INTRODUCTION

The efficiency of ionizing radiation in producing a range of biological effects depends not only on absorbed dose but also on radiation quality and its associated track structure, with biological effectiveness per unit absorbed dose typically increasing with increasing linear energy transfer (LET). High-LET alpha particles are particularly important with respect to human exposures, most notably from radon and its progeny, which contribute approximately 50% of the annual effective dose to the UK population and with a similar contribution worldwide. The effective dose to individuals from radon and its progeny can vary by many orders of magnitude depending on location (1). The main decay chain for ²²²Rn produces α particles ranging in energy from 5.3–7.7 MeV with corresponding ranges in tissue (assumed here to be water) of approximately 40 and 71 μm, respectively. As an alpha particle slows down and loses energy, the LET of the particle rises from between ~70–90 keV μm⁻¹ at the start of the track (depending on initial energy) to a peak of ~ 237 keV μm⁻¹ at an energy of 0.65 MeV towards the end of the track before falling again at the very end of its range (Fig. 1). Therefore, the cells traversed by the alpha particle are irradiated with a range of energies and LETs. The resulting human exposures are normally at very low dose rates to these very short densely ionizing tracks, with an average cell nucleus in the bronchial epithelium typically receiving around 0.002–0.009 alpha-particle traversals per year (2). However, while many cells are not traversed, those cells that are traversed receive a substantial local dose of up to ~0.5 Gy. Therefore, the biological effectiveness of a single α-particle traversal is important.

FIG. 1.

Variation in energy and associated LET with distance traversed in tissue (assumed to be water) by a 5.5 MeV alpha particle (corresponding to energy of alpha particle emitted from a ²²²Rn decay). Also shown is the approximate variation in energy and LET through a 5 μm thick cell (represented by bars) for the various experimental conditions described in this article, defined by the thickness of helium traversed in the irradiator used.

With respect to medical exposures, there has been an increased interest in exploring the potential use of α-particle-emitting radionuclides in radiotherapy. In particular, there is current interest in the use of targeted alpha-particle therapy (TAT) for the treatment of micrometastases, which is achieved by targeting the α-particle-emitting radionuclide to cancer cells by attaching monoclonal antibodies (or other targeting molecules). This technique takes advantage of the short high-LET tracks produced, typically resulting in a high efficiency of cell inactivation and limiting damage to a small number of cells close to the point of decay, thereby minimizing damage to surrounding healthy tissue. As discussed in a recent review by Elgqvist et al. (3), a number of clinical trials have been reported for a variety of cancer types using a range of α-emitting radionuclides (including ²¹³Bi, ²¹¹At, ²²⁵Ac, ²¹²Pb/²¹²Bi and ²²³Ra), which emit either single or multiple α particles depending on the decay chain, with energies ranging from 5.7 MeV up to 8.8 MeV. Again, irradiated cells are traversed by a range of energies and LETs as each α particle slows down, resulting in a variation in efficiency of cell inactivation.

The biological effectiveness of high-LET charged particles as a function of energy is also an important factor with respect to the increasing use of accelerated charged particle for therapy. This is particularly true for carbon ion therapy, for which there is a significant variation in relative biological effectiveness (RBE) with depth of the treatment beam as carbon ions lose energy and with the increasing contribution due to secondary ions resulting from fragmentation processes. A range of radiobiological models have been developed for use in conjunction with treatment planning software to take this variation into account and enable the optimization of the beam with respect to biological effect rather than absorbed dose. For proton therapy, the RBE is routinely assumed to be 1.1, however, experiments have shown that the RBE does vary with energy and significantly increases towards the very end of particle range (4). To develop and test these radiobiological models [e.g., LEM IV (5), RCR (6) and MCDS (7) models] it is important that experimental data are available for a range of energies (ideally including monoenergetic energies), particles and cell types. Although clinical treatments are currently restricted to protons and carbon ions, the potential of a range of different ions is also being explored (6).

The conventional picture of RBE variation with LET for a range of biological end points (8), including cell inactivation (9, 10), mutation (11, 12) and transformation (13), shows a sharp rise in RBEs from low-LET radiation (~several keV μm⁻¹) up to a peak at about 100–200 keV μm⁻¹ and then decreasing at higher LET values. For a number of studies the experimental data are obtained using a range of particles to cover the required range in LET. However, it is known that the different particles of the same LET will have different RBE values as a result of changes in the local ionization density due to differences in the energy spectra and range of δ-ray electrons produced along the track (14). Therefore, it is important whenever possible to use only the particle of interest, although this can be problematic in some cases due to the limited particle range. A limited number of studies have investigated the variation of RBE as a function of energy/LET of α particles. However, the determination of RBEs in these types of experiments depends very critically on estimating the “true” absorbed dose to the cell and to sensitive structures within the cell. This is not a trivial problem, since the α-particle energy and LET may change rapidly over the dimensions of the cell, especially as the α particle approaches the Bragg peak. The dimensions of the cell and its nucleus must be accurately known. The energy and angular distributions of the incident α particles must be well characterized, and their subsequent behavior must be understood as they continue to traverse the cell.

The Medical Research Council (MRC) plutonium-238 irradiator (now located at the Oxford Institute for Radiation Oncology, Oxford) is particularly well suited for the study of the variation in biological effectiveness with α-particle energy (15). It allows the irradiation of a cell layer with an essentially parallel beam of nearly monoenergetic α particles, with energies varying from 4.2 MeV down to 1 MeV and below, with corresponding LETs rising from 103 keV μm⁻¹ up to the maximum of 237 keV μm⁻¹ within the irradiated cell. In this study, the facility was characterized for six different irradiation conditions enabling incident energies from 4.0 MeV down to 1.1 MeV, with the incident dose determined by fluence measurements for each energy and cell morphology measurements determined to calculate the average absorbed dose to the cells. The variation in clonogenic survival of V79-4 cells was determined as a function of incident energy along with the relative variation in the initial yields of DNA double-strand break (DSB) induction measured using the fraction of activity released (FAR) assay. The results will expand on the previous data obtained at Harwell for the same V79-4 cells and culture techniques using helium ions with energies ranging from 5.85 MeV up to 34.9 MeV. In addition to the variation in RBEs for cell inactivation as a function of α-particle (helium ion) energy, the variation in the probability of surviving a single α-particle traversal is explored.

MATERIALS AND METHODS

Tissue Culture and Cell Survival

For each irradiation experiment, an ampoule of V79-4 Chinese hamster cells was recovered from the MRC stock in liquid N₂ storage, thawed and grown in T75 Eagle's minimum essential media (Gibco®; Life Technologies, Grand Island, NY) with 10% fetal calf serum (Sigma-Aldrich Company Ltd., Dorset, UK), 1 mM L-glutamine (Sigma-Aldrich) and penicillin/streptomycin (Sigma-Aldrich). Exponential growth was maintained for 3 days at 37°C in 5% CO₂/95% air. Cells were harvested with trypsin/EDTA (Becton Dickinson, Franklin Lakes, NJ), centrifuged, resuspended in growth media and counted before dilution to the required density. Approximately 2 × 10⁵ cells in 2 ml were seeded into each irradiation dish. The dishes consisted of Pyrex® glass cylinders (25 mm deep, 30 mm internal diameter) with a thin Hostaphan® base (0.35 mg cm⁻² polyethylene terephthalate; Hoechst, Franford, Germany) to allow passage of the α particles. Dishes were incubated for approximately 40 h prior to irradiation as a subconfluent, exponentially growing cell monolayer, with two dishes irradiated per dose point and cells pooled. Immediately after irradiation a 26 mm diameter disc was cut out from the base of each dish using a custom-made tool (16) and placed in trypsin/EDTA to remove cells. The cells were centrifuged, resuspended in growth media, counted and respread into 9 cm diameter petri dishes at densities expected to yield about 200 colonies per dish. After 8 days of growth, the colonies were fixed, stained with methyl blue and surviving colonies (>50 cells) counted. The fraction of surviving cells was calculated with respect to the cloning efficiency of the sham-irradiated cells. The resulting curve was fitted to the equation,

where S/S₀ is the surviving fraction and D is the mean dose to an average cell.

DSB Induction

To determine the variation in initial yield of DSB as a function of α-particle energy, cells were prepared in irradiation dishes in an identical manner to the survival experiments and the assay performed is similar to that previously described (17). Approximately 24 h prior to irradiation, cells were labeled with tritiated thymidine (3 μl of stock solution consisting of 0.1 ml [³H]thymidine) (Amersham International, Little Chalfont, UK) in 50 μl of free thymidine (1 mg ml⁻¹ in water; Sigma-Aldrich) and 1.85 ml of phosphate buffered saline (PBS; Sigma-Aldrich). At least 1 h prior to irradiation the active media was replaced with fresh media and tritium free thymidine. Dishes were irradiated (mean doses ranging from 0 to ~20 Gy) at ~5°C and placed on ice immediately afterwards. Cells were harvested on ice and resuspended in 0.8% low-melting-point agarose in PBS and immediately pipetted into chilled plug molds and stored at 4°C, with 2 plugs per dose point. The plugs were subsequently removed from the molds and the cells lysed with ice-cold solution of proteinase K (0.05%), N-lauryl sulphate (2%) and 0.5 M dm⁻³ EDTA (Sigma-Aldrich) at pH 7.5 for 1 h at 4°C and then 24 h at 37°C. DSB induction was determined using the fraction of activity released (FAR) assay using pulsed-field gel electrophoresis (PFGE) (model no. CHEF DR11, Bio-Rad Laboratories Ltd, Hemel Hempstead, UK). The electrophoresis grade agarose gel (UltraPure™; Gibco) was prepared at 0.8% in TBE buffer (Sigma-Aldrich) and run for 96 h at 45 V with a 60 min switching time at 16°C. The gel was subsequently stained with 0.0 μg ml⁻¹ ethidium bromide (Sigma-Aldrich) then washed with water. With the aid of an ultraviolet transluminator, the well and lanes for each plug were separated with the lane cut into five pieces. These were transferred into glass scintillation vials containing 0.5 ml of 1 M dm⁻³ HCl (Sigma-Aldrich) and heated to melt the gel. Once cool, ~7 ml of liquid scintillant (OptiPhase HiSafe 3; PerkinElmer® Inc., Woodbridge, Canada) was added and mixed by vortexing. The activity in each vial was determined using a Beckman LS6500 scintillation counter [Beckman Coulter Ltd., High Wycombe, UK]. The relative yield of DSB was determined by the amount of DNA extracted from the cell given by the fraction extracted:

where C_lane and C_well represent the total counts per min in the lane and well, respectively.

Dose Determinations

Due to loss of energy of the α particles as they traverse the cell, and corresponding increase in LET, the mean dose to the cell is different to the dose on the incident surface of the cell. The dose at a given depth within the cell is given by:

where ρ is the density, Φ is the particle fluence and L is the LET [this corresponds to D (Gy) = 0.16 * Φ (particles μm⁻²) * L (keV μm⁻¹) assuming a cell density of 1 g cm⁻³].

The average dose within the cell nucleus was calculated as follows. Each energy distribution for a given helium distance (see below) was multiplied by the appropriate fluence (Table 1) for that energy distribution. In this way each channel of the energy spectrum gave the number of alpha particles entering the cell layer within a certain energy interval. For a given energy interval, alpha particles were followed through the cell layer calculating the energy loss at successive 0.1 μm intervals using the SRIM stopping power data for helium ions in water (22, 23) and so giving the variation in LET and absorbed dose as functions of depth in the cell layer. The resulting doses were summed over the entire energy spectrum and averaged throughout the depth of the cell nucleus:

where Φ(E_i) is the fluence of particles in the energy interval E_i to E_i+1 entering the cell layer covering the whole energy spectrum and L_ij is the LET for the depth interval j to j + 1 in the cell layer for incident energy E_i, which is summed from the incident surface to a distance corresponding to the average height of a cell.

TABLE 1

Beam Characteristic as a Function of Helium Distance Traversed by the α Particles

RESULTS

Cell Characteristics

The electron micrograph in Fig. 2A shows the attached V79 cells to have a flat, pancake shape with the nucleus approximately elliptic cylindrical in shape. From analysis of these micrographs the average ratio of nuclear to cell thickness was estimated to be 0.8. Electron micrographs by themselves cannot be relied upon to give accurate values of cell thickness because of shrinkage during preparation (18, 19). From the confocal laser-scanning microscope measurements on live cells, a mean cell thickness of 5.0 μm (standard deviation = 1.2 μm) was obtained from 111 cells analyzed, with the distribution in cell heights shown in Fig. 3A. Therefore, for the dose calculations detailed below an average cell thickness of 5.0 μm was assumed. A nuclear height of 4.0 μm, with 0.5 μm of cytoplasm above and below the nucleus was also assumed (Fig. 2B). A mean nuclear area of 113 μm² (standard deviation of 34 μm²) was obtained from the analysis of 469 cells, with Fig. 3B showing the wide distribution in the measured areas. Cell cycle analysis performed on cells grown in Hostaphan-based irradiation dishes at the time of irradiation showed that on average 55–60% of the cells were in G₀/G₁ phase, 10–12% were in G₂ phase and 28–35% were in the S phase.

FIG. 2.

Panel A: Examples of typical electron micrograph sections of attached V79-4 cells. Panel B: Schematic of cell geometry assumed for calculations.

FIG. 3.

V79-4 morphological measurements. Panel A: Cell height distribution from 111 cells, giving a mean height of 5.0 μm (SD = 1.2 μm). Panel B: Nuclear area distribution, giving a mean nuclear area of 113 μm² from 469 cells (SD = 34 μm²).

Beam Characteristics and Dose Determination

Figure 4 shows the spectra of α-particle energy as a function of helium traversed, with the calculated mean incident energies shown in Table 1. From the CR39 measurements, the calculated mean fluence of α particles across the central 26 mm of the irradiation dish per revolution of the wheel holding the dishes is shown in Table 1 for each of the helium depths used, along with the incident dose rate on the surface of the cells. For helium distances of 65 mm and above, the CR39 measurements showed minimal variation across the central 26 mm of the dish, while for 25 mm of helium a correction factor of 0.927 was used to account for the observed variation (see Supplementary Fig. S1;  http://dx.doi.org/10.1667/RR13835.1.S1 (10.1667_RR3835.1.S1.pdf)). Assuming a mean cell height of 5 μm (Fig. 3A) and the incident α-particle spectra shown in Fig. 4, the calculated mean dose rate to the cell and average LET are tabulated in Table 1 along with the range of LET values experienced by the cell as the α particle traversed the cell or nucleus.

FIG. 4.

Spectra of alpha-particle energies incident on the cell monolayer for the specified distances of helium traversed (25–140 mm). The associated mean energy and half-width at half-maximum values are shown in Table 1.

Clonogenic Survival

Figure 5 shows the V79-4 clonogenic survival curves as a function of α-particle energy, determined by the source position. Table 2 shows the corresponding fitting parameters. The results for the 3.2 MeV (helium depth of 65 mm) showed good agreement with experimental data previously published by Hill et al. (24) when plotted using mean dose to a 5 μm thick cell; the data were combined for subsequent analysis. While the survival curve for the two highest energies used (4.0 MeV and 3.2 MeV) resulted in similar survival curves (with 3.2 MeV slightly more effective), further decreasing the energy of the α particles (with associated increase in LET) resulted in an increase in survival for a given dose. The RBE was calculated by comparing to the previously published ⁶⁰Co γ-ray data obtained using the same V79-4 cells (25) and described by the equation:

FIG. 5.

Survival of V79-4 cells as a function of alpha-particle energy: () = 4.0 MeV (25 mm of He); () = 3.2 MeV (65 mm of He); () = 3.2 MeV [65 mm of He, previously published data (24)]; () = 2.4 MeV (100 mm of He); () = 1.8 MeV (120 mm of He); () = 1.7 MeV (130 mm of He); () = 1.1 MeV (140 mm of He). The previously published ⁶⁰Co γ-ray survival data for V79-4 cells is also shown (25). Errors bars represent standard error of the mean.

TABLE 2

Variation in the RBE for 10% Survival (RBE_10%), Maximum Low-Dose RBE (RBE_M) and Associated Parameters as a Function of Alpha-Particle (Helium Ion) Incident Energy and Average LET to the Cell

Table 2 shows the resulting RBE values obtained for the 10% survival, as well as the maximum low-dose RBE (RBE_M) determined as the ratio of the initial slopes of the dose-effect curves for the current radiation study to that of the γ-ray reference radiation. The RBEs increase with increasing level of survival due to the curvature of the γ-ray survival curve. Also included in Table 2 are the experimental data obtained for the same V79-4 cells (11, 25) performed using the Variable Energy Cyclotron or the Tandem Van de Graaff accelerator, which were formerly operated by the UK Atomic Energy Authority at the Harwell Research Establishment (26). The RBE values presented are also calculated with respect to the same ⁶⁰Co γ-ray data (25) used for the current data. The variation in RBE as a function of incident α-particle energy and LET for the combined data set are shown graphically in Fig. 6. The data shows a peak in the RBE value for incident energies of around 3.2 MeV and corresponding average LET of 131 keV μm⁻¹.

FIG. 6.

Variation in RBE for 10% survival (RBE_10%), maximum low-dose RBE (RBE_M) and the calculated probability, S₁, of surviving a nuclear traversal by a single alpha-particle track as a function of incident energy (panel A) and average LET (panel B) in the cell. Circles and triangles represent current data, while squares and diamonds represent original Harwell data.

For a random (Poisson) distribution of numbers of tracks per cell the average number of tracks, N, traversing a mean nuclear area A (in μm²) for a given dose D (in Gy) is given by

where L is the LET in keV μm⁻¹ and assuming unit density. The mean probability, S₁, of survival after a single α-particle traversal through the cell nucleus can be estimated (27, 28) with the mean nuclear area and the mean lethal dose D₀, using the equation,

As described by Goodhead et al. (28) the exponential survival curve S/S₀ = exp(–αD) can be represented by

where l₁ is the average number of lethal lesions randomly distributed along a track through a cell nucleus. Therefore, the average number of lethal lesions per track is given by

and the number of lethal lesions per unit absorbed dose as

It is worth noting that the quantities l₁ and l_D provide estimates of an “intrinsic radiosensitivity” to the α-particle-induced damage taking into account the effects of potential multiple lethal lesions along individual tracks. Based on an average nuclear area of 113 μm², the calculated number of nuclear traversals per Gy, the mean probability of surviving a single α-particle traversal, S₁, the average number of lethal lesions per track, l₁, and corresponding lethal lesions per unit dose, l_D, are tabulated for both the present data and the previous Harwell data in Table 2. The variation in S₁ as a function of energy and LET is shown graphically in Fig. 6. These data show that α particles with incident energy of around 1.8 MeV (average LET of 190 keV μm⁻¹) are the most lethal with respect to single α-particle traversals with the corresponding highest number of lethal lesions per track, l₁.

The variation in the relative yield of initial DSB measured using the FAR assay is shown in Fig. 7, normalized to the 3.2 MeV data (where the relative yield is defined as 1) for comparison with previous data from MRC Harwell obtained showing an RBE for DSB induction in V79-4 cells of between 0.85 using sucrose sedimentation (29) and 1.0 using PFGE (30) for α particles of similar energy. The data is plotted as both the relative number of DSB per Gy and the relative number of DSB per track. As the energy of incident α particles decreases, this data suggests that although there appears to be slight decrease in the measured yield of DSB per Gy, the yield of DSB per nuclear traversal increases.

FIG. 7.

Variation in the relative yields of initial DSB as a function of incident energy of the alpha particles (helium ions) per unit dose (panel A) and per nuclear traversal (panel B). The values have been normalized to 3.2 MeV.

DISCUSSION

A number of experimental studies have been performed over the years examining the variation in biological effectiveness of charged particles at inducing cell inactivation as a function of particle energy (LET) for a range of different particles and cell types. Much of this work has now been compiled in the Particle Irradiation Data Ensemble (PIBE) database, which contains information on over 800 cell experiments from laboratories around the world (31). The RBE is not a single value for a given LET, but is also dependent on the level of biological effect at which the comparison is made, the cell type, growth conditions and environment, reference radiation and the particle type (8). The variation in biological effectiveness for different particles with the same LET is due to differences in the track structure and variation in local dose along the track because of different ranges of the δ electrons.

As shown in Fig. 8, the variation in RBE for 10% survival with LET for the α-particle V79-4 survival data presented in this article [which includes the data previously obtained with the same cells for helium ions at Harwell (11, 25)] compares well with other published V79 survival data for helium ions (closed symbols joined by lines). The data presented for Folkard et al. (4) and Furusawa et al. (32) was produced using ³He ions rather than ⁴He ions. Folkard et al. irradiated cells as an unattached monolayer with minimal media and the cells were assumed to be partially flattened spheres approximately 10 μm thick. Although V79 cells are not of human origin, this and previous data (24) showed that these cells represent a sensitive and reproducible assay useful for mechanistic studies.

FIG. 8.

Variation in RBE for 10% survival as a function of LET of helium ions (³He and ⁴He) for a range of experimental data on V79 cells (closed symbols) and other cell lines (open symbols).

In addition, Fig. 8 shows good agreement with the variation in RBE for survival after helium ion irradiation for a range of other cell lines (open symbols). The data for primary human fibroblast HF19 cells (12, 25) were also obtained at Harwell. Interestingly, for these cells the corresponding 250 kVp X-ray survival curve was also observed to fit to the equation S/S₀ = exp(–αD) and therefore the RBE is independent of survival level used for the comparison (dose). Figure 8 also includes data on human kidney T-1 cells (9, 33) (cells assumed to be 4 μm thick), C3H 10T1/2 cells (13) and human salivary gland (HSG) tumor cells (32). The data for the 75 keV μm⁻¹ point for the C3H 10T1/2 cells and all the points for HSG cells were also obtained using ³He ions. Where values or fits were not readily available from the original data the survival curve fit coefficients in the PIDE database were used to calculate the RBE. For low-incident energy particles (high LET) towards the end of their range, the cell morphology can be a major factor in not only the biological response but also the calculated LET and associated calculation of dose due to the large variation in LET through the cell (see Fig. 1).

The data presented in this article for the variation of RBE as a function of LET shows the curve “curling back” on itself, indicating that more than one value of RBE may be obtained by particles with the same LET but different energies. Although difference may be expected from the fact that the energy and associated range of δ rays produced by the lower energy alpha particles will be less than those produced by higher energy α particles of a similar LET, there are additional considerations that need to be taken into account when interpreting the 1.1 MeV data. The LET for these α particles varies significantly through the cell, with an incident LET of ~200 keV μm⁻¹ rising to a maximum of ~237 keV μm⁻¹ before falling to ~116 keV μm⁻¹ on the exit surface of 5 μm thick cell (Table 1). In addition, because of the variability in the range of individual α particles together with the variation in thickness of individual cells (Fig. 3A), the exit LET will vary significantly and for a fraction of cells not all of the α particles will fully traverse the cell or nucleus.

As the energy of the α particle (helium ion) decreases, there is an increase in LET resulting in an increase in the average number of lethal lesions per track, l₁, and associated decrease in survival probability after a single particle traversal, S₁ (Fig. 6, Table 2). At low energies this increase is associated with an increase in the average absorbed dose per traversing track and increasing number of DSBs and associated complexity of the DSB [see Nikjoo et al. (34) and Fig. 7B]. As the α particle slows down to energies less than 0.65 MeV, the LET starts to fall with decreasing energy. This becomes important in this experiment when the incident energy is less than ~1.8 MeV when the peak of LET lies within the 5 μm thick cell (Fig. 1) leading to a reduction in l₁ values and associated increase in S₁ (Table 2). Although the effectiveness of individual tracks at inactivating the cell typically increases with decreasing incident energy, the number of traversals per unit absorbed dose decreases. The data show for the experimental set-up used, that the maximum number of lethal lesions per unit absorbed dose, l_D, is obtained for incident energies of 3.2 MeV, resulting in a peak in the RBE values.

Although the RBE-LET relationship for α particles is quite similar for the variety of cell lines shown in Fig. 8, it is likely to vary substantially for other cell types, particularly for those of greater intrinsic radiosensitivity and/or with compact nuclei (24, 35, 36). The variation will be reflected also in the single-track survival probabilities, S₁. For example, from previous experimental data with the same α-particle irradiator, it was estimated that, for α particles of LET 120 keV μm⁻¹ the survival probability, S₁, varied from a value indistinguishable from zero for primary pre-B cells from Balb/c mice to about 0.8 for C3H 10T1/2 mouse fibroblasts (37). Evaluation of the effectiveness of the α-particle tracks for a given cell line in terms of number of lethal lesions per track, l₁, and hence effectiveness per unit track length, provides a means for estimating the sensitivity of the cells per unit dose if irradiated in other geometries, including along the long axis of flattened cells.

Ionizing radiation deposits its energy along highly structured tracks of ionization and excitation events, with the pattern varying not only on the nanometer/DNA scale, but also on the scale of the cell and tissue, and differences across all of these three scales can have a significant impact on the ultimate biological response. As for DNA damage, the increasing LET of the α particle (helium ions) with decreasing energy not only increases the frequency of clustering damage to DNA, but also the degree of complexity of the damage at the site, and it has been proposed that this is the main factor driving the differences in biological effectiveness between different radiation qualities for a range of biological end points (38, 39) including cell survival. For low-LET radiation, such as X rays or γ rays, the majority of DSBs produced are simple. Monte Carlo calculations estimate that the percentage of DSBs that are complex by virtue of an additional strand break within 10 bp (defined as DSB_C) was ~20%, rising to ~50% (40, 41) if the possibility of a base damage within 10 bp is also included in the definition of complexity (defined as DSB_CB). Similar calculations performed on helium ions (34) show that as the energy decreases from 10 MeV down to 2 MeV there is an increase in the local clustering of damage with the percentage of DSBs classified as complex (DSB_C) increasing from 45–73% (corresponding values for DSB_CB increase from 84–96%). This difference in the complexity of DSBs produced by α particles compared to gamma rays is reflected in the significant reduction in the extent of DSB rejoining. Previous experiments with V79-4 cells showed that ~90% of the DSB were rejoined 3 h after gamma-ray exposure, while only 30% were rejoined 3 h after exposure to 3.3 MeV α particles for a dose giving the same number of initial DSB (42).

Due to the deposition of energy along narrow densely ionizing tracks, and the additional complexity of damage produced at a single site on DNA, α particles have the potential to produce correlated sites of damage across higher orders of DNA packing. Such packing includes DNA wrapped around nucleosome (~10 nm diameter), packaging of these nucleosomes into chromatin fiber (~30 nm wide) and these fibers arranged in loops. Correlated DSBs can lead to the production of DNA fragments, the existence of which have been experimentally demonstrated (43, 44) and supported by theoretical calculations (45). Most DSB assays, including the FAR assay used in our study, are not able to resolve these correlated events, and thus have a tendency to underestimate the actual yield of DSB (44–46). To some extent this can be addressed using fragmentation analysis to separate out and quantify as many of the small fragments as possible, however, to obtain the required sensitivity a very high dose is needed, which complicates the analysis of the results because of the multitrack effects. Studies by Belli et al. (47) on V79 cells using helium ions (123 keV μm⁻¹) showed that while the conventional FAR assay gave an RBE for DSB induction between 0.8 and 1.0, analysis using fragmentation counting gave an RBE of 1.2. Similar studies by Newman et al. (48) again irradiating V79 cells with α particles (110 keV μm⁻¹), for conventional FAR analysis or applying random-breakage fits gave an RBE for DSB induction of 0.6–0.8, while summing the number of DNA fragments in the range of 10 kpb–5.7 MBp gave an RBE of 1.7. Therefore, although Fig. 7A suggests a slight increase in the yield of DSBs per nuclear traversal with decreasing energy, the DSB yield is expected to increase significantly.

It has also been shown that track structure is important on the scale of the cell. Within the interphase nucleus of a human cell, the DNA is in the form of 23 pairs of chromosomes and each of these has been shown to occupy discrete domains with limited overlap (49) and typically with a nonrandom relative distribution of these within the nucleus (50, 51). High-LET α particles deposit energy along discrete, narrow, densely ionizing tracks with 90% of the energy deposited within ~10 nm (52) of the particle path due to the short range of the δ electrons. This very heterogeneous pattern of energy deposition results in highly correlated damage sites, including DSBs along the path of the alpha particle within and between the traversed chromosome domains. The resulting radiation-induced chromosome aberrations are critically dependent on the spatial distribution of these DNA damage sites and therefore the track structure. The movement of DSB within the nucleus has been shown to be limited to less than ~1 μm (53, 54). Studies in human peripheral lymphocytes (PBLs) have shown that the passage of individual alpha particles is efficient at inducing complex chromosome rearrangements (involving 3 or more breaks in 2 or more chromosomes) in contrast to low-LET X rays, which produce mainly simple rearrangements (no more than 2 breaks in 2 chromosomes) at low doses (55, 56). The vast majority of the complex aberrations produced are nontransmissible and therefore are lethal to the cell. A similar response was observed in human CD34⁺ hemopoietic cells irradiated with a mean of ~1 alpha-particle track per cell with incident energies of 3.8 MeV, 3.3 MeV and 2.2 MeV and corresponding mean LETs through the cell of 119, 135 and 182 keV μm⁻¹. It was observed that with increasing LET of the alpha particle, there was an increase in the ratio of complex to simple aberrations (57). This increase in the proportion of complex aberrations with LET per alpha-particle traversal is a result of the increasing number of DSBs per track (support by the data shown in Fig. 7, which will be an underestimate of the actual increase due to the production of DNA fragments) and corresponding decrease in the average separation distance between these breaks (57). In addition, this may be aided by an increase in the complexity of the damage at individual sites and also a potential increase in probability of production of short DNA fragments. At higher doses, where there are multiple nuclear traversals, there is an increasing probability of rearrangements between sites of DNA damage produced by independent tracks or two tracks traversing the same chromosome domain producing two independent exchanges but connected by the common chromosome (58). However, both PBL and CD34⁺ cells have a spherical geometry and relatively small nuclear volumes. The yield and complexity of the resulting aberrations are dependent on nuclear geometry and the orientation of the traversing high-LET particle (59, 60). This results in a variation in the number of DSBs induced within the nucleus per track and the number of chromosome domains traversed.

SUMMARY

The MRC ²³⁸Pu α-particle irradiator has been calibrated to enable studies investigating the variation in biological response with incident energies ranging from 4.0 MeV down to 1.1 MeV. This facility has initially been used to investigate the variation in effectiveness of alpha particles at inactivating V79-4 cells as a function of energy, extending the previously published data obtained at Harwell using the same cells irradiated with helium ions with energies ranging from 34.9 MeV to 5.85 MeV. These studies were performed in conjunction with cell morphology measurements on live cells, enabling precise determination of absorbed dose and calculation of the average LET. The increased biological effectiveness per absorbed dose above that of gamma rays for these alpha particles (helium ions) is a result of the increased ability of these densely ionizing particles to produce clustered damage to DNA and associated decrease in repairability of the damage. As the particle slows down and LET increases this can lead to an increase in the number of DSBs (after accounting for correlated breaks resulting in short DNA fragments) and in the complexity of the DSBs produced. In addition, correlation of damage, such as DSBs, along the path of the particle leads to an increased probability of producing complex chromosomal rearrangements that, more often than not, are nontransmissible at mitosis and lead to cell death. The results show an increase in RBE for cell inactivation with decreasing helium ion energy (increasing LET), reaching a maximum for incident energies of ~3.2 MeV and corresponding average LET of 131 keV μm⁻¹, above which the RBE is observed to fall at lower energies (higher LETs). The variation in RBE with LET shows good agreement with the limited published data for helium ions for the range of energies/LETs studied. The effectiveness of single alpha-particle traversals (relevant to low-dose exposure) at inducing cell inactivation was observed to increase with decreasing energy to a peak of ~ 68% survival probability for incident energies of ~1.8 MeV (average LET of 190 keV μm⁻¹) producing ~0.39 lethal lesions per track through the cell nucleus. However, the efficiency of single traversals will also vary significantly with cell morphology and angle of incidence, as well as cell type.

SUPPLEMENTARY INFORMATION

Fig. S1. Variation in fluence with distance from the center of the dish for irradiations performed with a helium depth of 25 mm (panel A) perpendicular to the wheel motion and (panel B) tangential to the wheel motion.