Mice lacking the Gadd45a gene are susceptible to ionizing radiation-induced tumors. Increased levels of Gadd45a transcript and protein are seen after treatment of cells with ionizing radiation as well as many other agents and treatments that damage DNA. Because cells deficient in Gadd45a were shown to have a partial defect in the global genomic repair component of the nucleotide excision repair pathway of UV-induced photoproducts, dimethylbenzanthracene (DMBA) carcinogenesis was investigated because this agent produces bulky adducts in DNA that are also repaired by nucleotide excision repair. Wild-type mice and mice deficient for Gadd45a were injected with a single i.p. dose of DMBA at 10–14 days of age. The latency for spontaneous deaths was slightly decreased for Gadd45a-null mice compared with wild-type mice. At 17 months, all surviving animals were killed, and similar percentages of each genotype were found to have tumors. However, nearly twice as many Gadd45a-null than wild-type mice had multiple tumors, and three times as many had multiple malignant tumors. The predominant tumor types in wild-type mice were lymphoma and tumors of the intestines and liver. In Gadd45a-null mice, there was a dramatic increase in female ovarian tumors, male hepatocellular tumors, and in vascular tumors in both sexes. In wild-type mice, this dose of DMBA induced a >5-fold increase in Gadd45a transcript in the spleen and ovary, whereas the increase in liver was >20-fold. Nucleotide excision repair, which repairs both UV- and DMBA-induced DNA lesions, was substantially reduced in Gadd45a-null lymphoblasts. Mutation frequency after DMBA treatment was threefold higher in Gadd45a-null liver compared with wild-type liver. Therefore, lack of basal and DMBA-induced Gadd45a may result in enhanced tumorigenesis because of decreased DNA repair and increased mutation frequency. Genomic instability, decreased cell cycle checkpoints, and partial loss of normal growth control in cells from Gadd45a-null mice may also contribute to this process.

Gadd45a is a Mr 21,000 protein that has been linked to many important cellular processes such as DNA repair, chromatin accessibility, cell cycle checkpoints, and genome stability (1, 2, 3, 4, 5). GGR2, a subtype of NER was decreased in mouse embryo fibroblast cells lacking Gadd45a, whereas TCR was unchanged (2). In vitro assays have shown that addition of Gadd45a can prevent assembly of histones onto DNA and can specifically bind to UV-induced lesions in nucleosome complexes (3). It was, therefore, proposed that Gadd45a may enhance DNA repair by allowing accessibility of repair complexes to damaged DNA.

In the human genetic diseases XP and CS, one of many genes involved in NER, which confers sensitivity of these individuals to UV radiation-induced carcinogenesis and/or sunburn, is mutated or missing (reviewed in Ref. 6). Mouse models of XP have been generated that show this same predisposition to UV-induced carcinogenesis as well as carcinogenesis by other agents that produce damage repaired by NER (7, 8). The NER defects in XP often involve the GGR subpathway and are associated with increased tumors. In contrast, a defect in TCR, such as in CS, is associated with sun sensitivity, yet no increased carcinogenesis has been observed (9). GGR is defective in Gadd45a-null cells, and therefore, increased carcinogenesis by agents producing damage repaired by NER was anticipated in mice lacking this gene.

Gadd45a-null mice were shown to have a decreased latency for IR-induced tumors (5). However, IR produces predominantly DNA strand breaks and base damage that are not repaired by NER. Like XPA and XPC-null mice, most Gadd45a-null mice appear normal and do not show a significant increase in spontaneous tumors. However, laboratory mice live in very controlled environments in the absence of UV radiation or carcinogens. It is likely that these genes are not required to prevent spontaneous tumors when there is no exogenous damage to the DNA. In the presence of damage, lack of any one of many DNA repair genes could lead to increased mutagenesis and consequently carcinogenesis. To determine the effect of Gadd45a deletion on tumorigenesis by an agent whose damage is repaired by NER, young mice were injected i.p. with DMBA and monitored for tumors. There was a small increase in tumor-induced mortality and prevalence of tumors in Gadd45a-null mice compared with wt mice. However, there were far more Gadd45a-null mice with multiple tumors and a dramatic increase in vascular, ovarian, and hepatocellular tumors.

Animals.

Mice were housed in Plexiglas cages and given autoclaved NIH 31 diet and water ad libitum. NIH is an Association for Assessment and Accreditation of Laboratory Animal Care-accredited animal facility, and all experiments were done under an approved NCI animal study protocol.

Mice were treated at 10–14 days of age with a single i.p. injection of DMBA in corn oil. The study consisted of 11 female and 15 male Gadd45a-null mice and 11 female and 13 male wt mice. Animals exhibiting obvious tumors or who were moribund, cachectic, or nonresponsive were killed for necropsy, and at 17 months after DMBA injection, surviving animals were killed for necropsy. Tumors and abnormal tissue were taken for histopathological analysis. Tumor and tissue sections were collected, stained with H&E, and evaluated by a board-certified veterinary pathologist.

DNA Repair and Mutation Assays.

Splenic lymphocytes were isolated from spleens of 4- to 6-week-old mice by disruption between two sterile glass slides and grown as previously described (5). The relative percentage of 6-4 photoproducts in total genomic DNA was determined using an ELISA as previously described (10, 11). Briefly, splenic cells in culture were UV (254 nm) irradiated (10 J/m2), and DNA was prepared 3 to 24 h later. Genomic DNA was extracted using a Blood Kit (Qiagen) according to the manufacturer‘s recommendations. Polyvinyl chloride flat-bottomed microtiter plates precoated with 1% protamine sulfate (Sigma) were incubated with 300 ng of DNA in PBS at 37°C for 20 h. After drying, the plates were washed five times with PBS containing 0.05% Tween (PBST). The plates were blocked with 2% fetal bovine serum in PBS for 30 min at 37°C. After five washings with PBST, the plates were incubated with 64M-2 (11) anti-(6-4) photoproduct antibodies (in quadruplicate) diluted 1:1000 in PBS. Another five PBST washings were followed by two consecutive incubations (30 min, 37°C) with goat antimouse IgG F(ab′)2 fragment conjugated with biotin (1:2000 dilution in PBS; Zymed) and then with streptavidin-peroxidase conjugate (1:10000 dilution in PBS; Zymed). Finally, after five PBST washings and one citrate-phosphate buffer (pH 5.0) washing, 100 μl of substrate solution (0.04% o-phenylene diamine and 0.0075% H2O2 in citrate-phosphate buffer) were added to each well and incubated for 30 min at 37°C. Reactions were stopped with 50 μl 2 m H2SO4. The absorbance at 490 nm was measured using an E-max microplate reader (Molecular Devices).

Gadd45a-null mice were crossed into the BigBlue strain of mice, which harbor a λ shuttle vector that can be used for mutation detection. Six-week-old mice that carried the λ integration and were either wt or null for Gadd45a were injected i.p. with 20 nmol/g body weight DMBA, and tissues were harvested for DNA isolation 7 days later. Genomic DNA was prepared using the RecoverEase DNA isolation kit, and bacteriophage were packaged using Transpack packaging extract (Stratagene). At least 2 × 104 plaques were screened for λcII mutations, according to the manufacturer‘s instructions in the λcII Mutation Assay Detection Kit (Stratagene).

RNA Analysis.

For RNA isolation, 6-week-old mice were injected i.p. with 20 nmol of DMBA/g body weight, and tissues were harvested 4 h later. Tissues were immediately homogenized in RNAzol (Life Technologies), and RNA was prepared as directed by the manufacturer. RNA was dot-blotted onto nylon membranes (Nytran; Schleicher & Schull). cDNA probes were labeled with [32P]dCTP using Prime-It random primer labeling (Stratagene). Gadd45a RNA was normalized to poly(A) content, which was estimated by hybridization to a labeled polyuridylic acid probe (12, 13). Radioactivity for RNA dot blots was counted on a PhosphorImager (Molecular Dynamics).

Because Gadd45a is involved in NER, it was expected that there would be decreased DNA repair leading to increased DMBA-induced mutations in Gadd45a-null animals, resulting in more tumors. Because of the involvement of Gadd45a in cell cycle checkpoints, growth control, and genome stability, there was also the possibility that these features of the Gadd45a-null phenotype could contribute to decreased latency and increased incidence. The dose of DMBA used was expected to produce tumors in wt animals so that tumor types, incidence rates, and latency could be compared between Gadd45a-null and wt mice. At 16 months posttreatment, there was a slight increase in the incidence of deaths in the Gadd45a-null mice (Fig. 1), primarily attributable to killing of moribund Gadd45a-null males. For males, 6 (40%) of 15 Gadd45a-null died early compared with 2 (15%) of 13 wt, whereas in the females, 3 (27%) of 11 Gadd45a-null died early compared with 4 (36%) of 11 wt. Only one early killing each in the wt males and Gadd45a-null males and two early sacrifices in the Gadd45a-null females were not attributable to neoplasia. All surviving mice were necropsied at 17 months.

Total tumor burden was greater in the Gadd45a-null mice of both sexes than in wt mice (Table 1). This increase was attributable to a greater number of male Gadd45a-null mice with neoplasm as well as to increased multiplicity of tumors per animal, both benign and malignant, in both the male and female Gadd45a-null when compared with the wt animals. Almost twice as many Gadd45a-null had more than one tumor, and more than three times as many Gadd45a-null mice had two or more malignant tumors (malignancy being based on presence of metastasis, invasion, or atypia), when compared with the wt mice (Fig. 2). The average number of all tumors per affected mouse was 1.53 and 2.37 for wt and Gadd45a-null mice, respectively [significantly different (P < 0.03) by t test and one-way ANOVA].

There was a notable increase in incidence of hepatocellular tumors in the male Gadd45a-null, ovarian tumors in the female Gadd45a-null, and hemangiomas/hemangiosarcomas in both male and female Gadd45a-null mice when compared with the wt (Table 1). Hemangiomas/hemangiosarcomas were most prevalent in the intestinal tract of both Gadd45a-null and wt mice; however, in the Gadd45a-null mice these vascular tumors tended to occur in multiple sites and in additional tissues including the spleen, pancreas, adipose tissue, and skeletal muscle.

Gadd45a RNA is increased in cell cultures and in vivo after treatment with a variety of agents that damage DNA. This increase might provide a protective function because Gadd45a is involved, either directly or indirectly, in NER, the DNA repair pathway that repairs bulky lesions such as those produced by DMBA. Wt mice were injected with the same dose of DMBA that was used for carcinogenesis, and Gadd45a RNA levels were measured after 4 h, the time when Gadd45a RNA levels are often maximal after treatment with DNA-damaging agents. From 5- to >20-fold induction was seen in various tissues (Fig. 3). The highest induction was seen in the liver, which might be expected because this is probably the major organ that metabolizes DMBA. Induction was seen both in organs in which tumors were more prevalent in Gadd45a-null mice as well as those in which there was no increase in tumors in Gadd45a-null mice.

Decreased repair has been shown in Gadd45a-null mouse embryo fibroblasts (2). To determine whether repair is attenuated in tissues from Gadd45a-null mice, splenic lymphocytes were treated with UV radiation and allowed to repair the DNA damage for up to 24 h. Although UV and DMBA produce different types of DNA lesions, they are repaired by the same mechanism, NER. UV radiation was used here instead of DMBA because antibodies to 6-4 UV photoproducts are available to measure DNA lesions. As predicted, repair was diminished in Gadd45a-null lymphocytes compared with wt lymphocytes (Fig. 4).

Decreased repair of DMBA lesions would be expected to result in increased mutations, which might lead to tumor formation. Gadd45a-null mice were bred into the BigBlue mouse strain, which uses a bacteriophage λ shuttle vector for mutation detection. Livers were taken from DMBA-treated mice 7 days after treatment, and mutation frequencies of the λ cII gene were determined (Fig. 5). Gadd45a-null livers exhibited three times more DMBA-induced mutations than wt livers. Gadd45a-null and wt livers showed a 3.1- and 2.3-fold increase in mutations after DMBA treatment, respectively. Basal mutation frequency in wt livers was similar to published data but was increased 1.7-fold in Gadd45a-null livers.

Many human tumor-prone disorders result from mutation of known DNA repair genes. In many cases, loss of DNA repair capacity is associated with family history of specific tumor types, such as mismatch repair defects in hereditary nonpolyposis colon cancer (14). In some cases, tumors are known to result from environmental exposure, such as UV radiation-induced tumors in XP patients (9). In addition, mouse models of XP show increased tumorigenesis by other agents that produce damage repaired by NER (7, 8). However, not all DNA repair defects lead to increased tumorigenesis. For example, the CS group B gene (CSB) is involved in transcription-coupled repair, but patients with this disorder do not develop UV radiation-induced tumors, possibly because of increased apoptosis of affected cells (15). Gadd45a-null cells have a partial defect in the GGR subpathway of NER (2), similar to that observed in XP group C and group E mutants (9). Mice lacking Gadd45a develop ionizing radiation-induced tumors, although the DNA lesions produced are not repaired by GGR. The mechanism for this tumorigenesis is unknown but may be related to the induction of Gadd45a by IR in a p53-dependent fashion. In addition, Gadd45a may be involved in other pathways of DNA repair because a role for the Gadd45a protein was proposed in chromatin accessibility to repair complexes (3).

DMBA produces bulky DNA adducts that are repaired via NER. Because NER is defective in cells derived from Gadd45a-null mice, it was expected that unrepaired DMBA mutations would lead to more tumors in these mice. Gadd45a-null mice did indeed develop more DMBA-induced tumors than wt mice (Fig. 2). This increase in tumors was associated with decreased NER in splenic lymphocytes (Fig. 4), and an increase in DMBA-induced mutations was found in Gadd45a-null liver, consistent with decreased repair (Fig. 5). Therefore, a mechanism for increased tumorigenesis in these mice can be ascertained by which critical mutations lead to tumor formation.

XP and CS are two rare human photosensitive disorders (9). Nine different genes were found to be responsible for various forms of these diseases, all of which are involved in NER. XPA, XPB, XPD, XPF, and XPG are all defective in both subpathways of NER, GGR, and TCR. Only XPC and XPE are involved solely in GGR. In contrast, CSA and CSB are involved solely in TCR. In animal models, GGR is the major determinant of UV-induced skin cancer, whereas TCR is the major determinant of sunburn (16). This is consistent with the relative lack of tumor formation in CS patients, whose cells are defective only in TCR. Like XPC, Gadd45a is not required for efficient TCR but is essential for maximal GGR.

Other similarities exist between XPC, XPE, and Gadd45a. All three have affinity for UV-induced DNA lesions in the context of chromatin and have postulated roles in lesion accessibility. XPC is the earliest factor involved in the initial recognition of damage in reconstituted in vitro assays. XPC changes the DNA conformation around lesions, and this has been suggested to facilitate binding of other NER proteins to the lesion (17). Likewise, Gadd45a can bind to damaged DNA in chromatin, the natural state of DNA in the cell. In vitro experiments suggest that Gadd45a may alter chromatin structure, perhaps allowing access to XPC or other repair proteins in vivo. XPE forms a complex in vivo that tightly associates with chromatin after DNA damage, suggesting that, like Gadd45a, it is also involved in recognition of chromatinized DNA damage (18). In addition to established or potential roles in damage recognition, XPC, XPE, and Gadd45a are all activated by the tumor suppressor p53 (19, 20).3 p53 regulation of these three genes, which may have similar functions in damage recognition, supports the involvement of p53 in the NER pathway.

XPA-deficient mice are sensitive to UV-induced skin tumors as well as to benzo(a)pyrene-induced internal tumors (7, 21). Like Gadd45a-null mice, none of these mouse models for human NER deficiency syndromes show high levels of spontaneous tumors. Gadd45a deletion in mice, therefore, resembles mouse models of XP. Mouse models of XP, however, have generally less severe phenotypes than their human counterparts, which also show neurological symptoms (9). This is not surprising because GGR in mice is much less robust than in humans. Human mutations in Gadd45a have not been found in human tumor cell lines (Ref. 22 and data not shown), although few tumor types have been examined. A human inactivation of Gadd45a would be expected to confer increased carcinogeninduced tumorigenesis, perhaps similar to XP.

DMBA-treated Gadd45a-null mice had a dramatically higher multiplicity of tumors than did wt mice, with many developing multiple different malignant tumors. The reason for this increase in tumorigenicity and malignancy may result from the growth and transformation properties observed for Gadd45-null cells in culture. Gadd45a-null MEF grow more rapidly and are immortal. These cells are transformed by a single oncogene (activated ras) and exhibit genomic instability (5). Therefore, in a multistage model of carcinogenesis, Gadd45a-null cells may already have compromised growth control mechanisms and hence may be even more susceptible to malignant transformation by additional cellular events.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

2

The abbreviations used are: GGR, global genomic repair; NER, nucleotide excision repair; TCR, transcription-coupled repair; CS, Cockayne‘s syndrome; XP, xeroderma pigmentosum; IR, ionizing radiation; DMBA, dimethylbenzanthracene; wt, wild type.

3

Sally Amundson, personal communication.

We thank Dr. Toshio Mori for the kind gift of anti-(6-4) photoproduct monoclonal antibody.

1
Smith M. L., Chen I. T., Zhan Q., Bae I., Chen C. Y., Gilmer T. M., Kastan M. B., O‘Connor P. M., Fornace A. J., Jr. Interaction of the p53-regulated protein Gadd45 with proliferating cell nuclear antigen [see comments].
Science (Washington DC)
,
266
:
1376
-1380,  
1994
.
2
Smith M. L., Ford J. M., Hollander M. C., Bortnick R. A., Amundson S. A., Seo Y. R., Deng C. X., Hanawalt P. C., Fornace A. J., Jr. p53-mediated DNA repair responses to UV radiation: studies of mouse cells lacking p53, p21, and/or gadd45 genes.
Mol. Cell. Biol.
,
20
:
3705
-3714,  
2000
.
3
Carrier F., Georgel P. T., Pourquier P., Blake M., Kontny H. U., Antinore M. J., Gariboldi M., Myers T. G., Weinstein J. N., Pommier Y., Fornace A. J., Jr. Gadd45, a p53-responsive stress protein, modifies DNA accessibility on damaged chromatin.
Mol. Cell. Biol.
,
19
:
1673
-1685,  
1999
.
4
Zhan Q., Antinore M. J., Wang X. W., Carrier F., Smith M. L., Harris C. C., Fornace A. J., Jr. Association with Cdc2 and inhibition of Cdc2/Cyclin B1 kinase activity by the p53-regulated protein Gadd45.
Oncogene
,
18
:
2892
-2900,  
1999
.
5
Hollander M. C., Sheikh M. S., Bulavin D. V., Lundgren K., Augeri-Henmueller L., Shehee R., Molinaro T. A., Kim K. E., Tolosa E., Ashwell J. D., Rosenberg M. P., Zhan Q., Fernandez-Salguero P. M., Morgan W. F., Deng C. X., Fornace A. J., Jr. Genomic instability in Gadd45a-deficient mice.
Nat. Genet.
,
23
:
176
-184,  
1999
.
6
de Boer J., Hoeijmakers J. H. Nucleotide excision repair and human syndromes.
Carcinogenesis (Lond.)
,
21
:
453
-460,  
2000
.
7
de Vries A., van Oostrom C. T., Hofhuis F. M., Dortant P. M., Berg R. J., de Gruijl F. R., Wester P. W., van Kreijl C. F., Capel P. J., van Steeg H., et al Increased susceptibility to ultraviolet-B and carcinogens of mice lacking the DNA excision repair gene XPA.
Nature (Lond.)
,
377
:
169
-173,  
1995
.
8
Friedberg E. C., Bond J. P., Burns D. K., Cheo D. L., Greenblatt M. S., Meira L. B., Nahari D., Reis A. M. Defective nucleotide excision repair in xpc mutant mice and its association with cancer predisposition.
Mutat. Res.
,
459
:
99
-108,  
2000
.
9
van Steeg H., Kraemer K. H. Xeroderma pigmentosum and the role of UV-induced DNA damage in skin cancer.
Mol. Med. Today
,
5
:
86
-94,  
1999
.
10
Matsunaga T., Mori T., Nikaido O. Base sequence specificity of a monoclonal antibody binding to (6-4) photoproducts.
Mutat. Res.
,
235
:
187
-194,  
1990
.
11
Mori T., Nakane M., Hattori T., Matsunaga T., Ihara M., Nikaido O. Simultaneous establishment of monoclonal antibodies specific for either cyclobutane pyrimidine dimer or (6-4) photoproduct from the same mouse immunized with ultraviolet-irradiated DNA.
Photochem. Photobiol.
,
54
:
225
-232,  
1991
.
12
Hollander M. C., Fornace A. J., Jr. Estimation of relative mRNA content by filter hybridization to a polythymidylate probe.
Biotechniques
,
9
:
174
-179,  
1990
.
13
Koch-Paiz C. A., Momenan R., Amundson S. A., Lamoreaux R., Fornace A. J., Jr. Estimation of relative mRNA content by filter hybridization to a polyuridylic probe.
Biotechniques
,
29
:
706
-714,  
2000
.
14
Buermeyer A. B., Deschenes S. M., Baker S. M., Liskay R. M. Mammalian DNA mismatch repair.
Annu. Rev. Genet.
,
33
:
533
-564,  
1999
.
15
Ljungman M., Zhang F. Blockage of RNA polymerase as a possible trigger for u.v. light-induced apoptosis.
Oncogene
,
13
:
823
-831,  
1996
.
16
Berg R. J., Rebel H., van der Horst G. T., van Kranen H. J., Mullenders L. H., van Vloten W. A., de Gruijl F. R. Impact of global genome repair versus transcription-coupled repair on ultraviolet carcinogenesis in hairless mice.
Cancer Res.
,
60
:
2858
-2863,  
2000
.
17
van der Spek P. J., Eker A., Rademakers S., Visser C., Sugasawa K., Masutani C., Hanaoka F., Bootsma D., Hoeijmakers J. H. XPC and human homologs of RAD23: intracellular localization and relationship to other nucleotide excision repair complexes.
Nucleic Acids Res.
,
24
:
2551
-2559,  
1996
.
18
Otrin V. R., McLenigan M., Takao M., Levine A. S., Protic M. Translocation of a UV-damaged DNA binding protein into a tight association with chromatin after treatment of mammalian cells with UV light.
J. Cell Sci.
,
110 (Pt 10)
:
1159
-1168,  
1997
.
19
Kastan M. B., Zhan Q., el-Deiry W. S., Carrier F., Jacks T., Walsh W. V., Plunkett B. S., Vogelstein B., Fornace A. J., Jr. A mammalian cell cycle checkpoint pathway utilizing p53 and GADD45 is defective in ataxia-telangiectasia.
Cell
,
71
:
587
-597,  
1992
.
20
Hwang B. J., Ford J. M., Hanawalt P. C., Chu G. Expression of the p48 xeroderma pigmentosum gene is p53-dependent and is involved in global genomic repair.
Proc. Natl. Acad. Sci. USA
,
96
:
424
-428,  
1999
.
21
Ide F., Iida N., Nakatsuru Y., Oda H., Tanaka K., Ishikawa T. Mice deficient in the nucleotide excision repair gene XPA have elevated sensitivity to benzo[a]pyrene induction of lung tumors.
Carcinogenesis (Lond.)
,
21
:
1263
-1265,  
2000
.
22
Blaszyk H., Hartmann A., Sommer S. S., Kovach J. S. A polymorphism but no mutations in the GADD45 gene in breast cancers.
Hum. Genet.
,
97
:
543
-547,  
1996
.