function of their size. As expected, the energy of electrons emanating from the GNPs is inversely proportional to their diameter. Many models [206–211] take into account localized effects of Auger‐electron cascades. They consider the huge enhancement of energy deposited in the vicinity of GNPs, as arising from the considerable increase in photoelectric absorption cross section of gold in comparison to that of tissue [200, 208, 210, 211]. The increase in this cross section produces an additional local generation of photoelectrons, Auger electrons, and characteristic X‐rays [208, 212]. The major portion of the energy absorbed by the GNPs is converted into electrons, most of which escapes the GNPs with low energy (0–30 eV) [213–215].

The indirect effect of emitted electrons was investigated in water solutions containing GNPs, where the nanoparticle‐induced OH concentration from radiolysis was measured. Relevant literature and details can be found in the paper of Sicard‐Roselli et al. [189], who also proposed a new mechanism for hydroxyl radical production in irradiated GNP solutions.

The direct effect of high‐energy radiation on DNA, resulting from the presence of GNPs, was first investigated by Zheng and coworkers [35, 214–218]. Relatively thick (~0.3 and 2.9 μm) films of plasmid DNA with or without electrostatically bound GNPs were bombarded with 60 keV electrons. The probabilities of formation of SSBs and DSBs from the exposure of 1:1 and 2:1 GNP–plasmid mixtures to fast electrons increased by a factor of about 2.5, compared to DNA alone. It was suggested that the additional damage in the presence of GNPs was generated by LEEs escaping the nanoparticles. This hypothesis was later verified experimen‐ tally by the work of Xiao et al. [214]. These authors investigated the radiosensitization efficiency in terms of DNA damage as a function of the length of a ligand bound at one end to the surface of the GNP and at the other to DNA. They used the same DNA film preparation as in the experiments of Zheng et al. [215] and measured the ratio of induced damage with GNPs to that without GNPs (i.e., the enhancement factor, EF) for different lengths of the ligand. As indicated in **Figure 8** from their work, the corresponding EFs induced by 60 keV electrons on plasmid DNA bound to GNPs of various coatings range from 2.3 to 1.6 and 1.2, depending on the length of ligand separating the gold surface from the plasmid. This length ranged from 0 to 2.5 and 4 nm, respectively. The attenuation by the coating of short‐range LEEs emitted from the GNPs could explain the decrease in radiosensitization with increasing length of the ligand [214]. Since the attenuation range of LEEs is shorter than about 10 nm, it is obvious that the emission of LEEs from the GNPs and LEE‐interaction with DNA plays a major role in the mechanism of GNP radiosensitization.

Later, similar DNA–GNP films were bombarded with electrons of energies below the ioniza‐ tion potential of DNA. In this case, essentially no secondary LEEs were emitted from the DNA and the gold surface, so that Yao et al. [218] could investigate the purely chemical radiosensi‐ tization induced by GNPs. They showed that even without the emission of photoelectrons, direct electrostatic binding of an average of 0.2–2 GNPs to DNA could increase sensitization to LEEs by factors varying from 1.5 to 4.

Since GNPs increase the local density of LEEs and cisplatin enhances LEE interactions with DNA and damage to the molecule, it seemed likely that binding GNPs to a cisplatin–DNA complex would further boost radiosensitization and DNA damage induced by cisplatin [216]. This hypothesis was verified by irradiating with 60 keV electrons, GNPs electrostatically bound to a cisplatin–DNA complex [216]. Dry films of bare plasmid DNA and DNA–cisplatin, DNA–GNP, and DNA–cisplatin–GNP complexes were irradiated [216]. The yields of SSBs and DSBs were measured as described in the protocol established by Zheng et al. [215]. When the ratio of GNP to DNA was 1:1 and that for cisplatin to DNA was 2:1, the EFs for SSBs were between 2 and 2.5. With a cisplatin to GNP to plasmid ratio of 2:1:1, the EF increased to 3. This small increase could only be additive and unrelated to the interaction of additional LEEs with cisplatin. For DSB formation, however, the binding of both GNPs and cisplatin to a DNA molecule produced an impressive increase in the EF, that is, DSBs were increased by a factor of 7.5 with respect to pure DNA. It appeared quite obvious that the additional DSBs in the cisplatin–DNA–GNP complex arose from the generation of additional secondary electrons from the GNPs. The synergy between GNPs and cisplatin could arise from a number of basic phenomena, including the possibility of two or multiple event processes triggered by the interaction of a single 60 keV electron with a GNP. Within 10 nm of its location, a single gold atom increases the density of LEEs by a large factor [207, 212], and hence, a GNP that contains thousands of gold atoms is expected to generate a dramatic increase in this density [213]. Combined with the fact that cisplatin considerably lowers the energy threshold for DSB formation, a single or multiple LEE interactions on opposite strands within a distance of 10 base pairs could increase considerably the number of DSBs formed in GNP–cisplatin–DNA complexes.

The indirect effect of emitted electrons was investigated in water solutions containing GNPs, where the nanoparticle‐induced OH concentration from radiolysis was measured. Relevant literature and details can be found in the paper of Sicard‐Roselli et al. [189], who also proposed

The direct effect of high‐energy radiation on DNA, resulting from the presence of GNPs, was first investigated by Zheng and coworkers [35, 214–218]. Relatively thick (~0.3 and 2.9 μm) films of plasmid DNA with or without electrostatically bound GNPs were bombarded with 60 keV electrons. The probabilities of formation of SSBs and DSBs from the exposure of 1:1 and 2:1 GNP–plasmid mixtures to fast electrons increased by a factor of about 2.5, compared to DNA alone. It was suggested that the additional damage in the presence of GNPs was generated by LEEs escaping the nanoparticles. This hypothesis was later verified experimen‐ tally by the work of Xiao et al. [214]. These authors investigated the radiosensitization efficiency in terms of DNA damage as a function of the length of a ligand bound at one end to the surface of the GNP and at the other to DNA. They used the same DNA film preparation as in the experiments of Zheng et al. [215] and measured the ratio of induced damage with GNPs to that without GNPs (i.e., the enhancement factor, EF) for different lengths of the ligand. As indicated in **Figure 8** from their work, the corresponding EFs induced by 60 keV electrons on plasmid DNA bound to GNPs of various coatings range from 2.3 to 1.6 and 1.2, depending on the length of ligand separating the gold surface from the plasmid. This length ranged from 0 to 2.5 and 4 nm, respectively. The attenuation by the coating of short‐range LEEs emitted from the GNPs could explain the decrease in radiosensitization with increasing length of the ligand [214]. Since the attenuation range of LEEs is shorter than about 10 nm, it is obvious that the emission of LEEs from the GNPs and LEE‐interaction with DNA plays a major role in the

Later, similar DNA–GNP films were bombarded with electrons of energies below the ioniza‐ tion potential of DNA. In this case, essentially no secondary LEEs were emitted from the DNA and the gold surface, so that Yao et al. [218] could investigate the purely chemical radiosensi‐ tization induced by GNPs. They showed that even without the emission of photoelectrons, direct electrostatic binding of an average of 0.2–2 GNPs to DNA could increase sensitization

Since GNPs increase the local density of LEEs and cisplatin enhances LEE interactions with DNA and damage to the molecule, it seemed likely that binding GNPs to a cisplatin–DNA complex would further boost radiosensitization and DNA damage induced by cisplatin [216]. This hypothesis was verified by irradiating with 60 keV electrons, GNPs electrostatically bound to a cisplatin–DNA complex [216]. Dry films of bare plasmid DNA and DNA–cisplatin, DNA–GNP, and DNA–cisplatin–GNP complexes were irradiated [216]. The yields of SSBs and DSBs were measured as described in the protocol established by Zheng et al. [215]. When the ratio of GNP to DNA was 1:1 and that for cisplatin to DNA was 2:1, the EFs for SSBs were between 2 and 2.5. With a cisplatin to GNP to plasmid ratio of 2:1:1, the EF increased to 3. This small increase could only be additive and unrelated to the interaction of additional LEEs with cisplatin. For DSB formation, however, the binding of both GNPs and cisplatin to a DNA molecule produced an impressive increase in the EF, that is, DSBs were increased by a factor

a new mechanism for hydroxyl radical production in irradiated GNP solutions.

mechanism of GNP radiosensitization.

208 Radiation Effects in Materials

to LEEs by factors varying from 1.5 to 4.

**Figure 8.** Enhancement factors (EFs) for the formation of SSB, DSB, and loss of supercoiled DNA induced by 60 keV electrons, obtained with GNP–DNA complexes of ratio 1:1. The groups of three histograms represent the respective damages. In each group, the EF corresponds to the damage when the GNP alone is bound to DNA or when the GNP has been coated with ligands 2.5 and 4 nm in lengths corresponding to GNP@C11H23 or GNP@DTDTPA (i.e., dithiolat‐ ed diethylenetriaminepentaacetic acid), respectively in the figure.

As shown by Zheng et al., only one GNP per DNA molecule is on average necessary to increase DNA damage considerably [216]. Thus, as long as the nanoparticles reach the DNA of cancer cells, the amount to be administered to patients to obtain significant radiosensitization should be at most the same as that of the Pt‐drugs routinely administered in chemotherapy [176, 177]. In recent in vitro experiments, GNPs were targeted to the DNA in the cell nucleus by linking peptides to the gold surface [197, 202]. Such vectored GNPs, targeting the DNA of cancer cells, should be applicable in the clinic and may accordingly offer a new approach to radiotherapy treatments. However, this type of radiotherapy is expected to be limited to superficial tumors, owing to the requirement for that low‐energy (<100 keV) X‐rays be used to optimize LEE production and hence radiosensitization by the photoelectric effect. To treat deep tumors, a radioactive source may have to be encapsulated inside a gold nanoparticle (i.e., in a gold nanocage) [219]. Furthermore, if DNA specificity cannot be achieved in patients, suc‐ cessful treatment may still be possible by intratumor injection of GNPs, as recently shown by Shi et al. [220] and Bobyk et al. [204].
