Figure 1.
Comparison between Simulated and Experimental Images for Different Types of Radiation
(A) Shows RIF within a nucleus traversed by high energy Fe ions (1 GeV/amu Fe).
(B) shows a nucleus exposed to photons.
Shown at the top are RIF images of nuclei taken from a microscope (DAPI in blue and DNA damage marker γH2AX in green) 5–10 min post-IR. Middle images are pRIF microscopic images at the same magnification. These pRIF images are generated by blurring DNA damage simulations for equivalent doses of radiation with the PSF of the optic used. For pRIFs, the blue channel shows the resulting nuclear density and the green channel shows DSB. If the PSF is omitted from the simulation, DSB locations can be better resolved, as shown in the gray images at the bottom of each panel (brightness proportional to the number of breaks within each pixel). Scaling bar shown in white is 1 μm wide.
Table 1.
RIF Frequencies in Experimental Data and Simulations for High-LET and Low-LET Radiation
Figure 2.
Illustration of Image Manipulation to Predict the Average DNA Damage Pattern along a Track for a Given DNA Density Profile
(A) Shows a typical image of cells that have been traversed with 1 Gy of 1 GeV/amu Fe ions. After manually selecting a region that contains a clear track, foci identification and reshuffling is done as depicted by the cartoon. Foci detection is done automatically via in-house image algorithm (see Materials and Methods).
(B) Further illustrates the mathematical approach used (i.e., Monte Carlo concept), where the probability of damage at a pixel location is proportional to the DNA density at the same location. This process is done iteratively (i.e., 50 randomizations per nucleus analyzed) to give a reasonable average break distribution. For each iteration, RIF position is determined by a probability less than that determined by DNA density (blue line).
Figure 3.
Comparison of Image Prediction and Monte Carlo Simulation for DNA Damage Distance Distribution
(A) shows the distribution of distance between consecutive foci along the track for a set of 197 simulated nuclei exposed to a theoretical 1 GeV/amu Fe track. (B) illustrates a simulated nucleus: DSB are shown in red, DNA densities and DSB blurred with the PSF in blue and green, respectively. pRIF, blurred DSB, are identified by detecting maxima along the intensity profile (C) sampled over a narrow strip of the image in (B). These profiles are obtained by computing the maximum intensity projection of a 0.8-μm thick line aligned with the particle track. Some of the distances reported for this illustrated track are also shown in (B) and (C) and labeled correspondingly in (A). The average DSB distance distribution over all 197 nuclei is shown by the red dotted curve showing an expected Poisson-like distribution. The corresponding distance distribution for pRIF is shown as the dashed green curve and is similar to the DSB distribution except for the frequency of close-by foci that have diminished (i.e., need at least more than two-pixel gap to be separate, which corresponds to 0.48 μm). We could reproduce this behavior (dark solid curve) by simply randomizing pRIF along the track using the DNA profile as a probability for DNA damage (as described in Figure 2).
Figure 4.
Comparing Theoretical and Experimental DNA Damage Pattern along 1 GeV/amu Fe Track
Average distributions of distances between consecutive foci along Fe tracks are plotted at 4.5 min and 35 min following 1 Gy exposure (blue solid lines, (A) and (B), respectively). Error bars are standard errors based on four independent experiments. For each individual track analyzed from real data, foci were counted and their positions were then randomized based on DNA profiles, described previously, to generate a theoretical distribution pattern (red dashed lines). Measuring the correlation between theoretical and experimental distributions, we observe a decrease of correlation between these two time points, from 0.6 to 0.45. These data indicate that as early as 4.5 min following exposure to radiation, foci positions deviate from a theoretical random behavior by 40% and this tendency increases over the next 30 min with a 55% loss of correlation.
Figure 5.
Spatial Foci Pattern Increasingly Deviates from Prediction Following 1 Gy of 1 GeV/amu Fe Exposure
(A) Plots the correlation between theoretical and experimental distributions of distances between consecutive foci along Fe tracks over the first hour following exposure to radiation. All DNA damage markers used (i.e., γH2AX, ATMp, 53BP1) show the same loss of correlation to randomness over time.
(B) Shows the corresponding foci frequencies, depicting a rapidly decreasing curve indicative of DNA repair.
Figure 6.
Illustration of Rdna and Rgrad Measurements
Three hypothetical foci patterns over the same nucleus are illustrated with their corresponding Rdna and Rgrad values. Upper images (A,C,E) are overlays of the DAPI image with the center of hypothetical foci (in red). Lower images (B,D,F) are overlays of the foci location with the gradient image of DAPI. The gradient operator is often used in imaging as an edge detector. To illustrate this, the green arrow in (C) delineates the contour of the edge of a bright DAPI region. One can see in the corresponding gradient image in (D) that the same contour correlates to a bright gradient region. Rdna measures the ratio of the mean nuclear intensity at the foci locations over the mean intensity of the full nucleus. Rgrad measures the same ratio on the gradient image. Because the boundary of the nuclear image creates a strong gradient intensity, a conservative contour is used for nuclear segmentation (shown in blue) to avoid an edge effect when calculating Rdna and Rgrad. In (A) and (B), foci are placed in areas of surrounding high nuclear density. The surrounding high density keeps the foci distal from areas of density change, thus we see the foci lie in low-intensity regions in the corresponding gradient image. This results in Rdna above 1 and Rgrad below 1. By manually placing foci at different locations with respect to DNA density regions, we show that Rdna is high when foci are located in bright regions of the nucleus (A) and (B); Rgrad is high when foci are located at the interface of bright and dim regions of the nucleus (C) and (D); and Rdna is low when foci are located in dim regions of the nucleus (E) and (F).
Table 2.
Reshuffling Simulated Foci Positions along Synthetic Tracks Lead to the Same Relative DNA Values as Simulations
Figure 7.
Rdna and Rgrad Computation Confirm Rapid Relocation to Dim–Bright Nuclear Interfaces with a Lower Proportion of Foci in the High DNA Density Regions after Exposure to 1 Gy of 1 GeV/amu Fe
Measured Rdna and Rgrad normalized to predicted values are graphed in (A–C). For all DNA damage markers analyzed here, all Rgrad ratios are above 1 and Rdna ratios are below one. This indicates a tendency of RIF to locate themselves at the interface between high and low DNA density regions and preferably in the low DNA density regions. This tendency is stronger within the first 10 min following exposure to radiation and statistically significant for γH2AX and 53BP1 for the first 30 min post-IR (significance is labeled by an asterisk with the number of independent experiments in parentheses, statistical test based on t-test between measured averages and predicted ones. Predictions are based on reshuffling original RIF. Based on that test, a 95% confidence interval for expected normalized ratios is shown as the gray area). For ATMp, only the earliest time point was statistically significant, indicating a return to normality much faster than the other markers. A representative nucleus 3 min post-IR is shown in (D), with γH2AX RIF appearing as a green signal and DAPI shown as blue. The white dashed arrow indicates the traversal of one Fe particle, and small solid-color arrows indicate specific RIF. The same nucleus is seen in (E) with the DAPI intensity displayed in a 3-D topographic blue surface and segmented γH2AX RIF shown as green beads. (Rendering done with Bitplane, http://www.bitplane.com/). (F) shows the same topographic view, sectioned along the particle trajectory to better appreciate the position of RIF with respect to the DAPI intensity profile. For orientation purposes, the same RIF shown with solid color arrows in (D) are shown in (E) and clearly illustrate the preferred location of RIF at the interface between high and low DNA density regions.
Figure 8.
Representative RIF Distribution Following Low-LET Radiation Exposure
DAPI is shown in blue, ATMp in green, and 53BP1 in red. (A) shows a representative 3-D image of a nucleus in orthogonal cross-section views. The same nucleus is seen in (B) as a 3-D surface rendering with only DAPI and ATMp RIF being segmented (i.e., blue and green surfaces, respectively; rendering done with Bitplane). This nucleus clearly shows the preferred location of ATMp RIFs at the interface between high and low DNA density regions. Different ATMp RIFs located at these interfaces are shown by colored arrows. (An arrow of the same color represents the same focus.) (C) and (D) overlay the identified locations of ATMp and 53BP1 RIFs (green disk and red circles, respectively) with the DAPI intensity and gradient images, respectively. Rdna and Rgrad values are given for each panel with colored text corresponding to the protein. One can see in (D) that the green RIFs shown by arrows are along high-gradient contours, which are reflected by a high Rgrad value. In contrast, the red RIF seem to locate themselves fairly randomly over the full nucleus, as reflected with Rdna and Rgrad values close to 1.
Table 3.
Experimental Relative DNA Measurements Normalized to Predicted-for Exposure for 1 Gy of γ-Rays
Figure 9.
Increased Co-Localization of DNA Damage Markers Rapidly after Exposure to Radiation
(A) is an analysis summary of γH2AX and ATMp co-localization with 53BP1 done in 3-D. Two RIFs are considered co-localized if the distance between their centers is less than or equal to 0.48 μm. As described in Materials and Methods, RIF centers are determined as the brightest pixel within each spot. The γH2AX co-localization is illustrated in (B) with representative images of different time points and different type of radiations. γH2AX and 53BP1 foci locations are shown as green disks and red circles, respectively. Co-localized centers are circled by a yellow contour on the image, and the amount of γH2AX co-localization is reported for each image. Both exposure to γ-rays (Cs) and 1 GeV/amu Fe are illustrated for the two different time periods considered after exposure to radiation (1–4 min and 5–10 min). Averages for Cs are based on two independent experiments, whereas averages for Fe are based on four independent experiments. A t-test was performed, and a p-value of 0.01 was computed between the first and second time period considered for γH2AX (statistical significance noted on the graph as *).
Figure 10.
No Decondensation of Chromatin following 5 Gy of Low LET
HeLa cells transfected with H1.2-GFP (see Materials and Methods) were imaged before and after exposure to 5 Gy of X-rays (A). Control cells that were not exposed to X-rays were also monitored (B). Intensity correlation within the nucleus (contour shown in blue on images) was computed between consecutive time points. To correct for noise which is unique to each image, correlation values were normalized to the average correlation measured for the three first time points (i.e., before radiation). These relative correlations are graphed in (C) for both groups (i.e., each group represents averages from five different nuclei). Correlation values before and after radiation (20–50 min) are shown on the graph by dashes on the bars for both control and irradiated specimen.