Deletion of Mouse Rad9 Causes Abnormal Cellular Responses to DNA Damage, Genomic Instability, and Embryonic Lethality

Abstract

The fission yeast Schizosaccharomyces pombe rad9 gene promotes cell survival through activation of cell cycle checkpoints induced by DNA damage. Mouse embryonic stem cells with a targeted deletion of Mrad9, the mouse ortholog of this gene, were created to evaluate its function in mammals. Mrad9^−/− cells demonstrated a marked increase in spontaneous chromosome aberrations and HPRT mutations, indicating a role in the maintenance of genomic integrity. These cells were also extremely sensitive to UV light, gamma rays, and hydroxyurea, and heterozygotes were somewhat sensitive to the last two agents relative to Mrad9^+/+ controls. Mrad9^−/− cells could initiate but not maintain gamma-ray-induced G₂ delay and retained the ability to delay DNA synthesis rapidly after UV irradiation, suggesting that checkpoint abnormalities contribute little to the radiosensitivity observed. Ectopic expression of Mrad9 or human HRAD9 complemented Mrad9^−/− cell defects, indicating that the gene has radioresponse and genomic maintenance functions that are evolutionarily conserved. Mrad9^+/− mice were generated, but heterozygous intercrosses failed to yield Mrad9^−/− pups, since embryos died at midgestation. Furthermore, Mrad9^−/− mouse embryo fibroblasts were not viable. These investigations establish Mrad9 as a key mammalian genetic element of pathways that regulate the cellular response to DNA damage, maintenance of genomic integrity, and proper embryonic development.

Exposure of cells to radiation or chemicals can damage DNA, but repair mechanisms are usually available to mend lesions. In addition, cells can delay cycling, primarily in the G₁ or G₂ phases of the cell cycle, or temporarily stop synthesizing DNA after damage is incurred. These transient events, termed cell cycle checkpoint controls (34, 57), are thought to provide extra time for repair before entry into or completion of progression through critical phases of the cell cycle. A meitoic checkpoint has also been described (58). In spite of these mechanisms, the enzymes responsible for DNA repair do not always properly fix damaged DNA, and some lesions may remain unrepaired or be misrepaired. Residual lesions can lead to apoptosis (71).

Similarities in several aspects of yeast and mammalian DNA repair pathways suggest that studies with yeast can provide numerous insights into mammalian systems, including a detailed description of the molecular and biochemical basis of the cellular processes that respond to DNA damage and promote survival (85). Schizosaccharomyces pombe cells that contain an alteration within rad1, rad3, rad9, rad17, rad24/rad25, rad26, hus1, chk1 (rad27), or chk2/cdc1 are sensitive to ionizing radiation, certain chemicals, and/or UV light. Mutations within these genes eliminate or dramatically diminish the ability of cells to delay cycling transiently in G₂ after irradiation or in S phase after hydroxyurea treatment, which is thought to be at least partly responsible for the sensitivity of these mutants to DNA damage (2, 3, 21, 22, 65, 77). As such, these genes provide a link between radiation or chemical exposure and cell cycle control.

S. pombe cells containing alterations in rad9, one of the checkpoint genes, are extremely sensitive to ionizing radiation, chemicals, and UV light, indicating that the normal gene product plays a key role in promoting cell survival after exposure to DNA-damaging agents. Furthermore, rad9 mutants are completely deficient in the radiation-induced G₂/M checkpoint and the hydroxyurea-induced DNA replication checkpoint. In addition, the rad9 gene product probably plays a more direct role in repairing damaged DNA. In support of this, extragenic suppressors of the radiation and hydroxyurea sensitivity of S. pombe rad9::ura4⁺ mutants enhance resistance without restoring checkpoint control function (46), suggesting that the protein might also participate in DNA repair. These results support multiple cell cycle-dependent and -independent functions for rad9 in modulating cell survival after DNA is damaged.

Many of the checkpoint control genes first described in S. pombe are highly conserved throughout evolution. Mammalian versions of several have been isolated, providing strong evidence that checkpoint control mechanisms are also highly conserved (24, 40, 44, 47, 53, 61, 62, 67, 73). Human and mouse cDNAs encoding proteins highly homologous to the S. pombe rad9 gene product have been cloned (30, 47). The cDNAs are capable of partially rescuing the hydroxyurea sensitivity, radiosensitivity, and associated checkpoint deficiencies of S. pombe rad9 mutant cells. Studies using small interfering RNA additionally support a role for the human protein in promoting resistance to DNA damage and in checkpoint control (36). Checkpoint genes in mammals are thought to maintain genomic stability, especially in the presence of DNA damage (6). Consistent with this, mutations in these kinds of genes are associated with high levels of spontaneous chromosome aberrations as well as with carcinogenesis (33).

HRAD9 is a nucleoprotein and is extensively phosphorylated posttranslationally, constitutively as well as in response to DNA damage (66, 69, 70, 76). ATM can phosphorylate HRAD9 on serine 272, and there is evidence that the event is important for G₁ checkpoint control (11). Furthermore, HRAD9 protein contains a BH3-like domain at its N-terminal region that can bind the antiapoptotic proteins Bcl-2 and Bcl-xL and cause apoptosis when overexpressed in human cells (43). Interestingly, the S. pombe rad9 gene product has similar properties (42) even though cognates of many of the known mammalian apoptosis-related proteins are not found in yeast (38). Phosphorylation of the tyrosine residue in the BH3 domain of HRAD9 by c-Abl is critical for the apoptotic function (83). In addition, the checkpoint control proteins HHUS1 and HRAD1 bind HRAD9 at its C-terminal region (29), and, as mentioned above, antiapoptotic proteins bind near its N terminal, suggesting that HRAD9 has at least two functional domains regulated by different phosphorylation events. It is thought that HRAD9 forms a proliferating-cell nuclear antigen-like heterotrimer with HHUS1 and HRAD1 and that this structure is loaded onto DNA via HRAD17 to serve as part of a DNA damage-sensing mechanism. Formation of the heterotrimer has been demonstrated in vitro and in vivo (7, 19, 26, 64, 86). The same is thought to be true for the S. pombe versions of these proteins (64, 75). However, it remains possible that HRAD9 functions in cells not just as part of this complex but also either alone or when associated with other proteins. Interestingly, phosphorylation of HRAD9 by protein kinase Cdelta is essential for the formation of the HRAD9-HRAD1-HHUS1 complex, as well as for the regulation by HRAD9 of the apoptotic response to DNA damage (84).

In this report, we focus on defining the function of Mrad9, the mouse version of the gene, by creating and characterizing cells and animals with a targeted gene deletion. Our studies demonstrate that Mrad9 is involved in regulating genomic stability, since homozygous mutant embryonic stem (ES) cells demonstrate a high frequency of spontaneous chromosome aberrations and hprt mutations. Furthermore, these cells are extremely sensitive to DNA damage and demonstrate unique checkpoint control defects. Also, interestingly, although Mrad9^−/− ES cells are viable, mouse embryo fibroblasts (MEFs) with the same genotype cannot be grown in culture. Mrad9^+/− mice are viable, but homozygous mutant embryos die in midgestation, revealing that the gene function is essential for embryonic development.

MATERIALS AND METHODS

Targeting and expression vector construction.

A targeting vector was made to produce a conditional deletion in Mrad9, based on the action of CRE recombinase to recognize loxP sites placed in the 5′-untranslated and second-intron regions of the gene. In addition, the NEO selection marker was flanked by FRT sites to provide the option of deleting it with FLP recombinase if desired. The construction process involved the production of five intermediate plasmids to obtain the final targeting vector. pKSloxpNT (32) was cut with HindIII and Xbal (to excise NEO), and the ends were filled in with Klenow, generating blunt ends. pFRT2 (18) was cut with AvrII, and the ends were filled in with Klenow before it was ligated to the NEO fragment from pKSloxpNT to create construct 1, a plasmid with NEO between two FRT sites (FRT-NEO-FRT). PCR with primers 5′-GACCTCAGATCTCATAATATGTCCTATCCCGT-3′ and 5′-AACCTCAGATCTCTCCCCATTCACCATTGAAT-3′ was used to amplify exon 1, intron 1, exon 2, and part of intron 2 of Mrad9 genomic DNA previously isolated from 129 SvEv mice (30). This PCR product was cut with BglII. Construct 1 was cut with BamHI, and the two fragments were ligated, resulting in construct 2, which puts Mrad9 exons 1 and 2 5′ of FRT-NEO-FRT. pKSloxpNT was cut with Xbal and PvuII to obtain the 3′ loxP site. The ends were filled in with Klenow. Construct 2 was cut with NotI and then filled in with Klenow. Ligation of these two fragments places a loxP site 3′ of Mrad9-FRT-NEO-FRT (construct 3). The latter plasmid was digested with HindIII and then filled in with Klenow. pKSloxpNT was digested with HindIII and then filled in with Klenow, and the two fragments were ligated. This places the herpes simplex virus thymidine kinase gene (tk) and a loxP site 5′ of Mrad9-FRT-NEO-FRT-loxP (construct 4). PCR in conjunction with primers 5′-CGGTCCAGATCTACTGTGTCTGGACAGTGAAA-3′ and 5′-GTCCATAGATCTAGTCCCAGCAGCTACTACGA-3′ was used to amplify a 1-kb Mrad9 genomic DNA fragment 5′ of exon 1. This fragment was cut with BglII. Construct 4 was digested with BamHI, and the two were ligated. This generates construct 5, containing the Mrad9 sequence 3′ to tk and 5′ of loxP-Mrad9-FRT-NEO-FRT-loxP. Mrad9 genomic DNA was cut with AvrII and HindIII, and the ends were filled in with Klenow. Construct 5 was cut with SalI, the ends were filled in with Klenow, and the two fragments were ligated. This creates the final targeting construct (tk-Mrad9-loxP-Mrad9-FRT-NEO-FRT-loxP-Mrad9); (see Fig. 1A), which was linearized with HindIII prior to transfection of ES cells.

FIG. 1.

Targeted deletion of Mrad9. (A) Structure of the targeting vector to make Mrad9 knockout ES cells and mice. The foundation is a genomic fragment of DNA containing Mrad9 from 129 SvEv mice. Other sequences: TK, tk; L, loxP sites, L; white boxes, Mrad9 exons 1 to 5; NEO; modified NEO; F, FRT sites. (B) Southern blot analysis of the Mrad9 gene in mouse ES cells. Alleles: +, wild type; Tar, containing targeting vector; −, deletion. (C) Western blot analysis of Mrad9 protein in mouse ES cells. (D) PCR to assess genotypes. The primer pair 5′-CCGGGTGAACCAATAAGGAA-3′ and 5′-AAGGAAGCAGGCATAGGCAG-3′ was used. Experimental details are found in the text.

Mrad9 and HRAD9 expression constructs were also made. To clone Mrad9, primers (5′-TCGGTGAAGCTTACAATGAAGTGCCTGATCAC-3′, 5′-TTTAGAGCGGCCGCCCTTCACCATCACTGTCTT-3′) were used to PCR amplify the previously cloned Mrad9 cDNA (30) and TA clone it into pGEM-T (Promega). This plasmid was then cut with HindIII and NotI for subcloning the cDNA into the corresponding sites of pZeoSV2 (Invitrogen). To clone HRAD9, pCMV2-FLAG-HRAD9 (29) was cut with HindIII and BamHI and the cDNA was subsequently ligated into the same sites in pZeoSV2.

Growth of ES cells, gene targeting, and production of Mrad9-deficient cells and mice.

Mouse ES cells were cultured by established methods (52). Antibiotics to select stable transfectants were added as needed (i.e., ganciclovir at 2 μM, puromycin at 3 μg/ml, G418 at 200 μg/ml, or zeocin at 25 μg/ml).

The targeting vector (see Fig. 1A) was linearized with HindIII and electroporated into mouse 129 SvEv ES cells (4). The population was then enriched for cells undergoing a single targeted event by a subsequent challenge with G418 to select for receipt of the plasmid and ganciclovir to counterselect the herpes simplex virus tk gene and enrich for homologous targeting events. Only 1 in 1,384 doubly resistant colonies examined contained the targeted event, as judged by Southern blotting and PCR. One explanation is that the Mrad9 region is inherently a “cold spot” for homologous recombination, as has been observed for parts of the genome of numerous organisms (23, 28, 41, 63, 74, 81).

For construction of Mrad9 heterozygous or homozygous mutant ES cells, the initial targeted ES Mrad9^Tar/+ isolate was grown in 300 μg of G418 per ml to select derivative cells that have two targeted Mrad9 events. Cells capable of growing in this higher concentration of G418 were examined by Southern blot and PCR analyses to assess whether two Mrad9 alleles had indeed been targeted. These cell populations were transiently coelectroporated with HS-CRE (a CRE expression vector [15]) and pPUR (a puromycin expression vector; Clontech) to mediate single and double deletions (17). Cells were then selected in medium containing puromycin and examined for deletion in one or two alleles of Mrad9, because of recombination between loxP sites. They were assayed by a replica-plating technique to identify the predicted G418-sensitive clones (17). For this procedure, initially about 100 clones from each transfection were picked into 96-well plates and then replica plated into two others. One set was grown in G418 to identify drug-sensitive colonies. These clones were subsequently expanded from the drug-free replica and examined by Southern analyses and related techniques to confirm the predicted deletion of one or both alleles of Mrad9. The single and double knockouts were verified using the techniques described to analyze the single-knockout cells mentioned above (52). Mrad9^−/− ES cells were transfected with pZeoSV2-Mrad9 and pZeoSV2-HRAD9 and then challenged with zeocin to generate stable mutant cells ectopically expressing Mrad9 or HRAD9, respectively.

The singly targeted ES cells were used to make heterozygous knockout mice, as follows. Targeted ES cell lines were used first to make chimeric mice by aggregation with Swiss Webster embryo cells (55) or by injection of C57BL/6 blastocysts (60). Chimeric offspring were born and mated with black Swiss Webster mice to test for successful germ line transmission of the targeted Mrad9 allele. DNA from the tails of agouti (the coat color encoded within the ES cells) progeny mice were analyzed by Southern blotting, and the targeted Mrad9 animals were identified. Such animals were mated with TK-CRE mice to mediate Mrad9 deletions in all cells and tissues (Mrad9^+/−). Subsequently, heterozygous Mrad9 knockout mice, which have been verified by Southern blot analyses and/or PCR, were bred to attempt to produce homozygotes (Mrad9^−/−).

For the isolation of MEFs, embryos (8.5 embryonic days [E8.5]) were dissected from mice pregnant as the result of timed matings between pairs of Mrad9^+/− p21^+/+ or Mrad9^+/− p21^−/− animals. The procedure to isolate fibroblasts was as published (39), with the exceptions that glass beads were not used to disrupt the embryos, trypsinization was performed for only 5 to 10 min, and the cell culture dishes were coated with gelatin. Normal-size embryos were plated into 24-well dishes, while smaller embryos were plated into 96-well dishes. Fibroblasts that proliferated were passaged up to 10-cm dishes. During passage of the cells, portions were used for determining the genotype of Mrad9 and p21. For embryos which did not proliferate, cell debris was used to determine the genotype. To isolate DNA from the embryos and cells, direct PCR (yolk sac) (Viagen Biotech, Los Angeles, Calif.) was used. PCR primers used to genotype adult mice were also employed for the same purpose with the embryos.

Southern, Western, and PCR assays to assess genotypes.

For Southern blotting, genomic DNA was isolated from ES cells by methods described previously (52). DNA was digested with AvrII, separated on a 0.7% agarose gel, transferred to a nylon membrane, and hybridized to a ³²P-labeled probe, which corresponds to sequences downstream of the targeting site. It was generated by PCR using primers 5′-TTCTGTCCTTTCCCCTTGCA-3′ and 5′-GGAGGAAAGCAACAAGTCCT-3′ in conjunction with mouse genomic DNA as the template.

For Western blots, 10⁶ pelleted ES cells were resuspended in 1 ml of sodium dodecyl sulfate sample buffer and lysed by sonication. After the lysates had been centrifuged for 5 min, 30-μl volumes were boiled for 5 min and subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis (10% polyacrylamide) (Invitrogen). Samples were then transfected to a polyvinylidene difluoride membrane by electroblotting. The blots were probed with rabbit anti-mouse Rad9 (Santa Cruz Biotech., Santa Cruz, Calif.) or rabbit anti-human actin (Santa Cruz) followed by horseradish peroxidase-conjugated goat anti-rabbit immunoglobulin G (Santa Cruz). The ECL Western blotting system (Amersham, Piscataway, N.J.) was used to visualize the results.

For PCR genotyping, genomic DNA was isolated from tail fragments, ES cells, or dissected embryos using the DNeasy tissue kit (Qiagen, Valencia, Calif.). To detect deletions of the first two exons of Mrad9, primers 5′-CCGGGTGAACCAATAAGGAA-3′ and 5′-AAGGAAGCAGGCATAGGCAG-3′ were used. To detect the presence of the targeting sequence, primers 5′-TTCGGGTGGGAGAATCAGAC-3′ and 5′-GGATCTCTCCCCATTCACCA-3′ were used. PCR conditions were 95°C for 5 min followed by 35 cycles of 95°C for 30s, 55°C for 1 min, and 72°C for 1 min, with a final extension at 72°C for 2 min.

Chromosome analysis.

Mouse ES cells were treated with colcemid (0.1 μg/ml) for 2 h, trypsinized, and centrifuged for 8 min at 120 × g. After undergoing hypotonic swelling in 0.03 M sodium citrate for 20 min at 37°C, the cells were fixed in methanol-acetic acid (3:1). After two or three additional changes in fixative, cell suspensions were dropped on wet, clean slides and dried. Metaphase chromosomes were stained with DAPI (4′,6-diamidino-2-phenylindole dihydrochloride; Sigma) and viewed with a Zeiss Axioplan 2 Imaging microscope equipped with a charge-coupled device camera. A minimum of 100 metaphase spreads per sample were acquired and analyzed using ISIS software (MetaSystems, Altussheim, Germany).

HPRT mutation assay.

ES cells were grown for 7 days in ES cells medium containing 1× HAT supplement (0.1 mM sodium hypoxanthine, 0.4 μM aminopterin, and 16 μM thymidine; Invitrogen) and then for 7 days in ES cell medium without 1× HAT before being plated at a density of 10⁵ per 10-cm dish. Then, 6-thioguanine (Sigma) was added at a concentration of 1 μg/ml. After 10 days, the cells were stained and colonies were counted.

ES cell survival assays.

Mouse ES cells were seeded at various concentrations in triplicate onto gelatinized tissue culture dishes. To test hydroxyurea sensitivity, various concentrations of the drug were added to the medium. After 24 h, the medium was removed, the cells were washed in phosphate-buffered saline (PBS), and fresh medium was added. To determine sensitivity to gamma rays, the plates were exposed to radiation from a Gammacell 40 ¹³⁷Cs irradiator (dose rate, 1 Gy/min) 4 h after seeding. To establish sensitivity to 254-nm UV light, medium was removed 24 h after seeding, the cells were washed with PBS, and the plates were exposed to various doses of 254-nm UV light emitted from a germidical bulb at a fluence of 3.5 J/m²/s. The medium was replaced, and the plates were incubated at 37°C for 7 days; colonies that formed were stained with Giemsa and counted. Percent survival was calculated as 100 × (number of colonies in treated dishes/number of colonies in mock-treated control dishes). Final numbers represent the mean (and standard deviation) of three independent trials, each with three dishes per point.

Assays for detecting radiation-induced cell cycle checkpoint control.

Flow cytometry was used to monitor cell cycle profiles of populations before or after exposure to 8 Gy of gamma rays. Cells (2 × 10⁵) were plated into each well of a six-well gelatinized dish and incubated at 37°C for 24 h. One set of cells was then irradiated. At this time, nocodazole (final concentration, 50 ng/ml) was added to one of the wells. At various times postirradiation, the cells were washed with PBS, trypsinized, pelleted by centrifugation, washed twice with PBS, and then resuspended in 0.5 ml of cold PBS. Cold ethanol (1.5 ml) was slowly dropped onto the cells while they were being gently vortexed. The cells were fixed in ethanol overnight at −20°C, centrifuged, and washed once with PBS. A 1-ml volume of propidium iodide solution (10 mM Tris, 0.7 mg of RNase per ml, 10% NP-40, 1M NaCl, 0.05 mg of propidium iodide per ml) was then added, and the cells were stained for 2 h at 4°C. Flow cytometry was performed with a FACSCalibur System (Becton Dickinson) in conjunction with accompanying CellQuest software.

To assess the S phase population for checkpoint control induced by gamma rays, cells were grown to 70% confluence, and bromodeoxyuridine (BrdU) was added to medium at a final concentration of 10 μM. Cells were incubated for 10 min at 37°C and washed twice with PBS. Prewarmed medium was added back, and cells were exposed to 8 Gy of gamma rays. After further incubation for specified times, cells were probed with fluorescein isothiocyanate-conjugated anti-BrdU antibodies and stained with PI solution. BrdU-positive cells were gated and analyzed for DNA content by flow cytometry.

The ability of UV light to induce delays in DNA synthesis was assayed as described previously (59). Cells (1 × 10⁵ to 2 × 10⁵) were plated in each well of a six-well gelatinized dish. The medium was changed the next day, and 0.02 μCi of [¹⁴C]thymidine/ml (Amersham; 50 to 62 mCi/mmol) was added. The next day, the medium was removed and cells were washed twice with PBS (Invitrogen) before being exposed to 20-J/m² UV light. Preconditioned medium was then added to the cells, and 20 μCi of [³JH]thymidine (Amersham; 25 Ci/mmol) per ml was added at 10, 40, 130, and 220 min after irradiation. At 20 min later, the medium was removed and the cells were washed twice with PBS and trypsinized for 5 min. The cells were pelleted by centrifugation for 5 min, washed with cold 1× SSC (0.15 M NaCl plus 0.015 M sodium citrate; Invitrogen), pelleted again by centrifugation, resuspended in cold 1× SSC containing 50 μg of thymidine per ml, and filtered through fiberglass filters premoistened with 4% perchloric acid. The filters were then washed with cold 4% perchloric acid, 70% ethanol, and finally 100% ethanol. They were air dried before being placed in scintillation counting fluid. Radioactivity was quantified using a scintillation counter. DNA synthesis capability was calculated as the ³H/¹⁴C ratio in irradiated versus unirradiated cells.

Detection of cell cycle checkpoint gene expression in mouse ES cells.

Northern blot analysis was used to assess gene expression. Total RNA was prepared from ES cells using TRIzol (Life Technologies, Inc.) as specified by the manufacturer. Poly(A)⁺ mRNA was then isolated with Oligotex (Qiagen). A 2-μg portion was fractionated by electrophoresis in a 1.2% (wt/vol) formaldehyde-agarose gel and transferred to a Hybond-N membrane. Hybridization was carried out in QuikHyb hybridization solution (Stratagene). To make RNA probes for Mhus1, Mrad1, Mrad17, and mouse p21, cDNAs were synthesized using the Ambion Lig'nscribe No-cloning promoter addition kit, in combination with the following primers: Mhus1, 5′-AAACTGCCCAGAACTCCAGA-3′ and 5′-AGTCTGGGATGGAGGGTTCT-3′; Mrad1, 5′-CCTCTCCTAACCCAGTACAA-3′ and 5′-TTCTTCATCAGGGCAGCAGT-3′; Mrad17, 5′-GAGCCGTGGGTGGATAAATA-3′ and 5′-CACATCCTGGAGGACCTGTT-3′; and p21, 5′-CGGTGGAACTTTGACTTCGT-3′ and 5′-CAGGGCAGAGGAAGTACTGG-3′. These cDNAs were then used to make RNA probes by being labeled with [α-³²P]dUTP (Amersham) using the Strip-EZ RNA labeling kit (Ambion). β-Actin cDNA was obtained from Clontech and used as a double-stranded DNA probe via the NeBlot kit (New England Biolabs, Beverly, Mass.) for labeling with [α-³²P]dCTP (Amersham). The filter was used to expose an X-ray film.

Morphological and histological analysis of mouse embryos.

Mouse embryos derived from Mrad9^+/− × Mrad9^+/− crosses were obtained at several stages of gestation, including E6.5, E7.5, E8.5, E9.5 and E12.5, where E0.5 was taken to be noon of the day when a vaginal plug was detected. At least 20 embryos were collected per stage. All dissections were performed in 1× PBS. Whole embryos were rinsed with 1× PBS, and photographs were taken. Samples for histological analysis were fixed for 24 h in 10% neutral buffered formalin at room temperature, processed, and embedded in paraffin using methods described by Chapman and Wolgemuth (10). Histological sections were cut at 5-μm thick and stained with hematoxylin. Pictures of whole embryos and sections were taken while viewed by a Wild Heerbrugg dissecting microscope.

Apoptosis and cellular proliferation in mouse embryos.

Embryos from timed Mrad9 heterozygote matings were isolated at E8.5 to E9.5. Following fixation for 12 h in 10% neutral buffered formalin at room temperature, they were processed and embedded in paraffin as described previously (10). Paraffin blocks were sectioned at 5 μm thick. Terminal deoxynucleotidyltransferase-mediated dUTP-biotin nick end labeling (TUNEL) staining was performed with the ApopTag peroxidase in situ apoptosis detection kit (Roche Diagnostics Corp.) and used as specified by the manufacturer. The staining was visualized using 0.2 mg of diaminobenzidine (DAB) per ml plus 0.01% hydrogen peroxide in 0.1 M Tris (pH 7.2). as the peroxidase substrate, and counterstaining was performed with hematoxylin. To analyze BrdU incorporation, BrdU was injected intraperitoneally into pregnant females at 100 μg per g of body weight (79). At 1 h after BrdU injection, embryos were isolated, processed, and sectioned as described above. BrdU staining was performed with a biotinylated mouse anti-BrdU antibody, using a BrdU staining kit (Oncogene Inc.) as specified by the manufacturer. DAB was used as the peroxidase substrate, and counterstaining was performed with hematoxylin.

RESULTS

Mutation in Mrad9 causes sensitivity of ES cells to DNA-damaging agents.

The Mrad9 gene was deleted in mouse ES cells by using the targeting strategy illustrated in Fig. 1A. The targeting construct contains two loxP sites inserted into the 5′ untranslated region and second intron of the gene, such that Cre-mediated recombinase causes a recombination event yielding a deletion of Mrad9. NEO flanked with FRT sites was inserted into the intron. This deletion is predicted to generate a null mutation, based on previous studies with human and fission yeast orthologs indicating the biological significance of this region (31, 42, 43). Singly targeted cells (see Materials and Methods) were expanded and then treated with a higher concentration of G418 to select for homozygous targeted cells. Three of four colonies surviving this challenge had both copies of Mrad9 targeted. A plasmid encoding Cre recombinase (HS-cre [15]) was then cotransfected with pPUR into the homozygous targeted ES cells, and several colonies that formed after selection in puromycin were randomly chosen for further analyses. Multiple heterozygous and homozygous Mrad9-deleted populations were identified. Figure 1B shows Southern blot analyses of one of each of these populations, illustrating the Mrad9 gene deletions. The presence of a Mrad9⁺ allele corresponds to a 4.3-kb DNA fragment. After targeting, an 8.7-kb band was produced. Treatment with Cre deleted the region between loxP sites, resulting in a 6.2-kb DNA fragment. Figure 1C depicts a Western blot demonstrating that the Mrad9^Tar/− cells (containing one Mrad9 floxed [i.e., flanked by loxP sites] allele plus NEO/FRT and the other deleted) have about half as much protein as the wild-type control and that the Mrad9^−/− cells have no detectable corresponding protein, as predicted. Homozygous targeted cells demonstrated normal levels of the protein. Thus, insertion of NEO in the intron had little effect on the level of Rad9 protein made. Figure 1D shows the PCR strategy used to genotype the ES cells (or mice) with respect to the status of Mrad9. At least two homozygous and two heterozygous independent Mrad9 deletion mutants were used for many ES cell studies cited to ensure reproducibility. None of these cells retained the plasmid HS-cre, as judged by PCR and Western blotting to detect the Cre protein.

Fission yeast S. pombe rad9::ura4⁺ cells are very sensitive to gamma rays, UV light, and the DNA synthesis inhibitor hydroxyurea. Therefore, we examined whether this function of yeast rad9, to promote resistance to DNA damage, is conserved in the mouse cognate gene, Mrad9. As indicated in Fig. 2A, mouse ES cells homozygous for a deletion of Mrad9 are highly sensitive to gamma rays relative to an isogenic Mrad9^+/+ ES cell population. Moreover, the Mrad9^−/− cells were also very sensitive to UV light (Fig. 2B) and hydroxyurea (Fig. 2C). Ectopic expression of Mrad9 (from the SV2 promoter in pZeoSV2-Mrad9) in the Mrad9^−/− ES cells at levels found in Mrad9^+/+ cells (data not shown) restored wild-type radioresistance (Fig. 2A and B) and hydroxyurea resistance (Fig. 2C) to the population, indicating that this gene does play an important role in promoting resistance to DNA damage. When the human version of the gene, HRAD9, was similarly expressed from pZeoSV2-HRAD9 in Mrad9^−/− cells at levels comparable to those in the wild type, resistance also was fully complemented (Fig. 2), thus demonstrating that this survival-promoting function is conserved between Mrad9 and HRAD9. Mrad9^Tar/− cells demonstrated levels of resistance to gamma rays and hydroxyurea between those observed for the Mrad9^+/+ and Mrad9^−/− cells, but they had wild-type levels of resistance to UV. This result indicates that the concentration requirements of Mrad9 for mediating UV resistance is different from that needed to confer resistance to the other two agents. Mrad9^Tar/Tar cells demonstrated wild-type levels of resistance to all three DNA-damaging agents (data not shown), indicating that the targeted alleles do not alter Rad9 function.

FIG. 2.

Sensitivity of mouse ES cells to DNA-damaging agents. (A) Gamma rays; (B) UV; (C) hydroxyurea. Points are the means of three trials. Error bars indicate standard deviations.

Inability of Mrad9-deficient cells to maintain ionizing radiation-induced G₂ checkpoint control.

The sensitivity of Mrad9^−/− ES cells to gamma rays, as well as the similar sensitivity and defective checkpoint response of S. pombe rad9::ura4⁺ cells to ionizing radiation, suggests that Mrad9 functions in regulating delays in progression through the G₂ phase of the cell cycle in response to gamma-ray exposure. This was tested by using flow cytometry to examine cell cycle profiles of populations representing different status of Mrad9, i.e., mock irradiated or exposed to gamma rays. Figure 3 shows the results of the study in a graphic format, and Table 1 lists the percentage of each population in each phase of the cell cycle. As shown in Fig. 3 and Table 1, an unirradiated asynchronously dividing Mrad9^+/+ cell population had 30% of its cells in G₁, 36% in S, and 33% in G₂/M. At 4 h after exposure to 8 Gy of gamma rays, this population showed an increase in the number of cells in G₂ (56%). After 8 h, more cells were found in G₂ (nearly 80%) but less were found in S and G₁. This pattern was further pronounced at 12 h. By 24 h, the cells were beginning to cycle more normally and there were increased numbers in G₁ and G₂. This is a typical response of wild-type ES cells to gamma-ray treatment, with a lack of G₁ delay but a radiation-inducible G₂ checkpoint (1, 68). Relative to the profile for Mrad9^+/+, a proliferating population of ES cells lacking Mrad9 had somewhat more cells in G₂ (40%) but slightly fewer in S (33%) and G₁ (27%). At 4 h postirradiation, the cells began to accumulate in G₂/M. By 8 h, 76% of these cells were in G₂/M, similar to, but slightly lower than, the percentage for the wild-type control. However, at 12 h, the Mrad9^−/− population only had 67% of cells in G₂/M, in contrast to the Mrad9^+/+ population, which maintained a high percentage of cells in this phase of the cell cycle. This suggests that the homozygous mutants can delay cycling in G₂ postirradiation but cannot maintain the delay as long as wild-type cells can. Mrad9^Tar/− cells, or homozygous mutants ectopically expressing Mrad9⁺, had flow cytometric profiles more similar to those of the Mrad9^+/+ population. This was also true for the Mrad9^−/− population containing HRAD9⁺. As indicated in Fig. 3 and Table 1, nocodazole, a chemical that blocks progression of the cell cycle in G₂/M, stopped the wild-type, heterozygous, and homozygous deleted Mrad9 cells as well as the latter ectopically expressing Mrad9 irradiated in G₂ (more than 95% of each population of cells are in this phase). This provides evidence that the radiation dose was not impairing the ability of the Mrad9^−/− cells to cycle during the time when cell samples were taken for analysis. Application of the drug shows that although the Mrad9^−/− cells did not maintain cycling delays in G₂ in response to irradiation alone, they were viable enough to continue cycling and be blocked dramatically in G₂ by a treatment that acts independently of checkpoint control mechanisms (i.e., nocodazole works by disrupting tubulin).

FIG. 3.

Flow cytometric analysis of mouse ES cells mock treated or after exposure to 8 Gy of gamma rays in the absence or presence of nocodazole. Regions of the profiles corresponding to G₁, S, or G₂/M are delineated above the first row of graphs and defined below the set of graphs.

TABLE 1.

Percentage of cells at different stages of the cell cycle mock treated or at different times after exposure to 8 Gy of gamma rays

Genotype	Time (h) postirradiation	% of population in cell cycle phase:
		G1	S	G2/M
Mrad9+/+	0	30.12	36.43	33.45
	4	8.06	35.93	56.01
	8	3.91	16.14	79.95
	12	6.88	9.73	83.39
	24	19.40	17.79	62.81
	12 + nocodazole	1.81	2.79	95.40
Mrad9Tar/−	0	27.08	36.45	36.47
	4	3.87	31.55	64.58
	8	2.34	9.14	88.52
	12	6.15	10.06	83.79
	24	20.84	22.32	56.84
	12 + nocodazole	1.56	2.44	96.00
Mrad9−/−	0	26.83	33.56	39.61
	4	5.58	31.12	63.30
	8	7.79	15.68	76.53
	12	12.00	20.88	67.12
	24	20.84	28.85	49.12
	12 + nocodazole	1.37	2.26	96.37
Mrad9−/− pZeoSV2-Mrad9	0	29.09	34.14	36.77
	4	3.95	29.77	66.28
	8	2.63	8.08	89.29
	12	5.72	10.96	83.32
	24	18.96	18.85	62.19
	12 + nocodazole	1.34	2.09	96.57
Mrad9−/− pZeoSV2-HRAD9	0	29.56	35.03	35.41
	4	4.13	30.49	65.38
	8	2.91	9.45	87.64
	12	6.11	10.48	83.41
	24	19.33	18.73	61.94
	12 + nocodazole	1.72	2.02	96.26

To confirm the results in Fig. 3 and Table 1, we pulse-labeled similar cell populations with BrdU and monitored the cell cycle progression of that subpopulation immediately and at the indicated times after exposure to 8 Gy of gamma rays. As shown in Fig. 4 and Table 2, Mrad9^+/+ cells accumulated in G₂/M, especially at 4 and 8 h after treatment, and those levels remained high at 12 h. Concomitantly, the percentage of the population in G₁ decreased at 4 and 8 h, but then increased at 12 and 24 h after treatment. A similar pattern was found for the Mrad9^Tar/− population. However, Mrad9^−/− cells began to accumulate in G₂/M at 4 h after treatment to the same extent as did Mrad9^+/+ or Mrad9^Tar/− cells, and there was also a reduction in the number of those cells in G₁. However, at 8 and especially at 12 h, fewer cells accumulated in G₂/M and significantly more were found in G₁ compared to the other populations. At 24 h, fewer cells were in G₂/M and more were in G₁ as well as S. Mrad9^−/− cells ectopically expressing Mrad9⁺ or HRAD9⁺ demonstrated cell cycle profiles similar to those of the wild-type control. Therefore, these experiments with BrdU-labeled cells confirmed studies with asynchronously growing unlabeled populations shown in Fig. 3 and Table 1, again indicating that Mrad9^−/− cells can initiate an ionizing radiation-induced G₂/M checkpoint but cannot maintain it.

FIG. 4.

Flow cytometric analysis of BrdU-labeled mouse ES cells mock treated or after exposure to 8 Gy of gamma rays. Regions of the profiles corresponding to G₁, S, or G₂/M are delineated above the first row of graphs and defined below the set of graphs.

TABLE 2.

Percentage of BrdU-labeled cells at different stages of the cell cycle mock treated or at different times after exposure to 8 Gy of gamma rays

Genotype	Time (h) postirradiation	% of population in cell cycle phase:
		G1	S	G2/M
Mrad9+/+	0	32.98	29.10	37.92
	4	5.12	31.48	63.40
	8	2.64	9.50	87.86
	12	11.26	3.01	85.73
	24	22.45	13.43	64.12
Mrad9Tar/−	0	28.03	28.35	43.62
	4	5.70	27.52	66.78
	8	2.04	6.80	91.16
	12	6.93	2.89	90.18
	24	21.05	12.52	66.43
Mrad9−/−	0	28.68	27.51	43.81
	4	8.01	25.75	66.24
	8	10.41	8.24	81.35
	12	21.29	9.43	69.28
	24	20.13	20.15	59.72
Mrad9−/− pZeoSV2-Mrad9	0	33.17	28.46	38.37
	4	5.06	31.55	63.39
	8	2.76	10.59	86.65
	12	5.88	4.46	89.66
	24	14.21	7.19	78.60
Mrad9−/− pZeoSV2-HRAD9	0	26.63	27.61	45.76
	4	10.53	29.70	59.77
	8	2.81	16.24	80.95
	12	6.11	5.89	88.00
	24	15.96	8.24	75.80

Mrad9^−/− cells can delay DNA synthesis after UV irradiation.

Since we demonstrated that Mrad9^−/− ES cells are sensitive to UV light and gamma rays (Fig. 2) and that gamma rays elicit an inefficient G₂ delay response in these mutants (Fig. 3; Table 1), the results suggest that UV-induced changes in DNA synthesis would also be aberrant. To test this hypothesis, we examined the ability of UV light to inhibit DNA synthesis in ES cells with different Mrad9 genotypes. As shown in Fig. 5, exposure of Mrad9^+/+ cells to UV light at 20 J/m² reduced DNA synthesis (incorporation of [³H]thymidine into DNA) at 30 min after treatment to 58% of the levels detected in unirradiated control cells. Similar experiments with a Mrad9^−/− population (population 1) demonstrated that there was a reduction in DNA synthesis only to 93% of unirradiated levels and that ectopic expression of Mrad9⁺ or HRAD9⁺ restored the ability of UV light to decrease [³H]thymidine incorporation to near wild-type levels in this mutant. However, a second Mrad9^−/− population (population 2) demonstrated a wild-type response to UV at 30 min and some delay, albeit very minor when considering statistical significance, at 1 and 2.5 h posttreatment. Therefore, Mrad9 does not regulate UV-induced alterations in DNA synthesis, at least not in a dramatic manner in these cell lines. Mrad9^Tar/− ES cells showed a level of UV-induced reduction in DNA synthesis similar to that observed for the Mrad9^+/+ population. Furthermore, UV light at 10 J/m² was also used to induce an S-phase cell cycle delay, and, again, no dramatically significant differences in delay kinetics were observed for these cells that differed in Mrad9 status (data not shown).

FIG. 5.

DNA synthesis in mouse ES cells, unirradiated or exposed to UV light at 20 J/m². Genotypes of cells are indicated. Bars represent the mean of three independent trials, indicating the ³H/¹⁴C ratio in irradiated versus unirradiated cells. Error bars indicate standard deviation.

Effect of Mrad9 status on the expression of other cell cycle checkpoint genes.

The influence of Mrad9 on the expression of cell cycle genes Hus1, Rad1, Rad17, and p21 was examined to attempt to gain mechanistic insight into the function of Mrad9. Northern blot analysis (Fig. 6) indicates that, relative to control β-actin RNA levels, heterozygous or homozygous deletion of the Mrad9 gene did not cause a dramatic shift in the level of expression of the other cell cycle-related genes, especially for a change that was reversed in the Mrad9^−/− cells after ectopic expression of Mrad9⁺ or HRAD9⁺. Interestingly, though, overexpression of Mrad9⁺ or HRAD9⁺ in Mrad9^−/− cells caused increased expression of Hus1. Nevertheless, the results indicate that the phenotypes associated with Mrad9^−/− are not caused by alterations in levels of Hus1, Rad1, Rad17, or p21 RNA.

FIG. 6.

Northern blot analyses to detect levels of mouse Hus1, Rad1, Rad17, and p21 RNA in mouse ES cells differing in the status of Mrad9. Samples in each lane are indicated. β-Actin serves as a loading control.

Mrad9 deletion leads to chromosome aberrations and increased frequency of spontaneous HPRT mutation but no significant change in growth rate.

The radiosensitivity and hydroxyurea sensitivity of Mrad9^−/− ES cells, as well as defects in radiation-induced checkpoint control, suggested that these mutant cells may not be able to properly maintain genomic integrity even in the absence of exposure to exogenous DNA-damaging agents. To test this hypothesis, we examined chromosome integrity in ES cells differing in the status of Mrad9 alleles and after a comparable number of passages. Cytogenetic analysis of DAPI-stained chromosomes revealed an elevated level of spontaneous chromosome aberrations in the Mrad9^−/− ES cells compared to the wild-type cells (Table 3). Of 106 metaphase spreads examined from Mrad9^−/− cells, 14 showed chromosome aberrations, and there was an increase in the frequency of chromosome and chromatid breaks compared to wild-type cells. Mrad9^Tar/− cells demonstrated a small increase in aberration frequency above control levels, since 3 of 110 spreads contained chromosome aberrations. Ectopic expression of Mrad9⁺ or the human cognate HRAD9⁺ in Mrad9^−/− cells reduced the aberration frequency to near wild-type levels, indicating that Mrad9 contributes to the maintenance of genomic integrity and that function is conserved between mice and humans. Interestingly, ES cells are intrinsically aneuploid (approximately 41 chromosomes), and the average number of chromosomes per metaphase was within the expected range and did not vary significantly among the different cell populations examined.

TABLE 3.

Chromosome aberrations

Genotype	No. of metaphases analyzed	No. of chromosomes per metaphase (mean ± SDsc)	No. of chromosome breaks	No. of chromatid breaks	No. of fragmentsa	No. of other aberrationsb	No. (%) of metaphases with aberrations
Mrad9+/+	106	41.40 ± 0.5	0	0	0	0	0
Mrad9Tar/−	110	41.2 ± 1.0	1 (0.01/cell)	1 (0.01/cell)	0	1 (0.01/cell)	3 (2.73)
Mrad9−/−	106	40.94 ± 1.0	6 (0.06/cell)	3 (0.03/cell)	8 (0.08/cell)	8 (0.08/cell)	14 (13.21)
Mrad9−/−, pZeoSV2-Mrad9	106	40.6 ± 0.9	0	0	1 (0.01/cell)	1 (0.01/cell)	2 (1.89)
Mrad9−/−, pZeoSV2-HRAD9	110	41.2 ± 0.7	2 (0.018/cell)	1 (0.009/cell)	1 (0.009/cell)	1 (0.009/cell)	2 (1.82)

^aCentric and acentric fragments and interstitial deletions included.

^bFusions included.

^cSD, standard deviation.

As an alternative strategy to assess genetic instability, the spontaneous HPRT mutation frequency was examined. No HPRT mutants were detected when 10⁷ Mrad9^+/+ or an equal number of Mrad9^+/− cells were assayed. However, 12 HPRT mutants were identified in a population of 10⁷ Mrad9^−/− cells. Furthermore, an equal number of Mrad9^−/− cells ectopically expressing Mrad9⁺ or HRAD9⁺ failed to yield HPRT mutants. Therefore, cells homozygous for the Mrad9 null allele demonstrate genomic instability, based on both the high spontaneous frequencies of chromosome aberrations and HPRT mutations.

Since spontaneous chromosome aberration and HPRT mutation frequencies were elevated in the Mrad9^−/− cells, the effect on the cell doubling time was examined. The average doubling times for three independent trials for log-phase-growing Mrad9^+/+, Mrad9^Tar/−, and two Mrad9^−/− populations derived independently from Mrad9^Tar/Tar cells were 12.5 ± 0.7, 13 ± 0, 13 ± 0, and 14.5 ± 0.7 h (mean ± standard deviation), respectively. Furthermore, Mrad9^−/− cells ectopically expressing Mrad9⁺ or HRAD9⁺ had doubling times of 12 ± 2.8 and 13 ± 2.0 h, respectively. Although there was some variation in doubling times for each population from experiment to experiment, changes in the Mrad9 status in particular did not significantly influence the growth rate.

Deficiency in Mrad9 causes midgestational embryonic death, accompanied by increased apoptosis and reduced cellular proliferation.

The targeted allele was transmitted into the mouse germ line by injecting mouse Mrad9^Tar/+ ES cells into blastocysts and producing chimeric mice. Subsequent matings produced heterozygous and homozygous Mrad9 targeted animals, and the former set were mated to transgenic mice expressing Cre recombinase, under the direction of the cytomegalovirus promoter, in most cells. Mice heterozygous for the deleted allele (Mrad9^+/−) were identified by Southern blot and PCR analyses (similar to that shown in Fig. 1B and D), demonstrating that Mrad9^+/− mice are viable. However, multiple Mrad9^+/− × Mrad9^+/− crosses failed to yield Mrad9^−/− mice. Of 419 pups produced from such matings, only 154 Mrad9^+/+ and 265 Mrad9^+/− pups were identified. These results show that homozygous deletion of Mrad9 causes embryonic death since mice with this genotype (Mrad9^−/−) were not obtained. The 1:1.72 ratio of wild type to heterozygous pups approximates the 1:2 ratio predicted if mice with the Mrad9^−/− genotype are inviable. There was no obvious difference in the viability of Mrad9^+/− and Mrad9^+/+ pups after birth.

To establish the time during embryonic development when Mrad9 function is critical, as well as the morphological consequences of a mutation in this gene, timed heterozygote matings were performed and embryos were retrieved at different stages of gestation. PCR was used to genotype DNA isolated from yolk sacs. At least 20 embryos were collected for each stage of gestation, including E6.5, E7.5, E8.5, E9.5, and E12.5. Table 4 shows the numbers and genotypes of the embryos obtained. Mrad9^−/− (as well as Mrad9^+/+ and Mrad9^+/−) embryos were identified at E6.5, E7.5, E8.5, and E9.5. However, homozygous mutants were not found at E12.5, although there was evidence of resorptions. These results demonstrate that Mrad9^−/− embryos die between E9.5 and E12.5.

TABLE 4.

Genotypes of embryos from timed Mrad9^+/− × Mrad9^+/− crosses

Stage	No. of embryos with indicated genotype
	Mrad9+/+	Mrad9+/−	Mrad9−/−
E6.5	9	5	6
E7.5	8	12	3
E8.5	8	16	3
E9.5	8	9	11
E12.5	8	24	7 resorptions

Gross morphology and histological analysis of embryos at the different stages of development are presented in Fig. 7. For all stages examined, the morphology of both wild-type and heterozygous mutant (not shown) Mrad9 embryos were normal and comparable. For E6.5 embryos, there were no distinguishable gross differences among the 20 embryos produced by Mrad9^+/− × Mrad9^+/− crosses (examples are shown in Fig. 7A). Wild-type and Mrad9^−/− embryos demonstrated a similar range of sizes and were essentially indistinguishable at E6.5. Although E6.5 embryos used for histological analysis could not also be genotyped because of limited material, examination of 10 such embryos from heterozygous intercrosses revealed no obvious differences (other than a range of sizes [data not shown]). Figure 7B shows one such representative histological section. Therefore, Mrad9 is not required for early embryonic development and implantation.

FIG. 7.

Gross morphology and histological sections of mouse embryos derived from Mrad9^+/− × Mrad9^+/− crosses. Photographs of intact or sectioned embryos from E6.5 (A and B), E7.5 (C to F), E8.5 (G to I), and E9.5 (J to L) are presented. Labeling for panel D is as follows: I, extraembryonic component of the chorion; II, allantois; III, neural ectoderm (neuroepithelium) in the primitive streak region. ^+/+, Mrad9^+/+; ^−/−, Mrad9^−/−.

By E7.5, there were clear morphological differences between wild-type and Mrad9^−/− embryos (Fig. 7C to F). The homozygous mutant embryos demonstrated a range of sizes but were overall consistently smaller than their wild-type littermates and were devoid of the many well-defined organ structures beginning to appear in Mrad9^+/+ (Fig. 7D as examples) and Mrad9^+/− (not shown) embryos at this time. The severity of the defects in the homozygous mutants was more extreme at later periods in development (i.e., E8.5 and E 9.5). At E8.5, most regions, such as the heart and brain, were underdeveloped and the caudal extremity of the embryo was stunted (Fig. 7G). Histological analysis (Fig. 7H and I) revealed well-defined tissues in Mrad9^+/+ embryos but unidentifiable structures in the homozygous mutants. At E9.5, size and structural differences in organs between Mrad9^+/+ and Mrad9^−/− embryos were more exaggerated (Fig. 7J to L). At E12.5, Mrad9^−/− embryos were resorbed and thus even an outline of the homozygous mutants was not detected.

Embryos were characterized with respect to apoptosis and cellular proliferation in order to understand in more detail why the Mrad9 null mutation negatively affects embryogenesis. As indicated, in Fig. 8A, B, E, and F, E8.5 and E9.5 Mrad9^+/+ embryos had no detectable cells, demonstrating an apoptotic phenotype. However, as illustrated in Fig. 8I and J, Mrad9^−/− embryos at the same times had an abundance of cells staining positive in the TUNEL assay and were therefore undergoing programmed cell death. Mrad9^+/+ and Mrad9^−/− also demonstrated differences in cellular proliferation. E8.5 or E9.5 Mrad9^+/+ embryos contained many cells that incorporated BrdU (Fig. 8C, D, G, and H), an indication of proliferative ability. In contrast, Mrad9^−/− embryos contained many fewer cells that were actively proliferating (Fig. 8K and L). Therefore, the Mrad9 null mutation negatively affects embryogenesis by increasing the frequency with which cells undergo apoptosis and by reducing proliferative activity.

FIG. 8.

TUNEL and BrdU uptake assays in mouse embryos derived from Mrad9^+/− × Mrad9^+/− crosses. Left set of panels: TUNEL assay to detect apoptotic cells in mouse embryos. (A) Mrad9^+/+, E8.5; (B) Mrad9^+/+, E9.5; (E) boxed region in panel A magnified; (F) boxed region in panel B magnified; (I) Mrad9^−/−, E8.5; (J) Mrad9^−/−, E9.5. Cells stained brown are undergoing apoptosis. Right set of panels: BrdU uptake assay to detect proliferation. (C) Mrad9^+/+, E8.5; (D) Mrad9^+/+, E9.5; (G) boxed region in panel C magnified; (H) boxed region in panel D magnified; (K) Mrad9^−/−, E8.5; (L) Mrad9^−/−, E9.5. Brown stain indicates cells that have taken up BrdU and are proliferating.

Inability to make MEFs null for Mrad9.

Attempts were made to construct Mrad9^−/− MEFs since mouse ES cells null for the gene are viable even though such a genotype leads to embryonic death. E8.5 embryos were obtained from Mrad9^+/− intercrosses, and attempts were made to establish MEFs. Such crosses produced 6 Mrad9^+/+, 10 Mrad9^+/−, and 6 Mrad9^−/− embryos. Viable MEFs were successfully established from the first two sets of progeny but not from the homozygous mutants. Since limited viability for Hus1^−/− MEFs in a p21^−/− background has been reported (79), we performed Mrad9^+/− intercrosses in a p21 null background. Again, E8.5 embryos were isolated and genotyped, and attempts were made to establish MEFs. For these experiments, 14 Mrad9^+/+, 24 Mrad9^+/−, and 15 Mrad9^−/− embryos (all p21^−/−) were identified, but viable MEFs were established successfully from only the first two sets of embryos.

As an alternative approach to establish Mrad9^−/− MEFs, we isolated E15 embryos that were Mrad9^Tar/Tar. We established MEFs and introduced a Cre expression vector to delete the Mrad9^Tar alleles in these cells. This approach also failed to yield Mrad9^−/− MEFs. Therefore, we conclude that ES cells and MEFs differ with respect to their requirement for a functional Mrad9 protein and that only MEFs devoid of Mrad9 are inviable.

DISCUSSION

S. pombe cells containing a rad9 null mutation are sensitive to ionizing radiation, UV light, and the DNA replication inhibitor hydroxyurea, as well as demonstrate defects in the cell cycle checkpoints normally induced by exposure to these agents (2, 65). In this investigation, the function of the mouse cognate gene was studied by creating and characterizing mouse ES cells and intact mice with a targeted deletion of Mrad9. Heterozygous and homozygous mouse ES cells were obtained, indicating that in mice (at least in this type of cell, but not in MEFs), as in fission yeast, the gene is not essential for cell viability. However, Mrad9^−/− cells demonstrated a high frequency of spontaneous chromosome aberrations and HPRT mutations, thus revealing a key role of the encoded protein in maintaining genomic integrity. Exactly how Mrad9 performs this function is not clear, although it might be due at least in part to stabilization of telomeres, since HRAD9 and S. pombe rad9, along with several other checkpoint control and DNA repair proteins, can bind telomere chromatin and maintain as well as protect telomeres (54, 56). Since genomic instability is one hallmark feature of cancer (51), this phenotype of Mrad9 null cells is consistent with the gene participating in carcinogenesis, perhaps as a tumor suppressor. However, no mutation in Mrad9 or the human equivalent, HRAD9, has yet been directly associated with this process. Nevertheless, the recent demonstration that HRAD9 can repress androgen receptor activation suggests a link to prostate cancer (78). Also, a human paralogue of the gene, HRAD9B, might be involved in testicular cancer (37).

As with mutations in S. pombe rad9, the Mrad9^−/− genotype causes sensitivity to gamma rays, UV light, and hydroxyurea, indicating that the mammalian gene product promotes the survival of cells exposed to these DNA-damaging agents. Homozygous Mrad9 mutant cells also demonstrate a novel checkpoint control phenotype, the inability to maintain G₂ delay induced by exposure to gamma rays. This was demonstrated by using asynchronously dividing cells, as well as by monitoring populations pulse-labeled with BrdU. Interestingly, S. pombe rad9 null cells are completely unable to delay cycling in G₂ after ionizing radiation exposure (2, 65) but two different point mutants can initiate yet fail to maintain the delays for the same period as in rad9⁺ cells (31). This raises the possibility that at least one function of Mrad9 in ionizing-radiation-induced G₂ checkpoint control in mammals is to maintain a cycling delay and that other proteins are responsible for the initiation of this checkpoint. These results suggest that defects in cell cycle control in response to gamma-ray exposure contribute to the sensitivity of the homozygous null mutant cells to this agent. However, since the defect is not dramatic, other functions of Mrad9 might also play a role in promoting ionizing-radiation resistance. Interestingly, Mrad9 appears not to play a major role in S-phase delay induced by gamma rays or UV light, and so a defect in that checkpoint is not responsible for the marked sensitivity to these types of radiation. Furthermore, the Mrad9 null mutant, relative to the Mrad9^+/+ control, did not show altered levels of Hus1, Rad1, Rad17, or p21. Thus, sensitivity of the mutant is not associated with aberrant levels of these mRNAs.

Mrad9^Tar/− mouse ES cells were sensitive to gamma rays and hydroxyurea at levels intermediate between those found for wild-type and homozygous mutant cells, but they were relatively resistant to UV light. Since the heterozygous mutant cells had approximately half as much Mrad9 protein as wild-type cells did, this level is sufficient to maintain a high degree of resistance to UV but not to the other two agents. Furthermore, no cell cycle checkpoint control defect was observed for ionizing radiation-treated Mrad9^Tar/− cells, even though they were sensitive to gamma rays. This lack of a strict correlation between checkpoint control and resistance to DNA damage indicates either that these checkpoints do not play a significant role in promoting resistance or that other compensatory mechanisms are sometimes in effect in Mrad9 heterozygous mutant cells. These results suggest multiple roles for Mrad9 in mediating resistance to DNA damage and are consistent with previous studies of S. pombe, where extragenic suppressors of rad9::ura4⁺ could uncouple resistance from checkpoints (46), again suggesting the existance of rad9-related pathways in addition to cell cycle checkpoints capable of promoting survival. It is already established that the human (and S. pombe) version of the gene has proapoptotic function in addition to a role in cell cycle checkpoint control (42, 43). Bessho and Sancar demonstrated that HRAD9 has a 3′-to-5′ exonuclease activity (8). Recently, the ability of HRAD9 to regulate the transcription of several downstream target genes was reported (82). Perhaps the function of this protein in the cellular response to DNA damage is not even limited to these pathways or activities.

Molecular modeling, coimmunoprecipitation, and yeast two-hybrid assays, as well as biochemical tests, indicate that human HRAD9 forms a PCNA-like heterotrimer with HHUS1 and HRAD1 (7, 26, 64, 86), which is loaded onto DNA by HRAD17 and serves multiple functions that include influencing the substrate specificity of ATR (19, 86). There is also strong evidence supporting a similar model for the S. pombe orthologous proteins (64, 75). This heterotrimeric structure may be important for the cell cycle checkpoint control activities of HRAD9, but it is not clear whether HRAD9 can also function independently alone or as part of a different protein complex. For example, the C-terminal region of the protein can bind the checkpoint control proteins HRAD1 and HHUS1, but the N-terminal region can bind the anti-apoptotic proteins Bcl-2 and Bcl-XL. The relationship and coordination of these complexes have yet to be determined.

Although Mrad9^Tar/− and Mrad9^−/− mouse ES cells are viable, only heterozygous and not homozygous mutant adult mice are viable. We demonstrate that Mrad9 is essential for embryonic development but is not critical until after implantation (after E6.5). Other checkpoint control genes are also essential for the proper development of mouse embryos, including Hus1, Chk1, Brca1, Brca2, and Atr, since animals with two null mutant alleles of any one of these genes are not viable (9, 12, 13, 25, 27, 48, 49, 50, 72, 79). However, since homozygous inactivating mutations in these genes cause embryonic death at different stages of development, unique to each gene, the encoded proteins cannot play the same role during embryogenesis. Furthermore, defects in checkpoint control per se might not be the cause of embryonic death since viable adult animals with homozygous null mutations in genes such as atm, chk2, p21, or p53, which neutralize multiple checkpoints, can be obtained (14, 16, 20, 35, 80). Furthermore, c-Abl can phosphorylate human HRAD9 in a tyrosine residue within its BH3 domain (83) and ATM can phosphorylate the protein at serine 272 (11). Presuming that such protein processing occurs in mice, these events involving Rad9 are also not essential for embryonic development, since elimination of these kinases does not cause embryonic death and since homozygous c-Abl or atm mutant animals are viable (5, 20, 45, 80). However, we found that a Mrad9 null mutation causes increased apoptotic activity and reduced cellular proliferation in embryos. Nevertheless, whether the critical function of Mrad9 protein (or some of the other essential checkpoint control proteins) relates to checkpoint control, apoptotic activity, maintenance of genomic stability, or an as yet undefined mode of action needs to be determined.

Interestingly, although Mrad9^−/− ES cells are viable, MEFs with the same genotype cannot be established. This cannot be attributed to overproduction of p21, since MEFs null for p21 are still inviable when also Mrad9^−/−. In contrast, the inviability of Hus1 null MEFs can be partially rescued if the cells are also p21^−/− (79). These results underscore the differences between Hus1 and Mrad9, as well as between ES cells and MEFs. The mechanisms underlying these differences need to be defined.

Mrad9 and HRAD9 encode proteins that are very similar structurally (30). Both bear a BH3-like domain and contain amino acids needed for interacting with Hus1 and Rad1. Commensurate with their structural relatedness, Mrad9^−/− ES cells can be rescued by ectopic expression of either Mrad9⁺ or HRAD9⁺ to demonstrate near wild-type levels of resistance to DNA-damaging agents and spontaneous frequencies of chromosome aberrations or HPRT mutation. Whether this relatedness extends to roles in embryogenesis has yet to be determined.

In this investigation, we have used Mrad9 null cells and animals to establish the role of the gene in maintaining genomic integrity, promoting resistance to DNA damage, and regulating a unique aspect of radiation-induced checkpoint control. Furthermore, these studies reveal that Mrad9 is essential for embryogenesis and the viability of MEFs. The function of this gene in so many processes, including its likely role as a proapoptotic protein, based on previous overexpression studies with the S. pombe and human orthologs, highlight the importance of understanding all the activities of Mrad9. In addition, since genes with characteristics similar to these are involved in carcinogenesis, deeper insight into the function of Mrad9 and its human cognate, as well as the molecular pathways it regulates, should have important implications for human disease.