Ku86 preserves chromatin integrity in cells adapted to high NaCl

Abstract

Cells adapted to high NaCl have many DNA breaks both in cell culture and in the renal inner medulla in vivo; yet they survive, function, and even proliferate. Here, we show that Ku86 is important for maintaining chromosomal integrity despite the continued presence of DNA breaks. The Ku heterodimer is part of DNA-dependent PK (DNA-PK), a complex that contributes by nonhomologous end joining to repair of double-strand breaks. We demonstrate that cells deficient in Ku86, but not cells deficient in DNA-PKcs (the catalytic subunit of DNA-PK), are hypersensitive to high NaCl as manifested by profound inhibition of proliferation, aberrant mitosis, and increased chromosomal fragmentation. Lower eukaryotes, including the soil nematode Caenorhabditis elegans, lack a DNA-PKcs homologue but are able to adapt to high NaCl. We show that cells of C. elegans adapted to high NaCl have many DNA breaks, similar to the mammalian cells adapted to high NaCl. Ku86 mutant C. elegans as well as C. elegans fed with cku86 dsRNA also display hypersensitivity to high NaCl, characterized by a reduced number of progeny and prolonged generation time in high NaCl. We propose that Ku86 ameliorates the effects of high NaCl-induced DNA breaks in adapted cells by supporting alignment of the broken ends of the DNA and thus maintaining integrity of the fragmented chromatin.

We previously found that mIMCD3 cells, when adapted to high NaCl, contain numerous DNA breaks; yet the cells survive and even proliferate rapidly in culture. Also, numerous DNA breaks exist in the renal inner medulla, where the cells are normally exposed to high NaCl as part of the operation of the urinary concentrating mechanism. The DNA breaks persist as long as NaCl remains high but are repaired rapidly when the level of NaCl is lowered both in culture and in vivo (1-3). Little is known about how high NaCl induces the DNA breaks, why so many remain unconnected, and what, if any, processing they undergo. In particular, it is mysterious how the cells survive and function as well as they do despite the DNA breaks. We hypothesized that there are mechanisms that reduce the rate at which the breaks rejoin, because the number of breaks remains high, and also mechanisms that ensure chromatin integrity despite numerous DNA breaks. Here, we screened several proteins known to bind to DNA breaks in an attempt to identify any that might be ameliorating the effects of the breaks. Finding that a disproportionate amount of Ku86 is retained in the insoluble DNA-bound fraction from cells adapted to high NaCl, we further investigated its function in that condition.

Ku is a heterodimeric DNA end-binding complex composed of two proteins, Ku70 (69 kDa) and Ku86 (83 kDa). It is a major component of the complex involved in nonhomologous end joining of DNA double-strand breaks (DSBs). It binds with high affinity to DNA ends, independent of their end sequence or structure. It is a part of the DNA-dependent PK (DNA-PK) complex, a multisubunit serine/threonine kinase that includes a catalytic subunit, DNA-PKcs, and a DNA end-binding component, Ku. Cells or animals lacking this function because one or more of the subunits is defective are profoundly deficient in V(D)J recombination and in protective response to ionizing radiation and various radiomimetic agents (reviewed in refs. 4-6). In addition, Ku86^-/- cells and mice exhibit growth defects and a marked increase in chromosomal aberrations (7-9). Biochemical studies and the structure of the Ku heterodimer suggest that it acts as an alignment factor to promote end joining (10).

We used mutant mammalian cell lines to study possible involvement of Ku in adaptation to high NaCl. Cells deficient in Ku86, but not cells deficient in DNA-PKcs, are hypersensitive to high NaCl as manifested by profound inhibition of proliferation, aberrant mitosis, and increased chromatin fragmentation. Similar to the mammalian cells, cells of Caenorhabditis elegans living in high NaCl contain numerous DNA breaks. Further, Ku86 mutant C. elegans as well as C. elegans fed with cku86 dsRNA are hypersensitive to high NaCl, characterized by a reduced number of progeny and prolonged generation time. These results suggest that, during exposure to high NaCl, Ku maintains chromatin integrity by tethering broken ends of DNA.

Methods

Cell Cultures. Cell lines. Mouse inner medullary collecting duct cells (mIMCD3) (11) were provided by S. Gallans (Harvard Medical School, Boston) and used in passages 12-20. The xrs5 cell line (no. CRL-2348, American Type Culture Collection) is an x-ray-sensitive CHO mutant cell line that was derived from CHO-K1 cells (no. CCL-61, American Type Culture Collection) by treating the cells with ethyl methanesulphonate. These cells belong to x-ray complementation group 5 and are mutant in the p86 subunit of the Ku autoantigen (12, 13). M059J cells (no. CRL-2366, American Type Culture Collection) were isolated from a tumor specimen taken from a 33-year-old male with untreated malignant glioblastoma. M059K cells (no. CRL-2365) were isolated concurrently from the healthy part of the same specimen as M059J. M059J cells lack DNA-PK activity due to failure to express the p350 DNA-PKcs and are ≈30-fold more sensitive to ionizing radiation than are M059K cells (14, 15). Ku86^-/- mouse embryonic fibroblasts (MEFs) are described in ref. 7. The cells were cultured in medium containing 45% DME Low Glucose (Irvine Scientific), 45% mF-12 (Invitrogen), and 10% FBS (HyClone) at 37°C in a humidified atmosphere of 95% air/5% CO₂.

DNA-PK^-/- MEFs. To generate the MEFs, DNA-PK^+/- mice were crossed, and embryos were isolated 13.5 days after plug formation as described in ref. 9. Fibroblasts prepared from DNA-PK^+/+ and DNA-PK^-/- embryos were used for the experiments. The cells were grown in DMEM (Invitrogen) containing 15% FBS (HyClone) and used in passage 1-5.

Osmolality of control (“isotonic”) medium was 300 milliosmolal (mosmol)/kg. Hypertonic medium was prepared by adding NaCl to the total osmolality indicated in all figures and in the text. To adapt cells to high NaCl, osmolality of the medium was increased every 2 days in increments of 50 mosmol/kg. Cells were passaged when ≈90% confluent to maintain a logarithmic rate of growth.

Isolation of DNA-Bound and Soluble Proteins, Western Blotting, and Immunodetection. Cells were rinsed with PBS, adjusted with NaCl to the same osmolality as the medium, and then lysed with RIPA lysis buffer [50 mM Tris·HCl/1% Nonidet P-40/150 mM NaCl/1 mM EDTA/1 mM NaF/1 mM Na₃VO₄/protease inhibitors (no. 1836170, Roche Molecular Biochemicals)]. The lysates were collected in 1.5-ml tubes and placed in ice for 10 min. DNA and the proteins bound to it were pelleted by centrifugation at 800 × g. Supernatant containing soluble proteins was transferred to separate tubes, and protein concentration was measured by using the BCA Protein Assay (Pierce). Pellets were washed three times with 500 ml of the lysis buffer. The final pellet, which contained DNA and insoluble proteins, including those bound to DNA, was boiled for 5 min in 40 μl of Laemmli sample buffer (no. 161-0737, Bio-Rad) to denature the proteins and release them from DNA. DNA was pelleted by centrifugation at 18,000 × g, and the proteins in the supernatant were identified by Western analysis. Loading of the gels was normalized to equal numbers of cells, calculated from the amounts of soluble proteins in the corresponding fractions. Because histones are the most abundant DNA-bound proteins, readily visible in Coomassie-stained gel between the 10- and 20-kDa areas, we used their abundance to verify equal loading. Samples were separated by SDS/PAGE. Immunoblots used antibodies against Mre11 (PC388, Oncogene Science), Ku86 (Santa Cruz Biotechnology), Rad51 (PC130, Oncogene Science), and Nbs1 [a rabbit polyclonal antibody against amino acids 733-751, (C)KEESLADDLFRYNPNVKRR (16)].

DNA Content and Mitotic Index. Mitosis was identified by immunostaining with anti-phosphohistone H3 antibody (a mitotic marker) (no. 06-570, Upstate Biotechnology, Lake Placid, NY). DNA was stained with propidium iodide. DNA content per cell and percentage of mitotic cells were determined by laser scanning cytometry as described in ref. 17.

Cell Number. Several methods were used to estimate cell number. Staining of DNA with SYBR Green. Cells grown on 96-well plates were fixed with 100% methanol, washed two times with PBS, and stained with SYBR Green (no. 4250-050-05, Trevigen, Gaithersburg, MD). Integral green fluorescence, which estimates total amount of DNA in each well, was measured with a Wallac 1420 multilabel counter.

Hemocytometer. To estimate doubling time, cells were counted over several passages by using a manual cell counter (Hausser and Son, Philadelphia). Cells were maintained in logarithmic growth.

Preparation and Analysis of Metaphase Chromosome Spreads. Standard methods were used. Briefly, cells were incubated with the microtubule inhibitor colcemide (0.01 μg/ml) for 16 h, trypsinized, and incubated in hypotonic medium (0.075 M KCl) at 37°C for 15 min. Then, cells were fixed by three changes (for 30 min each time) of methanol/acetic acid (3:1) and spread on slides. Cells were stained with DAPI to visualize DNA and observed with a fluorescence microscope. To estimate the degree of chromatin fragmentation, the number of intact chromosomes and chromosome fragments was counted in a given spread. Ten to 30 chromosome spreads were scored for each condition. Cells that had been adapted to high NaCl were returned to isotonic medium for 2 h before addition of colcemide because we were unable to obtain useable spreads directly from high NaCl.

C. elegans. Strains and culture. Bristol N2 (WT) and cku80(ok861) strains were provided by the Caenorhabditis Genetics Center (CGC) (Minneapolis). The cku80(ok861) strain contains a homozygous 1,646-bp deletion, including a large section of coding sequence, in the cku80 locus (the sequence name is R07E5.8) (Fig. 7A, which is published as supporting information on the PNAS web site). The deletion was confirmed by PCR (Fig. 7B). The worms were grown on nematode growth medium agar plates spread with Escherichia coli bacterial strain OP50 (obtained from the CGC). Cultures were maintained at room temperature (≈20°C). Control nematode growth medium contains 51 mM NaCl, 1 mM MgSO₄, 1 mM CaCl₂, 25 mM KPO₄, 5 μg/ml cholesterol, 2.5 g/liter peptone, and 17 g/liter agar (18). The osmolality of this medium was increased by addition of 300 mM/liter NaCl.

Adaptation to high NaCl. Chunks of agar containing worms at all stages of development were placed on control plates and on plates made hyperosmotic by adding 300 mM NaCl to the agar. After 5 days, chunks of agar from the control and high-NaCl plates were transferred to identical new plates. Several days later, worms were present at all stages of development, with those on the hyperosmotic plates evidently having adapted to the high NaCl.

Detection of DNA breaks in C. elegans adapted to high NaCl. Control worms and worms adapted to high NaCl were washed off of the plates with ice-cold 4% paraformaldehyde, frozen immediately on dry ice, thawed, and then left in 4% paraformaldehyde overnight at 4°C. Fixed worms were paraffin-embedded, and sections were cut and mounted on silanized slides (American HistoLab, Gaithersburg, MD). DNA breaks were detected by in vitro labeling of their 3′ OH ends with BrdUTP in a reaction catalyzed by terminal deoxynucleotidyltransferase, as described in ref. 2.

RNA interference (RNAi). cku80 expression was reduced by feeding WT worms with E. coli [strain HT115(DE3)(pL4440)], producing dsRNA (19) homologous to the cku80 ORF. The cku80 ORF-RNAi feeding clone was obtained from Open Biosystems (Huntsville, AL). Presence of the cku80 ORF in the RNAi feeding vector was verified by sequencing.

dsRNA feeding conditions. Five milliliters of LB broth containing 100 μg/ml ampicillin was inoculated with a single colony of HT115(DE3)(pL4440) bacteria grown on plates supplemented with ampicillin (100 μg/ml) and tetracycline (12.5 μg/ml) and then cultured overnight at 37°C. The culture was diluted 100-fold, grown for 2 h, and divided between two flasks. Isopropyl-β-d-thiogalactopyranoside (IPTG) was added into one flask to a final concentration of 1 mM to induce synthesis of dsRNA, and both cultures were incubated for additional 4 h. The bacteria were concentrated by centrifugation and applied onto agar plates containing 100 μg/ml ampicillin. Plates had four different supplementations: (i) none, (ii) 1 mM IPTG, (iii) 300 mM NaCl, or (iv) 300 mM NaCl plus 1 mM IPTG.

Estimation of number of progeny and generation time. L2/L3 control larvae or larvae adapted to high NaCl were transferred to new plates (one animal per plate). Three to eight plates were prepared for each condition. Each worm was monitored separately for development into a gravid adult, egg-laying, and production of the next generation of larvae (Fig. 5A). When larvae reached the L2/L3 stage, one larva from each plate was transferred to new plate. The number of progeny and the generation time (time between transfers to new plates) were recorded for every worm through three generations (F1-F3 in Fig. 5A).

Fig. 5.

cku80 deficiency compromises adaptation of C. elegans to high NaCl. Adaptation of C. elegans to high NaCl was assessed by the effects on number of progeny and generation time. (A) Experiment design. A single L2/L3 larva adapted to low or high NaCl was transferred to a new plate. Each worm was monitored separately for development to a gravid adult, laying eggs, and development of next generation of larvae. When larvae had reached the L2/L3 stage, one larva from each plate was transferred to a new plate. Number of progeny and generation time (time between transfers to new plate) were recorded for every worm during three generations (F1-F3). (B) cku80^-/- C. elegans grown on high-NaCl plates have decreased progeny number and increased generation time compared with WT worms [mean ± SE, n = 12 (three generations of four worms); ^*, P < 0.02, cku80^+/+ vs. cku80^-/-, Student's t test). (C) Effect of reducing cku80 expression by RNAi. WT C. elegans grown on high-NaCl plates containing E. coli producing cku80 dsRNA [cku80(RNAi)] have decreased progeny number and increased generation time compared with control worms not subjected to RNAi [mean ± SE, n = 4-8; ^*, P < 0.05, control vs. cku80(RNAi), Student's t test].

Results

To identify proteins that might be important for stabilizing the DNA breaks induced by high NaCl, we screened several proteins known to bind to DNA breaks for their retention in the insoluble, chromatin-bound fraction after exposure to high NaCl (Fig. 1). We reasoned that proteins bound to DNA would be retained in the insoluble pellet that is left after treating cells with a nondenaturing lysis buffer. High NaCl reduces the total amount of DNA-bound protein (Fig. 1A). However, DNA binding of individual candidate proteins differs greatly (Fig. 1B). Binding of Mre11 and its binding partner Nbs1, which lacks its own DNA-binding domain, is much reduced by high NaCl, consistent with the previously observed high NaCl-induced translocation of Mre11 from nucleus to cytoplasm (1). Rad51 binds to the ssDNA that is formed during homologous recombination, during the processing of DSBs, and in stalled replication forks (20). High NaCl decreases the amount of Rad51 in the insoluble chromatin-bound fraction (Fig. 1B). In contrast, Ku86 remains in the insoluble fraction from the cells exposed to high NaCl (Fig. 1B). Although high NaCl does not increase the amount Ku86 bound to chromatin, in relative terms, more of it than of the other candidate proteins is retained in the insoluble fraction, consistent with the possibility that it might be important for stabilizing the DNA breaks.

Fig. 1.

During exposure to high NaCl, the insoluble DNA-bound fraction of proteins contains relatively more Ku86 than other candidate proteins that might bind to DNA breaks. mIMCD3 cells were acutely exposed to high NaCl by changing the cell culture media to one made hypertonic by adding NaCl to a final osmolality of 500 mosmol/kg or were adapted to high NaCl by gradual increase of NaCl to the same final osmolality and maintaining them in the high NaCl for several passages. DNA-bound proteins were isolated and analyzed by Western blot as described in Methods. (A) DNA-bound proteins separated on 4-12% gel and stained with Coomassie blue. (B) Western blot of DNA-bound proteins.

We used several cell culture models in which Ku86 is mutated or knocked out to elucidate its possible role in adaptation to high NaCl. The first model was the radiosensitive xrs5 mutant cell line, derived from CHO-K1 cells by treating them with ethyl methanesulphonate (12) and later shown to be deficient in Ku86 (13, 21). xrs5 cells conspicuously fail to adapt to high NaCl. After elevation of NaCl, they stop proliferating and turn into giant multinucleated cells that die within 10 days (Fig. 2A).

Fig. 2.

Lack of Ku86, but not of DNA-PKcs, prevents adaptation to high NaCl. xrs5 (Ku86 mutant) and M059J (DNA-PK mutant) cells were acutely exposed to elevation of NaCl to the final osmolality indicated and were photographed at the indicated times. Relative cell number was estimated 24 h after adding NaCl by using WST-1, as described in Methods. (A) xrs5 cells enlarge, become multinucleated, and then die within several days after NaCl is increased. (B) MO59J cells adapt to high NaCl, continuing to proliferate and not changing their appearance. (C) Twenty-four hours after NaCl is increased, the number of Ku86 mutant cells decreases much more than does the number of control (CHO-K1) cells, but the number of DNA-PKcs mutant (MO59J) cells does not, compared with its control (MO59K) (data are representative of two independent experiments that gave similar results).

During nonhomologous end joining, the Ku heterodimer binds to DNA ends and serves as an alignment factor (10). Then, DNA-PKcs is attracted and activated, and it targets other repair activities to the sites of DNA damage (reviewed in refs. 5, 6, and 22). Activity of DNA-PKcs is entirely dependent on the Ku heterodimer and is absent in Ku-deficient cells (6, 23). Thus, the phenotype could conceivably be caused by deficiency of DNA-PK activity. However, MO59J cells, which are deficient in DNA-PK (15), are able to adapt and proliferate in high NaCl (Fig. 2B). Consistent with these observations, Ku86 deficiency decreases the number of cells as early as 24 h after NaCl is added, whereas DNA-PK deficiency does not (Fig. 2C). The decreased number of xrs5 cells is accompanied by long G₂ arrest, as is evident from accumulation of cells in G₂ phase of cell cycle and the absence of the histone H3 phosphorylation that normally occurs in mitotic cells (Fig. 8, which is published as supporting information on the PNAS web site).

Because the Ku86 mutation in xrs5 cells is chemically induced, it seemed possible that some additional, unrecognized mutation is responsible for their inability to adapt to high NaCl. To further test the specific importance of Ku86 for adaptation to high NaCl, we studied MEFs derived from Ku86^-/- mice (7). Adaptation to high NaCl of Ku86^-/- MEFs is not as severely impaired as that of xrs5 cells. Ku86^-/- MEFs can proliferate in high NaCl (Fig. 3A). However, their doubling time in high NaCl is much greater than that of WT MEFs (Fig. 3B), and >30% of Ku86^-/- MEFs adapted to high NaCl are giant multinucleated cells (Fig. 3 A and C) similar in appearance to xrs5 cells after 3 days in high NaCl (Fig. 2A). In contrast, none of the MEFs derived from DNA-PK^-/- mice enlarge and become multinucleated after adaptation to high NaCl (data not shown). Also, survival of DNA-PK^-/- MEFs after acute elevation of NaCl and their doubling time after adaptation to high NaCl are similar to those of WT MEFs (Fig. 9, which is published as supporting information on the PNAS web site). Thus, the function of Ku in adaptation to high NaCl is independent of DNA-PK activity.

Fig. 3.

Ku86^-/- MEFs do not adapt to high NaCl as well as WT MEFs do. Spontaneously immortalized embryonic fibroblasts derived from WT and Ku86^-/- mice were adapted to high NaCl by gradually increasing NaCl to a final osmolality of 500 mosmol/kg. (A) WT and Ku86^-/- MEFs adapted to high NaCl for 20 days. Note the large, multinucleated Ku86^-/- MEF cells. (B) Doubling time of Ku86^-/- MEFs is greatly increased by high NaCl (mean ± SE of three independent experiments; ^*, P < 0.01, Ku86^+/+ vs. Ku86^-/-, Student's t test). (C) Percentage of large multinucleated Ku86^-/- MEF cells is greatly increased by high NaCl (mean ± SE of three independent experiments; ^*, P <0.001, Ku86^+/+ vs. Ku86^-/-, Student's t test).

There is evidence that Ku has functions independent of its role as a component of DNA-PK. For example, the growth defects of Ku86^-/- mice are much more severe than those of DNA-PK^-/- mice (7, 24, 25). Despite the existence of Ku in all eukaryotes examined, DNA-PKcs has so far only been identified in vertebrates (26). However, DNA repair by nonhomologous end joining occurs in all organisms, from yeast to man (27). One possibility is that the functions of DNA-PKcs in higher eukaryotes are carried out in lower eukaryotes by other members of the phosphatidylinositol 3-kinase-related kinase family. Alternatively, the functions of DNA-PKcs might indeed be restricted to higher organisms. Our results with Ku86 and DNA-PKcs deficient mammalian cells indicate that Ku, but not DNA-PKcs, is central to adaptation to high NaCl. Consistent with this distinction, many invertebrates, including numerous marine species (28) and the soil nematode C. elegans (29), are able to adapt to and live in a high-NaCl environment despite the absence of DNA-PKcs. Induction of DNA breaks by high NaCl was only previously shown in mouse cells (1, 2, 30). To test how general this phenomenon might be, we used TUNEL assay to test for DNA breaks in the cells of C. elegans adapted to high NaCl. We treated sections of WT C. elegans with terminal deoxynucleotidyltransferase in the presence of BrdUrd, which becomes incorporated in extensions of any disconnected 3′ OH ends of DNA, and detected BrdUrd incorporation by immunocytochemistry (31). DNA breaks are not apparent in cells of C. elegans when the NaCl concentration is low (51 mM), but there are numerous DNA breaks in cells of C. elegans adapted to high (351 mM) NaCl (Fig. 4), similar to the result in the mouse renal inner medulla, where NaCl is normally high (2).

Fig. 4.

DNA breaks in C. elegans adapted to high NaCl. TUNEL assay of DNA breaks performed on cross sections of C. elegans growing on agar containing a low (51 mM) or high (351 mM) concentration of NaCl. DNA breaks are present only after adaptation to the high NaCl.

To test whether Ku86 contributes to adaptation of C. elegans to high NaCl, we used the cku80(ok861) strain of C. elegans, which contains a homozygous deletion in the cku80 gene (Fig. 7), called cku80^-/- here. We analyzed the effect of this cku80 deletion on progeny number and generation time of the C. elegans exposed to low vs. high NaCl (Fig. 5B). There is no significant difference between WT and cku80^-/- C. elegans grown on control plates containing low NaCl (Fig. 5B). However, cku80^-/- C. elegans grown on high-NaCl plates have fewer progeny and increased generation time compared with WT worms (Fig. 5B). This result suggests evolutionary conservation of the role of Ku in adaptation to high NaCl.

However, the cku80^-/- strain was created by chemical mutagenesis, leaving the possibility that some additional mutation might be present and be responsible for the increased sensitivity to high NaCl. To confirm that cku80 is specifically important for the adaptation, we tested whether adaptation to high NaCl is compromised by feeding WT C. elegans with cku80 dsRNA (RNAi) (19). In fact, cku80 dsRNA produces a phenotype similar to that found in the cku80(ok861) strain. When WT C. elegans are grown on low-NaCl plates, there is no significant difference in the number of progeny or generation time whether or not they are fed cku80(RNAi) (Fig. 5C). However, on high-NaCl plates, C. elegans subjected to cku80(RNAi) have significantly fewer progeny and increased generation time compared with controls (Fig. 5C).

How does Ku contribute to adaptation to high NaCl? Loss of Ku86, even at normal NaCl concentration, leads to chromosomal instability, growth retardation, and decreased proliferation in vitro (7, 9), which have been proposed to result from inability to repair DNA DSBs incurred during normal DNA metabolism (7). Further, it was proposed that the gross chromosomal rearrangements in Ku86^-/- mice and in Ku86^-/- cells in culture result from lack of the Ku-mediated tethering of broken DNA ends, which is necessary for their subsequent ligation (4, 32). Exposure of cells to high NaCl greatly increases the number of DNA breaks both in culture and in the renal inner medulla in vivo (2, 30). Because these breaks persist as long as NaCl remains high (2), viability could require some mechanism to tether the broken DNA, preserving chromatin integrity. The Ku heterodimer is an attractive candidate for performing this function. Ku70 and Ku80 form a dyad-symmetrical molecular ring that encircles DNA (10). In this way, the Ku heterodimer structurally supports broken DNA ends and acts as an alignment factor. This function of Ku is evolutionary conserved and does not require DNA-PKcs. We reasoned that, if Ku is tethering the broken DNA ends when NaCl is high, then absence of Ku should increase chromatin fragmentation. To test this hypothesis, we prepared metaphase chromosome spreads from WT and Ku86^-/- MEFs after exposure to high NaCl and examined their structure. Chromosomes from WT MEFs are intact regardless of the level of NaCl (Fig. 6A). However, metaphase spreads from Ku86^-/- MEFs contain many more small fragments of broken chromatids, and the fragmentation is greatly increased by high NaCl (Fig. 6A).

Fig. 6.

Ku86 deficiency increases chromosomal fragmentation, especially after exposure to high NaCl. (A) Number of small chromosomal fragments (arrow in Inset) is increased in Ku86^-/- MEFs after exposure to high NaCl compared with WT. Metaphase chromosome spreads were prepared from control cells and cells exposed to high NaCl (final osmolality = 500 mosmol/kg) for several passages (mean ± SE of three independent experiments; ^*, P <0.01, NaCl vs. control, Student's t test). (B) Model of the proposed role of Ku in response to high NaCl. Ku maintains chromatin integrity by tethering the ends of the DNA breaks induced by high NaCl.

Discussion

The discovery that cells adapted to high NaCl contain many DNA breaks yet can proliferate in culture and survive and function in the renal inner medulla in vivo raises many questions (2, 3). How does high NaCl cause the DNA breaks? Why they are not repaired immediately? Why they do not activate apoptosis? How do cells manage to replicate and transcribe their broken DNA? Are the high NaCl-induced breaks dangerous, or do they somehow confer an advantage in presence of high NaCl? Most of these questions still lack an answer. However, the results of the present study identify the importance of the DNA end-binding protein Ku86 for adaptation to high NaCl. Cells that lack Ku86 have retarded growth, aberrant mitosis, and increased chromatin fragmentation in response to high NaCl.

Ku has a central role in repair of DNA DSBs (33). It binds to DNA double-strand ends and recruits the additional factors that are required to repair DNA by means of nonhomologous end joining (reviewed in ref. 34). The Ku heterodimer binds to the free DNA ends at the site of a break, keeping them aligned. DNA-PKcs joins the DNA-bound Ku to form the DNA-PK complex and is activated by phosphorylation. In addition to tethering the DNA ends, this interaction protects them from nuclease digestion before recruitment of the XRCC4/ligase IV complex to the break site by DNA-PK and ligation of the DNA breaks (reviewed in ref. 34). Ku also functions at telomeres, protecting chromosomes from end-to-end fusion (35).

Thus, Ku apparently has at least two functions. On one hand, it promotes DNA end joining, leading to DNA repair. On the other hand, it protects against end-to-end fusion of chromosomes. Are either of these functions of Ku involved in adaptation to high NaCl? Because high NaCl induces DNA breaks, one might expect that Ku should bind to the breaks and originate their repair as it does after induction of DNA damage by other agents. However, this usual role of Ku is ruled out by observations that repair of DNA is inhibited and the breaks persist as long as NaCl remains high (1, 2). Another possibility is that the binding of Ku to telomeres is involved. In this case, absence of Ku should lead to shortening of telomeres and fusion of the ends of chromosomes, resulting in growth retardation and defects in mitosis similar to those we observe in Ku86 mutants in high NaCl. However, the major chromosomal defect that we observe after exposure of Ku86^-/- MEFs to high NaCl is production of small chromosomal fragments, not fused chromosomes (Fig. 6A). Taking this observation into consideration, we propose the model for Ku's function that is presented in Fig. 6B. We hypothesize that Ku's main protective function in the presence of high NaCl is to tether the broken ends of DNA and thus preserve chromosomal integrity, despite the continued presence of broken DNA. In this view, the small chromosomal fragments that occur when NaCl is high and Ku is not present (Fig. 6A) might be formed as a consequence of chromatin fragmentation at the nonaligned ends of broken DNA.

Given that cells adapted to high NaCl have numerous DNA DSBs (2), it is unclear how they still manage to replicate and transcribe DNA. Our present findings implicate Ku in the process. The fact that DNA-PKcs deficiency does not prevent the adaptation to high NaCl but Ku86 deficiency does (Figs. 2 and 9) suggests that alignment of DNA ends is important (Fig. 6B), not DNA-PK kinase activity. One possibility is that local breaks are transiently ligated as needed, so that essential processes can occur at those locations despite the continued presence of many other breaks. Alternatively, there might be a mechanism by which the transcription and replication can proceed over the gaps. In either case, the process apparently is facilitated by tethering of the broken ends by Ku. Further, it seems likely that in addition to Ku, other proteins must participate, and it will be important to identify them. Mre11, which is an essential early participant in repair of DSBs caused by ionizing radiation, probably is not involved because high NaCl causes it to move out of the nucleus (1, 2) and reduces its binding to DNA (Fig. 1). The facts that high NaCl induces DNA breaks not only in mammalian cells but also in a lower eukaryote such as C. elegans (Fig. 4) and that Ku deficiency compromises their adaptation to high NaCl (Fig. 5) highlight the conservation of Ku's role in adaptation to high NaCl and its general importance in preserving DNA functions in cells exposed to high NaCl.