(Cambridge: Thomas Graham House), in press. than the value expected from traditional dose-. Geard, C. R., Loucas, B. D. and Randers±Pehrson, G., 1995,.
int. j. radiat. biol 1997, vol. 72, no. 4, 397 ± 407
Inactivation of individual mammalian cells by single a-particles M. PUGLIESE² ³ , M. DURANTE³ §, G. F. GROSSI³ §, F. MONFORTI¶d , D. ORLANDO§, A. OTTOLENGHI¶d , P. SCAMPOLI² ³ and G. GIALANELLA² ³ §* (R eceived 10 M arch 1997; accepted 16 J une 1997 ) Abstract. Purpose: To measure clonogenic death of Chinese hamster V79 cells following exposure to a de® ned number of 4´3 MeV aparticles (track-averaged LET=105 keV/mm). M aterials and methods: Cells were irradiated at the radiobiological facility installed at the TTT-3 Tandem accelerator in Naples by using a `Biostack’ approach, which allows the positions of incident tracks relative to cells to be carefully determined. Subcellular structure was identi® ed by ¯ uorescence microscopy, while tracks were visualized by LR-115 solid state nuclear track detectors. R esults: Particle hits in the cytoplasm did not signi® cantly aå ect cell survival, yet survival probability decreased exponentially as a function of the number of nuclear traversals. Measured probability of surviving to exactly one 4´3 MeV a-particle traversal in the cell nucleus was 0´67 Ô 0´10. Inactivation cross-section was substantially higher than expected from conventional survival curves. However, folding of the data with Poisson statistics showed that survival level expected if a mean of one a-particle goes through a nucleus is higher than the measured value after exactly one particle traversal. Conclusions : V79 cells have about 67% probability to survive a single a-particle traversal in the cell nucleus. Single-particle survival curves are consistent with conventional dose-survival relationships, once Poisson distribution of traversals is taken into account.
1. Introduction Human exposure to a-particles is characterized by low doses. Radon and its progeny represent the most important source of environmental radioactivity, health hazard resulting from the inhalation of these a-emitting radionuclides. For radon in dwellings, it has been estimated that in the average house-holder less than 1% of bronchial epithelium cells receive more than one a-particle traversal in a lifetime (Brenner 1994). Thus, risk from exposure to radon indoors is limited to single a-particle traversals through the target cells. Nonetheless, estimates of risk from radon indoors are *Author for Correspondence. ² Servizio di Radioprotezione, UniversitaÁ `Federico II’, Mostra d’Oltremare Pad.20, 80125 Napoli, Italy ³ INFN, Sezione di Napoli, 80125 Napoli, Italy § Dipartimento di Scienze Fisiche, UniversitaÁ `Federico II’, Mostra d’Oltremare Pad.20, 80125 Napoli, Italy. ¶ Dipartimento di Fisica, UniversitaÁ di Milano, Via Celoria 16, 20133 Milano, Italy. d INFN, Sezione di Milano, 20133 Milano, Italy.
derived from epidemiological studies at relatively high doses, e.g. from exposure of miners, and extrapolated from multiple- to single-traversals (Stidley and Samet 1993). Therefore, in vitro studies aimed to determine the biological response to single a-particle traversals are helpful in understanding risk associated to radon indoor exposure. Two diå erent approaches to the problem of measuring single-track eå ectiveness were identi® ed as early as the 1970s: microirradiation of single cells (reviewed in Michael et al . 1994) and broad® eld exposure of cells attached to solid state nuclear track detectors (reviewed in Durante et al . 1996). Microbeam irradiations are based on sophisticated methods for delivering, focusing, detecting and imaging particles. Microbeams can irradiate cells with a pre-determined number of particles in speci® c cellular substructures. Broad-® eld experiments are technically simpler and the amount of time per irradiation is much shorter, yet the number of traversals must be measured a posteriori rather than decided a priori . This approach has been adopted successfully in space-¯ ight studies, where biological resting-state samples were sandwiched between plastic detectors and carried aboard a spacecraft: the `Biostack’ approach (reviewed in Horneck 1992). A recent study developed a similar method to study the eå ectiveness of single charged particles, usable with cycling mammalian cells (Durante et al . 1994). In this approach, cells grow in a glass well with a mylar base, and a thin track-etch detector ® lm is stuck to the mylar base. Cells are exposed to accelerated helium ions, and processed immediately after exposure. Etching at 37ß C of the detector allows track visualization with no perturbation to the cells. This method has now been improved to visualize the cells by using ¯ uorescence microscopy. The method has been applied to measure inactivation of Chinese hamster V79 ® broblasts. 2. Materials and methods 2.1. Cell culture Chinese hamster V79 cell line is one of the most studied in vitro models of cytotoxicity of ionizing
0955± 3002/97 $12.00 Ñ 1997 Taylor & Francis Ltd
398
M . Pugliese et al.
radiation. Cells were kindly donated by Dr. M. Belli and grown in a-MEM medium at pH 7´2 supplemented with 10% foetal calf serum (Hyclone). Cells growing in tissue culture ¯ asks were incubated at 37ß C in 5% CO2 atmosphere. Cells were subcultured by incubation in 0´05% trypsin for 5 min at 37ß C and cell singlets were counted at a Coulter Counter (model Z1). Cells for irradiations were kept in con¯ uence for 24 h, and then plated on mylar. Mylar foils for cell growth were pre-cleaned in ethanol for 20 min. Cells were irradiated after about 6 h, when they appeared attached and ¯ at. The use of cells from the plateau-phase is important, because cell doublets must be discarded in these experiments. Cells coming from the G0 phase are far from mitosis, so they cannot duplicate in the time interval between seeding and observation at the microscope. In addition, cells were mostly in G1 when exposed to radiation, thus reducing biological variability. Even if no duplication occurs, it is important that cells are suæ ciently spaced, or colony overlapping will make it impossible to trace back the parent cell. Cells too close had to be discarded. Seeding density was set at 200 cells/well (base area=95 mm2 ). Plating eæ ciency ( PE ) was measured by seeding cells at low density, and colonies ® xed (10 min in methanol) and stained (10 min in Giemsa) following 1 week incubation at 37ß C. Nuclear area was measured in living cells growing on mylar by staining the nucleus in Hoechst 33258 (Calbiochem, Le Jolla, CA) at a ® nal concentration of 0´5 mg/ml. Nuclei were observed at an inverted ¯ uorescence microscope (Zeiss Axiovert Jene, Germany) and images obtained by a CCD camera were stored in a Quantimet 500 (Leica, Cambridge, UK) image analyser for area measurements. 2.2. Irradiation Glass wells (2 cm height, 1´1 cm internal diameter) were used to irradiate the cells. The base of the well is a 3 mm mylar ® lm. A LR-115 ® lm is stuck by araldite to the mylar base. White transparent Kodak LR-115 type II ® lms (12 mm thick) were provided by Dosirad Co. (Paris, France). It was observed that critical a-particle energy for track visualization following etching at 37ß C in these detectors was in the range 2´5± 3´5 MeV, as already observed by other authors after spark-counting of LR-115 (Bonetti et al . 1991). If the beam is shot at this energy through the LR-115, the residual range will be insuæ cient to cross the cells. For this very reason, in this study the beam was shot in front of the cell monolayer, after medium withdrawing from the well. After medium withdrawal, a thin layer of medium
is left on the cell monolayer. Thickness of V79 cells plus residual medium during irradiation was estimated by measuring beam energy loss as described elsewhere (Durante et al . 1993). Brie¯ y, beam energy spectrum was measured in air by a Ortec UltraCAM silicon detector after traversal of a 3 mm mylar sheet in front of the detector active area. A second spectrum was measured in the same position, but this time cells were grown on mylar. Diå erences in the spectra measured in the two cases provided information about the thickness of the cell monolayer plus the residual liquid on the cell surface after medium withdrawal. A total thickness of about 9 mm was estimated. Average thickness of V79 cells is about 6 mm (Townsend et al . 1990). Thus, residual liquid was about 3 mm. Helium ions were accelerated at the radiobiological facility of the Tandem TTT-3 accelerator in Naples. Details of the facility and dosimetry of accelerated helium ions have been published elsewhere (Napolitano et al . 1992). In the experiments described here, terminal voltage was set at 2´3 MV. The beam was extracted in air through an aluminized mylar window, and went through 2 cm air plus about 3 mm residual medium before hitting the cells. Cells were exposed for 10 s at a ¯ uence rate of 107 particles/cm2 ¯min. Immediately after irradiation, 1´5 ml fresh medium containing Hoechst 33258 (5 mg/ml) was added to the wells, and detectors were etched in a basin containing KOH 10N at 37ß C for 15 min in a 5% CO2 humid incubator. Etching at 37ß C is essential in these experiments, although signal/noise ratio decreases after low-temperature etching. Well base was then carefully washed for 15 min in water. 2.3. M icroscopy After washing in water, the medium containing Hoechst was withdrawn from the wells and replaced by fresh medium. Taking into account etching and washing time, cells were incubated in Hoechst 5 mg/ml for about 30 min. Cells were observed at an inverted ¯ uorescence microscope (Zeiss Axiovert Jene, Germany) connected to Quantimet 500 image analyser (Leica, Cambridge, UK). Wells were ® tted in a sample holder on the microscope stage and wedged through a scrap in the glass. Isolated cells were quickly localized with a 10 Ö objective in ¯ uorescence. A 32 Ö objective was then used for analysis. Coordinates of the cell on the microscope stage were stored. By adding low intensity visible light and a phase-contrast ® lter, details of the cell cytoplasm became visible together with the very bright ¯ uorescent nucleus. Particle
Survival to single a- particle tracks appeared as black dots in a focal plane slightly diå erent from the cell plane. Sometimes, three diå erent images (nucleus in ¯ uorescence, cell in phase contrast, and tracks in focus with the cell out of focus) were stored for each cell; other times, one single image already contained all the necessary details. Once all the cells in the well had been recorded, cells were incubated at 37ß C and the following well irradiated and processed as described above. 2.4. Image analysis Analysis of the images stored was performed at the Quantimet 500 image analyser. Diå erent images of the same cell were superimposed when necessary. Examples are given in Figure 1. Image analysis is necessary to determine the number of traversals and the impact parameter, in particular whether a nuclear or cytoplasmatic traversal occurred. Sensitivity of impact parameter measurements are limited by the following factors: (a) size of the etched tracks; (b) beam divergence; and (c) cell movements in the time interval between irradiation and observation (30± 60 min). Under the etching conditions, mean track diameter was 3´0 Ô 0´5 mm. Smaller tracks would improve sensitivity, but they are sometimes diæ cult to be unequivocally distinguished from the background. The impact point was identi® ed with the pixel corresponding to the track centroid, as calculated by the Quantimet computer. Beam collimation is essential in these experiments. If the particle hits the cell in a given point P , the orthogonal projection P ¾ of P on the nuclear ® lm should be detected. However, particles can enter the cells with incident angles diå erent from 0ß , and further scatter is produced by multiple collisions inside the cell and the mylar base. Thus, observed position O of the track will generally be diå erent from P ¾ . The diå erence d = O Õ P ¾ provides the lateral scattering of the particle (Figure 2). Accelerator provides a beam with an angular divergence of about 1 mrad. Further angular spreading is provided by multiple scattering in the exit window, in air, in the cells, and in the mylar base of the glass well. Beam lateral straggling was simulated by the TRIM91 Monte Carlo code (Biersack and Ziegler 1991). Cell movements have also been measured at the Quantimet 500. Images of over 50 cells have been stored at 15-min intervals for up to 2 h. It was found that V79 show complex movements, mostly translation and deformation. Measured parameters were coordinates x , y of the cell centroid and the largest
399
internal diameter L . Velocities of cell centroid vc and of deformation vL were calculated as: N 1 Ó [ x i(t+ Dt ) Õ x i(t)]2 + [ y i(t +Dt ) Õ y i(t)]2 vc = Dt i=1 N N 1 L i(t +Dt ) Õ L i(t) vL = N Dt i= 1 where N is the number of measured time intervals (8± 10) and Dt=15 min.
2.5. Survival assay Following 1 week incubation, colonies were ® xed in methanol and stained in Giemsa. Cells were observed again at the inverted microscope. Coordinates of individual cells were reproduced and the number of cells in the observed colony were measured. Notably, the coordinates need not be reproduced with very high ® delity, because less than one cell per microscope ® eld was observed and no colony superimposition was possible. Therefore, rather than microscopic reference marks used in some Biostack arrangements (Horneck et al . 1989, Weisbrod et al . 1992), Vernier scale on the microscope translator is suæ cient. 2.6. M onte Carlo calculations Monte Carlo calculations have been performed to calculate the average dose deposited per particle traversal and other biophysical parameters. Simulation of the particle beam energy deposition has been performed using a purposely written Monte Carlo code (Ottolenghi et al . 1995, 1997) based on ICRU 49 data (ICRU 1993). Biological target was an ellipsoid nucleus with a maximum thickness of 6´6 mm and cross-sectional area as measured in this experiment. Heterogeneity of cell population has been taken into account by generating cell populations whose sizes were randomly extracted from truncated Gaussian distributions peaked at these values. Further details of the calculations will be published elsewhere (Grossi et al . 1997, Ottolenghi et al . 1997). 3. Results and discussion 3.1. B iological parameters If residual medium does not dry out, PE will not be signi® cantly aå ected by medium withdrawal (Napolitano et al . 1992). It can be seen in Table 1 that etching procedure did not signi® cantly aå ect PE . It is well known that Hoechst has a cytotoxic eå ect,
400
M . Pugliese et al.
Figure 1. Visualization of tracks and cells in our experiments. (A) A V79 cell and a-particle tracks visualized in phase-contrast. (B) The same image as seen in UV light. Actual colour of the nucleus is blue. Electronic superimposition of the images has been performed by two functions: sum A + B (each pixel has the sum of the light intensities in the two images) and min(A,B) (each pixel has the minimum intensity between images A and B). A + B image is useful to identify position of the nucleus and cytoplasmatic traversals. Min(A,B) shows the nuclear traversal. Here one nuclear traversal is pointed by an arrow.
Survival to single a- particle
Figure 2. Schematic representation of an a-particle path in V79 cells in the experiment.
and even at non-toxic concentrations it can interfere with DNA repair following irradiation. Therefore the eå ects of the Hoechst stain on survival and radiation response was studied. Results on toxicity of Hoechst are again shown in Table 1. It can be seen that concentrations above 20 mg/ml are de® nitely toxic to the cells. On the other hand, concentrations of around 5 mg/ml do not signi® cantly aå ect PE nor interfere with radiation response. A histogram of measured nuclear area in a sample of 180 cells is shown in Figure 3. Distribution is approximately Gaussian with an average value of 106 mm2 and a standard deviation of 27 mm2. Speed of the cell centroid (centre-of-mass velocity) is plotted in Figure 4. Average speed of cellular Table 1. Growing surface Tissue culture plastic Mylar Mylar Plastic Mylar Plastic Mylar Mylar Mylar Mylar*
401
Figure 3. Frequency histogram of V79 cells nuclear crosssectional area. Cells from con¯ uence were plated at low density on mylar. A sample of 180 cells was analysed by ¯ uorescence microscopy as described in text.
translation was about 0´6 mm/h. Similar rate characterizes deformations. The eå ect of exposure to radiation was also determined, and it can be seen in Figure 4 that cell motility was not signi® cantly aå ected by radiation bombardment prior to the observation. 3.2. B eam parameters Incident energy on the cells was about 4´3 MeV. Energy was reduced to about 3´2 MeV by interactions with cell monolayer and mylar base. Beam energy spectrum in this position has been measured by an
Plating eæ ciency of V79 cells following diå erent treatments.
Etching in KOH 10 N (min)
a-particle radiation dose (cGy)
Hoechst concentration (mg/ml)
0 0 0 0 0 0 0 0 15 15
0 0 0 0 0 0 15 15 15 0
0 0 50 20 5 0´5 0 5 0´5 0´5
*Cells were carried at the accelerator and the medium withdrawn, as in actual irradiations.
Plating eæ ciency (%) Ô SE 82Ô 83Ô 3Ô 7Ô 77Ô 83Ô 74Ô 66Ô 72Ô 80Ô
5 7 1 3 6 4 5 4 5 3
402
M . Pugliese et al.
Figure 4. Frequency distribution of centroid speed in V79 cells growing on mylar averaged over diå erent time intervals from the ® rst observation. Sample size is 38 for control cells, and 27 for irradiated (2 Gy a-particles) cells.
Ortec Ultra-CAM detector, showing indeed a peak at 3´2 MeV with 300 keV full-width half-maximum (FWHM). Energy spectrum inside the accelerator peaked at 6´9 MeV with 50 keV FWHM. Trackaveraged LET of a-particles in V79 nuclei was calculated by Monte Carlo code. Mean value was 105 keV/mm. Measured energies were used as inputs for Monte Carlo calculations with TRIM91 transport code. The mean scattering angle in the exit window, air and medium covering the cell monolayer was about 10 mrad. Assuming a 9 mm distance between the top of the cell and the ® lm detector, data in Figure 5 show the lateral distribution in the Monte Carlo simulation. Over 95% of the tracks are within 0´5 mm from the actual point of impact in the cell. Because back-etching was used in LR-115 ® lms, a further lateral spread will be produced by traversal in the ® lm. A precise estimate is complicated by the etching procedure, where the latent track is enlarged in a conical shape. A rough estimate suggests about a doubling of displacements shown in Figure 5. Considering the diå erent contributions to impact parameter measurement uncertainty, it can be estimated that sensitivity in impact parameter measurements is in the order of 2 mm, as compared with an average nuclear and cellular size of about 12 and 20 mm, respectively. It would be possible, in principle, to calculate the absorbed dose in the nucleus per particle traversal. Yet, both nuclear size and incident particle energy
Figure 5. Monte Carlo simulation of a-particle traversals in V79 cells by TRIM91 code. In this simulation, 500 monoenergetic 4´3 MeV a-particles hit an average V79 cell (6´6 mm thick). Incident angle h on the cell is assumed to be 0 or 10 mrad.
are not constant. The absorbed dose is dependent upon the actual nuclear size, and on the impact parameter. In an ellipsoidal nucleus in track segment conditions the dose-distribution for exactly one traversal would be a triangular function (Charlton and Sephton 1991). However, cell nuclei have diå erent shapes and volumes, and energy loss is dependent upon the impact parameter. In addition, beam is not monoenergetic, and energy loss inside the nucleus produce a further spread in energy distribution. Simulated microdosimetric spectrum of local dose deposited by a particle energy beam peaked at 4´3 MeV with 200 keV FWHM is plotted in Figure 6. Dose-distribution is quite wide, with a modal value around 15 cGy. 3.3. Cell survival Results of survival assay are summarized in Table 2. Up to 11 traversals were scored in a single cell. Distributions of colony sizes are shown in Figure 7. Histograms are bi-modal for each class, suggesting that in most cases cells either live, or die after a few doublings. Therefore the classical 50-cells rule has been con® dently adopted (Puck and Marcus 1956), i.e. assumed that a survivor produces a colony with more than 50 cells, while a dead cell was scored if less than 50 cells were observed in the clone. To test whether cytoplasmatic traversals have an
Survival to single a- particle Table 2.
Figure 6. Simulated frequency distribution in V79 nuclei of radiation dose deposited by exactly one a-particle hitting the cell at an energy of 4´3 MeV. Monte Carlo calculation takes into account dependence from impact parameter, and variability in nuclear shape and beam energy.
in¯ uence on cell survival, a linear-quadratic surviving model was assumed: 2 2 S (n,T ) =PE ¯eÕ (an+b n ) ¯eÕ (cT +dT ) where n and T represent the number of nuclear and cytoplasmatic traversals, respectively. This function has a=5 ® tting parameters, namely a, b, c, d, and PE (plating eæ ciency). No interaction between nuclear and cytoplasmatic traversals is assumed. A likelihood function has been de® ned as:
403
Fate of V79 cells following traversals in the nucleus or cytoplasm.
Scored cells
Nuclear/ cytoplasmatic traversals
99 57 13 17 1 89 42 21 18 1 1 58 31 18 7 14 4 16 9 15 2 8 2 1 4 4 1 3 1 1 2 1 1
0/0 0/1 0/2 0/3 0/5 1/0 1/1 1/2 1/3 1/4 1/5 2/0 2/1 2/2 2/3 2/4 2/5 3/0 3/1 3/2 3/3 3/4 3/5 3/6 4/0 4/1 4/2 4/3 4/4 4/6 5/3 5/6 6/4
Clones >50 cells 66 26 9 12 0 36 16 10 8 0 0 12 10 3 1 2 0 3 1 2 0 2 0 0 1 0 0 0 0 0 0 0 0
Clones 50 (surviving cell) or