Double-strand break repair by homologous recombination in primary mouse somatic cells requires BRCA1 but not the ATM kinase

Abstract

Homology-directed repair (HDR) is a critical pathway for the repair of DNA double-strand breaks (DSBs) in mammalian cells. Efficient HDR is thought to be crucial for maintenance of genomic integrity during organismal development and tumor suppression. However, most mammalian HDR studies have focused on transformed and immortalized cell lines. We report here the generation of a Direct Repeat (DR)-GFP reporter-based mouse model to study HDR in primary cell types derived from diverse lineages. Embryonic and adult fibroblasts from these mice as well as cells derived from mammary epithelium, ovary, and neonatal brain were observed to undergo HDR at I-SceI endonuclease-induced DSBs at similar frequencies. When the DR-GFP reporter was crossed into mice carrying a hypomorphic mutation in the breast cancer susceptibility gene Brca1, a significant reduction in HDR was detected, showing that BRCA1 is critical for HDR in somatic cell types. Consistent with an HDR defect, Brca1 mutant mice are highly sensitive to the cross-linking agent mitomycin C. By contrast, loss of the DSB signaling ataxia telangiectasia-mutated (ATM) kinase did not significantly alter HDR levels, indicating that ATM is dispensable for HDR. Notably, chemical inhibition of ATM interfered with HDR. The DR-GFP mouse provides a powerful tool for dissecting the genetic requirements of HDR in a diverse array of somatic cell types in a normal, nontransformed cellular milieu.

DNA damage poses a threat to genomic integrity and must be repaired in an accurate and timely manner for the health and survival of the organism. A particularly cytotoxic lesion is a chromosomal double-strand break (DSB), which can arise from endogenous sources, including DNA replication and antigen receptor rearrangements in lymphocytes, as well as exogenous sources, such as ionizing radiation (IR) (1, 2). DSBs activate an elaborate cellular signaling network of proteins, a key component of which is the ataxia telangiectasia-mutated (ATM) protein kinase (3, 4). There are three major pathways for repairing DSBs in mammalian cells: (i) homology-directed repair (HDR), (ii) nonhomologous end joining (NHEJ), and (iii) single-strand annealing (SSA) (2, 5, 6). HDR, which is considered the most precise of the three repair pathways, uses a homologous donor sequence as a template for the repair event. Template preference is biased to the sister chromatid in mammalian cells, thus restoring the original sequence before damage. NHEJ refers to joining of ends without the use of extensive homology, and it is often accompanied by modification of the sequence surrounding the break site (1). SSA occurs when complementary strands from sequence repeats flanking the DSB anneal to each other, resulting in repair of the DSB but deletion of the intervening sequence (5). Consequently, both NHEJ and SSA are considered to be more error-prone than HDR. The importance of HDR to the organism is emphasized by the requirement for HDR factors for development, tumor suppression, and fertility (5).

HDR was established as a DSB repair pathway in mammalian cells using genomic reporters into which a DSB is introduced by the I-SceI endonuclease (7–9). Repair of the DSB from the homologous template on the sister chromatid or the same chromatid leads to expression of a scoreable marker, such as GFP from the DR-GFP reporter (10). Use of such reporters has led to direct evidence for the role of several proteins in HDR, including the breast cancer suppressors BRCA1 and BRCA2 (11, 12). However, these studies have been performed in transformed cell lines and immortalized ES cells, raising questions as to their relevance to normal somatic cells. Some studies have suggested that DSB repair pathways are developmentally regulated, with HDR being critical during embryonic cell cycles and NHEJ in differentiated tissues (13). Furthermore, a critical protein in HDR, the RAD51 recombinase, has been reported to be overexpressed in tumor cell lines, suggesting that HDR levels are altered in transformed cells (14, 15).

In the present study, we describe a mouse model to analyze the efficiency of HDR in primary somatic cell types. To this end, the DR-GFP reporter was targeted to a defined chromosomal locus in the mouse genome. We found that primary somatic cell types derived from different lineages from DR-GFP mice, including fibroblasts, brain (glial) cells, and mammary epithelial cells, undergo HDR at I-SceI–induced DSBs at similar frequencies. The frequency of HDR is lower in the tested somatic cell types compared with ES cells, which correlates with the lower fraction of cells in S phase in the more differentiated cell types. To examine the genetic requirements for HDR in somatic cells, we crossed the DR-GFP reporter into mice carrying mutations in the Brca1 and Atm genes. We found that BRCA1 is necessary for efficient HDR in somatic cells. By contrast, ATM is not required for HDR in primary fibroblasts, although chemical inhibition of ATM can interfere with HDR.

Results

Gene Targeting of the HDR Reporter DR-GFP to the Mouse Pim1 Locus.

Repair of DSBs by HDR in mitotically dividing cells occurs primarily by a non-crossover gene conversion mechanism, which in the DR-GFP reporter, restores a functional GFP gene (10). DR-GFP consists of two mutated GFP genes, SceGFP, which is disrupted at an LweI restriction site by the 18-bp recognition site for the I-SceI endonuclease, and iGFP, which is truncated at both the 5′ and 3′ ends (Fig. 1A). I-SceI cleavage of SceGFP followed by gene conversion with iGFP gives rise to a GFP+ gene, which is expressed from a β-actin promoter and CMV enhancer for broad expression in mouse cells (16). After an HDR event has occurred, cells are stably GFP+ and hence, constitutively express GFP, which is detectable by flow cytometry.

Fig. 1.

Generation of DR-GFP mice for the analysis of HDR in primary cells from various tissues. (A) The DR-GFP reporter was targeted to the Pim1 locus on chromosome 17 in ES cells, creating the Pim1^Drgfp allele. H, HincII. (B) Southern blot analysis of HincII-digested genomic DNA from Pim1^+/Drgfp mice using the Pim1 probe shown in A. (C) PCR analysis of genomic DNA from Pim1^+/Drgfp mice using Pim1 primers shown in A. (D) HDR is a prominent DSB repair pathway in primary fibroblasts from DR-GFP mice. Primary MEFs and ear fibroblasts from Pim1^+/Drgfp mice were transfected with an I-SceI expression vector and compared with Pim1^+/Drgfp ES cells for HDR levels by flow cytometry. Error bars in all figures represent SDs. The number of experiments for each cell type (independent transfections for ES cells, individual embryos for MEFs, and individual animals for ear fibroblasts) is provided under the graph. P values compared with ES cells were determined from an unpaired t test.

To analyze HDR in primary cells from the mouse, we introduced the DR-GFP reporter into a defined locus in the mouse genome (Fig. 1A). We chose the Pim1 locus on chromosome 17, because Pim1-deficient mice are viable and fertile (17, 18) and promoterless targeting vectors for this locus target at high efficiency (12, 19). We introduced the targeting vector into germ line-competent murine ES cells and after selection with hygromycin, isolated and expanded several clones for Southern blot analysis. Consistent with previous reports, ∼90% of the clones were correctly targeted, generating the Pim1^Drgfp allele. Chimeric mice derived with the targeted ES cell clones transmitted the Pim1^Drgfp allele through the germ line (Fig. 1 B and C). Pim1^Drgfp/Drgfp female and male mice were fertile and typically mated with nontransgenic mice to derive litters for analysis in which all progeny were hemizygous for the Pim1^Drgfp locus.

HDR in Primary Embryonic and Adult Fibroblasts.

Although HDR has been suggested to have a major contribution to DSB repair in rapidly cycling ES cells, the contribution of HDR to DSB repair in primary somatic cells is less clear. To determine if HDR contributes significantly to DSB repair in primary cells, we isolated mouse embryonic fibroblasts (MEFs) from embryonic day 12.5 Pim1^+/Drgfp embryos. Early passage (P2 or P3) cells were transiently transfected with the I-SceI expression vector and analyzed by flow cytometry 48 h posttransfection. Although GFP+ cells were not detected in mock-transfected controls (≤0.01%), a significant fraction of the transfected cell population was GFP+ (1.3 ± 0.6%), indicating DSB induction of HDR (Fig. 1D).

Ear fibroblasts from 2- to 4-mo-old Pim1^+/Drgfp mice were also examined for HDR. As with MEFs, we observed a distinct GFP+ cell population for ear fibroblasts after expression of I-SceI, such that 0.8 ± 0.3% were GFP+ (Figs. 1D and 2A). Spontaneous events were again low (≤0.01%), indicating that HDR events were DSB-dependent. PCR across the I-SceI site indicated that DSBs were also repaired by NHEJ in addition to HDR (Fig. S1A).

Fig. 2.

HDR varies over ∼2.5-fold range for primary mouse cell types. Primary cells from various tissues were transfected with an I-SceI expression plasmid, and HDR was assayed by flow cytometry 2 d (A; ear fibroblasts), 8 d (B; neonatal brain), 12–14 d (C; ovary), or 3 d (D; mammary epithelium) after transfection. Although I-SceI expression and hence, HDR are highest ∼24–48 h after transfection, the longer time in culture allowed neonatal brain and ovarian cells to recover from the transfection and express GFP. Representative fluorescence plots are shown for the indicated cell types, with the mean percent GFP+ cells indicated. Spontaneous homologous recombination events were rare in all cases, because GFP+ cells were ≤0.01% in the absence of I-SceI expression. The number of experiments for each cell type is as follows: ear fibroblasts, 11; neonatal brain, 3; ovary, 6; and mammary epithelium, 11. Each experiment used passage 2 or 3 cells derived from 2 to 10 mice.

By comparison, ES cells had a three- to fourfold higher number of GFP+ cells (3.7 ± 0.8%) (Fig. 1D). PCR across the I-SceI site indicated that total apparent DSB repair was no higher in ES cells than ear fibroblasts (Fig. S1A). However, ES cells have about two times as many cells in S and G2 compared with MEFs and ear fibroblasts (Fig. S2A), which likely contributes to higher HDR levels.

HDR in Other Primary Cell Types: Neonatal Brain, Adult Ovary, and Virgin Mammary Gland.

We also investigated HDR levels in primary cells from other tissues. In the nervous system, both HDR and NHEJ defects have been associated with developmental abnormalities and tumorigenesis (20). To assess HDR levels in the developing brain, primary cultures were derived from neonatal brains, which primarily consist of cells of glial origin (21). In keeping with a cycling cell population, the fraction of cells in G1 at early passage (P2–P4) was ∼60%, similar to early passage fibroblasts (Fig. S2A). Cells were transfected at early passage and assayed for HDR 8 d later. Although spontaneous recombination events were rare, DSB-induced HDR in these neonatal brain cultures was readily apparent (1.7 ± 0.1% GFP+ cells) (Fig. 2B).

Because defects in HDR are associated with breast and ovarian cancers, we also examined DSB repair in primary cultures of cells from the mammary gland and ovary (5). Ovarian cultures consisting of epithelial, stromal, granulose, and germ cells were established from 1- to 2-mo-old mice (22). Transient transfection of the ovarian cultures with the I-SceI expression vector gave rise to 0.9 ± 0.5% GFP+ cells as assayed 12–14 d posttransfection (Fig. 2C). The fluorescence spectrum for the ovarian cells was broader than for the fibroblasts or neonatal brain culture, and the experimental variation in HDR was also greater, which would be expected for a more complex cell population (Fig. 2C).

Primary mammary cultures were derived from the fourth inguinal mammary glands of ∼8-wk-old virgin female Pim1^+/Drgfp mice. Epithelial organoids were isolated by differential centrifugation of collagenase-disrupted glands. Transient I-SceI expression in early passage epithelial cultures (P1 or P2) gave rise to 0.65 ± 0.33% GFP+ cells 72 h posttransfection (Fig. 2D). PCR across the I-SceI site indicated that total apparent DSB repair in primary mammary epithelial cells was similar to total DSB repair in other cell types (Fig. S1A) along with I-SceI expression (Fig. S1B). Thus, although there is substantial variation in cell types derived from different tissues, the range of HDR is only ∼2.5-fold.

When analyzing HDR in the mammary gland, we found that nearly all cells from one animal (92%) were GFP+ (Fig. 3A). Analysis of the tail tip DNA from this mouse indicated that the DR-GFP reporter had undergone a spontaneous homologous recombination event in the germ line of one of its parents, such that this mouse was constitutionally GFP+. The high percentage of GFP+ cells indicates that GFP expression from the Pim1 locus is relatively ubiquitous in the tissue. Breeding led to the establishment of a Pim1^gfp/gfp mouse line, which was confirmed by gene conversion of the I-SceI site in DR-GFP to the LweI site present at the homologous position in the downstream iGFP gene (Fig. 3B). GFP expression was visible in the ears of these mice under blue light excitation (Fig. 3C). This recombined derivative of the Pim1^Drgfp mouse allele can, therefore, be used to assess GFP expression in various tissues.

Fig. 3.

Spontaneous recombination led to the establishment of a Pim1^gfp/gfp mouse line. (A) Mammary epithelial cells derived from an animal that contained a recombined DR-GFP reporter, resulting in constitutional GFP expression. By flow cytometry, 92% of the mammary epithelium is GFP+. (B) The I-SceI recognition site is converted to an LweI site in the recombined DR-GFP reporter. A colony of homozygous Pim1^gfp/gfp mice was established by crossing Pim1^Drgfp/gfp heterozygous animals. Genomic DNA extracted from tail tips of a Pim1^gfp/gfp and a Pim1^Drgfp/Drgfp mouse was PCR-amplified using primers surrounding either the LweI site present in the recombined GFP gene (Upper) or the I-SceI site at the same position in SceGFP. PCR products were digested by I-SceI or LweI and separated on an agarose gel (Lower). (C) A Pim1^gfp/gfp mouse exhibits green fluorescence on hairless skin (ears) upon excitation with blue light.

BRCA1 Is Required for Efficient HDR in Primary Cells.

Having established significant levels of HDR in primary cells from the animal, we next sought to determine the genetic requirements using the DR-GFP mouse model. BRCA1 has been shown to play an important role in homologous recombination, which was evidenced by reduced HDR in BRCA1-deficient ES cells and transformed human cell lines (11, 23, 24). Null alleles of Brca1 lead to early embryonic death in the mouse (25), with the inability to establish cell lines. Thus, to investigate the requirement for BRCA1 in HDR in primary cells, we took advantage of a viable Brca1 mouse model (26). The Brca1^tr allele, designed to mimic a mutation found in breast cancer patients, is predicted to encode a truncated protein of 924 aa, approximately one-half the size of full-length BRCA1 and lacking the C-terminal BRCT repeats. Homozygous Brca1^tr/tr mice are grossly normal, but they are cancer-prone and exhibit other phenotypes, including male infertility.

We tested the requirement for BRCA1 in both MEFs and ear fibroblasts. Brca1^tr/tr MEFs displayed a fivefold decrease in HDR (0.21% GFP+ cells) compared with MEFs derived from wild-type (WT) siblings (1.1%) (Fig. 4A), showing that BRCA1 plays an important role in HDR in these cells. Brca1^tr/tr ear fibroblasts also exhibited decreased HDR compared with those fibroblasts derived from WT littermates (threefold; 0.31% and 1.0% GFP+, respectively) (Fig. 4B). We did not observe significantly reduced HDR in cells heterozygous for the Brca1^tr allele.

Fig. 4.

BRCA1 is required for efficient HDR in somatic cells. (A and B) HDR is significantly lower in primary MEFs (A) and ear fibroblasts (B) from Brca1^tr/tr mice compared with control mice. Significant P values relative to Brca1^+/+ mice are shown. P values here and below were determined from an unpaired t test. n = the number of transfections. For MEFs, two transfections were performed for cells derived from each embryo; for ear fibroblasts, one or two transfections were performed from a total of three mice for controls and six mice for the mutant. (C) Brca1^tr/tr mice are hypersensitive to MMC. Survival curve of control (n = 7), Rad54^−/− (n = 7), and Brca1^tr/tr (n = 8) mice after a single i.p. injection of 3.0 mg MMC per kg body weight. Control mice are heterozygous for the Brca1^tr or Rad54 allele.

Transfection efficiency of the Brca1^tr/tr cells was not lower compared with transfection efficiency of control cells and thus, could not account for the reduced HDR. Furthermore, cell cycle analysis showed a similar S-phase fraction of Brca1 mutant ear fibroblasts compared with control ear fibroblasts (Fig. S2B), suggesting that the cell cycle distribution was not affecting HDR levels.

HDR mutants are characterized by sensitivity to DNA cross-linking agents like mitomycin C (MMC) (27). If primary cells in the animal require HDR, and specifically BRCA1, for repair of cross-links, then BRCA1-deficient mice should be sensitive to these agents. To test for sensitivity, adult mice were injected with 3 mg MMC per kg body weight into the peritoneal cavity and observed for 3 wk. Whereas all of the mice WT or heterozygous for Brca1 survived this relatively low dose of MMC, more than one-half of the Brca1^tr/tr mice were dead by day 10, and all were dead by day 16 after injection (Fig. 4C). By comparison, mice deficient for the HDR protein RAD54, which have been reported to succumb to higher doses of MMC (13), all survived the 3-mg/kg treatment. This high degree of sensitivity of mice harboring a hypomorphic allele of Brca1 is consistent with the HDR defect that we observed in primary fibroblasts from adult mice, and it points to a prominent role for HDR in the adult animal. These results also show that, although BRCA1 is a breast and ovarian cancer suppressor, it plays an important role more generally in tissues from adult mice.

ATM Is Not Required for HDR in Mouse Somatic Cell Types.

The ATM kinase is a key protein in cell cycle checkpoint activation in response to DSBs (28). More recently, ATM has also been implicated in DSB repair, including by regulating DNA end resection, a key early step in HDR (29). To examine the role of ATM in HDR, the Pim^Drgfp allele was crossed into mice containing an Atm truncation mutation, Atm^tm1Awb (simplified here as Atm^w) (30). ATM-deficient MEFs and ear fibroblasts from these animals have a clear defect in DNA damage signaling; phosphorylation of the ATM targets Chk2 and p53 was abrogated in response to ionizing radiation (Fig. 5A and Fig. S3A). HDR was not significantly reduced in Atm^w/w MEFs compared with cells from WT littermates (P = 0.144) (Fig. 5A). Similarly, HDR was unaffected in Atm^w/w ear fibroblasts (P = 0.380) (Fig. 5B). These results show that ATM is not critical for HDR. Consistent with these findings, Atm^w/w mice were not sensitive to low doses of MMC, which is in contrast to Brca1^tr/tr mice (Fig. 5C).

Fig. 5.

ATM loss does not affect HDR, whereas chemical inhibition of ATM impairs HDR. (A and B) HDR levels are not significantly different in primary MEFs (A) or ear fibroblasts (B) from Atm^w/w mice compared with control mice. Chk2 phosphorylation 1 h after 6 Gy IR is substantially reduced in ATM-deficient MEFs (A). n = the number of transfections. For MEFs, one or two transfections were performed from cells derived from each embryo (number of embryos: Atm^+/+, 2; Atm^+/w, 4; Atm^w/w, 7). For ear fibroblasts, two transfections were performed for cells derived from each animal. (C) Unlike Brca1^tr/tr mice, Atm^w/w mice show no sensitivity to low-dose MMC treatment. Survival curve of control (n = 7) and Atm^w/w (n = 6) mice after a single i.p. injection of 3.0 mg MMC per kg body weight. Control mice are WT or heterozygous for the Atm^w allele. Data for Brca1^tr/tr mice are from Fig. 4C. (D) DNA damage-induced phosphorylation of Chk2 and p53 (S15) is reduced in WT ear fibroblasts treated with the ATM kinase inhibitor KU55933. Atm^b/b fibroblasts, which have no detectable ATM protein, show no evidence of Chk2 or p53 phosphorylation after IR, even in the absence of the inhibitor. Cells were exposed to 5 Gy IR and harvested for analysis 1 h later. (E and F) Chemical inhibition of ATM kinase activity results in reduced HDR in WT ear fibroblasts (E) but has little effect on HDR in Atm^b/b fibroblasts (F). Cells were treated with varying concentrations of KU55933 immediately after transient transfection of the I-SceI expression vector. The number of GFP+ cells is normalized to untreated controls for each of four independent experiments. Significant P values relative to untreated are shown. (G) DNA damage-induced Chk2 phosphorylation is reduced in WT ES cells treated with KU55933, whereas Atm^b/b ES cells show near-negligible Chk2 phosphorylation after IR, even in the absence of the inhibitor. Cells were harvested as in D. (H and I) Chemical inhibition of ATM kinase activity results in a dose-dependent reduction in HDR in WT but not Atm^b/b ES cells. ES cells were treated with varying concentrations of KU55933 immediately after transient cotransfection of the DR-GFP reporter plasmid and the I-SceI expression vector. The number of GFP+ cells is normalized to untreated controls within each of five independent experiments (untreated: WT, 8.60 ± 2.57; Atm^b/b, 8.62 ± 4.99, P = 0.995). Significant P values relative to untreated are shown.

Chemical inhibitors have been developed that specifically interfere with ATM kinase activity, impairing DNA damage signaling and rendering cells sensitive to agents that cause DSBs (31). To test if inhibition of ATM kinase activity, as opposed to loss of ATM protein, affects HDR, primary ear fibroblasts from WT Pim1^+/Drgfp mice were treated with the inhibitor KU55933 immediately after transfection with the I-SceI expression vector. The ability of the ATM inhibitor to effectively block ATM kinase activity was shown by a dose-dependent reduction in ionizing radiation-induced Chk2 phosphorylation in WT cells (Fig. 5D and Fig. S3A). Increasing concentrations of inhibitor in WT cells resulted in a significant reduction in HDR (Fig. 5E). To determine if the effect on HDR was specifically caused by ATM inhibition, experiments were also performed in Atm^w/w ear fibroblasts. These cells had no detectable ATM protein or damage-induced phosphorylation of Chk2 or p53 (Fig. S3A). As with WT cells, treatment with KU55933 led to a significant reduction in HDR (Fig. S3B), suggesting that the inhibitor is not exerting its effect solely on ATM or that the Atm^w allele in these cells does not completely disrupt protein function. Previous work has suggested that residual levels of kinase-active ATM peptide can be expressed from an alternatively spliced mRNA in Atm^w/w neural cells (32). An alternatively spliced mRNA was detected in Atm^w/w ear fibroblasts, although the reading frame of the protein was not restored (Fig. S3C), consistent with the lack of detectable ATM protein (Fig. S3A). However, we cannot rule out that a low level of splicing occurs that restores the reading frame, leading to residual peptide that is affected by the inhibitor.

A different Atm mutant allele has been described that contains a targeted disruption specifically in the kinase domain, Atm^tm1Bal (simplified here as Atm^b) (33). Atm^b/b mice have nearly identical phenotypes to Atm^w/w mice, including impaired DNA damage signaling (Fig. 5D), although some differences have been noted in neural tissues (30, 32, 34). As with cells from Atm^w/w mice (Fig. 5B), HDR was not significantly reduced in Atm^b/b ear fibroblasts compared with WT fibroblasts (P = 0.677) (Fig. S3D), consistent with the conclusion that ATM is not required for HDR. Given that the Atm^b allele is disrupted in the kinase domain, we asked whether the ATM kinase inhibitor has any effect on HDR in cells with this allele. HDR in Atm^b/b ear fibroblasts was not significantly affected by KU55933 at low concentration (P = 0.399) (Fig. 5F). At a higher concentration of inhibitor, a small but significant reduction in HDR was observed (Fig. 5F), suggesting potential off-target effects of the inhibitor, although the level of reduction in Atm^b/b ear fibroblasts was significantly less than the reduction in WT fibroblasts (P = 0.002). The different effect of the inhibitor on HDR in WT and Atm^b/b cells cannot be accounted for by alterations in cell cycle distribution (Fig. S2C). Thus, loss of ATM and chemical inhibition of ATM kinase activity have distinct effects on HDR.

To determine whether HDR is affected by ATM kinase inhibition in other cell types, WT and Atm^b/b ES cells were treated with varying concentrations of KU55933. As expected, WT cells failed to phosphorylate Chk2 in response to ionizing radiation in the presence of KU55933, whereas Atm^b/b cells failed to phosphorylate Chk2, even without the inhibitor (Fig. 5G). HDR decreased with increasing doses of inhibitor in WT cells, such that, at the highest dose, HDR was reduced 2.7-fold (P < 0.0001) (Fig. 5H). By contrast, HDR was not substantially affected in Atm^b/b cells (Fig. 5I), indicating that the inhibitor is specifically targeting ATM to cause a reduction of HDR in WT ES cells.

Discussion

Recombination reporters consisting of repeat sequences have been used to interrogate spontaneous and DNA damage-induced homologous recombination in both the soma and germ line of mice, but these events are generally rare, except in the case where the reporter is genetically unstable (35–38). Moreover, interanimal variation can be high, requiring the use of large numbers of mice (∼50) to compare mutants (39). The establishment of the DR-GFP mouse line reported here shows that homologous recombination is efficiently detected when a DNA lesion, in this case, a DSB, is introduced into the reporter, which is in keeping with HDR as a major pathway to repair DSBs in mammalian cells (9). HDR was measured at frequencies of ∼1% after I-SceI expression in primary cell cultures from five tissues representing diverse cell types in the mouse. Thus, repeat sequences are reasonably stable in the somatic genome in the absence of damage, but a DSB leads to a ≥100-fold induction of gene conversion (HDR). Other DSB repair pathways are also detected in primary cell cultures, particularly NHEJ, consistent with the presence of additional pathways to repair this genome-threatening lesion. Although the DR-GFP reporter cannot be used to measure all types of HDR, its use for assaying intrachromosomal gene conversion events and identifying the factors that promote or suppress these events is well-proven (10, 40, 41); its incorporation into a mouse model now allows for the study of HDR in primary cell types rather than transformed cell lines, which may have altered repair phenotypes (for example, because of altered expression of repair factors like RAD51) (14, 15).

In the course of breeding DR-GFP mice, we identified a spontaneous recombinant in the colony, leading to the establishment of a GFP+ mouse line. Primary cell cultures from this line show nearly ubiquitous GFP expression (Fig. 3 A, mammary epithelium ≥ 90%, and C, ears), indicating that GFP expression after I-SceI expression is likely to be an accurate indicator of HDR levels.

Using the DR-GFP mice, HDR was examined in mutant backgrounds: a Brca1 hypomorph and two Atm mutants. Although not obviously affected by loss of ATM, HDR is significantly reduced in mice expressing a truncated BRCA1 peptide. This reduction is observed in primary MEFs and ear fibroblasts, showing that the role of BRCA1 in HDR is not restricted to tissue types targeted for tumorigenesis in patients with germ-line BRCA1 mutation. These results are consistent with BRCA1 having a key role in DNA repair in multiple tissue types, and the breast and ovary are predisposed for other reasons, such as altered frequency of loss of heterozygosity of the WT BRCA1 allele or survival of BRCA1 mutant cells (42).

ATM plays a central role in the cellular response to DSBs, localizing to sites of damage and phosphorylating numerous substrates involved in DNA damage signaling and DNA repair (43). Cells deficient in ATM are highly radiosensitive and genetically unstable, and they display defective cell cycle checkpoints in response to DSBs. Mouse models recapitulate many of the phenotypes seen in patients with ATM mutations, including tumor predisposition, radiosensitivity, and gonadal dysgenesis (30, 33, 34). Although some studies using transformed cell lines have reported that ATM is not required for HDR (44), others have suggested that ATM plays a critical role (45, 46). ATM has been reported to be necessary for DSB resection to generate Replication protein A (RPA)-coated ssDNA (29), an essential early intermediate in HDR.

Our results show that ATM is not required for HDR in mouse cells: normal levels of HDR of the I-SceI–induced DSB are observed in primary cells from either Atm^w/w or Atm^b/b mice. Thus, the role of ATM in end resection is not essential, perhaps because of redundancy of factors involved in the process (47). These experiments exemplify the use of the DR-GFP mouse for determining factors involved in HDR in primary somatic cells. However, a role for ATM in processing certain types of DSBs (e.g., with complex end structures or in heterochromatin) is still plausible (48).

Gonadal dysgenesis in Atm mice was initially thought to be caused by defects in meiotic recombination (30, 34), which is induced by DSBs generated by the SPO11 protein (49). Consistent with our results in mouse somatic cells, however, recent studies have shown that ATM is not required for meiotic recombination, like when Spo11 gene dosage is reduced (50), but rather, that ATM plays an important role in regulating DSB levels during meiosis (51). In mitotically dividing cells, ATM has been reported to suppress spontaneous homologous recombination (52, 53); increased recombination in the absence of ATM may result from increased numbers of lesions from defective DNA damage signaling or repair. Our system bypasses complications of altered lesion number, because a site-specific DSB is introduced in a controlled manner.

Chemical inhibition is frequently used to disrupt ATM kinase activity in cellular studies. We observed that a chemical inhibitor of ATM activity reduces HDR in WT ear fibroblasts and ES cells but not Atm^b/b cells, which contain a disruption in the ATM kinase domain. Our findings imply that inactivation of ATM kinase activity is not equivalent to loss of ATM protein for its effect on HDR. A similar conclusion has been reached for DNA damage-induced sister chromatid exchange in human fibroblasts (54). Possibly, the inactivated ATM kinase in WT cells interferes with recruitment of HDR factors to the I-SceI–generated DSB (55). Interference of HDR would also provide an explanation for embryonic lethality of mice with a kinase-dead allele of Atm (56, 57), similar to HDR mutants like Brca1 and Brca2 (25), whereas Atm null mice survive embryonic development.

The DR-GFP mouse described here offers a powerful model for dissecting HDR in a wide variety of somatic tissue types and identifying factors that differentially regulate repair in a normal, nontransformed cellular environment. In the future, tissue expression of I-SceI can be adapted to the system for in vivo studies, complementing the ex vivo analyses delineated here.

Materials and Methods

Detailed materials and methods are in SI Materials and Methods.

Animal Use and Generation of the DR-GFP Mouse Line.

Animal experiments were performed with the approval of the Memorial Sloan-Kettering Cancer Center (MSKCC) Institutional Animal Care and Use Committee in accordance with institutional guidelines. The DR-GFP6 vector was targeted to the Pim1 locus of 129/SvJ ES cells as previously described (12). Germ-line transmission of the targeted allele was assessed by Southern blotting and PCR. Mice were maintained on a mixed genetic background. MEFs were derived from 12.5-d embryos, and ear fibroblasts were derived from 2- to 4-mo-old animals. Adult ovary and brain cultures from neonates were prepared as described (21, 22). Mammary epithelial cells were derived from 8-wk-old virgin females. For MMC experiments, 2- to 4-mo-old mice received i.p. injections of 3 mg/kg body weight MMC (Sigma) and were monitored for 21 d.

Primary Cell Transfections.

ES cells, MEFs, ear fibroblasts, and brain cultures were electroporated with 50 μg pCBASce (58) or control DNA and harvested 2–8 d posttransfection for flow cytometry. Mammary epithelial and ovarian cultures were transfected with 4 μg pCBASce or control DNA using Lipofectamine (Invitrogen) LTX and 2000, respectively, and harvested 3–14 d posttransfection for flow cytometry. The ATM inhibitor KU55933 is from EMD Millipore.