High Molecular Crowding in Repair Foci Surrounding DNA Breaks, Measured by Fluorescence Lifetime Imaging Microscopy

ABSTRACT

The density of mammalian cells is determined primarily by the protein content. Local concentration of proteins in a cell is tightly controlled and varies between the cytoplasm, nucleoplasm, and nucleoli. We demonstrate that repair foci that are formed in response to DNA breaks are much more densely packed with proteins than the surrounding nucleoplasm. Using fluorescence lifetime imaging (FLIM), we demonstrated that the local concentration of all proteins (i.e., the residing and recruited ones) in double‐ and single‐strand DNA repair foci can be even 2.2 times higher than that in the surrounding nucleoplasm, which brings them close to the achievable maximum concentration. The highest protein density is found in the center of a repair focus and gradually decreases with distance from the DNA lesion. We hypothesize that a microenvironment characterized by such a high protein concentration may facilitate the formation of protein condensates, resulting in the stabilization of repair complexes.

Induction of a DNA break leads to recruitment of repair proteins and formation of a repair focus. Here, one of the recruited DNA repair proteins (53BP1) is tagged with eGFP and serves as a Fluorescence Lifetime Imaging Microscopy (FLIM) probe to demonstrate that the local concentration of all proteins residing in DNA double‐strand break repair foci can be even 2.2 times higher than in the surrounding nucleoplasm. This protein‐dense local environment may stabilize the DNA repair complexes.

Abbreviations

2D, 3D

2‐dimensional, 3‐dimensional

DDR

DNA damage response

DSBs

double‐strand breaks

eGFP

enhanced Green Fluorescent Protein

EtOH

ethanol

FA

formaldehyde

Fc

DNA repair focus

FCS

fluorescence correlation spectroscopy

FLIM

fluorescence lifetime imaging microscopy

FP

fluorescent protein

NER

nucleotide excision repair

PARP

poly (ADP‐ribose) polymerase

SSBs

single‐strand breaks

1. Introduction

Induction of DNA breaks activates a succession of events, beginning with the recognition of a DNA lesion, followed by recruitment of repair factors [1, 2], dismantling of nucleosomes [3] and the rearrangement of higher‐order chromatin structures at the damage site [4, 5]. The nuclear region surrounding the lesion, which becomes rich in recruited repair factors, is referred to as the DNA repair focus [6, 7]. In recent years, not only DNA damage response (DDR) proteins, but also noncoding RNAs have been shown to play a role in the formation of repair foci [8, 9], and may be involved in the repair process. Despite the seemingly stable nature of repair foci (DSB repair foci may last for hours, and SSB for several minutes or longer), repair proteins recruited to DNA damage are dynamic and continuously enter and leave a repair focus while the repair takes place.

In microscopy studies of repair processes, a DNA repair focus is usually detected by imaging only one of several types of recruited proteins, for example, XRCC1 for single‐strand breaks [10] or 53BP1 for double‐strand breaks [11]. In the case of DSBs, another effective way to visualize the repair focus is the detection of the phosphorylated histone H2A.X (γH2A.X). The phosphorylation of histone H2AX extends along the DNA molecule, on both sides of the DSB, and can reach hundreds of thousands of molecules; therefore, such repair foci can be readily detected by immunofluorescence.

53BP1 and XRCC1, which are at the center of this study, are recruited to DNA lesions in a large number of copies. 53BP1 functions as a scaffold protein that forms oligomers surrounding DSBs, and as a repair protein that is responsible for inhibiting DNA resection in a break [12]. XRCC1 is involved in SSB repair, has no known enzymatic activity, and functions as a scaffold protein [13].

It is reasonable to expect that the numerous repair factors that are recruited to a damaged site transiently create a distinct local microenvironment surrounding a DNA lesion. Several reports suggested that the local concentration of the recruited proteins was sufficiently high to induce a liquid–liquid phase transition [11, 14, 15], although their concentration was not measured. A high local concentration of repair factors may be a key factor in the mechanism of the DNA damage response, since it may promote repair by stabilizing the repair complex and accelerating enzymatic reactions.

Thus, an intriguing question arises as to the actual total concentration of all proteins residing in a repair focus and the role of high molecular crowding in creating an environment conducive to effective repair. We describe live cell measurements of local concentrations of all proteins that reside in the repair foci that were formed in response to single‐ and double‐strand breaks. The local protein concentration was measured using fluorescence lifetime imaging microscopy (FLIM). We exploited the fact that there exists an inverse quadratic relationship between the fluorescence lifetime of a fluorescent protein, such as eGFP or tdTomato, and the local refractive index (RI, n), which, in turn, depends on the local concentration of all proteins of all types residing in close proximity (see Supporting Information for a detailed description of the method).

We used an eGFP tag attached to 53BP1 or XRCC1 as a fluorescent probe that reports local protein concentration in the DSB or SSB repair foci and the surrounding nucleoplasm. In other words, we used molecules of a protein of one class, out of many types of proteins that reside in the nucleoplasm and within a repair focus, as a probe reporting the local concentration of all proteins that reside near the DNA break. The principle of such a measurement is shown in Figure 1. In this text, for brevity, we refer to “all‐protein” concentration, which refers to all proteins of all types present in the subnuclear region under scrutiny.

FIGURE 1.

Schematic representation of a FLIM‐based technique to measure the local concentration of all proteins present in close proximity to a GFP fluorophore. The fluorescence lifetime of GFP (fused to one type of repair factor) that interacts with all proteins in this region depends on their total concentration.

2. Materials and Methods

2.1. Cell Culture, Transfection, and Fixation

HeLa cells were cultured in Dulbecco's Modified Eagle Medium (DMEM, Sigma, Pozna, Poland) supplemented with 10% fetal calf serum (FBS, Gibco) at 37°C in a humidified atmosphere containing 5% CO₂. Cells were seeded on 18 mm diameter coverslips (Menzel Glaser, Germany) and allowed to adhere for 24 h. After 24 h, cells were transfected with the eGFP‐53BP1 or eGFP‐XRCC1 plasmid using FuGene 6 transfection reagent (Roche; Basel, Switzerland) according to the manufacturer's instructions and cultured for another 24 h. Subsequently, live cells were imaged in DMEM/F12 (Sigma) supplemented with 2% FBS.

In some experiments, live cell imaging was followed by fixation of the imaged cells, and the same cells were fixed on the microscope stage using three different fixation methods. First, the cells were fixed with 4% formaldehyde solution for 30 min. In the second and third experiments, cells were first fixed with 4% FA for 30 min and then cold EtOH was added for 3 to 5 min to permeabilize the cell membranes and thus improve the accessibility of the cell interior to the BSA solution. Following fixation, the samples were washed with PBS three times and then left on the microscope stage for at least 20 min to reach equilibrium with the environment in the stage microincubator prior to imaging.

2.2. DNA Damage Induction

A focused beam of blue laser light, which is routinely used to image cells under a confocal microscope, was used to induce controllable DNA damage at a selected site in the cell nucleus, as described previously [10, 16]. Unlike other methods, which are frequently used to induce DNA damage (e.g., genotoxic drugs or ionizing radiation), the laser microirradiation technique allows the induction of DNA lesions at predetermined sites within the cell nucleus with high spatio‐temporal resolution. The dose of energy and thus the level of damage was easily controlled; therefore, the induction of individual lesions, or small or large clusters of DNA breaks, was at the discretion of the experimenter. Furthermore, no exogenous photosensitizers were used (as in [16]); therefore, photodamage inflicted on cells during time‐lapse imaging of live cells during the repair process is minimized.

Cells growing under physiological conditions on coverslips were mounted in a custom‐made steel holder, which was placed on a stage of the Leica TCS SP5 confocal microscope (Leica Microsystems, Germany) equipped with an incubation chamber (The Cube, Life Imaging Services, Switzerland). The Wizard implemented in LAS AF was used to expose a selected spot in the cell nucleus to a focused light beam. DNA double‐strand breaks were induced by 458 nm wavelength light (argon ion laser) parked at a defined position in the cell nucleus, away from the nucleoli. In all experiments, the total dose of energy delivered by the laser beam to the illuminated region was 400 or 80 μJ for DSB or SSB induction, respectively. This resulted in the induction of a few DSBs or SSBs within no more than two hundred nanometers (in the xy and z directions) from the center of the light beam.

2.3. Fluorescence Imaging

Fluorescence images were recorded using a Leica TCS SP5 II confocal laser scanning microscope equipped with a 63× Plan Apo oil immersion lens (NA 1.4). Imaging parameters were as follows: excitation at 488 nm (Argon laser), detection at 590 to 610 nm, pinhole set at 1 Airy disc, and PMT gain at 700 to 800 V. The images were acquired at 512 × 512 resolution of 512 512 at a scanning speed of 200 Hz with up to 4 frames averaged. Fluorescence intensity measurements were analyzed using ImageJ (NIH) software [17]. 3D images were deconvolved using Huygens Deconvolution & Analysis Software (Scientific Volume Imaging B.V., Hilversum, the Netherlands).

2.4. FLIM Imaging

FLIM images were recorded using a Leica TCS SMD SP5 scanning confocal microscope coupled with a PicoHarp 300 TCSPC module (PicoQuant, Berlin, Germany) controlled by SymPhoTime32 software (PicoQuant). A 470 nm pulsed laser (LDH‐P‐C‐470B, PicoQuant) operated at 40 MHz was used as the excitation source. The imaging was carried out with a 63× oil immersion objective (numerical aperture NA = 1.4) and a scanning speed of 200 Hz. The emission was collected through an ET525/25 bandpass filter on a SPAD detector. To avoid significant pile‐up errors in all our measurements, the peak count rate never exceeded the recommended 10% of the excitation rate. Signal acquisition time was ~1 to 2 min per image. The TCSPC data analysis was performed using the SymPhoTime64 software package (PicoQuant) as described in “Image processing”.

2.5. Image Processing

As the fluorescence decay of eGFP in most cases can be fitted with a single exponential function with high precision (a reduced χ ² value close to 1–1.4), a monoexponential model was used for lifetime calculations. To ensure that by choosing a single‐exponential fitting of our data (rather than biexponential fitting), the output results will not be distorted, we carefully compared the results of single‐ and biexponential fitting (Supporting Information Text, Figure S6, Table S1).

To check the background autofluorescence level in live cells, control experiments with nonlabelled samples were performed (Figure S7). The maximum autofluorescence signal detected in the cytoplasm was 30 to 40 cs/pixel, and that in the cell nucleus was 10 to 15 cs (Figure S7). Thus, during data analysis, the threshold was set to 40 cs to eliminate autofluorescence background contributions in all datasets (Figure S8A). We did not observe any correlation between the measured lifetime and the signal intensity (Supporting Information Text, Figure S2). The spatial binning factor was set to 2 to obtain sufficient photon counts to fit the lifetime decay curve for each pixel. Intensity‐based segmentation was applied to separate ROIs during the calculation of the averaged lifetime value within the repair focus and the rest of the nucleoplasm. In this approach, the intensity threshold was set to pick up only the signal corresponding to the repair focus or nucleoplasm (Figure S8A). For the lifetime calculation within the center of the repair focus and its periphery, the ROI was manually selected based on the lifetime distribution obtained from intensity‐based segmentation (see, e.g., Figure S8B). The instrument response function was automatically determined by the software. Subsequently, the fluorescence lifetime was calculated for each pixel in the image by a single exponential fitting, and the lifetime map was generated using SymPhoTime64 software (PicoQuant). To facilitate image analysis, the brightness levels were adjusted to show the distribution of fluorescence lifetimes at the sites with the highest photon count, for example, a repair focus.

All experiments were repeated at least three times to ensure reliability and consistency. Data from 62 cells were acquired and analyzed in experiments involving DSB repair foci, while data from 20 cells were used for SSB repair foci analysis.

2.6. FLIM for Monitoring All‐Protein Concentration

In the FLIM technique, the fluorescence lifetime (τ) is determined as a function of the radiative and nonradiative pathways of the fluorophore transition from excited to ground level. The radiative rate constant is known to be a function of the local refractive index (n) [18, 19]. According to the Strickler–Berg equation [18], τ ~ 1/n ². This inverse quadratic relationship between the fluorescence lifetime and the local refractive index of the eGFP microenvironment was previously exploited to develop a FLIM‐based approach that enables quantitative monitoring of changes in protein concentration in live cells [20]. Briefly, in this approach, the alteration in the fluorescence lifetime of eGFP is employed for measuring changes in the all‐protein concentration in the local environment of eGFP. According to the calibration performed in [20], all tested eGFP fusion proteins demonstrated the same response to changes in protein concentrations in their local environment, for example, an ~100 ps fluorescence lifetime of eGFP corresponded to an increase in local protein concentration of ~150 mg/mL [20]. In particular, the proteins studied in [20] were at most 576 amino acids long, while the 53BP1 protein is a larger protein consisting of 1972 amino acids. Therefore, to verify that the above‐mentioned calibration is also valid for the 53BP1 repair protein, we performed a control experiment according to the protocol described previously [20]. Cells expressing eGFP‐53BP1 were fixed in 4% FA/PBS for 15 min, permeabilized with 0.5% Triton X‐100, and immersed in 150 mg/mL of BSA/PBS (see Figure S3A). This assay immobilizes fluorescent proteins in cells and removes membranes to allow the BSA to reach the cell interior. This experiment confirmed the expected reduction in the fluorescence lifetime of eGFP by 100 ps after the addition of 150 mg/mL of bovine serum albumin solution (Figure S3B,C).

3. Results

3.1. Induction of DNA Double‐Strand Breaks and Formation of Repair Foci

We used a focused laser beam parked in a selected spot in the cell nucleus to induce DNA breaks [16] in HeLa cells expressing eGFP‐53BP1. Induced damage consisted of several DNA breaks located within a region of approximately 300 × 300 × 600 nm [21] and resulted in recruitment of eGFP‐53BP1 and the formation of microscopically detectable, subnuclear repair foci [12, 22, 23, 24]. Fluorescence intensity and lifetime images were recorded 15 min after damage induction, when a detectable repair focus was formed (Figure 2A–C and Figure S1).

FIGURE 2.

Induction of DNA DSBs by a laser beam and recruitment of eGFP‐53BP1 to the damaged region (A, blue lightning bolts). In some cells, 53BP1 was distributed homogeneously (except in nucleoli), as in panel A, cell 1, while in other cells several 53BP1 foci were present, most likely representing endogenous DNA damage (as in A, cell 2 and 3). Within 15 min, in both cell types, repair foci were formed at the sites of exposure to a laser beam (red arrows). Two types of 53BP1 repair foci were induced (C): Some showed microscopically resolvable individual subregions (as in B, cell 1), while in others, the internal architecture was not fully resolved by standard confocal imaging (as in B, cells 2 and 3) (additional information regarding the 3D structure of repair foci is provided in Supporting Information, Section 2). Scale bars 5 μm (A, B) and 3 μm (C).

3.2. Measurements of the All‐Protein Concentration in Double‐Strand Break Repair Foci by FLIM

To measure the “all‐protein” concentration in DSB repair foci, we used the FLIM‐based technique described above (Figure 1 and Figure S1). eGFP fused with 53BP1 repair factor served as a reporter for both the local concentration of eGFP‐53BP1 (based on the fluorescence intensity of eGFP) and the concentration of all proteins (based on fluorescence lifetimes) (Figure 3A,B).

FIGURE 3.

Intensity (A) and lifetime (B, scale in ns) of eGFP‐53BP1 fluorescence in a DNA double‐strand break repair focus (Fc). DSBs were induced only in cell 1, while cell 2 served as a control. Before light exposure, 53BP1 was distributed throughout the nuclei of both cells (A, 0 min). Fifteen minutes after damage induction, recruited eGFP‐53BP1 formed a repair focus (cell 1), while no changes were detected in cell 2 (A, 15 min). The fluorescence lifetime of eGFP‐53BP1 that accumulated at the damage site (B, red arrow) was significantly shorter than in the nucleoplasm. Scale bars, 10 μm. (C) Distribution of the fluorescence lifetimes of free eGFP in the nucleoplasm and eGFP‐53BP1 in the nucleoplasm and repair foci. (D) Differences between fluorescence lifetimes (Δτ_NL‐Fc), in the nucleoplasm and repair foci (pale cyan) (N = 40), and between the nucleoplasm adjacent to the repair foci and the interior of these foci (orange) (N = 22). The dashed and short dashed lines in the violin plots represent the median and quartiles, respectively. Scale bar 5 μm.

We observed a significant shortening of the eGFP lifetime within the nuclear site where local DNA damage was induced (Figure 3B, cell 1), while there was no change in the fluorescence lifetime in the nucleoplasm of intact cell 2. This observation suggested that the concentration of proteins residing in the repair focus was significantly greater than in the surrounding nucleoplasm. To substantiate this claim and provide a quantitative assessment of this difference, we first demonstrated that this is not an artifact arising from the substantially greater number of photons detected in the repair focus than in the nucleoplasm (this analysis is described in detail in Supporting Information and Figure S2). We observed significant differences between the average lifetime of eGFP‐53BP1 in the nucleoplasm (outside of repair foci) of various cells. In other words, the baseline lifetime, with which the lifetimes of the repair foci are compared, varied from cell to cell. Therefore, the elevated local protein concentration in the repair foci was accurately represented by the difference between the average lifetime of the repair focus and the nucleoplasm in the same cell (Δτ_NL‐Fc, see below).

A careful study of baseline lifetimes in the nucleoplasm of various cells showed that they varied from 2.25 to 2.52 ns for eGFP‐53BP1 and from 2.38 to 2.49 ns for free eGFP (Figure 3C). This range of lifetime values is likely associated with cell cycle stages, as demonstrated before [25, 26, 27]. The somewhat shorter fluorescence lifetimes of eGFP‐53BP1 than free eGFP may reflect the binding of some eGFP‐53BP1, but not free eGFP, to endogenous DNA lesions. Such binding within endogenous repair foci is expected to shorten the lifetimes of eGFP‐53BP1.

Interestingly, the range of fluorescence lifetimes recorded in various repair foci in various cells was quite wide (2.13 to 2.44 ns). Despite the diversity of fluorescence lifetimes encountered in different cells, a difference of 70 ps between the mean lifetimes in the repair foci and their surrounding nucleoplasm was consistently observed (Figure 3C). The difference between the average lifetimes of the repair foci and the adjacent nucleoplasm was slightly lower than between the foci and the entire nucleoplasm (Figure 3C). More precisely, analysis of several individual foci confirmed that the fluorescence lifetime of eGFP‐53BP1 was always shorter in the repair focus than in the adjacent nucleoplasm, and this difference ranged between 20 and 125 ps in various cells (with a median value of 50 ps, Figure 3D).

Subsequently, we used the method developed by Pliss [20, 28] and calculated the difference in all‐protein concentrations between the repair foci and the nucleoplasm. According to the calibration described in [20], and according to the control experiment performed by us (see Supporting Information, Figure S3), a ~100 ps reduction in the fluorescence lifetime of eGFP‐53BP1 corresponds to a 150 mg/mL increase in protein concentration. Therefore, the all‐protein concentration in the DSB repair foci is ~30 to 170 mg/mL greater than in the surrounding nucleoplasm. This means that DNA damage induces the formation of a microenvironment surrounding a lesion, which is characterized by a significantly higher protein packing density than the surrounding nucleoplasm.

3.3. Measurement of All‐Protein Concentration in Single‐Strand Break Repair Foci

To determine whether a high protein concentration is also a feature of repair foci formed in response to single‐strand breaks, we performed FLIM measurements of eGFP‐tagged XRCC1 recruited to damage sites. Repair foci formed within seconds after damage induction (Figure 4A). Repair of SSBs is relatively fast; therefore, detectable dissociation of XRCC1 and disappearance of repair foci were usually observed within 15 min after damage induction (Figure 4B).

FIGURE 4.

Intensity and lifetime (scale in nanoseconds) of fluorescence of XRCC1‐eGFP recruited to SSB repair foci in two representative cells. (A) Before inducing local DNA damage, XRCC1 was evenly distributed in the nuclei of both cells (some XRCC1‐containing subnuclear bodies were often present (A, 0 min)). SSBs were induced only in cell 1, while cell 2 served as a control. The fluorescence lifetimes of XRCC1‐eGFP, which formed repair foci (Fc) at the damage sites, were 40 to 140 ps shorter than those of the nucleoplasm. The recruitment of XRCC1‐eGFP to the damage site reached a maximum within 5 min after damage induction (cell 1), while no new XRCC1 foci formed in cell 2 (A, 5 min). (B) 15 min after damage, a significant reduction in the local XRCC1 concentration (top row) and the total protein concentration (bottom row, red arrows) in repair foci decreased substantially. Scale bars 5 μm.

In rare cases, shortening of the fluorescence lifetime within the repair focus persisted for up to 15 min following damage induction (see Figure S4C). In these cells, which most likely suffered more extensive damage that requires longer repair times, signs of general stress were also observed, as many mobile XRCC1‐containing subnuclear bodies formed throughout the nucleus. This phenomenon was described previously [10].

We also detected some XRCC1 repair foci that persisted well beyond 15 min (30 min, occasionally even 80 min; Figure S4). Their fluorescence lifetimes were short (e.g., 40–70 ps or 80–140 ps shorter than those in the nucleoplasm), suggesting that all‐protein concentrations were significantly higher than those in the nucleoplasm (60–100 or 120–210 mg/mL, respectively) (Supporting Information, Figure S4). Interestingly, the decrease in the lifetime of eGFP fluorescence observed in repair foci, which in some cases can reach 140 ps, is comparable to the changes detected in the lifetime of eGFP fluorescence following hyperosmotic shock [29].

3.4. Molecular Crowding Limits Access to Protein‐Dense Repair Foci

The data described above demonstrate that repair foci exhibit a much higher all‐protein concentration than that of the surrounding nucleoplasm. This difference (reflected in Δτ_NL‐Fc) varied widely between foci in different cells, suggesting that the foci that had the shortest fluorescence lifetime of eGFP‐53BP1 had the highest protein density. We sought to verify this notion by investigating the accessibility of the interior of the repair foci to bovine albumin molecules in fixed cells.

BSA is a protein that is frequently used as a crowding agent [30] and is not expected to be involved in any enzymatic reactions in our experimental system. It is a medium‐size protein (approx. 600 amino acids, over 66 kD). Based on a previous report [20] and the calibration performed in this study (Figure S3), we know that an increase in the BSA concentration of 150 mg/mL corresponds to an ~100 ps in the fluorescence lifetime of eGFP in live cells. We hypothesized that, due to the postulated higher protein density within the repair foci, the interior of the foci would be less accessible to BSA molecules. Introducing BSA should therefore result in a relatively small shortening of the lifetime of eGFP.

Following induction of local DNA damage, cells were fixed with a protein/protein and protein/DNA cross‐linking reagent (FA) or FA/EtOH that causes cross‐linking, protein precipitation, and lipid extraction. This treatment retained the cellular proteins, permeabilized the cell membranes, and opened the cellular interior to BSA. Next, we equilibrated the fixed cells with a 150 mg/mL BSA solution. Regardless of the fixation method, the lifetime of eGFP‐53BP1 in the nucleoplasm of fixed cells was 40 to 80 ps shorter than in live cells. The introduction of BSA shortened τ_eGFP in the nucleoplasm by an additional ~50 to 70 ps compared to fixed cells (Figure 5). Thus, as expected, τ_eGFP in the nucleoplasm decreased in response to the addition of BSA. However, the observed changes were smaller than the changes that could be expected based on the calibration data. We note that the calibration was performed in cells after permeabilization with Triton X‐100 (Figure S3); therefore, we hypothesize that exposure to Triton X‐100 results in greater accessibility of the cellular interior for BSA molecules than EtOH fixation.

FIGURE 5.

Accessibility of the repair focus interior to proteins. Fixed cells were classified into two groups: cells with low (A) and high (B) all‐protein concentrations in repair foci, as indicated by long or short eGFP lifetimes (2.38 and 2.33 ns, respectively). In cells with low protein concentrations in repair foci, the lifetime of eGFP decreased significantly in both the nucleoplasm and repair foci after FA fixation and introduction of BSA. In contrast, in cells with high protein concentrations in the repair foci, the lifetime of eGFP changed substantially only in the nucleoplasm, whereas the change within the repair foci was negligible, demonstrating limited access of BSA molecules to the interior of the already dense repair foci. Scale bars 5 μm.

As expected, the influence of albumin on the lifetime of eGFP‐53BP1 varied depending on the original protein density of the foci. The effect was significant in the repair foci that were characterized by a low initial all‐protein concentration, but not in the foci with a high concentration. In foci with long τ_eGFP (e.g., with low initial all‐protein concentration), a slight increase in fluorescence lifetime was observed after fixation with FA (~20 ps), which was followed by a substantial decrease after the addition of BSA solution (~100 ps) (Figure 5A). On the contrary, no significant changes in τ_eGFP were detected either after FA fixation (< 10 ps) or after the introduction of BSA introduction (~10–20 ps) in foci with short τ_eGFP (i.e., those with high initial protein concentrations). These results suggest that in fixed cells, BSA molecules have access both to the nuclear interior and to repair foci characterized by low protein concentration, while repair foci with high local protein concentration remained virtually inaccessible to BSA due to the already highly crowded environment within their interior. This observation confirms our notion that repair foci with the shortest fluorescence lifetimes observed in live cells exhibit the highest protein density, which significantly exceeds the protein concentration in the nucleoplasm.

3.5. Distribution of All‐Protein Concentrations Within a DSB Repair Focus

53BP1 foci always consist of several subregions, likely representing individual DNA breaks. Despite the differences between the architectures of various repair foci (compact and showing discernible individual subregions) that are formed in response to DNA damage (Figure S5), a decrease in the lifetime of eGFP‐53BP1 in repair foci was detected in both subgroups. We noticed that the periphery of the foci frequently exhibited a longer lifetime than the center, but was still shorter than the rest of the nucleoplasm. We hypothesize that proteins are not uniformly distributed but are more likely to form highly concentrated central regions within an individual repair focus and are surrounded by a zone of diminishing protein density. To verify this hypothesis, we focused on compact repair foci with clearly distinguishable individual subregions (Figure 6).

FIGURE 6.

All‐protein distributions within a DSB repair focus. Representative fluorescence intensity and lifetime images of repair foci consisting of discernible subregions (marked with arrows). Each individual subregion is surrounded by an area whose fluorescence lifetime is longer than that of the interior. Thus, the subregions that were imaged in a plane close to the top (polar region) demonstrated longer lifetimes (magenta arrows), whereas those imaged in the equatorial plane demonstrated shorter lifetimes (blue arrows). The differences between the lifetimes within the periphery and the interior ranged from 20 to 90 ps. Scale bars, 2 μm.

Each individual subregion of the repair focus (marked with blue arrows, Figure 6) contains two regions: “a core” with a shorter fluorescence lifetime (green and blue, Figure 6) and “a shell” with a longer lifetime (red, Figure 6). The difference between the core and the shell areas can range from 20 to 90 ps, which means that the concentration of all proteins in the repair foci core may be 135 mg/mL higher than that of the shell. Some of the individual repair foci (marked with magenta arrows, Figure 6) exhibited a uniform lifetime. These images represent confocal sections that are cut close to the surface of a subregion, away from the equatorial plane of a repair focus, that is, such an image does not include the core. For detailed information on the 3D structure of the eGFP‐53BP1 repair foci in our experiments and to what extent the 2D FLIM images are related to the 3D architecture of the DSB repair foci, see the Supporting Information (Figure S1).

Based on the observations described above, we propose a model that depicts the characteristic all‐protein density distribution in and around a DSB repair focus (Figure 7). In this model, four zones that differ in all‐protein concentration can be distinguished: (1) the nucleoplasm outside the repair focus, (2) a thin layer adjacent to the repair foci, (3) the “shell,” and (4) the “core” of each individual repair subregion. The most protein‐crowded areas are formed in the centers of repair foci. The difference in protein densities between the center and the nucleoplasm reaches 135 mg/mL. At the same time, the all‐protein concentration is relatively uniform throughout the area of the surrounding nucleoplasm, with a slight increase close to the center of the repair focus (~30 mg/mL on average).

FIGURE 7.

Schematic representation of the distribution of all‐protein concentration in DSB repair foci. Representative FLIM images of fluorescence lifetimes within the nucleoplasm surrounding a repair focus (A) and within a DSB repair focus (B). (C) Schematic representation of the all‐protein density distribution within a DSB repair focus imaged in (B), based on FLIM data.

Together, these data demonstrate that DNA damage induces dense packing of nuclear proteins within a double‐strand break repair focus, creating the most protein‐dense center and a surrounding less‐dense region. A protein‐enriched crowded local microenvironment extends for a few hundred nanometers around a DNA break.

Interestingly, the “core/shell” organization may exist in other subcellular structures. A similar structural model was proposed for stress granules [31, 32, 33]. Stress granules were shown to be organized into two distinct regions: a central, stable core with a size of up to ∼150–200 nm and a surrounding, more dynamic shell [31]. The core is characterized by a high concentration of specific proteins and mRNAs, whereas the shell surrounding the core is less concentrated and more dynamic in nature [31, 32]. The authors implied that this type of organization may facilitate the function of stress granules [31].

4. Discussion

4.1. Control of Protein Density in Animal Cells

Proteins are the major contributor to mass density in animal cells. Their share in the dry mass of the cell amounts to approximately 60% (in contrast, nucleic acids contribute only some 5%) [34, 35].

Mass densities vary between the cytoplasm, nucleoplasm, and nucleoli, and animal cells appear to maintain a local concentration of macromolecules under tight control. Optical diffraction tomography has shown that the relative distribution of mass densities between the cytoplasm, nucleus, and nucleolus remained unchanged when actin and microtubule depolymerization and chromatin condensation were induced [36]. This suggests that, despite various environmental cues, tight control of the cell mass and molecular crowding applies to whole cells as well as major cellular subcompartments. Interestingly, cell density remains constant during the cell cycle (except during mitosis) [35, 37, 38, 39]. Small changes in cell density were detected during differentiation [40]. Cell density was also changed after exposure of mouse leukemia cells to staurosporin, a drug that induces apoptosis, but even then the change was small, below 2% [41]. The tendency of cellular macromolecular crowding to return to the set value was also demonstrated by FLIM and FCS studies of eGFP in cells exposed to osmotic shock. Although cell density initially increased in response to hyperosmotic shock, it steadily decreased despite the remaining high salt concentration in the surrounding medium [29].

The cellular mechanisms responsible for maintaining cell mass and molecular crowding at a steady level are not fully understood; therefore, phenomena leading to a transient local dramatic increase of molecular crowding and protein density, as seen during the formation of stress granules, subnuclear bodies, and DNA repair foci, are of particular interest.

4.2. Protein Density in DSB and SSB Repair Foci

The findings described in this report support the notion that a unique molecular microenvironment is formed in a small region surrounding DNA double‐ and single‐strand breaks. A protein concentration that is significantly higher in the repair focus than in the surrounding nucleoplasm emerges as a feature of the repair foci formed in response to DNA breaks. This phenomenon is not obvious since, following damage induction, repair factors are recruited, but at the same time, other proteins, including histones, detach from DNA, and chromatin undergoes decompaction. How can such a protein‐dense local environment be created, and what might be its role?

Some DNA repair factors are recruited in large numbers and form dense subnuclear bodies. Two (non‐exclusive) mechanisms can be involved in inducing a high concentration of DNA repair factors in a small region of the cell nucleus: one is based on the creation of a large number of binding sites in the immediate vicinity of a DNA lesion (PBM, polymer bridging model), and the other involves liquid–liquid phase transition mechanisms (LPM—liquid phase model) [7]. The first mechanism assumes a change in the molecular features of the DNA and proteins surrounding a lesion; the other involves modification of the repair factor itself, leading to the formation of a liquid droplet. In both cases, the local concentration of at least some repair factors in a repair focus (primarily scaffold proteins) is expected to be much higher than that in the rest of the nucleus. Our observations of high local concentrations of all proteins in repair foci support this notion and demonstrate that the local density of all proteins can increase by up to 30 to 170 mg/mL compared to that of the surrounding nucleoplasm, where the protein density is approximately 100 mg/mL and is uniform throughout the nucleus (except the nucleolus). This phenomenon may be fundamentally important for maintaining genome stability, considering that the high number of some of the recruited DNA repair factors and damage‐induced changes in the structure of chromatin are observed from yeast to plant and human cells.

4.3. High All‐Protein Concentration in Repair Foci Versus Microviscosity

A high concentration of proteins in repair foci resulting in significant macromolecular crowding is expected to increase the microviscosity of the repair foci. This supposition is supported by the available data. The dynamics of Rad52, a protein involved in the repair of DSB in yeast, was studied by single‐molecule tracking [14], and these experiments demonstrated that the diffusion of Rad52 molecules (fluorescently labeled with HaloTag) was significantly slower at the repair site than in the nucleoplasm. The microviscosity within a repair focus was estimated to be 8 to 20 times greater than that in the nucleoplasm [14]. In view of these facts, an intriguing question arises as to whether the increase in local concentration of a specific factor, such as HP1, in repair foci in human cells or Rad52 in yeast, is sufficient to induce a liquid–liquid phase transition of these repair factors and whether, under conditions of such high protein‐induced macromolecular crowding, other proteins may also participate in the formation of protein droplets.

4.4. Molecular Crowding in DSB Repair Foci

FLIM measurements demonstrated that the local concentration of all proteins was always significantly greater within a repair focus than in the nucleoplasm. This difference ranges from 30 to 170 mg/mL, which is quite high considering that the concentration of proteins in the surrounding nucleoplasm is on the order of only 100 mg/mL [42, 43, 44]. Based on the available data and our measurements of the all‐protein concentrations in DSB repair foci, one can roughly estimate that the mean distance between protein molecules in the nucleoplasm is on the order of 10 nm (i.e., the distance between the surfaces of protein molecules is only 2–3 times the protein diameter), while in the repair focus, proteins may be packed so densely that such a mean distance is reduced to less than one protein diameter. Two potential consequences need to be considered: such a high protein concentration may mean that proteins within a repair focus exist at the maximum achievable solubility and that the role of weak interactions between proteins may be dramatically greater than in the nucleoplasm. Both may have a profound influence on the ability to form liquid droplets and on the mechanisms of reactions occurring in repair foci [45]. This notion is in agreement with the published data and the hypotheses put forward by others [11, 15, 46, 47].

4.5. Boundaries of Repair Foci

The data presented here demonstrate that DNA repair foci are densely packed structures and, in this sense, resemble subnuclear bodies such as PML and nucleoli, which are also known for their high molecular density. However, in repair foci, the protein concentration gradually increases from the periphery, reaching a maximum in the center. This rule applies to small subregions that are discernible components of a large repair focus, as well as to repair foci in which small subregions are not visually resolved in confocal images. Such a protein distribution suggests that the highest protein density is found in the immediate vicinity of DNA breaks and that it gradually falls away at a distance from a DNA lesion. This architecture of repair foci distinguishes them from subnuclear bodies, which are known as structures that exhibit well‐defined boundaries.

4.6. The Role of High Protein Density in DSB Repair Foci

NER has been shown to repair UV damage 10 times slower in nucleosomal DNA than in naked DNA [48]. Clearly, the complex architecture of chromatin and the molecular crowding of the cell interior are factors that limit the rate of accumulation of repair proteins at a DNA lesion. The need to reorganize the chromatin to allow access and assembly of repair complexes is another factor that limits the rate of DNA repair. However, once a large number of scaffold proteins have been recruited and other DNA repair factors have accumulated, a local dynamic equilibrium is established between molecules diffusing in and out of the repair focus. Thus, we postulate that a high local all‐protein concentration and high microviscosity may stabilize the repair complex and thus accelerate the repair process [49, 50].

Author Contributions

S.M.L. and J.W.D. formulated the aim and planned the experiments, and wrote the manuscript; S.M.L. performed the experiments and data analysis.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Data S1: Supporting Information.

Levchenko S. M. and Dobrucki J. W., “High Molecular Crowding in Repair Foci Surrounding DNA Breaks, Measured by Fluorescence Lifetime Imaging Microscopy,” The FASEB Journal 39, no. 17 (2025): e71010, 10.1096/fj.202501727R.

Data Availability Statement

All data will be deposited in the Jagiellonian University repository and available upon request.