Live cell microscopy analysis of radiation-induced DNA double-strand break motion

Abstract

We studied the spatiotemporal organization of DNA damage processing by live cell microscopy analysis in human cells. In unirradiated U2OS osteosarcoma and HeLa cancer cells, a fast confined and Brownian-like motion of DNA repair protein foci was observed, which was not altered by radiation. By analyzing the motional activity of GFP-53BP1 foci in live cells up to 12-h after irradiation, we detected an additional slower mobility of damaged chromatin sites showing a mean square displacement of ≈0.6 μm²/h after exposure to densely- or sparsely-ionizing radiation, most likely driven by normal diffusion of chromatin. Only occasionally, larger translational motion connected to morphological changes of the whole nucleus could be observed. In addition, there was no general tendency to form repair clusters in the irradiated cells. We conclude that long-range displacements of damaged chromatin domains do not generally occur during DNA double-strand break repair after introduction of multiple damaged sites by charged particles. The occasional and in part transient appearance of cluster formation of radiation-induced foci may represent a higher mobility of chromatin along the ion trajectory. These observations support the hypothesis that spatial proximity of DNA breaks is required for the formation of radiation-induced chromosomal exchanges.

Double-strand breaks (DSBs) are considered the most critical DNA lesions induced by ionizing radiation. Even more deleterious is DNA damage generated by heavy nuclei, because it combines DSBs with additional lesions in so-called multiple damaged sites, because of the dense spacing of ionization events (1–3). Relatively little is known about the spatiotemporal organisation of repair events after the generation of DSBs, although these events are clearly decisive in determining the “fate” of the damaged cells. In fact, the interaction of DNA lesions leads to chromosomal aberrations, which are eventually responsible for early and late cellular effects. UV micropore or laser microirradiation have been recently used to generate localized DNA damage (reviewed in refs. 4 and 5). Heavy nuclei also produce highly-localized DNA damage (6), but in contrast to UV lasers the physical nature of damage generation is the same as for sparsely-ionizing irradiation (X- or γ-rays). Furthermore, the density of DNA lesions (i.e., the number of DNA breaks per unit track length) can be scaled according to the energy and charge of the particle used. These features make heavy ion irradiation a useful tool for the spatiotemporal analysis of repair processes. In addition, understanding of lesion processing following heavy ions is of great interest for hadrontherapy, where charged particles are used to treat solid cancers (7), and radiation protection in space, because heavy ions are present in the galactic cosmic radiation and represent a major risk for human space exploration (8, 9).

Most of our current knowledge is derived from studies on fixed cells and immunostaining of DSB-flanking γ-H2AX histone or other damage response proteins (10–14). Recently, a limited “movement” of DNA DSBs associated with chromatin decondensation and protrusion of the DSBs into low-density chromatin was described to occur after γ-rays or local UV-laser irradiation (15, 16). Furthermore, Aten et al. (11) discussed the formation of repair clusters after α-particle irradiation, which requires a substantial motion of damaged chromatin sites.

In contrast, experiments using focused ultrasoft X-rays revealed apparently immobile DSBs (17), and our recent data of pattern changes of γ-H2AX foci along heavy ion tracks in fixed human cells point to a stable positioning of DSBs (14).

However, data derived from fixed samples have severe limitations for tracking the dynamical behavior of repair events. Images of fixed and immunostained samples offer only a static view of a selected point in time, thus making the analysis of dynamic changes difficult. To circumvent these limitations, we elected to use live cell imaging covering time regimes from a few minutes up to several hours after irradiation. We used human osteosarcoma cells (U2OS) stably-expressing GFP-tagged 53BP1. 53BP1 is one of the components of the genome surveillance network activated by DNA DSBs via phosphorylation by ataxia telangiectasia mutated (ATM) protein (18). After irradiation, 53BP1 is readily recruited to the surrounding of DSBs and binds dimethylated lysine 79 of histone H3 via its Tudor domain (19). 53BP1 had been shown to colocalize with γ-H2AX-stained chromatin domains along heavy ion tracks (13, 14). The same osteosarcoma cell line (U2OS) stably-expressing GFP-tagged NBS1 (a member of the DSB processing MRN complex; ref. 20) or HeLa cells expressing Aprataxin-GFP (21) were used for the short time mobility measurements at the beamline. These proteins were chosen because they are rapidly recruited to sites of DNA damage after irradiation.

Results

Fast Confined Movement of Damaged Chromatin Within Minutes After Irradiation.

For the short time mobility analysis, we selected the GFP-tagged proteins Aprataxin and NBS1, that had shown a fast accumulation at sites of ion impacts reaching maximum intensities ≈1 and 2 min after irradiation, respectively. In addition, the selected proteins showed a more pronounced and therefore better traceable formation of ionizing radiation-induced foci (IRIF) at ion-induced lesions compared with the also fast accumulating DNA-PK_cs (6). These facts made these proteins especially suitable for following up the early kinetics of the damaged DNA. Time-lapse imaging starting shortly before irradiation showed motion of IRIF already 5 s after irradiation. In Fig. 1, the motion of individual NBS1 or Aprataxin foci is visualized with colored trajectories. Note that each “focus” in Fig. 1 actually corresponds to a short track perpendicular to the cell plane including several IRIF and comprising varying numbers of DNA lesions. An estimate of the density of the DNA DSBs is given in Table 1 based on the number of DSBs/Gy from ref. 22. No major radiation-induced dislocation of damaged chromatin could be observed, with the foci showing only very limited spatial transitions. For that reason, the superimposed trajectories appear as a spot at this spatial resolution during the first 5 min (Fig. 1 A–C). The observed fast dislocation was in the order of <0.5 μm, confirming previous results (5, 23). The motional extent did not depend on the chosen protein because Aprataxin and NBS1 (Fig. 1 A–C) showed a very similar behavior. There was also no significant difference in foci movement for the different tumor cell lines used (U2OS or HeLa). During analysis of this small-scale displacement in more detail up to 30 min, the path of the dislocation of the damage sites marked by NBS1 showed quivering motion (Fig. 1D). This behavior resembled a fast nondirected confined random motion. The dislocation covered a space of ≈0.5 μm in diameter (Fig. 1D, red trajectories) with jumps of ≈0.075 μm between each frame (10 s). The extent and path of motion were more clearly visualized by enlarging a sample trajectory (Fig. 1E). The motional trajectories of radiation-induced foci of NBS1-GFP were similar in respect to confinement and speed to those of observable preirradiation aggregates in the same nucleus (Fig. 1D, green trajectories, and F). In addition, similar values have been described in the literature for the motion of undamaged chromatin locations (24, 25).

Fig. 1.

Damaged chromatin sites show a fast constrained motion during the first minutes after irradiation. Live cell imaging directly at the beamline allowed measuring the motion of radiation-induced foci at sites of ion traversals (perpendicular to the cell layer). Time-lapse imaging started shortly before irradiation. Motional tracks were recorded 5 s after irradiation up to 30 min. (A–C) The small-scale spatially confined quivering movement of NBS1-GFP or Aprataxin-GFP foci (superimposed motional tracks; different colors and numbers represent individual measured trajectories) in U2OS or HeLa-cells, respectively, during a 5-min observation time (one frame every 5 s) after the irradiation with charged particles. Motion of foci was within 0.5 μm in all cases (A: U2OS NBS1-GFP; Sm 10290 keV/μm; B: U2OS-NBS1-GFP; Cr 2630 keV/μm; and C: HeLa Aprataxin-GFP; U 14,300 keV/μm). (D) Motion of NBS1-GFP foci in U2OS cells during a time course of 30 min (1 frame every 10 s) after the irradiation with Ca ions (1,850 keV/μm). Superimposed red trajectories indicate movement of radiation-induced foci; green trajectories indicate movement of preirradiation aggregates. (E and F) Enlargement of 2 trajectories of D (marked by circle) showing the quivering nondirected motion on a scale <1 μm. Motion of the whole nucleus was compensated by alignment of the center of intensity (indicated by the blue dot-like trajectory in D). No significant difference in motional scale could be observed between the traversal-induced foci containing DSBs and preirradiation aggregates.

Table 1.

Irradiation parameters

Ion	Energy, MeV/n	LET, keV/μm	Dose, Gy	DSBs/ μm
48Ca	7.4	1,850	5.9	24
52Cr	6.5	2,630	8.4	33
59Ni (low-angle irradiation)	6.0	3,430	11	45
150Sm	4.2	10,290	33	132
238U	3.0	14,300	46	184

Particle fluence was 2 × 10⁶ particles/cm² in all experiments. Calculation of DSBs per μm of track length is based on the assumption of 35 DSBs/Gy (22) and a nuclear volume of 500 μm³.

Slow Migration of Damaged Chromatin.

To facilitate the dynamic analysis of damaged chromatin domains, we generated long damage streaks in cell nuclei by using low-angle Ni-ion irradiation as described by Jakob et al. (10).

The use of live cell microscopy in a temperature-controlled Focht chamber and iterative revisiting of selected fields at a specialized microscope remote from the irradiation facility allowed the frequent observation of ion trajectories inside the cell nuclei for several hours, limited mainly by cell migration. 53BP1-GFP showed irradiation-dependent foci formation along the ion trajectories in living cells (representative images in Figs. 2A, 3A, and 4) reaching a maximum intensity within 10 min. The streaks in the live cell experiments clearly showed a nonhomogeneous, punctuate pattern of 53BP1 foci, similar to fixed and immunostained samples (14). At a linear energy transfer (LET) of 3,430 keV/μm, a traversing Ni-ion is expected to produce ≈45 DSBs per μm of track (Table 1). The presence of a gap structure after Ni irradiation in living cells thus confirms that this pattern is not introduced by fixation artefacts, but represents a biological phenomenon.

Fig. 2.

Motion of damaged chromatin sites inside individual trajectories generated by low-angle Ni irradiation (LET 3,430 keV/μm). Live cell imaging of 53BP1-GFP in U2OS cells in a Focht chamber allowed measuring the motion of radiation-induced foci along ion trajectories >12 h. (A) Representative images of 53BP1-GFP expressing U2OS nuclei traversed by Ni-ions 24–30 min after irradiation. Along the sites of ion trajectories, accumulation of foci can be observed forming a streak pattern. (B) Individual motional trajectories of foci inside these nuclei are represented by different colors and numbers and are superimposed on the image shown in A. These overlays indicate the course of individual foci inside an ion track during the observation time of 12 h. In most cases only moderate displacements were observed, indicating the quite stable positioning of damaged chromatin. (C) Magnification of the motion of individual foci inside the traversal-induced streaks (Left) of a typical nucleus (Right) during the 12-h observation time. The tracks show a largely nondirected motion with a slight preference along the particle path. The overall displacements were in the range of 2 to 4 μm (12 h).

Fig. 3.

Quantitative analysis and comparison of the motion of DSBs after low and high LET irradiation. (A) Time-dependent changes of a single ion-induced 53BP1-GFP streak showing the typical motional behavior of individual foci along the trajectory over the time course of 12 h after irradiation. From the displacements of individual foci the MSD values were calculated after correcting for nuclear translocation and rotation. (B) Calculation of the MSD from the experiments described in Figs. 2 and 6. Motion of 140 or 60 foci were analyzed for the Ni (○) or X-ray (●) experiment, respectively. Error bars represent the 95% confidence interval. Calculation of the linear fit gives values of 0.55 ± 0.04 μm²/h for high LET nickel and 0.69 ± 0.04 μm²/h for X-rays. The data points used for the calculation of the MSD show a slight bending in the case of the Ni ion irradiation at later times, potentially indicating a larger-scale confinement set by chromosomal territories or the nuclear volume itself. If we restrict the values for the nickel experiments to the time frame of the first 6 h (like the X-ray-data), the linear fit (dashed line) yields a MSD of 0.64 ± 0.08 μm²/h, very similar to the X-ray data.

Fig. 4.

The migrational activity of damaged DNA domains after irradiation covers a broad range in individual nuclei. Samples of 53BP1-GFP-expressing U2OS cells displayed as consecutive images over time (12 h) after low-angle Ni irradiation. (A) Rare case of observed larger displacements of 53BP1-GFP foci without major deformations of the whole nucleus or changes in nuclear morphology. The long-range migration from a peripheral site to the interior of 1 single spot is marked by the yellow trajectory. Other foci like the one marked in red remained stably positioned. Variations in signal intensities of the tracked spot are due to migration out of the focal planes. (B) This particular nucleus shows only very minor changes of the short streak of Ni-induced 53BP1 foci. Both the absolute position and the internal pattern are maintained during the observation time.

The 53BP1 streaks could be identified beside a substantial migration (and partial rotation) of the whole cell nucleus, which was compensated by rigid body transformation (see SI Text), indicating a general positional stability of the foci. However, substantial motion of substructures could be frequently monitored (Fig. 2C). For each analyzed focus of a trajectory in a nucleus, a number and a color were assigned as indicated in Fig. 2B (colored overlays). The foci in these trajectories yielded a displacement of <3 μm during the first 10 h of observation for >80% of the observed foci and only ≈2% of the foci displayed a migration exceeding 5 μm. The behavior of a typical ion-induced 53BP1 streak is displayed in Fig. 3A, showing a moderate displacement of individual foci in the range of 2 to 3 μm during the 12-h observation time. The mean square displacement (MSD) of the damaged chromatin sites calculated by linear regression of the displacement vectors was 0.55 ± 0.04 μm²/h in the U2OS cells (Fig. 3B). From this MSD, a diffusion constant of ≈0.31 × 10⁻¹² cm²/s can be estimated. Larger displacements or significant compaction events were generally accompanied by morphological changes of the shape of the nucleus, leading to an internal rearrangement or a loss of observed foci during the observation period (Fig. S1). Nevertheless, in some rare cases, large-scale motion of damaged sites could be observed without dramatic changes of the nuclear outline (Fig. 4A). In this particular case, a single spot from the peripheral region moved to the interior over a distance of ≈7 μm (Fig. 4A, yellow trajectory), whereas other foci (e.g., Fig. 4A, red trajectory) remained quite stable. The intensity variations of these tracked foci are caused by motion in z-direction and a slight focal drift. Other nuclei, as the one displayed in Fig. 4B, showed no variations during the observation period, with only very minor changes in both the nuclear positioning and appearance of the 53BP1 signal.

Repair Clusters.

As no large displacement of damaged chromatin domains were generally observed after heavy ions, we tested whether merging of nearby foci would occur (“repair clusters” or “repairosomes”). The living U2OS cells showed no general tendency to group the individual damaged domains into repair clusters (Figs. 2C, 3A, and 4). Nevertheless, a trend for an enhanced migration along the ion trajectory could be observed (evident in Fig. 2C, lower track). This behavior led in some cases to the observation of cluster formation (Fig. 5 and Fig. S2). In the selected nucleus of Fig. 5A, the development of a more compact form of the 53BP1 aggregations resembling the formation of repair clusters could be observed. In the nucleus in Fig. 5B, after a transient compaction (e.g., 3–9 h) the foci pattern turned back to more or less the original streaks of the ion trajectory (Fig. 5B, most obvious at 11 h; streaks marked by arrows). Thus, the clustering/merging of distinct foci along the tracks was only apparent during the intermediate times in this particular nucleus. These observations indicate a higher mobility of damaged chromatin sites along the ion trajectory compared with the perpendicular (radial) direction.

Fig. 5.

Motion of damaged chromatin sites inside individual Ni-ion trajectories points to a preferential motion within the track extension. 53BP1-GFP in U2OS cells after low-angle Ni-ion irradiation are displayed as consecutive images over time. Besides the preservation of the general local nuclear positioning of the streaks, motion could be observed affecting the track structure during the 12-h observation time. (A) In addition to the separation of focal structures in some cases this leads to the visualization of clustering. (B) A transient condensation of the 53BP1 foci resulting in the deformation of the original streaks (arrows at 30′) into compact structures (3–9 h). After 9 h a decondensation phase is apparent, establishing again a streak pattern (arrows at 11 h) very similar to the original one.

Extent of Motion of Damaged Chromatin Does Not Depend on DSB Density.

To assess the influence of DSB density on the mobility of damaged chromatin domains, we compared densely-ionizing Ni-ions (Table 1) to sparsely-ionizing X-rays. After 2 Gy X-rays, most radiation-induced 53BP1 foci gained intensity during the first 20–40 min after irradiation but, in contrast to ion-induced foci, they were not yet clearly visible at the first time point (10 min) (Fig. 6B). Analysis of the motion of selected X-ray-induced 53BP1 foci yielded a similar dynamical behavior (Fig. 6; motion of individual foci indicated by superimposed colored trajectories) compared with the irradiation with low-energy Ni-ions (Fig. 2). Nevertheless, loss of foci, most probably caused by repair, reduced the number of foci observable over longer times after X-rays. Notably, MSD (0.69 ± 0.04 μm²/h; Fig. 3B, solid line) and diffusion coefficient (0.38 × 10⁻¹² cm²/s) for X-rays were also similar to the heavy ion data, especially if we restrict the analysis to the first 5–6 h (Fig. 3B, dashed line), as in the X-ray experiment. In this case a MSD of 0.64 ± 0.08 μm²/h and a diffusion constant of ≈0.36 × 10⁻¹² cm²/s were calculated for the Ni ion data.

Fig. 6.

Motion of damaged chromatin after sparsely ionizing radiation. Live cell observation of 53BP1-GFP in U2OS cells after irradiation with 2 Gy X-rays imaged >5 h after irradiation incubation. (A) Superimposed trajectories on the images of representative nuclei (20 min after irradiation) indicate the course of individual foci during the observation time. In most cases only a moderate displacement was observed indicating the quite stable positioning of DSBs. Displacements were similar to the ones observed after high LET irradiation. (B) Consecutive images of 1 nucleus (A, *) showing the slower accumulation of 53BP1 after X-rays that is hardly visible 10 min after irradiation (first image). Later, most foci disappear because of ongoing repair processes, thus restricting the observation time.

The results for the motional analysis obtained in this live cell study do not depend on the selected 53BP1-GFP protein construct. Similar results could be obtained by using NBS1-GFP-expressing U2OS cells after irradiation with krypton ions (LET 5,100 keV/μm; Fig. S2), thus confirming that the experimentally-addressed motion is associated with the motion of chromatin sites damaged by irradiation.

Discussion

Current knowledge on chromatin mobility has been recently reviewed (26), but still leaves several open questions regarding the dynamical behavior of damaged chromatin domains. As shown by our group and others (10–13, 27), traversals of charged particles lead to the localized accumulation of several repair-related proteins and the formation of phosphorylated H2AX in the nucleus, referred to as IRIF. Charged particles produce similar biological damage as X- or γ-rays, but the lesion density increases by increasing the LET (3, 8): the resulting multiple damaged sites are thought to be involved in misrepair and represent the main lesions leading to early and late radiation effects (2, 9).

As the quantification of small spatial changes is difficult to address in static snapshots obtained from fixed cells because of the inter- and intranuclear variability of observable patterns, here, we used live cell microscopy as a more direct approach for visualizing and measuring the dynamics of damaged DNA sites after charged particle and sparsely-ionizing irradiation. We directly visualized the stable positioning of damaged DNA sites by observation of foci of the GFP-tagged repair related proteins Aprataxin and NBS1 (Fig. 1), during the first seconds and minutes after ion impacts. The lack of a long-range motion up to 30 min postirradiation (Fig. 1 D–F) for different proteins indicates that this result does not depend on the particular chosen protein, but rather represents a general feature of the damaged chromatin site, where the proteins are bound to. Noteworthy, the small range of motional activity observed allows only for a rather small extent of chromatin remodeling during repair, like a local decondensation as described (15, 16). However, we could observe a very fast spatially-confined motion on a scale of <1μm, most likely representing the normal Brownian-type motion of undamaged chromatin (25, 28, 29), but demonstrating that our experimental system is able to detect and resolve this range of small dynamical responses. A similar fast confined and radiation-independent motion was observed in preirradiation aggregates of NBS1-GFP (Fig. 1F), which could not be discriminated from irradiation-induced foci in terms of motional behavior. These results are in agreement with our earlier observations (23), where this type of fast radiation-independent confined quivering motion was also monitored by fluorescently-marked DNA sequences (scratch replication labeling) at both putative sites of ion traversals and more remote areas of the nucleus. They also support recent reports (15, 30) in mouse embryonic fibroblasts exposed to γ-rays. Imaging of multiple 53BP1-GFP foci in close proximity, frequently interacting and separating again, yielded a MSD of ≈0.9 μm²/h during the 50-min observation time (15). In another study, the mobility of telomeres was found to be unaffected by irradiation (30).

The streaks of particle-induced foci in the live cell experiments are similar to those observed for endogenous 53BP1 in fixed and immunostained cells (14), indicating a physiological response of the 53BP1-GFP construct. In that previous study, despite a very similar appearance of traversal-induced streaks at a short time after irradiation, different cell types (normal human fibroblasts and HeLa cells) revealed differences in the spreading and diversity of patterns during longer incubation times, with fibroblasts showing a greater positional stability of the damaged DNA. Using live cell imaging, we were now able to quantify this motion in U2OS cells, yielding MSDs and diffusion coefficients that were similar for densely and sparsely ionizing radiation, and within the range of diffusion constants found in the literature for undamaged chromatin (0.1 to 0.6 × 10⁻¹² cm²/s for HeLa and neuroblastoma cells) (31, 32). Recently, Wiesmeijer et al. (33) described the mobility of nondamaged chromatin of U2OS cells measured by photoactivation of Pa-GFP H4. The maximal dispersion after 2 h was described to be ≈2 μm and chromatin with different degrees of folding showed equal mobility. In addition, no significant differences in chromatin movement between cells in G₁, S, and G₂ were observed (33). Our data are in full agreement with these results and point to a nonaltered migration behavior of damaged chromatin after ionizing irradiation. Occasionally, we observed foci displacements after Ni irradiation exceeding 5 μm in the 12-h interval, but those events occurred in only <2% of the observed foci. We have to point out here that because of the restriction of the optical resolution we are not able to measure the motion of individual DSBs inside a single stained Mbp domain represented by a focus. Nevertheless, this type of movement would not contribute significantly to the long-range motional activity of the whole damaged domain. Analyzing the changes in the 53BP1 streaks formed along the trajectories of the traversing ions, we can draw the conclusion that DSB processing is not associated with a directed transition of damaged sites over a larger distance. This finding is in accordance with earlier data showing that after γ-irradiation the majority of foci stayed near sites of their origin and did not noticeably move (16). Whether the slightly-enhanced motion along the ion trajectory observed in our experiments (Fig. 2C) was induced by 53BP1 binding, as recently reported for unprotected telomeres (30), remains to be elucidated.

The lack of evidence for large-scale movements of DSBs in live cell imaging is consistent with the current models of nuclear architecture and formation of exchange-type chromosome aberrations, which require interaction of distinct DSBs (34). The irradiated cell nucleus was erstwhile regarded as a “bag of broken chromosomes” with the severed ends free to move around and find partners with which to form illegitimate reunions (35). This view was challenged by the studies on nuclear architecture, providing evidence that chromosomes occupy localized domains with limited movement (e.g., ref. 36). Moreover, the increased translocation frequency involving genes that are in close proximity in the interphase nucleus, such as the RET/PTC inversion in thyroid cells (37), lend support to the so-called “contact-first” hypothesis, where it was supposed that the damaged chromatin must colocalize at the time of exposure for the exchange to occur (reviewed in ref. 38). Our results in live cells show that the DNA breaks produced by ionizing radiation have limited diffusion in the nucleus. Is this consistent with the formation of radiation-induced chromosomal rearrangements? In fact, under the assumptions of limited movements and small interaction distance of DSBs, mathematical models of chromosome aberration formation are able to reproduce accurately the experimental data obtained after exposure to both sparsely-ionizing radiation (39–42) or heavy ions (43, 44). Besides, recent data on the human interphase chromosome territories suggest that the intermingling of different domains may be much higher than previously thought (45). Therefore, increased mobility of the damaged chromatin is not required to reproduce the dose- and radiation-quality dependence of the yield of chromosomal exchanges introduced by ionizing radiation.

Materials and Methods

Cell Culture and Irradiations.

Human osteosarcoma cells (U2OS) stably expressing 53BP1-EGFP were grown on round glass coverslips (40-mm diameter) submersed in 60-mm Petri dishes at 37 °C with 100% humidity and 5% CO₂ in DMEM (Biochrome) supplemented with 10% FCS and 1% penicillin/streptomycin. Cells were regularly checked to be free of mycoplasma contamination. For the measurement of protein recruitment and short time dynamics at the beamline microscope U2OS cells stably expressing tandem GFP-tagged NBS1 or HeLa cells expressing Aprataxin-EGFP were used. These cells were cultivated on polycarbonate foil (18-mm diameter; 40-μm thickness) and irradiated at the beamline microscope as described (23). The irradiation was performed at the accelerator facility of the GSI Helmholtzzentrum für Schwerionenforschung with low-energy heavy ions and a fluence of 2 × 10⁶ particles/cm² (Table 1).

Live Cell Imaging.

Live cell observation after irradiation was done in a temperature-controlled Focht chamber. Motional analysis was done in ImageJ (http://rsb.info.nih.gov/ij). Briefly, live cell imaging datasets were loaded into image5D macro (Joachim Walter, BioImaging Zentrum, Universität München, Planegg, Germany) and maximum z projection was performed. Details are described in SI Text.