DNA Ligase I Deficiency Leads to Replication-Dependent DNA Damage and Impacts Cell Morphology without Blocking Cell Cycle Progression

Drugs, cell lines, and cell treatments.Human simian virus 40-transformed fibroblast MRC-5V1 and 46BR.1G1 (European Collection of Cell Cultures no. CB2577) lines were maintained in monolayer culture in Dulbecco modified Eagle medium supplemented with 10% fetal bovine serum, 4 mM glutamine, and 50 μg/ml gentamicin (Sigma). 46BR.1G1 derivatives 7A3, M10, M13, and M83 were grown in complete Dulbecco modified Eagle medium supplemented with 300 μg/ml Geneticin (Sigma). Wild-type human LigI and LigI-4D mutant cDNAs were C terminally fused to the muscular actin epitope recognized by the HUC1-1 monoclonal antibody as previously described (8) and subcloned into the mammalian vector pcDNA3 (Invitrogen Corporation). After selection for resistance to Geneticin, the level of tagged LigI protein in each clone was determined by immunoblotting with HUC1-1 antibody. 46BR.1G1 cells were treated with 100 μM etoposide (Sigma) for 1 h. In all experiments 7A3, M10, M13, and M83 clones have been analyzed at early passages (<15) after their isolation.

Cell lysate and Western blotting.Cell extracts were prepared as previously described (23). The primary antibodies were obtained as follows: anti-phospho-histone H2AX (Ser¹³⁹) (clone JBW301) from Upstate Biotechnology, HUC1-1 (clone MSA06) from Diapath, anti-α-tubulin from Sigma, anti-Chk2 (clone 7) from Upstate Biotechnology, anti-Chk1 (clone G-4) from Santa Cruz Biotechnology, anti-phospho-Chk2 (Thr⁶⁸) and anti-phospho-Chk1 (Ser³⁵⁴) from Cell Signaling Technology, anti-ATM pS¹⁹⁸¹ (clone 10H11.E12) from Rockland, and anti-ATM from Rockland.

Immunofluorescence.Cells grown on glass coverslips were fixed with cold methanol as previously described (23). For the analysis of morphology, cells were fixed with paraformaldehyde and permeabilized in 0.5% Triton X-100 as previously described (12). Primary antibodies were anti-phospho-histone H2AX (Ser¹³⁹) monoclonal antibody (clone JBW301; Upstate Biotechnology), anti-cyclin A polyclonal antibody (clone H-432; Santa Cruz Biotechnology), anti-phospho-histone H3 polyclonal antibody (Upstate Biotechnology), antibromodeoxyuridine (anti-BrdU) monoclonal antibody (clone BMC9318; Chemicon), and antivinculin monoclonal antibody (clone 7F9; Chemicon); secondary antibodies were fluorescein isothiocyanate-conjugated goat anti-mouse and tetramethyl rhodamine isocyanate (TRITC)-conjugated goat anti-rabbit immunoglobulin G (Jackson ImmunoResearch Laboratories). Nuclei were stained with 0.1 μg/ml 4′,6-diamidino-2-phenylindole (DAPI; Sigma). To detect sites of DNA synthesis, cells were grown in 100 μM BrdU (Sigma) and processed as previously described (23). Actin filaments were decorated with TRITC-conjugated phalloidin (Sigma).

Comet assay.The alkaline comet assay was performed according to the method in reference 25. For the comet assay under neutral conditions, after lysis, slides were washed for 10 min in 1× Tris-borate-EDTA and transferred to the electrophoresis chamber. The electrophoresis was carried out in Tris-borate-EDTA at 1 V/cm for 17 min. Slides were neutralized, and DNA was stained with DAPI.

The BrdU comet assay was performed according to the method in reference 14. After DNA labeling with 100 μM BrdU (Sigma), cells were combined with low-melting-point agarose at 37°C and spread onto a standard comet assay slide. After electrophoresis, gels were incubated with an anti-BrdU antibody (Chemicon International) in the dark at room temperature for 1 h. The primary antibody was detected with fluorescein isothiocyanate-conjugated anti-mouse secondary antibody (Jackson ImmunoResearch Laboratories). The cells were counterstained with DAPI.

The image analysis software Comet Score (TriTeck Corporation) was used to quantify the different parameters of the images.

Isolation of neo-synthesized DNA.Medium was removed from exponentially growing cells and replaced with fresh, serum-free medium containing 5 μCi/ml [³H]thymidine at 37°C for 15 min. After being washed with phosphate-buffered saline, cells were collected by trypsinization and nuclei were isolated as previously described (20). Total genomic DNA was extracted with the standard phenol-chloroform-isoamyl alcohol procedure, precipitated with isopropanol alcohol, and resuspended in TNE buffer (10 mM Tris-HCl, pH.8.0, 100 mM NaCl, 1 mM EDTA). DNA was denatured at 95°C for 10 min, kept on ice for 10 min, loaded onto 5 to 30% linear neutral sucrose gradients in TNE, and centrifuged at 25,000 rpm (Beckman SW28 rotor) for 13 h at 20°C. In parallel, a reference tube with a double-stranded size marker DNA was denatured and loaded on an identical gradient. Gradients were collected in fractions of 1 ml each. The distribution of DNA size markers was determined by agarose gel electrophoresis. An 0.5-ml portion of each fraction was loaded onto a 24-well sample plate (1 ml) with the recommended scintillator solution (Perkin-Elmer). Quantification of radioactivity was achieved by scintillation counting (MicroBeta Trilux by Perkin-Elmer).

LigI deficiency leads to accumulation of DNA breaks.Because of the relevance of LigI in maintaining the integrity of the double helix, the LigI defect in 46BR.1G1 cells is expected to lead to a higher level of DNA damage. To verify this prediction, we determined the level of the phosphorylated form of the H2AX histone variant (γH2AX), which is an early marker of the cellular response to DNA breaks. In the absence of any exogenous source of DNA damage, the level of γH2AX in 46BR.1G1 cell extracts is higher than that in control MRC-5V1 fibroblasts (Fig. 1). Moreover, bright γH2AX foci are visible in 46BR.1G1 nuclei (Fig. 1B). However, the intrinsic level of DNA damage in 46BR.1G1 cells is far from being saturating and phosphorylation of H2AX further increases after treatment with the topoisomerase II poison etoposide. Increased phosphorylation of H2AX is a direct consequence of the LigI defect and in fact is strongly reduced upon stable overexpression of the wild-type epitope-tagged enzyme (clone 7A3) (Fig. 1A and B). Complementation of the LigI defect results also in a shorter proliferation time (cell doubling time of ≈20 h in 7A3 versus ≈30 h in 46BR.1G1).

• Open in new tab
• Download powerpoint

FIG. 1.

DNA breaks accumulate in LigI-defective cells. (A) Total cell extracts of the indicated cell lines were analyzed by Western blotting with anti-γH2AX and anti-α-tubulin antibodies. Extracts in the right panels were prepared from 46BR.1G1 cells before (−) or after (+) treatment with etoposide. (B) Representative images of γH2AX foci. 7A3, 46BR.1G1, and M10 cells were analyzed in indirect immunofluorescence assays with an anti-γH2AX antibody and counterstained with DAPI. Bar, 10 μm. (C and D) Alkaline (C) and neutral (D) comet assay analysis of 7A3, 46BR.1G1, and M10 cells. Distribution of the tail moment of comets in 7A3 (light gray bars), 46BR.1G1 (medium gray bars), and M10 (dark gray bars) cells. Data are calculated by the analysis of 450 to 550 cells for each cell line in four experiments. Results show means ± standard errors of the means. Cells are grouped in seven categories based on the value of comet tail moment.

We have previously shown that LigI is progressively phosphorylated on four residues (Ser 51, 66, 76, and 91) according to a precise temporal order during the cell cycle (8, 24). So far, the functional implications of these modifications are unclear. Phosphorylated residues fall in the N-terminal regulatory domain that is dispensable for the enzyme activity. This observation along with the fact that the hyperphosphorylated protein occurs in M phase, when replication complexes are no longer active, led us to propose a role in modulating the association of LigI with other replication factors. Indeed, replacement of the four serine residues with the phosphomimetic aspartic acid (LigI-4D) drastically impaired the association of LigI with replication factories in transiently transfected COS7 cells (8). We decided to exploit 46BR.1G1 cells to evaluate the effect of serine/aspartic acid substitutions, and we selected three independent clones (M10, M13, and M83) that stably express LigI-4D. In vitro assays with immunopurified enzymes from 7A3 and M10 cells show that the substitutions in LigI-4D do not appreciably affect the DNA joining activity (see Fig. S1B in the supplemental material) or the interaction with PCNA (see Fig. S1C in the supplemental material). Moreover, in contrast to what is observed in transient transfection (8), stably expressed LigI-4D localizes at the replication factories throughout S phase (data not shown). This difference probably reflects a lower expression level of the stably expressed LigI-4D, whose level in M10 cells is five- to sevenfold higher than that of the endogenous protein (data not shown) and comparable to the level of the wild-type enzyme in 7A3 cells (see Fig. S1A in the supplemental material). In order to avoid problems due to transgene inactivation, all the experiments described below were performed with transfected cells at early passages (less than 15). In contrast to the overexpressed wild-type enzyme, LigI-4D is unable to prevent the occurrence of DNA damage in 46BR.1G1 cells as indicated by the level of γH2AX and by the presence of γH2AX foci in M10 cells (Fig. 1A and B). Similar results were obtained in all three clones, arguing against the possibility that the observed phenotype was a clone-specific feature (data not shown). Intriguingly, M10, M13, and M83 have cell doubling times (≈36 h) even longer than that of the parental 46BR.1G1 cells. This longer duplication time, however, does not reflect the accumulation of the cells in any particular phase of the cell cycle that is comparable in 7A3, M10, and 46BR.1G1 cells (see Fig. S2 in the supplemental material). Moreover, LigI-4D expression does not result in an increased level of apoptosis (less than 1% of cells).

Phosphorylated H2AX accumulates in chromatin regions surrounding DSBs. We decided to verify the occurrence of SSBs and DSBs in 46BR.1G1, 7A3, and M10 cells by single-cell gel electrophoresis (comet assay). Cells were embedded in agarose on a microscope slide, lysed, and subjected to electrophoresis under alkaline conditions. DNA was then stained with a fluorescent dye and visualized under a fluorescence microscope. Undamaged DNA remains within a “core,” while broken DNA migrates from this core toward the anode, forming an image suggestive of a comet, with a “head” and a “tail” (see Fig. S3 in the supplemental material). The morphometric measurements were expressed in terms of “tail moment,” which positively correlates with the amount of DNA breaks (18). Figure 1C shows the distribution of proliferating 46BR.1G1, 7A3, and M10 cells in seven classes with increasing values of tail moment. Broken DNA accumulates in 46BR.1G1 cells, and only ≈50% of the cells are found in class 0 (tail moment < 1). Expression of the wild-type enzyme in 7A3 cells significantly reduces the occurrence of DNA damage, as indicated by a higher percentage of cells in class 0 (83%). In contrast, expression of the phosphomimetic mutant results in only ≈25% of M10 cells being in class 0 and increases the frequency of cells with a tail moment of >10 (40%), significantly higher than that observed in 46BR.1G1 (10%) and 7A3 (0%) cells. Thus, the LigI-4D mutant raises the level of DNA damage in 46BR.1G1 cells and behaves as a dominant-negative mutant. We then performed the comet assay under neutral conditions in order to reveal only DSBs (see Fig. S3B in the supplemental material). DSBs accumulate in LigI-deficient cells, and only 45% of 46BR.1G1 and 30% of M10 cells are found in class 0 compared to 80% of 7A3 cells (Fig. 1D). The difference between parental and M10 cells is less impressive under neutral than under denaturing conditions, as if the presence of LigI-4D mutant is associated with a higher level of SSBs.

LigI deficiency affects the maturation of nascent DNA.To understand whether accumulation of DNA breaks in 46BR.1G1 and M10 cells could be due to a defect in DNA replication, we combined the comet assay with BrdU pulse-labeling of replicating DNA as described in reference 14. This assay exploits the immunological localization of BrdU within the heads and tails of the comets in order to assess replicative integrity on a single cell. 7A3, 46BR.1G1, and M10 fibroblasts were pulse-labeled with BrdU and either processed directly for the comet assay or chased with normal nucleotides for increasing time periods. As shown in Fig. 2A, in both 7A3 and M10 cells nascent nonligated DNA was in the tail of the comet (a and g). After a 1-h chase, mature BrdU-DNA in 7A3 cells was retained in the head of the comet (d) while in M10 cells replicative intermediates still migrated in the tail (l). The shift of BrdU-DNA from the tail to the head (i.e., the reduction in percentage of BrdU-DNA in the tail) after the chase is indicative of maturation of replicative intermediates. The frequency distribution of BrdU-DNA in the comet tail is shown in Fig. 2B. As expected, a shift toward the head rapidly occurs in 7A3 cells, and after 10 h no BrdU-DNA is in the tail. In accordance with previous in vitro observations (11), the time required for the maturation of nascent DNA is longer in 46BR.1G1 cells, and even after 10 h, about 25% of the cells contain BrdU-labeled DNA in the tail. Notably, this effect is exacerbated in M10 cells, suggesting that the expression of LigI-4D deeply impacts the maturation of nascent DNA and DNA integrity.

• Open in new tab
• Download powerpoint

FIG. 2.

Replication intermediates accumulate in LigI-defective cell lines. BrdU comet analysis of 7A3, 46BR.1G1, and M10 cells. Cells were pulse-labeled with BrdU for 15 min and either immediately processed or chased for 1, 5, or 10 h prior to comet processing. (A) Representative images of 7A3 and M10 cells. Identical fields are shown with green fluorescein (BrdU) and blue DAPI (DNA) staining. (a and g) BrdU staining after the pulse; (d and l) 1-h chase; (b, h, e, and m) the same cells stained with DAPI; (c, f, i, and n) merged images. (B) Quantification of the BrdU comet assay at the indicated chase periods. Results are based on the analysis of 50 to 60 comets/point in four experiments. Bars show means ± standard errors of the means.

In order to confirm this observation, we examined the size of nascent DNA in 7A3 and M10 cells. Nascent DNA was pulse-labeled with [H³]thymidine and immediately purified. Following heat denaturation to free small newly synthesized fragments, DNA was separated through a sucrose gradient (Fig. 3). Nascent DNA molecules from M10 and 7A3 cells have distinct sedimentation profiles. The proportion of newly synthesized fragments shorter than 23,000 nt is higher in M10 (42%) than in 7A3 (27%) cells, and molecules with a size in the range of Okazaki fragments (150 nt) are detectable only in M10 cells. Okazaki fragments are no longer detectable after a 10-h chase when 74% of labeled DNA in M10 cells is longer than 23,000 nt, compared to a value of 83% in 7A3 cells (data not shown). The sedimentation profile in 7A3 cells is consistent with the idea that maturation of newly synthesized DNA fragments is extremely fast and that replication intermediates are promptly incorporated into long molecules. This process is affected by LigI deficiency, as in M10 cells.

• Open in new tab
• Download powerpoint

FIG. 3.

Analysis of replication intermediates. 7A3 and M10 cells were pulse-labeled for 15 min with [³H]thymidine. Purified DNA was heat denatured and fractionated through 5 to 30% neutral sucrose gradients. Fractions were taken from the top of the gradient. Aliquots of each fraction were counted to determine the distribution of nascent DNA in 7A3 and M10 cells. Arrows, DNA size markers in kilobases. The top of the gradient is on the left.

LigI deficiency leads to chronic activation of the ATM-mediated DNA damage checkpoint.The peculiarity of the damage resulting from LigI deficiency (“behind” instead of “at” the replication fork) prompted us to investigate the checkpoint pathways involved in sensing and repairing the damage. To this end, we examined the activation status of core protein components of the DNA damage checkpoints (3). In particular we considered the central transducers of DNA damage signaling, the Chk1 and Chk2 kinases, which are, respectively, phosphorylated by apical ATR and ATM kinases in response to replication stress and DNA damage. Constitutive phosphorylation of H2AX in LigI-deficient cells (Fig. 1) is accompanied by phosphorylation of ATM on Ser 1981 and Chk2 on Thr 68 (Fig. 4A). This pattern is almost completely reversed in 7A3 cells. Surprisingly, we never observed Chk1 activation in 46BR.1G1 and M10 cells, indicating that, in spite of the defect in the maturation of replicative intermediates, the type of DNA damage induced by the LigI deficiency does not trigger the S-phase-specific ATR-dependent checkpoint. This behavior is not due to a compromised ATR-Chk1 pathway in 46BR.1G1 cells since, as described for other cell lines (23), Chk1 phosphorylation is observed after etoposide treatment. In agreement with the lack of ATR activation, we failed to observe in both M10 and parental 46BR.1G1 cells phosphorylation of replication protein A (RPA) and cohesin (SMC1), two targets of ATR-Chk1. Moreover, we did not detect phosphorylation of Nbs1, which is a hallmark of the activation of pathways involved in repair of DSBs during S phase (see Fig. S4 in the supplemental material).

• Open in new tab
• Download powerpoint

FIG. 4.

Chronic activation of DNA damage signaling in LigI-defective cells. (A) Western blotting of MRC-5V1, 46BR.1G1, 7A3, and M10 total cell extracts with antibodies against the indicated proteins or phosphoepitopes. The same analysis was performed in 46BR.1G1 cells after etoposide treatment (+ et). (B) Asynchronous MRC-5V1, 7A3, 46BR.1G1, and M10 cells were costained with anti-γH2AX and anti-cyclin A antibodies. γH2AX-positive cells were scored as having more than five distinct foci in the nucleus. For each cell line, the fraction of nuclei displaying γH2AX foci was determined in 100 randomly selected cells in three experiments. Bars show means ± standard errors of the means. (C) Quantitation of γH2AX-positive cells in G₁ phase (cyclin A negative) and in S/G₂ phases (cyclin A positive) in MRC-5V1, 7A3, 46BR.1G1, and M10 cells. Bars show means ± standard errors of the means.

LigI deficiency increases the frequency of γH2AX foci throughout the cell cycle.The LigI defect delays the maturation of replication intermediates. However, it does not elicit the activation of the S-phase-specific checkpoint, which raises some questions about the moment during the cell cycle when DNA damage is generated. In order to address this issue, we followed γH2AX foci during the cell cycle. To avoid synchronization protocols that per se induce DNA damage and checkpoint activation, we performed the analysis in asynchronous cells by costaining with an antibody against cyclin A, which specifically marks S and G₂ cells (see Fig. S5 in the supplemental material). In accordance with the results of the fluorescence-activated cell sorting analysis (see Fig. S2 in the supplemental material), comparable fractions of cyclin A-positive cells are detectable in 46BR.1G1, 7A3, and M10 cells, further indicating that the LigI defect does not perturb the distribution throughout the cell cycle. As shown in Fig. 4B, ≈40% of 46BR.1G1 cells and 60% of M10 cells show γH2AX foci, compared to only ≈10% of both 7A3 cells and MRC-5V1 control fibroblasts. This result is consistent with the levels of γH2AX foci detected in Fig. 1. There is a general increase of nuclei with γH2AX foci throughout the cell cycle (Fig. 4C). In S/G₂ 70 to 80% of 46BR.1G1 and M10 nuclei show γH2AX foci, compared to 10 to 20% in control MRC-5V1 fibroblasts and 7A3 cells. In G₁ cyclin A-negative cells, foci are detectable in 12.5% of 46BR.1G1 and 32.5% of M10 cells, ≈10-fold and more than 20-fold the frequency, respectively, observed in MRC-5V1 fibroblasts (Fig. 4C).

LigI-deficient cells enter a new cell cycle with broken DNA.Replication-dependent DNA damage in LigI-deficient cells does not trigger the S-phase-specific DNA damage checkpoint and does not block cell cycle progression. We speculated that at least a fraction of the γH2AX foci in G₁ could originate from a defect in the maturation of nascent DNA during the previous S phase. To verify this prediction, we examined the level of γH2AX foci in mitotic 7A3, 46BR.1G1, and M10 cells and in control MRC-5V1 fibroblasts. Dividing cells were costained with antibodies against γH2AX and the phosphorylated form of histone H3 (Fig. 5A and B), which specifically marks mitotic cells. Mitotic cells were then grouped in four classes based on the number of γH2AX foci (Fig. 5C). More than 70% of mitotic MRC-5V1 cells fall in the first class (zero to three foci), and only ≈4% are in the fourth class (>10 foci). A slightly different profile is observed in 7A3 cells (≈50% in the first class and ≈16% in the fourth), which probably reflects an incomplete complementation of the LigI deficiency. In contrast, 46BR.1G1 and M10 cells show a completely different distribution with ≈56 and 60% of the cells being in class 4 (>10 foci), respectively. This analysis demonstrates that most LigI-defective cells enter mitosis in the presence of chromatin marks specifically associated with damaged DNA. Notably, foci are detectable also in telophase, indicating that mitosis can be completed even in the presence of damaged DNA. Intriguingly, in these cells there is little overlap between phosphorylated H3 and γH2AX staining, as if the two modified histones were associated with different chromosomal domains (Fig. 5B).

• Open in new tab
• Download powerpoint

FIG. 5.

LigI-deficient cells enter a new cell cycle with signs of unrepaired DNA breaks. (A and B) Asynchronous M10 cells were costained with anti-γH2AX and anti-phospho-histone H3 (H3-p) antibodies. DNA was stained with DAPI. Confocal images. Bars, 10 μm. Panel B shows a cell in telophase. (C) Quantitation of γH2AX foci in mitotic MRC-5V1, 7A3, 46BR.1G1, and M10 cells. Cells were stained as in panel A, and mitotic cells were scored for the number of γH2AX foci. About 100 mitotic cells were scored for each cell line. Results are pooled from two independent experiments. (D) Cells were pulse-labeled with BrdU for 15 min and either immediately processed (pulse) or chased for 10 h prior to fixation (chase). Representative images of M10 cells, where identical fields are shown with green fluorescein staining (BrdU) and red TRITC staining (cyclin A). Nuclei were stained with DAPI. Arrowheads, BrdU-positive nuclei. Bar, 10 μm. (E) 7A3, 46BR.1G1, and M10 cells were pulse-labeled with BrdU as described for panel D and scored for cyclin A-negative nuclei (G₁) with mid/late patterns of BrdU replication foci. Frequency distribution plots of cyclin A-negative cells with mid/late replication foci after a chase period of 10 h. Results show means ± standard errors of the means of three experiments.

The BrdU comet assay (Fig. 2) indicates that BrdU-labeled DNA in M10 nuclei fails to move from the tail to the head of the comet during a chase period of 10 h. This opens the possibility that nonligated, newly replicated DNA can pass through mitosis and enter the successive G₁ phase. To verify this possibility, we took advantage of the fact that the spatial distribution of replication foci is stable for many cell cycles. Thus, in vivo labeling of replication foci with BrdU pulses in any short interval of the S phase allows the visualization of the same replicons during the following G₁ phase (10). However only S/G₂ nuclei are stained by antibodies against cyclin A. We pulse-labeled proliferating 46BR.1G1, 7A3, and M10 cells for 15 min, and after a chase period of 10 h, we measured the percentage of cells displaying the easily recognizable mid/late patterns of labeled foci. As shown in Fig. 5D, immediately after the pulse all the BrdU-labeled nuclei are decorated with the anti-cyclin A antibody. Instead, BrdU-positive/cyclin A-negative cells after a 10-h chase represent the fraction of cells labeled in the previous S phase that passed through mitosis and entered G₁. The results in Fig. 5E clearly indicate that during the 10-h chase comparable fractions of 7A3, 46BR.1G1, and M10 cells are BrdU positive/cyclin A negative (G₁). The vast majority of these cells show mid/late replication patterns. Thus, LigI-defective cells enter the succeeding G₁ phase in spite of the massive presence of γH2AX foci in mitosis and incompletely matured BrdU-DNA.

The expression of the LigI-4D mutant affects cell morphology.During this analysis, we noticed that the expression of the phosphomimetic LigI-4D mutant is associated with a drastic morphological change. Indeed, M10, M13, and M83 clones differ from both 46BR.1G1 and wild-type 7A3 cell lines in the length and shape of the cytoplasm protrusions. In particular 7A3 cells show the expected fibroblast-like morphology with long cytoplasmic extensions forming a network on the plate. In contrast M10 cells have a reduced size and exhibit fewer cellular extensions when plated at low density (Fig. 6A). To further characterize these morphological differences, we stained actin fibers with rhodamine-conjugated phalloidin. A reorganization of cytoskeleton was visible, and actin filament bundles (stress fibers) and filopodia that are easily detectable in 7A3 cells are almost undetectable in M10 cells, where actin mainly localizes around the cell membrane (Fig. 6B). An intermediate phenotype is observed in 46BR.1G1 cells. The difference between M10 and 7A3 cells is revealed also by the distribution of vinculin, which in 46BR.1G1 and 7A3 marks focal adhesions where bundles of actin filaments terminate at the plasma membranes. As shown in Fig. 6C, in M10 cells vinculin relocalizes from the focal contacts to the cytoplasm. This defect in establishing focal contacts may account for the fact that M10 cells appear less flat than 46BR.1G1 parental cells.

• Open in new tab
• Download powerpoint

FIG. 6.

The expression of LigI-4D mutant affects cell morphology. Morphology of 46BR.1G1, 7A3, and M10 cells. (A) Cells were grown on coverslips, and images were taken by phase-contrast microscopy at subconfluent cell density. (B) Cells were decorated with TRITC-conjugated phalloidin. Nuclei were counterstained with DAPI. Arrows show bundles of actin filaments (stress fibers). (C) Cells were decorated with antivinculin antibody and TRITC-conjugated anti-mouse secondary antibody. Nuclei were counterstained with DAPI. Arrowheads show focal contacts. Bars = 10 μm.

LigI-mediated DNA damage during the cell cycle.LigI-defective cells are characterized by a replication-dependent DNA insult that occurs “behind” the replication fork due to a delay in the maturation of Okazaki fragments. Consequently, the LigI defect is associated with an increased number of SSBs and DSBs in newly replicated DNA molecules and leads to the formation of γH2AX foci. Inspection of γH2AX foci during the cell cycle of asynchronous cell populations has revealed that a basal, low level of these foci arises during the unperturbed S phase of control MRC-5V1 fibroblasts, most likely reflecting occasional problems encountered by the replication apparatus. However, the fraction of cells with γH2AX foci drastically increases (up to 70%) in 46BR.1G1 and M10 cells. We can only speculate on the origin of DSBs in LigI-deficient cells. The increased abundance of DSBs may simply reflect the inability of 46BR.1G1 and M10 cells to efficiently deal with damage occurring at replication forks. However, the lack of activation of the ATR/Chk1 pathway, which takes care of problems encountered by the replication machinery (2), argues against this possibility. We propose that a significant fraction of DSBs in LigI-deficient cells originate during the attempt to deal with SSBs produced by the delayed maturation of Okazaki fragments. The mechanism involved in the repair of these SSBs is as yet undetermined, even though it has been suggested that unligated Okazaki fragments are eventually released by strand displacement DNA synthesis (21). Notably, we did not observe phosphorylation of RPA (see Fig. S4 in the supplemental material) and the presence of RPA foci (not shown), two hallmarks of extended single-stranded regions frequently produced during the processing of DSBs and in the presence of stalled forks.

In spite of DNA damage generated during the replication process, 46BR.1G1 and M10 cells have a normal S phase and reach mitosis. Interestingly, similar fractions of cells with γH2AX foci are detectable in S/G₂ and M phases, as if foci in mitosis originate from DNA damage in the preceding S phase. It is unclear whether γH2AX foci during mitosis actually reflect the presence of damaged DNA or only remnants of the altered chromatin structure needed for efficient processing of the damage (26). There are conflicting reports on the presence of γH2AX foci in mitosis of cultured mammalian cells in the absence of exogenous sources of DNA damage (15). Indeed, some γH2AX foci are visible in a small fraction of mitotic MRC-5V1 fibroblasts. However, their frequency drastically increases in LigI-deficient cells.

The fraction of cells with γH2AX foci decreases in G₁. Probably this reduction reflects the activity of DNA repair pathways that involve LigIII and LigIV. Notably, the level of LigIII, which is involved in SSB repair, is enhanced in 46BR.1G1 cells compared to normal cells (17). We speculate that repair of SSBs and DSBs before the G₁/S border may avoid collapse or stalling of replication forks during the ensuing S phase, a condition that would lead to activation of the ATR/Chk1 pathway, a block of cell cycle progression, and ultimately to apoptosis. It is plausible that the activation of the ATM/Chk2 pathway observed in LigI-defective cells is required to delay initiation of DNA synthesis until damage originating in the previous S phase is repaired. Notably, one of the two branches of the ATM checkpoint, namely, the one involving the MRN complex and SMC1 (19), is not activated in LigI-defective cells. This pathway has been suggested to operate in cis by controlling origin firing in the presence of damaged DNA.

LigI defect, checkpoint, and cell morphology.It is difficult to understand how LigI-defective cells can progress through mitosis with an unusual chromatin organization, marked by a phosphorylated H2AX histone variant, or, even worse, in the presence of damaged DNA. Under these conditions, one would expect a block of cell cycle progression through the activation of the G₂/M checkpoint that involves the ATR/Chk1 pathway. However, we did not observe constitutive activation of Chk1 in M10 and 46BR.1G1 cells, even though this kinase is still activated in response to DNA-damaging agents such as etoposide (16). Failure to activate Chk1 is associated with a poorly understood phenomenon, called “adaptation,” in which cells can divide despite the presence of unrepaired DNA damage (27). Adaptation may facilitate elimination of cells with excessive DNA damage by allowing their progression through mitosis with the ensuing activation of apoptotic pathways. The risk exists, however, that some cells carrying small amounts of DNA damage might survive and eventually give rise to tumors. Interestingly, 46BR.1G1 and M10 cells do not show an increased apoptotic index. Thus, defective processing and maturation of Okazaki fragments may represent a potentially dangerous challenge for the organism. In this regard, it is worth noting that the murine model of LigI deficiency is characterized by an increased incidence of spontaneous cancers with a diverse range of epithelial tumors, particularly cutaneous adnexal tumors that are rare in mice (9). Defective DNA replication has been also observed at early stages of colon carcinogenesis by using the BrdU comet assay (14).

Intriguingly, LigI deficiency impacts cell morphology, another aspect relevant for cancer progression. The effect is particularly evident if one compares 7A3 with M10 cells. Cells expressing the LigI-4D mutant, in fact, are smaller and show a reduction of filopodia. These morphological changes are accompanied by the reorganization of actin filaments, with the reduction of stress fibers, and by the redistribution of vinculin that leads to the loss of focal contacts. A less dramatic difference from 7A3 is detectable in the case of 46BR.1G1 cells and mainly concerns the organization of the actin in stress fibers. The nature of the signal that triggers these reorganizations is still a matter of speculation. We hypothesize that the amount and type of DNA damage could impact the cytoskeleton reorganization through the activation of checkpoint kinases. Remarkably, a significant fraction of ATM substrates correspond to proteins involved in cell structure and motility (13). Recently, DNA damage has been shown to induce cell senescence (6). Most M10 cells, though, do not show the typical morphological markers associated with DNA damage-induced cell senescence. However, senescence-associated β-galactosidase is expressed in a fraction (<10%) of M10 cells (data not shown). As expected, these cells are flatter and larger than most M10 cells. It is presently unclear whether cell senescence is the final outcome of M10 cells or whether a threshold in the amount of DNA damage dictates the choice between two alternative programs, i.e., the morphological changes described in this article or cell senescence.