Postnatal Growth Failure, Short Life Span, and Early Onset of Cellular Senescence and Subsequent Immortalization in Mice Lacking the Xeroderma Pigmentosum Group G Gene

Abstract

The xeroderma pigmentosum group G (XP-G) gene (XPG) encodes a structure-specific DNA endonuclease that functions in nucleotide excision repair (NER). XP-G patients show various symptoms, ranging from mild cutaneous abnormalities to severe dermatological impairments. In some cases, patients exhibit growth failure and life-shortening and neurological dysfunctions, which are characteristics of Cockayne syndrome (CS). The known XPG protein function as the 3′ nuclease in NER, however, cannot explain the development of CS in certain XP-G patients. To gain an insight into the functions of the XPG protein, we have generated and examined mice lacking xpg (the mouse counterpart of the human XPG gene) alleles. The xpg-deficient mice exhibited postnatal growth failure and underwent premature death. Since XPA-deficient mice, which are totally defective in NER, do not show such symptoms, our data indicate that XPG performs an additional function(s) besides its role in NER. Our in vitro studies showed that primary embryonic fibroblasts isolated from the xpg-deficient mice underwent premature senescence and exhibited the early onset of immortalization and accumulation of p53.

Xeroderma pigmentosum (XP) is a rare autosomal recessive disease clinically characterized by hypersensitivity to sunlight, abnormal pigmentation, and predisposition to skin cancers, especially on sun-exposed areas, and is caused by genetical defects in an early step(s) of the nucleotide excision repair (NER) pathway (4). Cell fusion studies have revealed the presence of seven complementation groups in XP (XP-A to XP-G) (1). So far, genes encoding the XPA, XPB, XPC, XPD, XPF, and XPG proteins that are involved in NER have been isolated (2, 23, 39, 41–43, 48, 49). Besides XP, Cockayne syndrome (CS) is also known as a repair-deficient human disease characteristic of postnatal failure of growth, a limited life span, and progressive neurological dysfunction. CS has two complementation groups (CS-A and CS-B), whose corresponding genes (CSA and CSB) have been molecularly cloned (21, 44). Cells from CS patients are moderately sensitive to UV radiation (13, 33) and are defective in one subpathway for NER involving a transcription-coupled repair process capable of removing particular lesions from transcribed strands of active genes. However, CS cells are normal in another subpathway involving the genome overall repair process (17). Rare patients in three complementation groups of XP (XP-B, XP-D, and XP-G) also show characteristic features of CS, the so-called XP/CS complex (3, 14, 16, 24, 29, 37, 46, 47).

XP patients in group G are rare, with symptoms ranging from very mild cutaneous abnormalities to severe dermatological impairments. A combination of clinical hallmarks of XP and CS has been observed in several XP-G patients (14, 16, 24, 29). The XPG gene encodes an acidic protein with a predicted molecular mass of 133 kDa that shares two regions of extensive homology with the yeast DNA repair protein RAD2 as well as a number of prokaryotic and eukaryotic endonucleases (19, 39, 41). The XPG protein is reported to have a structure-specific DNA endonuclease activity and to function as a 3′-incision nuclease in a dual-incision reaction of the NER (5, 26, 27, 35, 36). However, this function deduced from in vitro experiments does not explain the complex clinical phenotypes associated with XP-G.

In the present study, to gain an insight into the functions of XPG protein, we generated mice carrying the nonfunctional xpg (the mouse counterpart of the human XPG gene [18]) alleles by using gene-targeting and embryonic stem cell technology. Mice with the nonfunctional xpg gene showed postnatal growth failure and premature death, similar to the clinical manifestations of CS. Since the XPA-deficient mice, which are totally defective in NER, do not show such symptoms (10, 32), our data indicate that XPG performs an additional function(s) besides its role in NER. Primary embryonic fibroblasts isolated from xpg-deficient mice underwent premature senescence and showed the early onset of immortalization and accumulation of p53 protein. These results suggest that the mouse genome is genetically unstable in the absence of the xpg gene, indicating that the second xpg gene function may be involved in genome stability. This putative second function may explain the characteristic phenotypes of growth retardation and short life span observed with the xpg-deficient mice.

MATERIALS AND METHODS

Construction of the xpg targeting vector.

Mouse genomic clones containing several exons of the xpg gene were isolated from a lambda DASH II (Stratagene) phage library constructed with genomic DNA from D3 (an ES cell line derived from mouse strain 129/Sv). A 4.8-kb BamHI-BglII fragment obtained from one of the genomic clones was subcloned into pUC118 and was used to generate the targeting vector. The 4.8-kb BamHI-BglII fragment contained exon 3 (Fig. 1A). Exon 3 corresponds to nucleotide residues 264 to 380 downstream of the ATG start codon of the xpg cDNA (18, 25). An insertional mutation was generated by inserting the 1.1-kb XhoI-BamHI neo cassette from pMC1neo (Stratagene) into the XhoI site of exon 3 in the same transcriptional orientation. A 3.5-kb BamHI-EcoRI fragment of the herpes simplex virus thymidine kinase (HSV-TK) cassette was positioned at the 3′ end of the construct for negative selection. The targeting vector thus constructed was designated as pMER5/TV2.

FIG. 1.

Gene targeting at the xpg locus. (A) Schematic representation of the insertional mutation at the mouse xpg locus. Two exons of the xpg gene, exons 3 and 4, are represented as black boxes. PCR primers are shown as arrows. The 3′ external probe used for Southern blot analysis is shown as a solid bar corresponding to the S-E fragment on the wild-type map on top of panel A, and the diagnostic fragments of 22.0 and 25.5 kb are shown as solid lines on the bottom of panel A. B, BamHI; X, XhoI; Bg, BglII; S, SphI; E, EcoRI. (B) PCR and Southern blot analyses of the targeted clone GG5. The predicted PCR products were PCR1 (with the neoS1 and TV2R4 primers) and PCR2 (with the TV2FD and TV2R4 primers) (shown in panel A). Southern blot analysis using the 3′ external probe also detected the predicted restriction fragments shown in panel A. M, size markers; ES, ES cells as a control. (C) PCR and Southern blot analyses of offspring from intercrosses between the chimeric males and C57BL/6J females. +/+, wild type; +/−, heterozygote; −/−, homozygous mutants. (D) Northern blot analysis of total RNA from newborn mice derived from a heterozygous intercross, using xpg cDNA as a probe. A β-actin cDNA probe was used as an internal control for estimation of total mRNA in each sample.

PCR and Southern and Northern blot analyses.

Homologous recombination between the targeting vector and mouse chromosome 1 was examined by PCR using the neoS1 and TV2R4 primers (see PCR1 in Fig. 1A). Genotypes of the xpg mutant alleles were determined by PCR with the TV2FD and TV2R4 primers (see PCR2 in Fig. 1A). The sequences of the primers were as follows: ATCGCCTTCTATCGCCTTCTT for neoS1, TGGTGACAGGGAAACGAACC for TV2FD, and AGAGCCAAGTACACTGAGAAG for TV2R4. PCRs were performed with an ExTaq PCR kit (Takara, Tokyo, Japan) according to the manufacturer’s recommendations. Following an initial denaturation step (94°C for 1 min), 30 cycles of PCR were performed (98°C for 20 s, 64°C for 30 s, and 75°C for 5 min) with a 480 Perkin-Elmer thermal cycler. Homologous recombination was also examined by hybridization with a probe containing a 0.5-kb SphI-EcoRI fragment (Fig. 1A). Genomic DNA isolated from ES cells or mouse tails was digested with BamHI, separated through 0.6% SeaKem GTG agarose gels (FMC BioProducts) with Tris-acetate-EDTA buffer at pH 7.5, and transferred onto Hybond-N+ membranes (Amersham). The absence of additional random integration of the targeting construct was examined with a neo probe. For Northern blotting, total RNA (20 μg) from fibroblasts of newborn mice obtained from intercrosses of the heterozygotes was separated on a 1% SeaKem GTG agarose gel (FMC BioProducts) containing 0.66 M formaldehyde. After electrophoresis, the RNA was blotted onto a Hybond-N+ membrane (Amersham) and hybridized with the ³²P-labeled mouse xpg cDNA. Embryos and newborn mice were sexed by analysis of the Sry locus as reported previously (15).

Gene targeting in ES cells and generation of mutant mice.

The 129/Sv-derived ES cell line D3 was maintained in Dulbecco’s modified Eagle medium (Nissui Seiyaku Co., Ltd., Tokyo, Japan) containing heat-inactivated 20% fetal bovine serum (GIBCO), 0.1 mM 2-mercaptoethanol, 2 mM glutamine, and 1,000 U of leukemia-inhibiting factor per ml (GIBCO). The cells were trypsinized, resuspended at a concentration of 10⁷ cells/ml in phosphate-buffered saline, and electroporated at room temperature with 50 μg of SalI-linearized vector DNA at 2,000 V/ml for 1 s with an electroporator (somatic hybridizer SSH-10; Shimadzu Co., Ltd., Kyoto, Japan). After electroporation, cells were kept at 37°C for 30 min and transferred onto 10-cm-diameter culture dishes coated with 0.1% gelatin in the medium described above. Twenty-four hours later, cells were cultured with a selection medium containing 200 μg of G418 per ml (GIBCO) and 2 μM ganciclovir (Syntex Research). The targeting event occurred at a frequency of 15.4%. Chimeras were constructed by injection of targeted ES cells into C57BL/6 blastocysts collected at day 3.5 postcoitum. Approximately 10 to 15 ES cells from each cell clone with a normal karyotype and carrying the homologous recombination were microinjected into the recipient blastocysts, and five embryos were transferred into each uterine horn of ICR pseudopregnant foster mothers. Five chimeras derived from two independent ES clones were crossed with C57BL/6 females to transmit the mutant allele through the germ line. All five chimeric males generated offspring with pigmentation, which is the hallmark of derivation from D3-derived germ cells.

Primary cell culture.

Heterozygous females and males were mated, and from the resulting embryos, fibroblasts were prepared at day 14.5 after mating. Cells from individual embryos were grown in Dulbecco’s modified Eagle medium supplemented with 10% heat-inactivated fetal bovine serum for 7 days and frozen at this point. The frozen stocks were later used for survival assays and growth experiments.

Cell survival assays.

Embryonic fibroblasts were plated in wells of 96-well plates and cultured overnight. For the UV exposure assay, cells were washed with phosphate-buffered saline, irradiated with UV light at the doses indicated, and cultured for another 5 days. For the H₂O₂ treatment assay, the growth medium was replaced with a medium containing various concentrations of H₂O₂, and the cells were further cultured for 5 days. Cell survival was measured by the MTT (3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyltetrazolium bromide) method according to the protocol of the manufacturer (Boehringer cell proliferation kit). For X-ray survival experiments, cells were irradiated with X rays at the indicated doses and further cultured for 5 days. Cell survival was measured by the bromodeoxyuridine-incorporation method as described by the manufacturer (Boehringer 5-bromo-2′-deoxyuridine labelling and detection kit III).

Other methods.

Direct binding of monoclonal antibodies to thymine dimers (TDM-2) or 6-4 photoproducts (64M-2) was measured by enzyme-linked immunosorbent assay (ELISA) as described before (28). Cellular p53 levels were measured by using a detection kit (Boehringer p53 pan ELISA kit) according to the protocol of the manufacturer. Tissue specimens from xpg-knockout mice as well as from the wild-type or heterozygous littermates were fixed with 10% buffered formalin and were embedded in paraffin. Sections (3 or 4 μm thick) were stained with hematoxylin and eosin.

RESULTS

Generation of xpg-deficient mice.

The pMER5/TV2 targeting vector was designed to generate an insertional mutation in the 3rd exon of the mouse xpg gene (Fig. 1A). Targeted clones with the predicted insertion were identified by PCR and Southern blot analyses (Fig. 1B). The targeted ES cells were injected into C57BL/6 blastocysts to generate chimeric mice capable of transmitting the mutant allele to F₁ offspring, which were later identified by both PCR and Southern blot analyses (Fig. 1C). To examine the effects of the insertional mutation on xpg gene expression, total RNA from newborn mice was analyzed by Northern blotting. No stable xpg transcript, either intact or truncated, was detected in the −/− homozygous mice. In the heterozygous (+/−) mice, the xpg mRNA content was approximately half of that in the wild-type (+/+) mice (Fig. 1D). These results indicated that the xpg gene was disrupted successfully.

Severe growth failure and short life span of the mutant homozygous mice.

The heterozygous mice were interbred to obtain homozygote mutants. Out of 163 pups born, 35 (21.5%) mice exhibited growth failure. PCR analysis revealed that all the 35 mice with growth failure were homozygous for the disrupted xpg gene. These mice died by 23 days postpartum (Fig. 2A). The 128 survivors were either wild type (39 mice [23.9%]) or heterozygotes (89 mice [54.6%]). Analysis of embryos at 16 and 21 days postcoitum revealed the three genotypes at the expected Mendelian ratio (9 wild type, 19 heterozygotes, and 10 mutant homozygotes) with no sexual bias, indicating that disruption of the xpg gene by our method did not affect embryonic viability. Although the sizes of the mutant homozygotes at birth were indistinguishable from those of the wild-type or heterozygote pups, growth of the xpg mutant homozygotes was severely retarded thereafter (Fig. 2B and C). In contrast, the heterozygous mice showed neither obvious physical abnormalities nor pathological alterations when compared to the wild type (Fig. 2C).

FIG. 2.

Growth characteristics and life span of the xpg mutant mice. (A) Survival curve of mutant mice postpartum. (B) Average body weights of the xpg mutant mice (solid circles). The body weights of males and females from the normal group (i.e., wild-type and heterozygous mice) were combined (open circles). (C) Gross phenotypic appearance of an xpg mutant (−/−) (right), a heterozygous (+/−) littermate (middle), and a wild-type (+/+) littermate (left) at 16 days postpartum.

Although we observed the −/− pups very closely, we were unable to detect any obvious physical changes or abnormal behaviors involving suckling. The mutant homozygotes seemed to cling normally to the teats of their dams. However, the normal littermates might have interfered with the mutant homozygotes’ clinging to the mother. Thus, to eliminate this possibility, we removed all of the normal-size pups from the nest and kept only the small pups (presumably the mutant homozygotes) at 10 days postpartum together with their mother. Nonetheless, all of the small pups (i.e., the mutant homozygotes) died off before weaning (data not shown).

UV sensitivities of xpg-deficient cells and kinetic analysis of UV damage removal.

To verify inactivation of NER in cells from the xpg^−/− mice, sensitivity to UV and removal kinetics of UV-induced DNA lesions (thymine dimers and 6-4 photoproducts) were examined. As shown in Fig. 3A, the primary embryonic fibroblasts from the −/− mice were hypersensitive to UV (254 nm) irradiation like cells obtained from a severe XP-G patient and rodent ERCC5 mutant cells (41). In contrast, cells from the heterozygotes were as resistant to UV irradiation as the wild type. On the other hand, the fibroblasts from the −/− mice were not hypersensitive to X rays or to H₂O₂, compared with the cells from +/− or +/+ mice (Fig. 3B and C). We compared the kinetics of removal of major UV-induced DNA lesions in these fibroblasts by ELISA using monoclonal antibodies against cyclobutane-type thymine dimers and 6-4 photoproducts. Neither thymine dimers nor 6-4 photoproducts were removed in the −/− fibroblasts, but they were removed in the +/− and +/+ fibroblasts (Fig. 3D and E). These results indicate that the NER activity is defective in the xpg^−/− mice.

FIG. 3.

Survival curves, removal kinetics of UV-induced DNA damage, and genetic instability tests for embryonic fibroblasts derived from xpg-deficient mice. UV survival curves (A), X-ray survival curves (B), and H₂O₂ survival curves (C) for cells derived from xpg-deficient mice are shown. Each point is an average of triplicate wells. (D and E) Removal kinetics of 6-4 photoproducts (D) and cyclobutane pyrimidine dimers (E). Ab, antibody. Each point represents an average of triplicate wells. (F) Growth properties of embryonic fibroblasts. Five independent experiments were carried out, and one of experimental data is shown in this figure. The timings of crisis and immortalization were somewhat different among experiments, but their tendencies were highly reproducible. (G) Accumulation of p53 in embryonic cells. All of the experiments were carried out with the embryonic fibroblasts from the wild type (open circles) and heterozygous (open squares) mice and from two homozygous xpg mutant mice (solid symbols) that originated from one litter at a low passage number (2 to 3).

Growth properties of cells from xpg-deficient mice.

To characterize the growth failure further, we examined growth properties of embryonic primary fibroblasts derived from xpg-deficient mice. With continuous culture, cells from −/− embryos ceased growing by 4 to 5 weeks after the start of in vitro culture, while cells from the +/+ or +/− embryos kept their growth capability 3 to 4 weeks longer (Fig. 3F), indicating that −/− cells underwent premature replication senescence. After 3 to 4 weeks of nongrowing periods, −/− cells started to grow again and appeared to gain an immortal phenotype. Normal (+/+ or +/−) cells regained their growth capability after somewhat longer periods of latency (5 weeks) (Fig. 3F). Furthermore, after 15 weeks of culture, xpg-deficient cells started to accumulate p53 (Fig. 3G), which is strongly associated with transformed phenotypes, such as fast growth, loss of contact inhibition, and changes in cellular morphology (12, 20, 22). These phenotypes were observed in −/− cells that accumulated p53 (data not shown). These results indicate that cells from xpg-deficient mice have a short replication life span and readily gain immortality and malignant phenotypes, suggesting that the −/− cells are genetically unstable, since some genetic changes are needed for acquisition of these phenotypes (12, 20, 22).

Histological and anatomical analyses.

We examined several organs (liver, stomach, intestines, spleen, kidney, and brain) of xpg mutant mice at 0, 5, 16, and 21 days postpartum. At day 0, the small intestines of the mutant homozygotes were apparently smaller in diameter than those of the wild-type mice and heterozygotes (Fig. 4A and D). At 5 days, very immature small intestines were observed in the mutant homozygotes (Fig. 4B and E). The number and size of villi were reduced compared with those in the wild-type mice and heterozygotes. However, in other organs, obvious defects were not observed at 0 and 5 days postpartum (data not shown).

FIG. 4.

Histological and anatomical analyses of the xpg-deficient mice. Cross-sections from the small intestines of wild-type mice and mutant homozygotes were stained with hematoxylin and eosin at 0, 5, and 16 days postpartum. (A) Wild type (+/+) at day 0. (B) Wild type (+/+) at 5 days. (C) Wild type (+/+) at 16 days. (D) Mutant homozygote (−/−) at day 0. (E) Mutant homozygote (−/−) at 5 days. (F) Mutant homozygote (−/−) at 16 days. (A to F) Magnification, ×25. (G) Appearance of the stomach and intestines in a heterozygote (+/−) and a homozygote (−/−) at 21 days postpartum. Yellow arrows point to the stomachs. (H) Appearance of spleens from a heterozygote (+/−) and a homozygote (−/−) at 21 days postpartum. The spleens were very small in the −/− mice. (I) Sections from the livers of a heterozygote (+/−) and a homozygote (−/−) at 16 days postpartum stained with hematoxylin and eosin (magnification, ×150).

At 16 and 21 days postpartum, abnormalities in the mutant homozygotes were observed, not only in the small intestines (Fig. 4C and F), but also in other organs (except the brain), which was consistent with the miniature size observed for the whole body. As shown in Fig. 4G, the stomach and the intestines in the mutant homozygous mice were relatively small, and many gas bubbles were present inside the intestines, suggesting dysfunction of the intestines. The spleens were also very small in all of the mutant homozygous mice (Fig. 4H), and the average spleen weight per body weight at 16 days was about 28% of that in the normal littermates (+/− or +/+). The average weights of the other organs were about half of those of the normal littermates. Liver cells in mutant homozygous mice were remarkably smaller than those in the heterozygous mice (Fig. 4I). These results suggest that abnormality in the small intestines caused insufficient digestion and ingestion of milk, resulting in severe starvation atrophy.

DISCUSSION

Patients suffering from XP-G show complex clinical phenotypes. Some patients exhibit the signs and symptoms of both XP and CS (XP/CS complex) (14, 16, 24, 29). The reason for this combined phenotype is not known at present. Mutations in five genes, CSA, CSB, XPB, XPD, and XPG, can cause the CS phenotype. Of the proteins coded for by these genes, CSA and CSB function exclusively in transcription and are required for transcription elongation and transcription-coupled repair (17). These are not essential genes for cell survival, and thus humans or mice with defects in these genes can grow to an average age of 12 years or to adulthood, respectively (33, 45). The XPB and XPD genes encode the subunits of the general transcription-repair factor TFIIH (11, 40), and hence only missense mutations in these genes are compatible with life (7, 8). Apparently, some of the mutations in these genes impair transcription to a significant level to cause the XP/CS complex in a subset of XP-B and XP-D patients (38).

In contrast to the other four genes which have been implicated in CS, at present, there is no direct evidence that the XPG gene plays a role in transcription. The XPG protein has been found to bind to TFIIH with moderate affinity (30); however, TFIIH preparations free of XPG are active both as a general transcription factor and a general repair factor (31). While these results do not exclude the possibility that XPG may play a role in transcription, they strongly suggest that postnatal growth failure and premature death of xpg mice are not caused by defective transcription, but may be related to an XPG function independent of its role in NER and a potential role in transcription.

A clue to a potentially vital role of XPG in survival was provided by the recent findings that XPG is required for transcription-coupled repair of oxidative DNA lesion thymine glycol by base excision repair and stimulates the general genome repair of this lesion (6). Furthermore, it was found that missense mutations that inactivate the NER nuclease function of XPG did not affect thymine glycol repair or cause CS. Only mutations which gave rise to severely truncated XPG reduced the rate of thymine glycol repair and caused CS, which is associated with growth retardation and short life span (34). In light of these findings, then, a likely cause of early senescence and death of xpg-deficient embryonic mouse cells and of xpg mice is the accumulation of oxidative damage, including thymine glycol, in the genome of the xpg mutant cells and mice. These lesions may cause the observed phenotypes by blocking replication and transcription or by causing mutations in important regulatory genes. The fact that we did not find increased sensitivity of xpg null cells to ionizing radiation and H₂O₂ is not necessarily in disagreement with this reasoning. A 10 to 20% reduction in the repair of thymine glycol (and other oxidative lesions, such as 8-oxoG [9]) may confer increased sensitivity that is difficult to detect in acute treatments. However, even marginally perceptible decrease in repair of oxidative damage could lead to lethal phenotype over the long haul. In this regard, it is of relevance to note that the XPG/CS cell lines with reduced thymine glycol repair capacity were not reported to have increased sensitivity to ionizing radiation or oxidative stress. Thus, a careful consideration of existing data on XPG mutants both in humans and in mice led us to speculate that the premature senescence and death of xpg mice are caused by genomic instability induced by oxidative lesions, which are repaired at a considerably slower rate in these mutants. It must be noted, however, that the pathologies of xpg mutants in mice and humans show significant differences. For example, the abnormally small intestines and the accompanying intestinal dysfunction, which may contribute to lethality in xpg mice, have not been reported in XPG/CS complex patients. Perhaps the backup repair systems in humans play a more significant role in some organs, such as intestines, and thus mitigate some of the clinical symptoms in human patients. We believe that the xpg-null mice we have generated will be useful in answering this and related questions regarding the role of repair in senescence and death.