Targeted Disruption of the Mouse Ing1 Locus Results in Reduced Body Size, Hypersensitivity to Radiation and Elevated Incidence of Lymphomas

Oncogene (2006) 25, 857–866 & 2006 Nature Publishing Group All rights reserved 0950-9232/06 $30.00 www.nature.com/onc ORIGINAL ARTICLE Targeted disruption of the mouse ing1 locus results in reduced body size, hypersensitivity to radiation and elevated incidence of lymphomas

JV Kichina1,2,6, M Zeremski2,6,7, L Aris2,6, KV Gurova1,3, E Walker4, RFranks 2, AY Nikitin5, H Kiyokawa2 and AV Gudkov1,3

1Department of Molecular Genetics, Lerner Research Institute, Cleveland, OH, USA; 2Department of Biochemistry and Molecular Genetics, University of Illinois at Chicago, Chicago, IL, USA; 3Cleveland BioLabs, Inc., Cleveland, OH, USA; 4Department of Quantitative Health Sciences, The Cleveland Clinic Foundation, Cleveland, OH, USA and 5Department of Biomedical Sciences, Cornell University, Ithaca, NY, USA

Ing1 belongs to the family of evolutionary conserved genes neoplastic transformation of mouse mammary epithelial encoding nuclear PHD finger-containing proteins impli- cells: the GSE that encoded an antisense RNA against a cated in a variety of processes, including tumorigenesis, previously unknown gene promoted transformation of replicative senescence, excision repair and response to fibroblasts and epithelia cells (Garkavtsev et al., 1996). genotoxic stress. We have generated mice deficient in all Since overexpression of cDNA for the identified gene the isoforms of Ing1 by targeted disruption of the exon resulted in growth repression of human fibroblasts, it that is common for all ing1 transcripts. Embryonic was named Ing1 (abbreviation from ‘INhibitor of fibroblasts from ing1-knockout mice were similar to the Growth’; Garkavtsev et al. (1996)). Decreased expres- wild-type cells in their growth characteristics, replicative sion of ING1 was found in several breast carcinomas lifespan in culture, p53 induction and sensitivity to various and the Ing1 gene was found rearranged in one cytotoxic treatments with minor alterations in cell cycle neuroblastoma cell line pointing at its putative tumor- distribution in response to genotoxic stress. ing1-deficient suppressor function (Garkavtsev et al., 1996). It was animals are characterized by reduced size with no obvious later shown that ING1 expression is increased in morphological, physiological or behavioral abnormalities, senescent human diploid fibroblasts and that expression indicating that ing1 function is dispensable for the viability of Ing1-derived antisense RNA prolongs the lifespan of of mice under normal physiological conditions. Loss of these cells in culture (Garkavtsev and Riabowol, 1997). ing1 was associated with earlier onset and higher incidence In addition, inhibition of Ing1 was associated with a of lymphomas. Consistent with the possible involvement of reduced apoptotic response in some cellular systems Ing1 in DNA repair, ing1-deficient mice were more (Helbing et al., 1997). sensitive to total body gamma radiation. Our observations Ing1 belongs to a family of evolutionally conserved are well in line with the suggested role of ing1 as a genes found in animals, plants and fungi that encode candidate tumor suppressor gene involved in control of nuclear PHD finger proteins. The Ing1 gene family in DNA damage response. humans includes five paralogs named Ing1, Ing2, Ing3, Oncogene (2006) 25, 857–866. doi:10.1038/sj.onc.1209118; Ing4 and Ing5 (He et al., 2005). Both mouse and human published online 19 September 2005 Ing1 genes encode several protein isoforms translated from different transcripts that are initiated from Keywords: ING1; gene knockout; lymphoma; mouse; alternative promoters and consist of different 50-and radiation sensitivity common 30- exons. The three alternative transcripts of mouse ing1 encode two proteins: p37 (Ing1b) and p31 (Ing1a and Ing1c), one of which is a truncated version of Introduction the other (Zeremski et al., 1999). The human Ing1 gene has similar general structure and expression with one ING1a The Ing1 gene was originally identified in the search for more isoform identified, p47 , which has not so far genetic suppressor elements (GSEs) that promoted been found in mice (Gunduz et al., 2000; Saito et al., 2000). ING1 is expressed in established cell lines and in all tissues of adult mice, with higher levels found in Correspondence: Dr AV Gudkov, Department of Molecular Genetics, thymus and testis (Zeremski et al., 1999). NC20, Lerner Research Institute, Cleveland Clinic Foundation, 9500 Euclid Avenue, Cleveland, OH 44195, USA. ING1 was found to be functionally and physically E-mail: [email protected] linked to p53 tumor suppressor (Garkavtsev et al., 1998; 6These authors contributed equally to this work. Zeremski et al., 1999; Leung et al., 2002), further 7Current address: Center for the Study of Hepatitis C, Weill Medical supporting the putative tumor suppressor function of College of Cornell University, 525 E.68th Street, New York, NY 10021, USA. ING1. There is experimental evidence indicating the Received 19 May 2005; revised 28 July 2005; accepted 4 August 2005; involvement of ING1 in repair of UV-induced DNA published online 19 September 2005 damage (Cheung, Jr et al., 2001). Consistently, the Phenotype of ing1 knockout mice JV Kichina et al 858 deficiency in Ing1 homologues in yeast interferes with be functionally distinct may complicate interpretation of normal cell cycle control and drug sensitivity (Loewith previously published clinical studies since none of the et al., 2000). In keratinocytes (Cheung, Jr et al., 2000) latter distinguished among the different ING1 protein and melanocytes (Cheung, Jr et al., 2001), the levels of forms. ING1 protein increase upon UV irradiation and its Thus, despite accumulation of a considerable body of accumulation in nucleoli confirms ING1 involvement in evidence concerning structure, evolution, expression of genotoxic stress response (Scott et al., 2001a). Ing1 gene and biochemical characteristics of its pro- While functional analyses suggested that ING family ducts, little is understood about physiological signifi- members play a role in cell response to genotoxic stress cance of ING1 at the organismal level. Evolutionary and determination of sensitivity to DNA damage, conservation and the association of ING family proteins biochemical analysis of ING proteins revealed their with important regulators of gene expression and DNA involvement in the enzymatic complexes determining repair implied the existence of basic physiological protein acetylation and deacetylation (Skowyra et al., functions for these proteins which could be revealed 2001; Kuzmichev et al., 2002; Vieyra et al., 2002). by translation into severe phenotypic defects resulting Moreover, another mammalian member of the ING from deficiency in Ing genes. To directly address this family, ING2, was found to take part in acetylation of intriguing possibility, we generated mice with targeted p53 (Nagashima et al., 2001). Interestingly, p33ING1b was disruption of the ing1 locus. The resulting animals reported to affect the degree of physical association appeared viable and fertile, although smaller than the between proliferating cell nuclear antigen (PCNA) and wild-type (wt) ones, and were characterized by an p300 (Scott et al., 2001b). Importantly, cooperation of increased sensitivity to ionizing radiation and elevated ING1 with p53 and its involvement in histone acetyla- incidence of lymphomas, observations consistent with tion/deacetylation complexes are different for different tumor-suppressor function of ING1 and its proposed ING1 isoforms (Skowyra et al., 2001; Vieyra et al., role in DNA repair. 2002) pointing to a functional interplay among the products of Ing1 locus. In summary, although many mechanistic details are still missing, the accumulated data suggest that ING1 proteins are likely to function as Results a link between DNA repair, apoptosis, and chromatin remodeling via HAT/ING1/PCNA protein complexes Ing1À/À-deficient mice have reduced body weight, but no (Campos et al., 2004a; He et al., 2005). gross abnormalities Although the Ing1 gene was mapped to the near- The mouse ing1 gene is composed of four exons. The telomeric region of the long arm of human chromosome first three exons (1a, 1b and 1c) form the 50 ends of three 13 (Garkavtsev et al., 1997; Zeremski et al., 1997), which alternative ing1 transcripts initiated from different is frequently rearranged in squamous cell carcinomas of promoters. The fourth exon is common for all tran- head and neck (Maestro et al., 1996), no cancer-specific scripts and contains most of the ing1 protein-coding rearrangements of Ing1 sequences were found in these sequences including the conserved PHD finger domain. types of tumors (Sanchez-Cespedes et al., 2000) or in Three mRNAs give rise to two Ing1 proteins p31ing1 and colorectal cancers (Sarela et al., 1999), myeloid leuke- p37ing1 (Zeremski et al., 1999). To investigate the mias (Ito et al., 2002) and lymphoid malignancies biological function of ing1, this gene was disrupted by (Ohmori et al., 1999). Rare somatic mutations in Ing1 homologous recombination. A 3 kb genomic fragment were reported in esophageal squamous cell carcinomas containing the entire common ing1 exon was replaced (Gunduz et al., 2000; Chen et al., 2001), breast tumors with a neomycin resistance (neo) gene abolishing (Toyama et al., 1999; Tokunaga et al., 2000) and production of both p31ing1 and p37ing1 protein isoforms gastrointestinal carcinoma (Oki et al., 1999); Ing1 (Figure 1a) in embryonic stem (ES) cells in vitro using a mutations are more common in melanoma, occurring standard gene disruption approach. Two selected clones in up to 20% of tumors. Moreover, these mutations are with dual resistance to G418 and gancyclovir were found in functionally important domains of the protein injected into C57BL/6 blastocysts to generate chimeric (Campos et al., 2004b). Expression of ING1 was found mice. Male chimeras were backcrossed with C57BL/6 to be decreased in a subset of breast (Toyama et al., females, and ing1 heterozygous F1 mice were interbred 1999), hepatocellular (Ohgi et al., 2002), lymphoid to produce the F2 generation. PCRanalysis of tail DNA (Ohmori et al., 1999; Chen et al., 2001) and gastric from these mice revealed transmission of the disrupted malignancies (Oki et al., 1999). In contrast, elevated allele and generation of mice with all three expected levels of ING1 were reported in melanomas (Campos genotypes (Figure 1b). To test whether the disruption of et al., 2002; Nouman et al., 2002; Kichina et al., 2003). the ing1 gene results in its complete inactivation, we Moreover, ING1 was found to be excluded from the analysed expression of ing1 mRNA and proteins in nucleus in such cancers as melanoma, seminoma, mouse embryo fibroblasts (MEFs). Neither ing1 mRNA papillary thyroid carcinoma, ductal breast carcinoma nor its protein products were detected in ing1À/À MEFs and acute lymphoblastic leukemia (Nouman et al., by Northern and Western analyses (Figure 1c and d, 2002); the biological meaning of this phenomenon respectively). though remains unclear. It is noteworthy that the Thorough morphological, histopathological and existence of multiple isoforms of ING1 that appear to hematological examination performed at the Mouse

Oncogene Phenotype of ing1 knockout mice JV Kichina et al 859

Figure 1 Targeted disruption of ing1 gene. (a) Structure and restriction map of the mouse ing1 gene, targeting vector and the locus following recombination. A 3 kb-long genomic area containing ing1 common exon is replaced with neo gene by homologous recombination. Homologous recombination does not introduce TK gene in the genome. Transcriptional directions of neo and TK are shown. Indicated XhoI–NotI fragment was used as a probe for screening for homologous recombinants in ES cells following XhoI digestion of genomic DNA. This probe is detecting 6.2 kb fragment in recombinant alleles, and 15 kb fragment in wt allels. (b) Genotyping of mice by PCR. Two sets of primers, corresponding to neo or to the common ing1 exon, were used. PCRreactions detect 248 bp neo fragment or 344 bp ing1 fragment. (c) Analysis of ing1 mRNA in ing1À/À, ing1 þ /À and ing1À/À mouse embryonic ﬁbroblasts. (d) Analysis of Ing1 proteins in ing1À/À, ing1 þ /À and ing1À/À mouse embryonic ﬁbroblasts.

Phenotyping Facility at Ohio State University (Colum- of 13 litters in ing1À/À crosses vs one out of 19 for the wt bus, OH, USA) revealed no apparent abnormalities crosses). in ing1-deficient animals as compared to ing1 þ / þ To check whether the loss of ing1 had an effect on littermates with all phenotypic parameters tested embryonic and early postnatal viability, we measured (see Materials and methods) being within normal the proportion of ing1À/À and ing1 þ /À animals among the limits. Animal size was the only morphological para- progeny of heterozygous crosses. Of 143 genotyped meter slightly different in ing1À/À animals: analysis of offspring, the occurrence of ing1 þ / þ , ing1 þ /À and ing1À/À littermates originating from three pairs of ing1 hemi- genotypes (30, 53, 17%, respectively) was slightly zygous parents indicated that ing1 deficiency was different from the expected Mendelian distribution associated with the reduction in body weight (P ¼ 0.06, w2 test), indicating that the loss of ing1 may (Po0.0001, Wilcoxon two sample test, n ¼ 21) in both affect survival during development. males (20% reduction) and females (10% reduction) (Figure 2). Animals of either genotype did not display any Sensitivity of ing1À/À MEFs to cytotoxic treatments obvious signs of gross behavioral abnormalities; how- A number of reports have implicated ING1 in stress- ever, their ability to care for their progeny may be induced apoptosis and growth arrest reviewed in He somewhat impaired since ing1À/À females showed a et al. (2005). We, therefore, investigated the sensitivity tendency to neglect or destroy their progeny (six out of ing1-deficient MEFs to various damaging treatments

Oncogene Phenotype of ing1 knockout mice JV Kichina et al 860 MEFs in UV-induced p53 activation (Figure 4). Both showed p53 accumulation after UV treatment accom- panied by the accumulation of cyclin-dependent kinase inhibitor p21 encoded by p53-responsive gene p21/waf1/ cip1 (el-Deiry et al., 1993). The amounts of accumulated p53 and its ability to induce its transcriptional target p21 were essentially indistinguishable between the MEFs of either genotype. Hence, small differences in UV sensitivity of ing1-deficient fibroblasts were not accom- panied by obvious differences in their p53 response. This, however, does not mean that p53 is not involved in modulation of cell sensitivity to UV by ing1 since the latter may affect G2 checkpoint control activity of p53, a possibility that remains to be experimentally tested. Figure 2 ing1 deficiency leads to reduction in body size. Relative weight represents weight of females and males normalized Another function of p53, control of replicative senes- according to average weight of wt animals of the particular gender. cence, was also unchanged in ing1-deficient MEFs, Each group consists of three litters originating from three different which were indistinguishable from the wt cells in their pairs of mice. lifespan in culture (data not shown).

including UV-B irradiation and chemotherapeutic Ing1 deficiency results in reduced resistance to total body drugs. Primary MEF cultures were prepared from the gamma irradiation embryos resulting from crossing ing1 heterozygous mice; To see whether animal response to genotoxic stress the genotype of embryos was determined after plating in changed after the deletion of ing1 gene, we compared the tissue culture. The experiment was repeated in three sensitivity to gamma irradiation of ing1À/À and wt mice. independent sets of cultures obtained from different First, males and females of both genotypes were exposed / females. No differences were observed between ing1À À, to a single dose of total body gamma radiation and their / / ing1 þ þ and ing1 þ À fibroblasts after treatment with survival was monitored. We used two radiation doses taxol, doxorubicin, colcemid or gamma radiation that cause animal death through different mechanisms: (Figure 3a). Marginal but reproducible (in three 11 Gy (Figure 5a), that kills animals primarily by independent experiments) differences were found in induction of hematopoietic syndrome, and 15 Gy, that sensitivity to UV-B as ing1-deficient cultures contained causes more rapid death from gastrointestinal syndrome a subpopulation of cells that were more sensitive to the (Komarova et al., 2004). p53-deficient mice, known to treatment (Figure 3a). Consistently, fluorescent-acti- be resistant to radiation-induced hematopoietic syn- vated cell sorting (FACS) analysis showed an increased drome (Komarov et al., 1999), were used for compar- number of cells in the sub-G1 zone indicative of a higher ison (data not shown). No significant differences proportion of apoptotic cells in the UV-B-treated were observed in the death rate of ing1-deficient and population (Figure 3b). wt mice at any dose applied (Figure 5b), while FACS analysis of the cell cycle distribution revealed p53-deficient animals demonstrated the expected radia- no differences between synchronized (48 h in serum-free tion resistance. medium) and unsynchronized cultures of wt and However, when mice of either ing1À/À or wt genotype / ing1À À MEF (data not shown). The application of were subjected to six consecutive daily doses of 2.34 Gy, different types of growth inhibitory treatments to MEF there was a remarkable decrease in survival among cultures followed by cell cycle analysis demonstrated ing1À/À animals of both genders (Figure 5c–e). The altered distribution of cells between G1 and G2 phases differences between the wt and knockout animals / in ing1À À MEFs: ing1 deficiency was associated with a remained significant after adjustment for gender and small (see numbers in Figure 3b), but reproducible (in body size (P ¼ 0.0001, Cox proportional hazard test). three independent experiments) decrease in proportion Moreover, the difference was more profound among of G2 after the treatments with UV-B, doxorubicine and females (P ¼ 0.0020, Cox proportional hazard test) as taxol. This apparent alteration in cell cycle checkpoint compared to males (P ¼ 0.0229, Cox proportional control, however, was not translated into significant hazard test). changes in sensitivity to radiation and drugs under the experimental conditions used. Increased incidence of lymphomas in aging ing1-deficient mice Ing1À/À MEFs are capable of p53 induction in response to To test the long-term effect of the absence of ing1 gene UV treatment expression, 118 ing1À/À mice and 54 of their littermates It has been reported previously that ING1 cooperates with wt genotypes were monitored for 23 months. with the transcriptional activation function of p53 During the period of observation 42 ing1-deficient (Garkavtsev et al., 1998). Nevertheless, we failed to (35.6%) and 11 wt littermates (20.8%), showed signs detect any difference between ing1-deficient and wt of morbidity or visible abnormalities, indicating a

Oncogene Phenotype of ing1 knockout mice JV Kichina et al 861

Figure 3 Sensitivity of ing1À/À MEFs to cytotoxic treatments. (a) ing1À/À, ing1 þ / þ or ing1 þ /À MEFs were treated with 100, 200 or 400 ng/ml of taxol; 50, 200 or 400 ng/ml of doxorubicin; 20, 50 or 200 ng/ml of colcemid; 30, 90 or 270 mJ cmÀ2of UV-B. Survival was measured using methylene blue staining/extraction method 5 days after the treatments. Survival of the treated cells was normalized for that of untreated cultures of the same genotype. (b) Cell cycle analysis of ing1À/À and ing1 þ / þ MEFs upon treatments with 10 Gy of gamma radiation, 210 mJ cmÀ2 sec of UV, 100 ng/ml of doxorubicin, 20 ng/ml taxol and 20 ng/ml of colcemid (representative result of one of three experiments performed). The proportions of cells in G2 phase are indicated. signiﬁcant difference between the genotypes (P ¼ 0.045, were composed of a pleomorphic population of small w2 test). The moribund animals were examined for gross lymphocytes, centrocytes, centroblasts and follicular pathology. In addition, a number of asymptomatic mice center cells, resembling those of a germinal center of all three genotypes were killed and examined as age- (Figure 6c). To a different, extent neoplastic lympho- matched controls. cytes were also detected in the liver, kidney, lungs and As opposed to wt animals, a number of ing1À/À mice mesentrial and mediastinal lymph nodes. Immunohisto- revealed splenomegaly upon necropsy. Enlarged spleens chemical evaluation demonstrated presence of CD45R had large white pulp areas (Figure 6a) suggesting the antigen in six tumors (Figure 6d), while two lymphomas possible presence of lymphomas in such mice. Histo- were bi-phenotypic, with the CD45R þ CD3 þ pheno- pathological analysis of 42 ing1À/À mice indeed showed type. Thus, the majority of lymphomas were classiﬁed as lymphomas in nine of them (21%, Figure 6c, Table 1). follicular center B-cell lymphomas. Importantly, none of 11 wt moribund littermates had Both ing1À/À and their wt littermates have developed lymphomas at comparable age. All but one lymphoma other neoplasms, including histiocytic sarcoma, soft was localized in the spleen. Most frequently, lymphomas tissue tumors and myeloid leukemia (Table 1) in

Oncogene Phenotype of ing1 knockout mice JV Kichina et al 862 agreement with earlier published rates of such tumors in aging C57BL/6/129 mice (Ward et al., 2000).

Discussion

The interest in the phenotypic consequences of Ing1 deﬁciency in a mammalian organism stems from multiple observations made both in mammalian cells and in yeast. The Ing1 gene is well conserved between the two taxa and so is the association of its products with Figure 4 ing1-deﬁcient cells are capable of activating p53 in histone acetylation/deacetylation complexes (reviewed response to UV irradiation. Wt and ing1À/À MEFs were irradiated with UV. Levels of p53 and p21 proteins were assayed at indicated in Campos et al., 2004a; He et al., 2005). In yeast, ING1- time points by immunoblotting with anti-p53 and anti-p21 related proteins are essential for the maintenance of antibodies, respectively. normal growth and morphology, as well as metabolism

Figure 5 Survival of ing1À/À and wt mice after gamma irradiation. Five females and ﬁve males of each genotype were subjected to 11 Gy (a)or 15 Gy of gamma irradiation (b). Survival was monitored for three months following the treatment. No additional mortality was observed after day 16 (a). (c). 21 ing1À/À and 32 wt mice were given six consecutive daily doses (2.34 Gy each) of gamma radiation, starting at day 1 (indicated by arrows). Percent survival is shown for all animals (upper panel) and separately for females and males. For survival, animals were monitored for 36 days.

Oncogene Phenotype of ing1 knockout mice JV Kichina et al 863

Figure 6 ing1-deﬁcient mice develop lymphomas. Follicular B-cell lymphoma of the spleen in ing1À/À mouse, 15-month-old. (a) Splenomegalic mouse. (b) Partial preservation of the nodular pattern in the spleen. (c) Lymphoma is composed of a mixed population of lymphoid cells, including small lymphocytes, centroblasts and follicular center cells. Note mitosis (arrowhead) and apoptotic cells (arrow). (d) Diffuse expression of CD45Rantigen in neoplastic cells. ( e) Expression of CD3 antigen is limited to reactive small lymphocytes with regular round nuclei and narrow rim of cytoplasm. (b and c), Hematoxylin and eosin. (d and e), Immunostaining with CD45Rand CD3 antibodies, respectively. Calibration bar: 200 mm(b), 20 mm(c) and 50 mm(d, e).

Table 1 Lesions in moribund ing1À/À mice and their wild-type ING1 in cell death (Helbing et al., 1997; Garkavtsev littermates et al., 1998; Shinoura et al., 1999; Cheung, Jr and Li, Lesion Ing1À/À Wild type 2002; Kuzmichev et al., 2002; Shimada et al., 2002) and repair (Cheung, Jr et al., 2001) prompted us to Numbera % Numbera % investigate the phenotypic consequences of ING1 Lymphomab 9210 0deletion at an organismal level. Histiocytic sarcoma 3 7 1 9 Ing1 encodes several protein isoforms that might be Soft tissue tumorsc 37218functionally different (Zeremski et al., 1999; Scott et al., Skin tumorsd 002182001b; Skowyra et al., 2001; Vieyra et al., 2002). Since Myeloid leukemia 3 7 1 9 there is some degree of uncertainty about the functional Extramedullary hematopoiesis 18 43 7 64 Nephropathy 11 26 1 9 interplay among the various ING1 isoforms, we chose Intestinal inflammation 3 7 0 0 an approach to constitutively eliminate all the products Periarteritis nodosa 1 2 0 0 of this locus via homozygous deletion of the conserved common exon that is shared by all Ing1 protein variants aNumber of mice with tumor (percentage) out of 42 Ing1À/À and 11 (Zeremski et al., 1999). The knockout succeeded in that wild-type mice, respectively. bLymphomas include B lymphoma (n ¼ 6), T lymphoma (n ¼ 1), bi-phenotypic lymphomas-(n ¼ 2) in no ing1 gene products were detectable in the cells and ing1À/À mice. cSoft tissue tumors include hemangiosarcoma (n ¼ 1), organisms carrying the homozygous mutation. Some- NOS (n ¼ 1), malignant fibrous histiocytoma (n ¼ 1) in ingÀ/À mice, and what unexpectedly for such an evolutionarily conserved hemangioma (n ¼ 1) and angiosarcoma (n ¼ 1) in their wild-type and functionally active gene, ing1-deficient animals did d littermates. Both skin tumors are kerathoacantomas. not exhibit major developmental or behavioral abnormalities. The most noticeable morphological change of nonfermentable carbon sources and resistance to accompanying the loss of ing1 was reduction in body various stress stimuli (Loewith et al., 2000). In size. mammalian cells, inhibition of Ing1 expression by a The deviation from Mendelian distribution in the GSE promotes transformation (Garkavtsev et al., 1996), progeny of heterozygous animals suggests that ing1- while ING1 proteins cooperate with p53 tumor sup- deficient individuals are preferentially lost during pressor (Garkavtsev et al., 1998) and are deregulated in embryogenesis or the early postnatal period. Interest- a variety of malignancies (Ohmori et al., 1999; Oki et al., ingly, contrary to what is expected from the Mendelian 1999; Toyama et al., 1999; Chen et al., 2001; Campos distribution, the occurrence of heterozygous pups in et al., 2002; Nouman et al., 2002; Ohgi et al., 2002). these crosses is also lower than double the occurrence These findings, as well as the suggested involvement of of their wt littermates. This may indicate that the

Oncogene Phenotype of ing1 knockout mice JV Kichina et al 864 heterozygous mice also have reduced viability in et al., 2005). Based on structural similarity one may comparison to their normal counterparts. predict certain degree of functional redundancy among Our experimental strategy was to characterize these proteins and, as the result, low penetrance of loss various phenotypes previously associated with ING1 of function mutations in individual members. Indeed, function at the cellular level. Barring small, albeit more profound ING1 association with cancer was reproducible, differences following UV treatment, shown for two other members of ING family, ING3 ing1-deficient and wt cells remained indistinguishable and ING4: reduced expression and allelic loss of ING3 in their growth rates, responses to various stress stimuli was frequently found in head and neck squamous cell and abilities to activate p53 tumor suppressor. It is carcinomas (Gunduz et al., 2002), while ING4 is noteworthy, that, according to the published reports downregulated in the majority of glioblastomas (Gar- (Cheung, Jr et al., 2000), fibroblasts, unlike melanocytes kavtsev et al., 2004). and keratinocytes fail to induce Ing1 in response to Second, earlier experiments with Ing1 in mammalian DNA damage; therefore further investigating the role of cells were conducted on the genetic background of stable Ing1 in DNA repair may require using other cell type cell lines, which are likely to contain multiple genetic models. Although suppression of Ing1 by antisense changes. It is possible that additional genetic alterations RNA was reported to extend lifespan of human have to cooperate with Ing1 in order to reveal the role of fibroblasts, suggesting involvement of Ing1 in the the latter in such processes as tumorigenesis and stress control of replicative senescence (Garkavtsev and response. In this case, the lack of a severe phenotype in Riabowol, 1997), we could not detect any difference in ing1À/À animals may be explained, for example, by the lifespan of MEFs derived from ing1À/À and wt functional complementation of otherwise structurally animals. unrelated proteins. The ing1-deficient animals showed a peculiar pattern Finally, direct functional comparison of the biological of sensitization to gamma irradiation. Although not activities of the long and short proteins encoded by significantly different after single-dose treatment, these mouse ing1 led to the conclusion that the two protein animals succumbed rapidly after frequent repetitive products have opposite biological activity with respect irradiation with lower doses. It is tempting to speculate to histone acetylation and the p53 signalling pathway, that this phenomenon is related to the previously with the longer one being a p53 inhibitor (Zeremski reported role of ING1 in DNA repair. In fact, relative et al., 1999). In this case, simultaneous elimination of hypersensitivity to repetitive low-dose DNA damage is both proteins could have milder consequences than consistent with a defect in DNA repair systems. In any individual knockouts. case, pronounced sensitization of ing1À/À animals The lymphoma-prone phenotype of the ing1À/À suggests that this function of ing1 cannot be compen- does not rule out the involvement of this locus in sated by other proteins. the function of other tissues, but rather reflects the At the organismal level, the differences between the wt situation where the natural redundancy of the cellular and the ing1-deficient animals became noticeable at an pathways is insufficient to compensate for the change in older age, when higher proportion of ing1À/À mice a certain tissue. All these possibilities can be experimen- developed lymphoid malignancies. This observation tally addressed by selective elimination of individual points at the potential involvement of the Ing1-p53 link ing1 isoforms and genetic crosses that combine ing1 in determination of this phenomenon: it is noteworthy deficiency with other genetic alterations. that the majority of spontaneous malignancies appear- In summary, the phenotype of mice with targeted ing in p53-deficient mice are also lymphomas, although disruption of ing1À/À is consistent with the original developing much earlier than those in ing1-deficient definition of this gene as a candidate tumor suppressor. mice (Donehower et al., 1992). In addition to the tumor Hypersensitivity of the knockout animals to repetitive phenotype, a number of ing1À/À mice (n ¼ 15, 36, vs low-dose DNA damage is consistent with ing1 involve- 1.9% in the control group) had intestinal inflammatory, ment in repair and genome maintenance. The animals nephropathy or periarteritis nodosa (Table 1). Poten- generated in this work provide a valuable model for tially, all those lesions may be indicative of a deregulated future studies on Ing1 functions on cellular and immune response. However, due to the limited number organismal levels. of observations, these results were not statistically significant, and the putative role of ing1 in the regulation of immune response requires further evaluation. Although we cannot exclude the possibility that the Materials and methods sensitivity or specificity of our assays was insufficient to detect some significant changes induced by ing1 defi- Generation of ing1-deficient mice ciency, the phenotype of ing1À/À animals appeared to be Genomic clones containing the entire ing1 sequences milder than what might be suggested based on the were isolated from 129/SVJ mouse genomic library (Strata- gene, CA, USA) using ing1 cDNA as a probe (Ramirez-Solis results of prior studies on this gene and its products. The and Bradley, 1994). ing1 was disrupted by replacing common À/À lack of prominent abnormalities in ing1 animals may ing1 exon for neo gene-expressing cassette using OS.DUP/DEL be a consequence of several, not mutually exclusive, vector. A 4 kb-long NotI–BamHI fragment of genomic biological phenomena. First, it is becoming clear that DNA positioned upstream from the ing1 common exon, Ing1 is a member of a conserved protein family (He containing ing1 exon 1c, was directly cloned into OS.DUP/

Oncogene Phenotype of ing1 knockout mice JV Kichina et al 865 DEL. Another 4 kb-long PstI–XbaI genomic fragment posi- with hematoxylin and eosin or with rabbit anti-human CD3 tioned downstream from the common ing1 exon was cloned polyclonal antibody (DACO, Denmark) at 1 : 400 dilution and first into a pBluescript and then cloned into OS.DUP/DEL with biotin-conjugated rat anti-mouse CD45R/B220 mono- vector using XhoI and NheI sites. To allow negative selection, a clonal antibody RA3-6B2 (BD Biosciences Pharmingen, CA, herpes simplex virus thymidine kinase gene was positioned USA) at 1 : 50 dilution. downstream of ing1 genomic fragment (Ramirez-Solis and Bradley, 1994). Targeting vector linearized by PvuI was introduced into CJ- Southern and Northern blot hybridizations 7 ES cells by electroporation (Swiatek and Gridley, 1993; Total RNA was extracted from cells and tissues by Trizol Kiyokawa et al., 1996). Transfected cells were selected in the Reagent (Invitrogen Co, CA, USA), while genomic DNA was presence of 0.5 mg/ml G418 and 0.2 mM gancyclovir and purified by DNAeasy kit (Qiagen Inc., CA, USA). DNA and analysed by Southern blot hybridization following XhoI RNA analyses were performed as described previously digestion. The 1 kb-long XhoI–NotI fragment, located up- (Zeremski et al., 1999). stream of the targeting vector fragment, was used as a hybridization probe which reveals a 6.2 kb fragment in ing1À/À, a 15 kb fragment in wt and both fragments in ing1 þ /À Western immunoblotting heterozygous cells. Cells were lysed in RIPA buffer (25 mM Tris HCl, pH7.2, Selected ES cells were injected into blastocyst-stage C57BL/ 125 mM NaCl, 1% NP40, 1% sodium deoxycholate, 1 mM 6 mice embryos and subsequently transferred into pseudo- EDTA) containing 1 mM PMSF (Sigma-Aldrich Co, MO, pregnant C57BL/6 recipient. Chimeric males were crossed to USA), 10 mg/ml of aprotinin ((Sigma-Aldrich Co, MO) and C57BL/6 females. Germline transmission of the injected ES 10 mg/ml of leupeptin (Sigma-Aldrich Co, MO, USA). Protein cells was monitored by identifying agouti mice among the F1 concentrations were determined by Dc protein assay kit offspring of chimeric animals crossed with C57BL6 mice and (BioRad Laboratories, CA, USA). Equal amounts of 30 mg subsequent PCRanalysis. F1 heterozygous mice were in- oftotal protein were separated on gradient 4–20% precast terbred to produce F2 generation. NuPage Novex gels (Invitrogen Co, CA, USA) and blotted Genotyping of mice was routinely performed by PCRusing onto PVDF membranes (Amersham Life Science, IL, USA). genomic DNA isolated from tails of weaning age pups. Neo- Ing1 protein isoforms were detected using monoclonal anti- specific primers (antisense: 50-GCTGTGCTCGACGTTGT 0 0 0 ING1 antibodies CAB1 (BD Biosciences Pharmingen, CA, CAC-3 ; sense: 5 -CTCTTCGTCCAGATCATCCTG-3 ) de- USA) at 1 : 1000 dilution. p53 and p21 were visualized by tect a 248 bp fragment from the knockout allele, while ing1- 0 mouse monoclonal antibodies DO1 and F-5 (Santa-Cruz specific primers (antisense: 5 -GCGTTCTCTCGATTCT Biotechnology Corp, CA, USA) at 1 : 1000 dilution, respec- CATTGTTT-30; sense: 50-CCTGAAGGAGCTGGACGAC 0 tively. Secondary goat anti-mouse IgG HRP conjugated TACT-3 ) amplify a 344 bp fragment from the wt ing1 locus antibodies used at 1 : 3000 dilutions were also obtained from and fail to do so in the knockout one (Figure 1b). PCRregime Santa-Cruz Biotechnology Corp. (CA,USA). included 3 min at 941C followed by 30 cycles consisting of 1 min at 941C, 1 min at 601C and 40 s at 721C completed by 5 min at 721C. Animals were maintained in a pathogen-free environment at Mouse embryonic fibroblasts Cleveland Clinic Animal Facility in accordance with regula- Primary cultures of mouse embryonic fibroblasts were isolated tions. Gamma radiation was performed using a Sheperd 137Cs from 12.5-day embryos according to standard protocol and source at a dose rate of 3.85 Gy/min. maintained in Dulbecco’s modified Eagle’s medium supple- mented with 10% fetal bovine serum. In total, 10 000 cells were plated per one well of 96-well plate, and subjected to a range of treatments with cytotoxic drugs or UV next day. After 5 days, Mouse phenotyping cell survival was measured by standard methylene blue Comparison of a pair wt and ing1 knockout littermates staining/extraction assay. In brief, cells were fixed with was performed at the Mouse Phenotyping Facility at methanol and stained with 1% methylene blue dye dissolved Ohio State University (Columbus, OH, USA; Director: Dr in 50% methanol solution. Methylene blue was then extracted Donna F Kusewitt). Males and females (6-week-old) were from the cells by 1% HCl and the absorbance was measured at analysed using morphological and X-ray examination of 600 nm. organs, estimation of gross animal weight and weight of individual organs, histopathological examination of HE- stained samples of major tissues (heart, diaphragm, tongue, soleus, lung, thymus, trachea, esophagus, thyroid, parathy- Analysis of cell cycle using FACS roid, kidney, salivary glands, cervical lymph nodes, adrenal, Cells were lifted from the plate by incubation with 0.5 M. pituitary, stomach, pancreas duodenum, Ileum, cecum, colon, EDTA, washed with PBS and fixed in 70% ethanol. After liver, gall bladder, spleen, testis, epididymis, seminal vesicle, 30 min incubation at À201C, ethanol was replaced by staining prostate, bladder, rectum, neck skin, inguinal skin, mammary solution (100 mg/ml of RNAse A and 40 mg/ml of propidium gland, preputial gland, knee, sternum, femur, spine, brain, iodide). After 2 h incubation at 371C, DNA content was head, eyes, ovary), and hematological and serum chemistry measured using FACS Calibur (Becton Dickinson) and analysis. analysed using CellQuest software.

Pathology diagnosis Acknowledgements Moribund mice were killed and examined grossly. Organs with We thank James Artwohl for help in animal experiments and pathological lesions were collected and ﬁxed in 10% buffered Natalia Tararova for assistance in FACS analysis. This work formalin. Fixed tissues were processed by the Imaging Core at was supported by grants from NIH CA60730 and CA17579 to the Cleveland Clinic. Parafﬁn-embedded sections were stained AVG.

Oncogene Phenotype of ing1 knockout mice JV Kichina et al 866 References

Campos EI, Cheung Jr KJ, Murray A, Li S, Li G. (2002). Br J Leung KM, Po LS, Tsang FC, Siu WY, Lau A, Ho HT et al. Dermatol 146: 574–580. (2002). Cancer Res 62: 4890–4893. Campos EI, Chin MY, Kuo WH, Li G. (2004a). Cell Mol Life Loewith R, Meijer M, Lees-Miller SP, Riabowol K, Young D. Sci 61: 2597–2613. (2000). Mol Cell Biol 20: 3807–3816. Campos EI, Martinka M, Mitchell DL, Dai DL, Li G. Maestro R, Piccinin S, Doglioni C, Gasparotto D, Vukosavl- (2004b). Int J Oncol 25: 73–80. jevic T, Sulfaro S et al. (1996). Cancer Res 56: 1146–1150. Chen L, Matsubara N, Yoshino T, Nagasaka T, Hoshizima N, Nagashima M, Shiseki M, Miura K, Hagiwara K, Linke SP, Shirakawa Y et al. (2001). Cancer Res 61: 4345–4349. Pedeux R et al. (2001). Proc Natl Acad Sci USA 98: Cheung Jr KJ, Bush JA, Jia W, Li G. (2000). Br J Cancer 83: 9671–9676. 1468–1472. Nouman GS, Angus B, Lunec J, Crosier S, Lodge A, Cheung Jr KJ, Li G. (2002). Exp Cell Res 279: 291–298. Anderson JJ. (2002). Hybrid Hybridomics 21: 1–10. Cheung Jr KJ, Mitchell D, Lin P, Li G. (2001). Cancer Res 61: Ohgi T, Masaki T, Nakai S, Morishita A, Yukimasa S, Nagai 4974–4977. M et al. (2002). Scand J Gastroenterol 37: 1440–1448. Donehower LA, Harvey M, Slagle BL, McArthur MJ, Ohmori M, Nagai M, Tasaka T, Koeffler HP, Toyama T, Montgomery Jr CA, Butel JS et al. (1992). Nature 356: Riabowol K et al. (1999). Am J Hematol 62: 118–119. 215–221. Oki E, Maehara Y, Tokunaga E, Kakeji Y, Sugimachi K. el-Deiry WS, Tokino T, Velculescu VE, Levy DB, Parsons R, (1999). Cancer Lett 147: 157–162. Trent JM et al. (1993). Cell 75: 817–825. Ramirez-Solis R, Bradley A. (1994). Curr Opin Biotechnol 5: Garkavtsev I, Demetrick D, Riabowol K. (1997). Cytogenet 528–533. Cell Genet 76: 176–178. Saito A, Furukawa T, Fukushige S, Koyama S, Hoshi M, Garkavtsev I, Grigorian IA, Ossovskaya VS, Chernov MV, Hayashi Y et al. (2000). J Hum Genet 45: 177–181. Chumakov PM, Gudkov AV. (1998). Nature 391: 295–298. Sanchez-Cespedes M, Okami K, Cairns P, Sidransky D. Garkavtsev I, Kazarov A, Gudkov A, Riabowol K. (1996). (2000). Genes Chromosomes Cancer 27: 319–322. Nat Genet 14: 415–420. Sarela AI, Farmery SM, Markham AF, Guillou PJ. (1999). Garkavtsev I, Kozin SV, Chernova O, Xu L, Winkler F, Eur J Cancer 35: 1264–1267. Brown E et al. (2004). Nature 428: 328–332. Scott M, Boisvert FM, Vieyra D, Johnston RN, Bazett-Jones Garkavtsev I, Riabowol K. (1997). Mol Cell Biol 17: 2014– DP, Riabowol K. (2001a). Nucleic Acids Res 29: 2052–2058. 2019. Scott M, Bonnefin P, Vieyra D, Boisvert FM, Young D, Gunduz M, Ouchida M, Fukushima K, Hanafusa H, Etani T, Bazett-Jones DP et al. (2001b). J Cell Sci 114: 3455–3462. Nishioka S et al. (2000). Cancer Res 60: 3143–3146. Shimada H, Liu TL, Ochiai T, Shimizu T, Haupt Y, Hamada Gunduz M, Ouchida M, Fukushima K, Ito S, Jitsumori Y, H et al. (2002). Oncogene 21: 1208–1216. Nakashima T et al. (2002). Oncogene 21: 4462–4470. Shinoura N, Muramatsu Y, Nishimura M, Yoshida Y, Saito He GH, Helbing CC, Wagner MJ, Sensen CW, Riabowol K. A, Yokoyama T et al. (1999). Cancer Res 59: 5521–5528. (2005). Mol Biol Evol 22: 104–116. Skowyra D, Zeremski M, Neznanov N, Li M, Choi Y, Uesugi Helbing CC, Veillette C, Riabowol K, Johnston RN, M et al. (2001). J Biol Chem 276: 8734–8739. Garkavtsev I. (1997). Cancer Res 57: 1255–1258. Swiatek PJ, Gridley T. (1993). Genes Dev 7: 2071–2084. Ito K, Kinjo K, Nakazato T, Ikeda Y, Kizaki M. (2002). Am J Tokunaga E, Maehara Y, Oki E, Kitamura K, Kakeji Y, Hematol 69: 141–143. Ohno S et al. (2000). Cancer Lett 152: 15–22. Kichina JV, Rauth S, Das Gupta TK, Gudkov AV. (2003). Toyama T, Iwase H, Watson P, Muzik H, Saettler E, Oncogene 22: 4911–4917. Magliocco A et al. (1999). Oncogene 18: 5187–5193. Kiyokawa H, Kineman RD, Manova-Todorova KO, Soares Vieyra D, Loewith R, Scott M, Bonnefin P, Boisvert FM, VC, Hoffman ES, Ono M et al. (1996). Cell 85: 721–732. Cheema P et al. (2002). J Biol Chem 277: 29832–29839. Komarov PG, Komarova EA, Kondratov RV, Christov- Ward JM, Anver MR, Mahler JF, Devor-Henneman DE. Tselkov K, Coon JS, Chernov MV et al. (1999). Science 285: (2000). Pathology of Genetically Engineered Mice. Ward JM, 1733–1737. Mahler JF, Maronpot RR, Sundberg JP (eds). Iowa State Komarova EA, Kondratov RV, Wang K, Christov K, University Press: Ames, pp. 161–179. Golovkina TV, Goldblum JR et al. (2004). Oncogene 23: Zeremski M, Hill JE, Kwek SS, Grigorian IA, Gurova KV, 3265–3271. Garkavtsev IV et al. (1999). J Biol Chem 274: 32172–32181. Kuzmichev A, Zhang Y, Erdjument-Bromage H, Tempst P, Zeremski M, Horrigan SK, Grigorian IA, Westbrook CA, Reinberg D. (2002). Mol Cell Biol 22: 835–848. Gudkov AV. (1997). Somat Cell Mol Genet 23: 233–236.

Oncogene